id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-06-27 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.19059 | Luan Hoang | Luan Hoang and Akif Ibragimov | Asymptotic estimates for solutions of inhomogeneous non-divergence diffusion equations with drifts | 31 pages, submitted for publication | null | null | null | math.AP math.DS | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We study the long-time dynamics of the nonlinear processes modeled by diffusion-transport partial differential equations in non-divergence form with drifts. The solutions are subject to some inhomogeneous Dirichlet boundary condition. Starting with the reduced linear problem, we obtain the asymptotic estimates for the solutions, as time $t\to\infty$, depending on the asymptotic behavior of the forcing term and boundary data. These are established in both cases when the drifts are uniformly bounded, and unbounded as $t\to\infty$. For the nonlinear problem, we prove the convergence of the solutions under suitable conditions that balance the growth of the nonlinear term with the decay of the data. To take advantage of the diffusion in the non-divergence form, we prove an inhomogeneous version of the Landis-typed Growth Lemma and apply it to successive time-intervals. At each time step, the center for the barrier function is selected carefully to optimize the contracting factor. Our rigorous results show the robustness of the model. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:18:37 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Hoang",
"Luan",
""
],
[
"Ibragimov",
"Akif",
""
]
] |
2506.19060 | Keith Merrill | Lior Fishman, David Lambert, Keith Merrill, David Simmons | Diophantine approximation on abelian varieties; a conjecture of M. Waldschmidt | 11 pages | null | null | null | math.NT | http://creativecommons.org/licenses/by/4.0/ | Following the work of Waldschmidt, we investigate problems in Diophantine approximation on abelian varieties. First we show that a conjecture of Waldschmidt for a given simple abelian variety is equivalent to a well-known Diophantine condition holding for a certain matrix related to that variety. We then posit a related but weaker conjecture, and establish the upper bound direction of that conjecture in full generality. For rank 1 elliptic curves defined over a number field $K \subset \mathbb{R}$, we then obtain a weak-type Dirichlet theorem in this setting, establish the optimality of this statement, and prove our conjecture in this case. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:22:25 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Fishman",
"Lior",
""
],
[
"Lambert",
"David",
""
],
[
"Merrill",
"Keith",
""
],
[
"Simmons",
"David",
""
]
] |
2506.19061 | Sam Spiro | Sean English, Sam Spiro | Rational Exponents for General Graphs | null | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | A rational number $r$ is a \textbf{realizable exponent} for a graph $H$ if there exists a finite family of graphs $\mathcal{F}$ such that $\mathrm{ex}(n,H,\mathcal{F})=\Theta(n^r)$, where $\mathrm{ex}(n,H,\mathcal{F})$ denotes the maximum number of copies of $H$ that an $n$-vertex $\mathcal{F}$-free graph can have. Results for realizable exponents are currently known only when $H$ is either a star or a clique, with the full resolution of the $H=K_2$ case being a major breakthrough of Bukh and Conlon.
In this paper, we establish the first set of results for realizable exponents which hold for arbitrary graphs $H$ by showing that for any graph $H$ with maximum degree $\Delta \ge 1$, every rational in the interval $\left[v(H)-\frac{e(H)}{2\Delta^2},\ v(H)\right]$ is realizable for $H$. We also prove a ``stability'' result for generalized Tur\'an numbers of trees which implies that if $T\ne K_2$ is a tree with $\ell$ leaves, then $T$ has no realizable exponents in $[0,\ell]\setminus \mathbb{Z}$. Our proof of this latter result uses a new variant of the classical Helly theorem for trees, which may be of independent interest. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:23:11 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"English",
"Sean",
""
],
[
"Spiro",
"Sam",
""
]
] |
2506.19062 | Giorgio Arcadi | Giorgio Arcadi, Stefano Profumo | WIMPs Below the Radar: Blind Spots and Benchmarks Beyond the Neutrino Floor | 23 Pages, 8 Figures | null | null | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate benchmark scenarios for Weakly Interacting Massive Particles (WIMPs) that naturally evade current direct detection constraints by featuring suppressed spin-independent cross-sections. Focusing on three representative models, the Singlet-Doublet fermion model, its extension to a Two-Higgs-Doublet plus pseudoscalar sector (2HDM$+a$), and a dark $SU(3)$ gauge model, we systematically analyze the interplay between thermal freeze-out, direct detection blind spots, and radiative corrections. In each case, we identify viable regions of parameter space where the predicted dark matter relic abundance is consistent with observations while elastic scattering rates lie below current exclusion limits and, in some cases, but now always, below the neutrino floor. Loop-induced effects are shown to play a critical role, particularly in scenarios with suppressed tree-level interactions. Our findings demonstrate that models with rich electroweak and scalar sectors can populate the experimentally challenging, yet phenomenologically motivated parameter space between existing constraints and the ultimate sensitivity of current-technology direct detection experiments. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:24:59 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Arcadi",
"Giorgio",
""
],
[
"Profumo",
"Stefano",
""
]
] |
2506.19063 | Xianyue Zhao | Nan Du, Ilia Polian, Christopher Bengel, Kefeng Li, Ziang Chen, Xianyue Zhao, Uwe Huebner, Li-Wei Chen, Feng Liu, Massimiliano Di Ventra, Stephan Menzel, Heidemarie Krueger | Mixed-Mode In-Memory Computing: Towards High-Performance Logic Processing In A Memristive Crossbar Array | null | null | null | null | physics.app-ph | http://creativecommons.org/licenses/by/4.0/ | In-memory computing is a promising alternative to traditional computer designs, as it helps overcome performance limits caused by the separation of memory and processing units. However, many current approaches struggle with unreliable device behavior, which affects data accuracy and efficiency. In this work, the authors present a new computing method that combines two types of operations,those based on electrical resistance and those based on voltage, within each memory cell. This design improves reliability and avoids the need for expensive current measurements. A new software tool also helps automate the design process, supporting highly parallel operations in dense two-dimensional memory arrays. The approach balances speed and space, making it practical for advanced computing tasks. Demonstrations include a digital adder and a key part of the encryption module, showing both strong performance and accuracy. This work offers a new direction for reliable and efficient in-memory computing systems with real-world applications. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:26:27 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Du",
"Nan",
""
],
[
"Polian",
"Ilia",
""
],
[
"Bengel",
"Christopher",
""
],
[
"Li",
"Kefeng",
""
],
[
"Chen",
"Ziang",
""
],
[
"Zhao",
"Xianyue",
""
],
[
"Huebner",
"Uwe",
""
],
[
"Chen",
"Li-Wei",
""
],
[
"Liu",
"Feng",
""
],
[
"Di Ventra",
"Massimiliano",
""
],
[
"Menzel",
"Stephan",
""
],
[
"Krueger",
"Heidemarie",
""
]
] |
2506.19064 | Francesco Concetti | Francesco Concetti, David Belius, Giuseppe Genovese | Variational formula for the logarithmic potential of free additive convolutions | 2 figures, 25 pages | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We establish a general variational formula for the logarithmic potential of the free additive convolution of two compactly supported probability measure on $\R$. The formula is given in terms of the $R$-transform of the first measure, and the logarithmic potential of second measure. The result applies in particular to the additive convolution with the semicircle or Marchenko-Pastur laws, for which the formula simplifies. The logarithmic potential of additive convolutions appears for instance in estimates of the determinant of sums of independent random matrices. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:32:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Concetti",
"Francesco",
""
],
[
"Belius",
"David",
""
],
[
"Genovese",
"Giuseppe",
""
]
] |
2506.19065 | Guang Yang | Guang Yang, Victoria Ebert, Nazif Tamer, Luiza Pozzobon, Noah A. Smith | LEGATO: Large-scale End-to-end Generalizable Approach to Typeset OMR | null | null | null | null | cs.CV cs.DL | http://creativecommons.org/licenses/by/4.0/ | We propose Legato, a new end-to-end transformer model for optical music recognition (OMR). Legato is the first large-scale pretrained OMR model capable of recognizing full-page or multi-page typeset music scores and the first to generate documents in ABC notation, a concise, human-readable format for symbolic music. Bringing together a pretrained vision encoder with an ABC decoder trained on a dataset of more than 214K images, our model exhibits the strong ability to generalize across various typeset scores. We conduct experiments on a range of datasets and demonstrate that our model achieves state-of-the-art performance. Given the lack of a standardized evaluation for end-to-end OMR, we comprehensively compare our model against the previous state of the art using a diverse set of metrics. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:35:59 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Yang",
"Guang",
""
],
[
"Ebert",
"Victoria",
""
],
[
"Tamer",
"Nazif",
""
],
[
"Pozzobon",
"Luiza",
""
],
[
"Smith",
"Noah A.",
""
]
] |
2506.19066 | Mirajul Islam | Mirajul Islam, Michael J. Daniels, Donald Lloyd-Jones, Juned Siddique | A Bayesian approach for unadjudicated events in cardiovascular disease cohort studies | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | An important issue in joint modelling for outcomes and longitudinal risk factors in cohort studies is to have an accurate assessment of events. Events determined based on ICD-9 codes can be very inaccurate, in particular for cardiovascular disease (CVD) where ICD-9 codes may overestimate the frequency of CVD. Motivated by the lack of adjudicated events in the Established Populations for Epidemiologic Studies of the Elderly (EPESE) cohort, we develop methods that use a related cohort Atherosclerosis Risk in Communities (ARIC), with both ICD-9 code events and adjudicated events, to create a posterior predictive distribution of adjudicated events. The methods are based on the construction of flexible Bayesian joint models combined with a Bayesian additive regression trees to directly address the ICD-9 misclassification. We assessed the performance of our approach by simulation study and applied to ARIC data. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:37:50 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Islam",
"Mirajul",
""
],
[
"Daniels",
"Michael J.",
""
],
[
"Lloyd-Jones",
"Donald",
""
],
[
"Siddique",
"Juned",
""
]
] |
2506.19067 | Hossein Taji | Hossein Taji, Jos\'e Miranda, Miguel Pe\'on-Quir\'os and David Atienza | MEDEA: A Design-Time Multi-Objective Manager for Energy-Efficient DNN Inference on Heterogeneous Ultra-Low Power Platforms | Submitted to ACM Transactions on Embedded Computing Systems (TECS) | null | null | null | cs.AR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The growing demand for on-device AI necessitates energy-efficient execution of DNN based applications on resource-constrained ultra-low power (ULP) platforms. Heterogeneous architectures, combining specialized processing elements (PEs), have emerged as a key solution for achieving the required performance and energy efficiency. However, optimizing energy while executing applications on these platforms requires efficiently managing platform resources like PEs, power features, and memory footprint, all while adhering to critical application deadlines. This paper presents MEDEA, a novel design-time multi-objective manager for energy-efficient DNN inference on Heterogeneous ULP (HULP) platforms. MEDEA uniquely integrates: kernel-level dynamic voltage and frequency scaling (DVFS) for dynamic energy adaptation; kernel-level granularity scheduling, suitable for specialized accelerators; memory-aware adaptive tiling to navigate severe memory constraints; and all within a timing constraint-based optimization strategy, which minimizes energy based on application deadline. To showcase practical viability, we evaluate MEDEA on HEEPtimize, a heterogeneous ULP platform (22 nm, FPGA-prototyped) featuring a RISC-V processor besides Near-Memory Computing (NMC) and Coarse-Grained Reconfigurable Array (CGRA) accelerators. Experimental results, using a biomedical seizure detection case study, demonstrate that MEDEA achieves overall energy reductions of up to 38% compared to representative state-of-the-art methods, while consistently meeting all timing and memory requirements. This effectiveness is attributed to its integrated features, with our analysis showing that kernel-level DVFS alone can be responsible for over 31% of the energy savings in specific scenarios. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:39:49 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Taji",
"Hossein",
""
],
[
"Miranda",
"José",
""
],
[
"Peón-Quirós",
"Miguel",
""
],
[
"Atienza",
"David",
""
]
] |
2506.19068 | Gabriel' Silva | Gabriel A. S. Silva, Camila P. Novaes, Carlos A. Wuensche, Eduardo J. de Mericia, Bruno B. Bizarria, Jiajun Zhang, Elcio Abdalla, Filipe B. Abdalla, Amilcar R. Queiroz, Thyrso Villela, Bin Wang, Chang Feng, Edmar C. Gurjao, Alessandro Marins | Detectability of the 21 cm signal with BINGO through cross-correlation with photometric surveys | 27 pages, 9 images | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | 21 cm intensity mapping (HI IM) can efficiently map large cosmic volumes with good redshift resolution, but systematics and foreground contamination pose major challenges for extracting accurate cosmological information. Cross-correlation with galaxy surveys offers an efficient mitigation strategy, as both datasets have largely uncorrelated systematics. We evaluate the detectability of the 21 cm signal from the BINGO radio telescope by cross-correlating with the LSST photometric survey, given their strong overlap in area and redshift. Using lognormal simulations, we model the cosmological signal in the BINGO frequency range (980 - 1260 MHz), incorporating thermal noise, foregrounds, and cleanup. The LSST simulations include uncertainties in photometric redshift (photo-z) and galaxy number density in the first three redshift intervals (mean redshift approximately equal to 0.25, 0.35, 0.45), corresponding to the expected performance after 10 years of the survey. We show that photo-z errors significantly increase the noise in the cross-correlation, reducing its statistical significance to levels comparable to those of the autocorrelation. Still, the HI signal remains detectable through the cross-correlation, even with photo-z uncertainties similar to those of the LSST. Our results corroborate the feasibility of this approach under realistic conditions and motivate further refinement of the current analysis methods. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:40:48 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Silva",
"Gabriel A. S.",
""
],
[
"Novaes",
"Camila P.",
""
],
[
"Wuensche",
"Carlos A.",
""
],
[
"de Mericia",
"Eduardo J.",
""
],
[
"Bizarria",
"Bruno B.",
""
],
[
"Zhang",
"Jiajun",
""
],
[
"Abdalla",
"Elcio",
""
],
[
"Abdalla",
"Filipe B.",
""
],
[
"Queiroz",
"Amilcar R.",
""
],
[
"Villela",
"Thyrso",
""
],
[
"Wang",
"Bin",
""
],
[
"Feng",
"Chang",
""
],
[
"Gurjao",
"Edmar C.",
""
],
[
"Marins",
"Alessandro",
""
]
] |
2506.19069 | Joost Van Hofslot | Joost W.F. van Hofslot (1,2), Izabella E. Thompson (1), Anno Touwen (1,2,3), Nithesh Balasubramanian (1,2), Roman Bause (1,2,5), Hendrick L. Bethlem (1,3), Anastasia Borschevsky (1,2), Ties H. Fikkers (1,2), Steven Hoekstra (1,2), Steven A. Jones (1,2), Jelmer E.J. Levenga (1,2), Maarten C. Mooij (2,3), Heleen Mulder (1,2), Bastiaan A. Nijman (1,2), Efion H. Prinsen (1,2), Bart J. Schellenberg (1,2), Lucas van Sloten (1,2), Rob G.E. Timmermans (1,2), Wim Ubachs (3), Jordy de Vries (1,4) and Lorenz Willmann (1,2) (for the NL-eEDM collaboration, (1) Van Swinderen Institute for Particle Physics and Gravity, University of Groningen, The Netherlands, (2) Nikhef, National Institute for Subatomic Physics, Amsterdam, The Netherlands, (3) Department of Physics and Astronomy, and LaserLaB, Vrije Universiteit Amsterdam, The Netherlands, (4) Institute of Physics and Delta Institute for Theoretical Physics, University of Amsterdam, The Netherlands, (5) Present address: Menlo Systems GmbH, Martinsried, Germany) | 2D transverse laser cooling of a hexapole focused beam of cold BaF molecules | null | null | null | null | physics.atom-ph | http://creativecommons.org/licenses/by/4.0/ | A cryogenic buffer gas beam, an electrostatic hexapole lens, and 2D transverse Doppler laser cooling are combined to produce a bright beam of barium monofluoride ($^{138}$Ba$^{19}$F) molecules. Experimental results and trajectory simulations are used to study the laser cooling effect as a function of laser detuning, laser power, laser alignment, and interaction time. A scattering rate of 6.1(1.4) $\times 10^{5}$ s$^{-1}$ on the laser cooling transition is obtained; this is $14 \%$ of the expected maximum, which is attributed to limited control of the magnetic field used to remix dark states. Using 3 tuneable lasers with appropriate sidebands and detuning, each molecule scatters approximately 400 photons during 2D laser cooling, limited by the interaction time and scattering rate. Leaks to dark states are less than 10$\%$. The experimental results are used to benchmark the trajectory simulations to predict the achievable flux 3.5 m downstream for a planned $e$EDM experiment. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:40:59 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"van Hofslot",
"Joost W. F.",
""
],
[
"Thompson",
"Izabella E.",
""
],
[
"Touwen",
"Anno",
""
],
[
"Balasubramanian",
"Nithesh",
""
],
[
"Bause",
"Roman",
""
],
[
"Bethlem",
"Hendrick L.",
""
],
[
"Borschevsky",
"Anastasia",
""
],
[
"Fikkers",
"Ties H.",
""
],
[
"Hoekstra",
"Steven",
""
],
[
"Jones",
"Steven A.",
""
],
[
"Levenga",
"Jelmer E. J.",
""
],
[
"Mooij",
"Maarten C.",
""
],
[
"Mulder",
"Heleen",
""
],
[
"Nijman",
"Bastiaan A.",
""
],
[
"Prinsen",
"Efion H.",
""
],
[
"Schellenberg",
"Bart J.",
""
],
[
"van Sloten",
"Lucas",
""
],
[
"Timmermans",
"Rob G. E.",
""
],
[
"Ubachs",
"Wim",
""
],
[
"de Vries",
"Jordy",
""
],
[
"Willmann",
"Lorenz",
""
]
] |
2506.19070 | Norbert Euler | Marianna Euler, Norbert Euler | From fully-nonlinear to semilinear evolution equations: two symmetry-integrable examples | 15 pages | null | null | null | nlin.SI | http://creativecommons.org/licenses/by/4.0/ | In this paper we derive two examples of fully-nonlinear symmetry-integrable evolution equations with algebraic nonlinearities, namely one class of 3rd-order equations and a 5th-order equation. To achieve this we study the equations' Lie-B\"acklund symmetries and apply multipotentialisations, hodograph transformations and generalised hodograph transformations to map the equations to known semilinear integrable evolution equations. As a result of this, we also obtain interesting symmetry-integrable quasilinear equations of order five and order seven, which we display explicitly. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:41:09 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Euler",
"Marianna",
""
],
[
"Euler",
"Norbert",
""
]
] |
2506.19071 | Olivia Griffith | Olivia Griffith, Grace Showerman, Sumit K. Sarbadhicary, Chelsea E. Harris, Laura Chomiuk, Jesper Sollerman, Peter Lundqvist, Javier Moldon, Miguel Perez-Torres, Erik C. Kool and Takashi J. Moriya | A Late-time Radio Survey of Type Ia-CSM Supernovae with the Very Large Array | 17 pages, 6 figures, 3 tables, submitted to ApJ | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Type Ia-CSM supernovae (SNe) are a rare and peculiar subclass of thermonuclear SNe characterized by emission lines of hydrogen or helium, indicative of a high-density circumstellar medium (CSM). Their implied mass-loss rates of $\sim 10^{-4}-10^{-1}$ M$_{\odot}$ yr$^{-1}$ (assuming $\mathrm{ \sim 100 \ km\ s^{-1}}$ winds) from optical observations are generally in excess of values observed in realistic SN Ia progenitors. In this paper, we present an independent study of CSM densities around a sample of 29 archival Ia-CSM SNe using radio observations with the Very Large Array at 6 GHz. Motivated by the late ($\sim$2 yr) radio detection of the Ia-CSM SN 2020eyj, we observed old ($>$1 yr) SNe where we are more likely to see the emergent synchrotron emission that may have been suppressed earlier by free-free absorption by the CSM. We do not detect radio emission down to 3$\sigma$ limits of $\sim$35 $\mu$Jy in our sample. The only radio-detected candidate in our sample, SN 2022esa, was likely mis-classified as a Ia-CSM with early spectra, and appears more consistent with a peculiar Ic based on later-epochs. Assuming a wind-like CSM with temperatures between $2 \times 10^4$ K and $10^5$ K, and magnetic field-to-shock energy fraction ($\epsilon_B$) = $0.01-0.1$, the radio upper limits rule out mass-loss rates between $\sim 10^{-4}-10^{-2}$ M$_{\odot}$ yr$^{-1}$ (100 km s$^{-1}$)$^{-1}$. This is somewhat in tension with the estimates from optical observations, and may indicate that more complex CSM geometries and/or lower values of $\epsilon_B$ may be present. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:42:07 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Griffith",
"Olivia",
""
],
[
"Showerman",
"Grace",
""
],
[
"Sarbadhicary",
"Sumit K.",
""
],
[
"Harris",
"Chelsea E.",
""
],
[
"Chomiuk",
"Laura",
""
],
[
"Sollerman",
"Jesper",
""
],
[
"Lundqvist",
"Peter",
""
],
[
"Moldon",
"Javier",
""
],
[
"Perez-Torres",
"Miguel",
""
],
[
"Kool",
"Erik C.",
""
],
[
"Moriya",
"Takashi J.",
""
]
] |
2506.19072 | Yimu Wang | Yimu Wang, Mozhgan Nasr Azadani, Sean Sedwards, Krzysztof Czarnecki | HAWAII: Hierarchical Visual Knowledge Transfer for Efficient Vision-Language Models | Work in progress | null | null | null | cs.CV cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Improving the visual understanding ability of vision-language models (VLMs) is crucial for enhancing their performance across various tasks. While using multiple pretrained visual experts has shown great promise, it often incurs significant computational costs during training and inference. To address this challenge, we propose HAWAII, a novel framework that distills knowledge from multiple visual experts into a single vision encoder, enabling it to inherit the complementary strengths of several experts with minimal computational overhead. To mitigate conflicts among different teachers and switch between different teacher-specific knowledge, instead of using a fixed set of adapters for multiple teachers, we propose to use teacher-specific Low-Rank Adaptation (LoRA) adapters with a corresponding router. Each adapter is aligned with a specific teacher, avoiding noisy guidance during distillation. To enable efficient knowledge distillation, we propose fine-grained and coarse-grained distillation. At the fine-grained level, token importance scores are employed to emphasize the most informative tokens from each teacher adaptively. At the coarse-grained level, we summarize the knowledge from multiple teachers and transfer it to the student using a set of general-knowledge LoRA adapters with a router. Extensive experiments on various vision-language tasks demonstrate the superiority of HAWAII, compared to the popular open-source VLMs. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:43:25 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Wang",
"Yimu",
""
],
[
"Azadani",
"Mozhgan Nasr",
""
],
[
"Sedwards",
"Sean",
""
],
[
"Czarnecki",
"Krzysztof",
""
]
] |
2506.19073 | Francielle Vargas | Jackson Trager and Francielle Vargas and Diego Alves and Matteo Guida and Mikel K. Ngueajio and Ameeta Agrawal and Flor Plaza-del-Arco and Yalda Daryanai and Farzan Karimi-Malekabadi | MFTCXplain: A Multilingual Benchmark Dataset for Evaluating the Moral Reasoning of LLMs through Hate Speech Multi-hop Explanation | Under Review | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Ensuring the moral reasoning capabilities of Large Language Models (LLMs) is a growing concern as these systems are used in socially sensitive tasks. Nevertheless, current evaluation benchmarks present two major shortcomings: a lack of annotations that justify moral classifications, which limits transparency and interpretability; and a predominant focus on English, which constrains the assessment of moral reasoning across diverse cultural settings. In this paper, we introduce MFTCXplain, a multilingual benchmark dataset for evaluating the moral reasoning of LLMs via hate speech multi-hop explanation using Moral Foundation Theory (MFT). The dataset comprises 3,000 tweets across Portuguese, Italian, Persian, and English, annotated with binary hate speech labels, moral categories, and text span-level rationales. Empirical results highlight a misalignment between LLM outputs and human annotations in moral reasoning tasks. While LLMs perform well in hate speech detection (F1 up to 0.836), their ability to predict moral sentiments is notably weak (F1 < 0.35). Furthermore, rationale alignment remains limited mainly in underrepresented languages. These findings show the limited capacity of current LLMs to internalize and reflect human moral reasoning. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:44:21 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Trager",
"Jackson",
""
],
[
"Vargas",
"Francielle",
""
],
[
"Alves",
"Diego",
""
],
[
"Guida",
"Matteo",
""
],
[
"Ngueajio",
"Mikel K.",
""
],
[
"Agrawal",
"Ameeta",
""
],
[
"Plaza-del-Arco",
"Flor",
""
],
[
"Daryanai",
"Yalda",
""
],
[
"Karimi-Malekabadi",
"Farzan",
""
]
] |
2506.19075 | Dan Garber | Dan Garber | First-Order Sparse Convex Optimization: Better Rates with Sparse Updates | null | null | null | null | math.OC cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In was recently established that for convex optimization problems with a sparse optimal solution (may it be entry-wise sparsity or matrix rank-wise sparsity) it is possible to have linear convergence rates which depend on an improved mixed-norm condition number of the form $\frac{\beta_1{}s}{\alpha_2}$, where $\beta_1$ is the $\ell_1$-Lipchitz continuity constant of the gradient, $\alpha_2$ is the $\ell_2$-quadratic growth constant, and $s$ is the sparsity of the optimal solution. However, beyond the improved convergence rate, these methods are unable to leverage the sparsity of optimal solutions towards improving also the runtime of each iteration, which may still be prohibitively high for high-dimensional problems. In this work, we establish that linear convergence rates which depend on this improved condition number can be obtained using only sparse updates, which may result in overall significantly improved running times. Moreover, our methods are considerably easier to implement. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:44:37 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Garber",
"Dan",
""
]
] |
2506.19076 | Carlos Hernandez-Suarez M | Carlos M Hernandez-Suarez | Fast and Accurate Reconstruction of Voronoi Generators in Large Tessellations | 7 pages, 2 Figs, 2 Tables, link to GitHub code | null | null | null | math.MG | http://creativecommons.org/licenses/by/4.0/ | A Voronoi diagram partitions the plane into convex cells, each containing the points closest to a single generator. Given such a tessellation, the inverse Voronoi problem seeks the generator set \( S \) that produced it. Our algorithm selects a single interior cell with \( k \) edges and solves a compact, consistent linear system with \( 2(k+1) \) unknowns and \( 4k \) scalar equations to recover that cell's generator together with the \( k \) generators of its neighbors in one step. The remaining sites follow by successive geometric reflections. The overall running time is \( O(n) \) for a diagram with \( n \) cells. Across \( 10^3 \) Monte Carlo simulations on diagrams of \( 10^4 \) cells, the method achieved an average RMSE of \( 10^{-12} \) and a worst-case individual reconstruction error of \( 10^{-8} \), demonstrating both efficiency and robustness. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:45:58 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Hernandez-Suarez",
"Carlos M",
""
]
] |
2506.19077 | Christoph Willibald | Christoph Willibald, Daniel Sliwowski, Dongheui Lee | Multimodal Anomaly Detection with a Mixture-of-Experts | 8 pages, 5 figures, 1 table, the paper has been accepted for publication in the Proceedings of the 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025) | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | With a growing number of robots being deployed across diverse applications, robust multimodal anomaly detection becomes increasingly important. In robotic manipulation, failures typically arise from (1) robot-driven anomalies due to an insufficient task model or hardware limitations, and (2) environment-driven anomalies caused by dynamic environmental changes or external interferences. Conventional anomaly detection methods focus either on the first by low-level statistical modeling of proprioceptive signals or the second by deep learning-based visual environment observation, each with different computational and training data requirements. To effectively capture anomalies from both sources, we propose a mixture-of-experts framework that integrates the complementary detection mechanisms with a visual-language model for environment monitoring and a Gaussian-mixture regression-based detector for tracking deviations in interaction forces and robot motions. We introduce a confidence-based fusion mechanism that dynamically selects the most reliable detector for each situation. We evaluate our approach on both household and industrial tasks using two robotic systems, demonstrating a 60% reduction in detection delay while improving frame-wise anomaly detection performance compared to individual detectors. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:52:38 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Willibald",
"Christoph",
""
],
[
"Sliwowski",
"Daniel",
""
],
[
"Lee",
"Dongheui",
""
]
] |
2506.19078 | Dipankar Paul | Dipankar Paul, Harekrushna Behera, Sukhendu Ghosh | Stability of soluble surfactant-laden falling film over a hydrophobic incline in the presence of external shear | 40 pages, 21 figures, comments are welcome | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | The hydrodynamic stability analysis of gravity-driven, soluble surfactant-laden fluid streaming down a slippery, slanted plane in the presence of external shear force is being explored in this article. The Navier-Stokes equations are considered for the fluid flow along with the appropriate advection-diffusion equations for the concentration of different surfactant species. The monomers considered here are anticipated to dissolve in the bulk flow and can be adsorbed at the interface of air-liquid as well. Also, the adsorption-desorption kinetics of the surfactants at the free space is taken into consideration. The motivation behind this work is to extend the work of Karapetsas and Bontozoglou[1] for flow over a slippery bottom and in the presence of externally imposed shear forces and observe their impact on the flow dynamics. The Orr-Sommerfeld eigensystem is obtained, then it is solved analytically using the longwave approximation method in the longwave regime ($k \ll 1$) and subsequently, the Chebyshev spectral collocation method is employed for numerical evaluation in the arbitrary wave regime. Using the analytical method, two longwave modes, viz, surface mode and surfactant mode, are detected. Alternatively, the numerical analysis substantiated the existence of temporal surface and surfactant modes. Moreover, another temporal mode named shear mode arises in the high modified Reynolds number region at a low inclination angle. Thereafter, the modified Reynolds-Orr energy equation is deduced under the normal mode conditions, and the behaviour of different energy components is investigated for various slip parameters and imposed shear force. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:56:14 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Paul",
"Dipankar",
""
],
[
"Behera",
"Harekrushna",
""
],
[
"Ghosh",
"Sukhendu",
""
]
] |
2506.19079 | Iosif Tsangko | Iosif Tsangko, Andreas Triantafyllopoulos, Adem Abdelmoula, Adria Mallol-Ragolta, Bjoern W. Schuller | Reading Smiles: Proxy Bias in Foundation Models for Facial Emotion Recognition | null | null | null | null | cs.CV cs.AI cs.HC | http://creativecommons.org/licenses/by/4.0/ | Foundation Models (FMs) are rapidly transforming Affective Computing (AC), with Vision Language Models (VLMs) now capable of recognising emotions in zero shot settings. This paper probes a critical but underexplored question: what visual cues do these models rely on to infer affect, and are these cues psychologically grounded or superficially learnt? We benchmark varying scale VLMs on a teeth annotated subset of AffectNet dataset and find consistent performance shifts depending on the presence of visible teeth. Through structured introspection of, the best-performing model, i.e., GPT-4o, we show that facial attributes like eyebrow position drive much of its affective reasoning, revealing a high degree of internal consistency in its valence-arousal predictions. These patterns highlight the emergent nature of FMs behaviour, but also reveal risks: shortcut learning, bias, and fairness issues especially in sensitive domains like mental health and education. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:56:30 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Tsangko",
"Iosif",
""
],
[
"Triantafyllopoulos",
"Andreas",
""
],
[
"Abdelmoula",
"Adem",
""
],
[
"Mallol-Ragolta",
"Adria",
""
],
[
"Schuller",
"Bjoern W.",
""
]
] |
2506.19080 | Dmitry Kazantsev | V.S. Ivchenko and D.V. Kazantsev and V.A. Ievleva and D.A. Matienko and E.A. Kazantseva and A.Yu. Kuntsevich | Absence of reflectivity of the phonon-polariton at the SiC surface from the metal mask edge | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Surface phonon polariton (SPhP) waves are excited at the silicon carbide ($SiC$) surface under irradiation of light close to the lattice resonance frequency. Metal mask at the surface blocks irradiation of certain areas and thus allows tuning standing or propagating wave pattern and, thus, open opportunities for surface polariton optic devices. In this study we show by means of scanning near-field microscopy that the edge of such a mask reflects SPhP waves negligibly. This condition differs dramatically from numerous recent observations of polariton reflections in 2D materials and makes the metallized $SiC$ platform advantageous in a sense of capability to calculate wavefield using a simple Green function-based approach. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:56:36 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Ivchenko",
"V. S.",
""
],
[
"Kazantsev",
"D. V.",
""
],
[
"Ievleva",
"V. A.",
""
],
[
"Matienko",
"D. A.",
""
],
[
"Kazantseva",
"E. A.",
""
],
[
"Kuntsevich",
"A. Yu.",
""
]
] |
2506.19081 | Francisca Guzm\'an-Lastra | J. Morales, P. Munoz, D. Noto, H.N Ulloa and F. Guzman-Lastra | Emergent collective dynamics from motile photokinetic organisms | 11 pages 5 Figures | null | null | null | cond-mat.soft | http://creativecommons.org/licenses/by/4.0/ | The day-night cycle drives the largest biomass migration on Earth: the diel vertical migration (DVM) of aquatic organisms. Here, we present a three-dimensional agent-based model that incorporates photokinesis, gyrotaxis, and stochastic reorientation to explore how individual-level swimming behaviors give rise to population-scale DVM patterns. By solving Langevin equations for swarms of swimmers, we identify four distinct regimes -- Surface Accumulation, Shallow DVM, Deep DVM, and Sinking -- governed by two key dimensionless parameters: the Peclet number (Pe), representing motility persistence, and the vertical swimming asymmetry ratio (W=wdown/wup), encoding photokinetic bias. These regimes emerge from nonlinear interactions between light-driven navigation and active noise, diagnosed through topological and statistical features of vertical distributions. A critical feedback is uncovered: upward-biased swimming (W<1) promotes surface aggregation, while excessive downward bias (W>1) leads to irreversible sinking. Analytical estimates link regime boundaries to gyrotactic alignment and velocity reversals. Together, our results provide a mechanistic framework to interpret DVM diversity and emphasize the central role of light gradients-beyond absolute intensity-in shaping ecological self-organization. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:57:51 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Morales",
"J.",
""
],
[
"Munoz",
"P.",
""
],
[
"Noto",
"D.",
""
],
[
"Ulloa",
"H. N",
""
],
[
"Guzman-Lastra",
"F.",
""
]
] |
2506.19082 | Nitish Nagesh | Nitish Nagesh, Ziyu Wang, Amir M. Rahmani | FairCauseSyn: Towards Causally Fair LLM-Augmented Synthetic Data Generation | Accepted to IEEE EMBC 2025 | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Synthetic data generation creates data based on real-world data using generative models. In health applications, generating high-quality data while maintaining fairness for sensitive attributes is essential for equitable outcomes. Existing GAN-based and LLM-based methods focus on counterfactual fairness and are primarily applied in finance and legal domains. Causal fairness provides a more comprehensive evaluation framework by preserving causal structure, but current synthetic data generation methods do not address it in health settings. To fill this gap, we develop the first LLM-augmented synthetic data generation method to enhance causal fairness using real-world tabular health data. Our generated data deviates by less than 10% from real data on causal fairness metrics. When trained on causally fair predictors, synthetic data reduces bias on the sensitive attribute by 70% compared to real data. This work improves access to fair synthetic data, supporting equitable health research and healthcare delivery. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:59:26 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Nagesh",
"Nitish",
""
],
[
"Wang",
"Ziyu",
""
],
[
"Rahmani",
"Amir M.",
""
]
] |
2506.19083 | Alexander Goldberg | Alexander Goldberg, Giulia Fanti, and Nihar B. Shah | A Principled Approach to Randomized Selection under Uncertainty | null | null | null | null | cs.GT cs.CY | http://creativecommons.org/licenses/by/4.0/ | Many decision-making processes involve evaluating and then selecting items; examples include scientific peer review, job hiring, school admissions, and investment decisions. The eventual selection is performed by applying rules or deliberations to the raw evaluations, and then deterministically selecting the items deemed to be the best. These domains feature error-prone evaluations and uncertainty about future outcomes, which undermine the reliability of such deterministic selection rules. As a result, selection mechanisms involving explicit randomization that incorporate the uncertainty are gaining traction in practice. However, current randomization approaches are ad hoc, and as we prove, inappropriate for their purported objectives. In this paper, we propose a principled framework for randomized decision-making based on interval estimates of the quality of each item. We introduce MERIT (Maximin Efficient Randomized Interval Top-k), an optimization-based method that maximizes the worst-case expected number of top candidates selected, under uncertainty represented by overlapping intervals (e.g., confidence intervals or min-max intervals). MERIT provides an optimal resource allocation scheme under an interpretable notion of robustness. We develop a polynomial-time algorithm to solve the optimization problem and demonstrate empirically that the method scales to over 10,000 items. We prove that MERIT satisfies desirable axiomatic properties not guaranteed by existing approaches. Finally, we empirically compare algorithms on synthetic peer review data. Our experiments demonstrate that MERIT matches the performance of existing algorithms in expected utility under fully probabilistic review data models used in previous work, while outperforming previous methods with respect to our novel worst-case formulation. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 19:59:30 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Goldberg",
"Alexander",
""
],
[
"Fanti",
"Giulia",
""
],
[
"Shah",
"Nihar B.",
""
]
] |
2506.19084 | Benjamin Watson | Maryam Riahi and Benjamin Watson | Am I Playing Better Now? The Effects of G-SYNC in 60Hz Gameplay | null | Proceedings of the ACM on Computer Graphics and Interactive Techniques (2001), Volume 4, Issue 1 Article No.: 6, Pages 1 - 17 | 10.1145/3451269 | null | cs.GR | http://creativecommons.org/licenses/by/4.0/ | G-SYNC technology matches formerly regular display refreshes to irregular frame updates, improving frame rates and interactive latency. In a previous study of gaming at the 30Hz frame rates common on consoles, players of Battlefield 4 were unable to discern when G-SYNC was in use, but scored higher with G-SYNC and were affected emotionally. We build on that study with the first examination of G-SYNC's effects at the 60Hz frame rate more common in PC gaming and on emerging consoles. Though G-SYNC's effects are less at 60Hz than they were at 30Hz, G-SYNC can still improve the performance of veteran players, particularly when games are challenging. G-SYNC's effects on emotion and experience were limited. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:00:04 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Riahi",
"Maryam",
""
],
[
"Watson",
"Benjamin",
""
]
] |
2506.19085 | Florian Gr\"otschla | Florian Gr\"otschla, Ahmet Solak, Luca A. Lanzend\"orfer, Roger Wattenhofer | Benchmarking Music Generation Models and Metrics via Human Preference Studies | Accepted at ICASSP 2025 | In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025 | 10.1109/ICASSP49660.2025.10887745 | null | cs.LG cs.SD | http://creativecommons.org/licenses/by/4.0/ | Recent advancements have brought generated music closer to human-created compositions, yet evaluating these models remains challenging. While human preference is the gold standard for assessing quality, translating these subjective judgments into objective metrics, particularly for text-audio alignment and music quality, has proven difficult. In this work, we generate 6k songs using 12 state-of-the-art models and conduct a survey of 15k pairwise audio comparisons with 2.5k human participants to evaluate the correlation between human preferences and widely used metrics. To the best of our knowledge, this work is the first to rank current state-of-the-art music generation models and metrics based on human preference. To further the field of subjective metric evaluation, we provide open access to our dataset of generated music and human evaluations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:01:29 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Grötschla",
"Florian",
""
],
[
"Solak",
"Ahmet",
""
],
[
"Lanzendörfer",
"Luca A.",
""
],
[
"Wattenhofer",
"Roger",
""
]
] |
2506.19086 | Rubayat Khan | Fahmida Hai, Saif Nirzhor, Rubayat Khan and Don Roosan | Enhancing Biosecurity in Tamper-Resistant Large Language Models With Quantum Gradient Descent | The conference schedule and details can be found here: https://www.insticc.org/node/technicalprogram/DATA/2025 | null | null | null | q-bio.MN | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces a tamper-resistant framework for large language models (LLMs) in medical applications, utilizing quantum gradient descent (QGD) to detect malicious parameter modifications in real time. Integrated into a LLaMA-based model, QGD monitors weight amplitude distributions, identifying adversarial fine-tuning anomalies. Tests on the MIMIC and eICU datasets show minimal performance impact (accuracy: 89.1 to 88.3 on MIMIC) while robustly detecting tampering. PubMedQA evaluations confirm preserved biomedical question-answering capabilities. Compared to baselines like selective unlearning and cryptographic fingerprinting, QGD offers superior sensitivity to subtle weight changes. This quantum-inspired approach ensures secure, reliable medical AI, extensible to other high-stakes domains. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:03:35 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Hai",
"Fahmida",
""
],
[
"Nirzhor",
"Saif",
""
],
[
"Khan",
"Rubayat",
""
],
[
"Roosan",
"Don",
""
]
] |
2506.19087 | Bowen Zhang | Bowen Zhang, Jesse T. Boulerice, Nikhil Kuniyil, Charvi Mendiratta, Satish Kumar, Hila Shamon, B.S. Manjunath | RareSpot: Spotting Small and Rare Wildlife in Aerial Imagery with Multi-Scale Consistency and Context-Aware Augmentation | Accepted to the CVPR 2025 Workshop on Computer Vision for Animal Behavior Tracking and Modeling (CV4Animals) | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Automated detection of small and rare wildlife in aerial imagery is crucial for effective conservation, yet remains a significant technical challenge. Prairie dogs exemplify this issue: their ecological importance as keystone species contrasts sharply with their elusive presence--marked by small size, sparse distribution, and subtle visual features--which undermines existing detection approaches. To address these challenges, we propose RareSpot, a robust detection framework integrating multi-scale consistency learning and context-aware augmentation. Our multi-scale consistency approach leverages structured alignment across feature pyramids, enhancing fine-grained object representation and mitigating scale-related feature loss. Complementarily, context-aware augmentation strategically synthesizes challenging training instances by embedding difficult-to-detect samples into realistic environmental contexts, significantly boosting model precision and recall. Evaluated on an expert-annotated prairie dog drone imagery benchmark, our method achieves state-of-the-art performance, improving detection accuracy by over 35% compared to baseline methods. Importantly, it generalizes effectively across additional wildlife datasets, demonstrating broad applicability. The RareSpot benchmark and approach not only support critical ecological monitoring but also establish a new foundation for detecting small, rare species in complex aerial scenes. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:03:43 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Zhang",
"Bowen",
""
],
[
"Boulerice",
"Jesse T.",
""
],
[
"Kuniyil",
"Nikhil",
""
],
[
"Mendiratta",
"Charvi",
""
],
[
"Kumar",
"Satish",
""
],
[
"Shamon",
"Hila",
""
],
[
"Manjunath",
"B. S.",
""
]
] |
2506.19088 | Fanny Lehmann | Fanny Lehmann, Firat Ozdemir, Benedikt Soja, Torsten Hoefler, Siddhartha Mishra, Sebastian Schemm | Finetuning a Weather Foundation Model with Lightweight Decoders for Unseen Physical Processes | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recent advances in AI weather forecasting have led to the emergence of so-called "foundation models", typically defined by expensive pretraining and minimal fine-tuning for downstream tasks. However, in the natural sciences, a desirable foundation model should also encode meaningful statistical relationships between the underlying physical variables. This study evaluates the performance of the state-of-the-art Aurora foundation model in predicting hydrological variables, which were not considered during pretraining. We introduce a lightweight approach using shallow decoders trained on the latent representations of the pretrained model to predict these new variables. As a baseline, we compare this to fine-tuning the full model, which allows further optimization of the latent space while incorporating new variables into both inputs and outputs. The decoder-based approach requires 50% less training time and 35% less memory, while achieving strong accuracy across various hydrological variables and preserving desirable properties of the foundation model, such as autoregressive stability. Notably, decoder accuracy depends on the physical correlation between the new variables and those used during pretraining, indicating that Aurora's latent space captures meaningful physical relationships. In this sense, we argue that an important quality metric for foundation models in Earth sciences is their ability to be extended to new variables without a full fine-tuning. This provides a new perspective for making foundation models more accessible to communities with limited computational resources, while supporting broader adoption in Earth sciences. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:03:53 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Lehmann",
"Fanny",
""
],
[
"Ozdemir",
"Firat",
""
],
[
"Soja",
"Benedikt",
""
],
[
"Hoefler",
"Torsten",
""
],
[
"Mishra",
"Siddhartha",
""
],
[
"Schemm",
"Sebastian",
""
]
] |
2506.19089 | Nathaniel Getachew | Nathaniel Getachew, Abulhair Saparov | Language Models Might Not Understand You: Evaluating Theory of Mind via Story Prompting | 14 pages, 11 figures | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | We introduce $\texttt{StorySim}$, a programmable framework for synthetically generating stories to evaluate the theory of mind (ToM) and world modeling (WM) capabilities of large language models (LLMs). Unlike prior benchmarks that may suffer from contamination in pretraining data, $\texttt{StorySim}$ produces novel, compositional story prompts anchored by a highly controllable $\texttt{Storyboard}$, enabling precise manipulation of character perspectives and events. We use this framework to design first- and second-order ToM tasks alongside WM tasks that control for the ability to track and model mental states. Our experiments across a suite of state-of-the-art LLMs reveal that most models perform better on WM tasks than ToM tasks, and that models tend to perform better reasoning with humans compared to inanimate objects. Additionally, our framework enabled us to find evidence of heuristic behavior such as recency bias and an over-reliance on earlier events in the story. All code for generating data and evaluations is freely available. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:06:53 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Getachew",
"Nathaniel",
""
],
[
"Saparov",
"Abulhair",
""
]
] |
2506.19090 | Seok-Hwan Park | Eunhyuk Park, Seok-Hwan Park, Osvaldo Simeone, Marco Di Renzo, Shlomo Shamai | SIM-Enabled Hybrid Digital-Wave Beamforming for Fronthaul-Constrained Cell-Free Massive MIMO Systems | Submitted to an IEEE journal | null | null | null | eess.SP cs.IT math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As the dense deployment of access points (APs) in cell-free massive multiple-input multiple-output (CF-mMIMO) systems presents significant challenges, per-AP coverage can be expanded using large-scale antenna arrays (LAAs). However, this approach incurs high implementation costs and substantial fronthaul demands due to the need for dedicated RF chains for all antennas. To address these challenges, we propose a hybrid beamforming framework that integrates wave-domain beamforming via stacked intelligent metasurfaces (SIM) with conventional digital processing. By dynamically manipulating electromagnetic waves, SIM-equipped APs enhance beamforming gains while significantly reducing RF chain requirements. We formulate a joint optimization problem for digital and wave-domain beamforming along with fronthaul compression to maximize the weighted sum-rate for both uplink and downlink transmission under finite-capacity fronthaul constraints. Given the high dimensionality and non-convexity of the problem, we develop alternating optimization-based algorithms that iteratively optimize digital and wave-domain variables. Numerical results demonstrate that the proposed hybrid schemes outperform conventional hybrid schemes, that rely on randomly set wave-domain beamformers or restrict digital beamforming to simple power control. Moreover, the proposed scheme employing sufficiently deep SIMs achieves near fully-digital performance with fewer RF chains in most simulated cases, except in the downlink at low signal-to-noise ratios. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:08:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Park",
"Eunhyuk",
""
],
[
"Park",
"Seok-Hwan",
""
],
[
"Simeone",
"Osvaldo",
""
],
[
"Di Renzo",
"Marco",
""
],
[
"Shamai",
"Shlomo",
""
]
] |
2506.19091 | Anastasiya Ponosova | Anastasiya Ponosova, Irina Zhluktova, Daria Ruzhitskaya, Daniil Trefilov, Anqi Huang, Alexey Wolf, Vladimir Kamynin, Vladimir Tsvetkov, and Vadim Makarov | Pulsed laser attack at 1061 nm potentially compromises quantum key distribution | 6 pages, 4 figures, 1 table | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum key distribution systems offer cryptographic security, provided that all their components are thoroughly characterised. However, certain components might be vulnerable to a laser-damage attack, particularly when attacked at previously untested laser parameters. Here we show that exposing 1550-nm fiber-optic isolators to 17-mW average power, 1061-nm picosecond attacking pulses reduces their isolation below a safe threshold. Furthermore, the exposure to 1160-mW sub-nanosecond pulsed illumination permanently degrades isolation at 1550 nm while the isolators maintain forward transparency. These threats are not addressed by the currently-practiced security analysis. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:13:37 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Ponosova",
"Anastasiya",
""
],
[
"Zhluktova",
"Irina",
""
],
[
"Ruzhitskaya",
"Daria",
""
],
[
"Trefilov",
"Daniil",
""
],
[
"Huang",
"Anqi",
""
],
[
"Wolf",
"Alexey",
""
],
[
"Kamynin",
"Vladimir",
""
],
[
"Tsvetkov",
"Vladimir",
""
],
[
"Makarov",
"Vadim",
""
]
] |
2506.19092 | J. Anthony Tyson | Phanindra Kandula, J. Anthony Tyson, Jacob Askari, Forrest Fankhauser | Simulated impact on LSST data of Starlink V1.5 and V2 satellites | To be submitted to the Astronomical Journal | null | null | Rubin PUB-188 | astro-ph.IM astro-ph.GA | http://creativecommons.org/licenses/by-sa/4.0/ | The new Starlink V2 satellites incorporate improvements to the chassis brightness through dielectric mirrors, off-pointing solar arrays, and black paint on exposed components. For the general case in which the reflectivities are initially unknown, we simulate LSST operations and repeated photometry of every satellite in simulated model constellations. We derive a brightness model of the Starlink V2 satellite and study the simulated apparent brightness as a function of the satellite position relative to the observer and the sun. We find that the V2 Starlink satellites appear brightest at two distinct positions in the sky: when oriented toward the sun at low elevations where light is specularly reflected, and nearly overhead where the satellite is closest to the observer.
A simulation of Starlink V2 satellites at 550 km height distributed across a series of Walker constellations with varying inclinations was analyzed to study the impact on the LSST observations.
Some bright satellites will be visible in LSST observations. For every thousand V1.5 Starlink satellites imaged by LSST in the first hour of a summer night, we find 1.2 of them will appear brighter than 7 AB magnitude. By comparison, for every thousand V2 Starlink satellites observed, we find only 0.93 of them will appear this bright. The off-pointed solar array and reduced diffuse reflection of the chassis mitigate the brightness.
Finally, we simulate lowering this Walker constellation to 350km. Only 0.56 V2 Starlink satellites per thousand brighter than 7 AB magnitude will be observed in the first hour at this height. This is a 40% reduction in number of bright satellites entering the focal plane compared to the constellation at 550km height. We find that a combination of factors yield an apparent surface brightness of these satellites for LSST operations only 5% brighter than at 550km orbit. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:14:37 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Kandula",
"Phanindra",
""
],
[
"Tyson",
"J. Anthony",
""
],
[
"Askari",
"Jacob",
""
],
[
"Fankhauser",
"Forrest",
""
]
] |
2506.19093 | Rubayat Khan | Fahmida Hai, Shriya Samudrala, Ijeoma Ezengwa, Rubayat Khan, Saif Nirzhor and Don Roosan | Harnessing Diet and Gene Expression Insights through a Centralized Nutrigenomics Database to Improve Public Health | Conference details can be found here: https://www.insticc.org/node/technicalprogram/DATA/2025 | null | null | null | q-bio.QM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nutrigenomics is an emerging field that explores the intricate interaction between genes and diet. This study aimed to develop a comprehensive database to help clinicians and patients understand the connections between genetic disorders, associated genes, and tailored nutritional recommendations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:14:46 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Hai",
"Fahmida",
""
],
[
"Samudrala",
"Shriya",
""
],
[
"Ezengwa",
"Ijeoma",
""
],
[
"Khan",
"Rubayat",
""
],
[
"Nirzhor",
"Saif",
""
],
[
"Roosan",
"Don",
""
]
] |
2506.19094 | Matthew Golub | Belle Liu, Jacob Sacks, and Matthew D. Golub | Accurate identification of communication between multiple interacting neural populations | null | Forty-second International Conference on Machine Learning (2025) | null | null | q-bio.NC cs.CE | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Neural recording technologies now enable simultaneous recording of population activity across many brain regions, motivating the development of data-driven models of communication between brain regions. However, existing models can struggle to disentangle the sources that influence recorded neural populations, leading to inaccurate portraits of inter-regional communication. Here, we introduce Multi-Region Latent Factor Analysis via Dynamical Systems (MR-LFADS), a sequential variational autoencoder designed to disentangle inter-regional communication, inputs from unobserved regions, and local neural population dynamics. We show that MR-LFADS outperforms existing approaches at identifying communication across dozens of simulations of task-trained multi-region networks. When applied to large-scale electrophysiology, MR-LFADS predicts brain-wide effects of circuit perturbations that were held out during model fitting. These validations on synthetic and real neural data position MR-LFADS as a promising tool for discovering principles of brain-wide information processing. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:15:29 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Liu",
"Belle",
""
],
[
"Sacks",
"Jacob",
""
],
[
"Golub",
"Matthew D.",
""
]
] |
2506.19095 | Fien van Wetten | Fien van Wetten, Aske Plaat and Max van Duijn | Baba is LLM: Reasoning in a Game with Dynamic Rules | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) are known to perform well on language tasks, but struggle with reasoning tasks. This paper explores the ability of LLMs to play the 2D puzzle game Baba is You, in which players manipulate rules by rearranging text blocks that define object properties. Given that this rule-manipulation relies on language abilities and reasoning, it is a compelling challenge for LLMs. Six LLMs are evaluated using different prompt types, including (1) simple, (2) rule-extended and (3) action-extended prompts. In addition, two models (Mistral, OLMo) are finetuned using textual and structural data from the game. Results show that while larger models (particularly GPT-4o) perform better in reasoning and puzzle solving, smaller unadapted models struggle to recognize game mechanics or apply rule changes. Finetuning improves the ability to analyze the game levels, but does not significantly improve solution formulation. We conclude that even for state-of-the-art and finetuned LLMs, reasoning about dynamic rule changes is difficult (specifically, understanding the use-mention distinction). The results provide insights into the applicability of LLMs to complex problem-solving tasks and highlight the suitability of games with dynamically changing rules for testing reasoning and reflection by LLMs. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:16:28 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"van Wetten",
"Fien",
""
],
[
"Plaat",
"Aske",
""
],
[
"van Duijn",
"Max",
""
]
] |
2506.19096 | Hanyu Cheng | Hanyu Cheng, Ziwen Yin, Eleonora Di Valentino, David J. E. Marsh, Luca Visinelli | Constraining exotic high-$z$ reionization histories with Gaussian processes and the cosmic microwave background | 16 pages, 7 figures, 6 tables. COST Actions CA21106 and CA21136. Codes available at https://github.com/Cheng-Hanyu/CLASS_reio_gpr | null | null | CA21106; CA21136 | astro-ph.CO hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The large-angle polarization anisotropies observed in the cosmic microwave background (CMB) are generated by Thomson scattering of CMB photons off free electrons in the post-recombination Universe. In the standard $\Lambda$ cold dark matter cosmological model, the free electron density becomes large at redshifts $z \lesssim 10$ due to the formation of the first stars, leading to reionization of the intergalactic medium. In this work, we use \emph{Gaussian processes} to give a model-independent reconstruction of the reionization history of the Universe constrained by the \textit{Planck} CMB measurements. This method gives a consistent reconstruction of the standard reionization at $z \lesssim 10$, and strongly constrains any additional high-$z$ reionization. We use our results to construct a new derived parameter expressing the high-$z$ contribution to the CMB optical depth, $\tau_{\mathrm{highz}}$. The posterior distribution on $\tau_{\mathrm{highz}}$ can be used to set accurate constraints on models for exotic energy injection, which we demonstrate for the case of decaying dark matter with particle mass in the range $\mathcal{O}(1\,\text{MeV})$. A companion paper demonstrates the use of this methodology to test ensembles of multi-axion models. Our results can be recast to different models using the provided data and code: \href{https://github.com/Cheng-Hanyu/CLASS_reio_gpr}{github.com/Cheng-Hanyu/CLASS\_reio\_gpr}. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:17:17 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Cheng",
"Hanyu",
""
],
[
"Yin",
"Ziwen",
""
],
[
"Di Valentino",
"Eleonora",
""
],
[
"Marsh",
"David J. E.",
""
],
[
"Visinelli",
"Luca",
""
]
] |
2506.19097 | Rubayat Khan | Don Roosan, Saif Nirzhor, Rubayat Khan and Fahmida Hai | Quantum Gradient Optimized Drug Repurposing Prototype for Omics Data | Conference details can be found here: https://www.insticc.org/node/technicalprogram/DATA/2025 | null | null | null | q-bio.GN | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel quantum-enhanced prototype for drug repurposing and addresses the challenge of managing massive genomics data in precision medicine. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:17:55 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Roosan",
"Don",
""
],
[
"Nirzhor",
"Saif",
""
],
[
"Khan",
"Rubayat",
""
],
[
"Hai",
"Fahmida",
""
]
] |
2506.19098 | Dmitry V. Bolotov | Dmitry V.Bolotov | Dual Thurston norm of Euler classes of foliations on negative curvature 3-Manifolds | arXiv admin note: text overlap with arXiv:2212.06807 | null | null | null | math.GT math.DG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we give an upper bound estimate on the dual Thurston norm of the Euler class of an arbitrary smooth foliation $\mathcal{F}$ of dimension one defined on a closed three-dimensional orientable manifold $M^3$ of negative curvature, which depends on the constants bounded the injectivity radius $inj(M^3)$, the volume $Vol(M^3)$, sectional curvature of the manifold $M^3$ and the mean curvature modulus of the leaves of the foliation $\mathcal{F}$. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:20:21 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bolotov",
"Dmitry V.",
""
]
] |
2506.19099 | Vaclav Vavrycuk | Vaclav Vavrycuk | Time dilation observed in Type Ia supernova light curves and its cosmological consequences | 14 pages, 7 figures | Galaxies, 13, 55 (2025) | 10.3390/galaxies13030055 | null | astro-ph.CO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The cosmic time dilation observed in Type Ia supernova light curves suggests that the passage of cosmic time varies throughout the evolution of the Universe. This observation implies that the rate of proper time is not constant, as assumed in the standard FLRW metric, but instead is time-dependent. Consequently, the commonly used FLRW metric should be replaced by a more general framework, known as the Conformal Cosmology (CC) metric, to properly account for cosmic time dilation. The CC metric incorporates both spatial expansion and time dilation during cosmic evolution. As a result, it is necessary to distinguish between comoving and proper (physical) time, similar to the distinction made between comoving and proper distances. In addition to successfully explaining cosmic time dilation, the CC metric offers several further advantages: (1) it preserves Lorentz invariance, (2) it maintains the form of Maxwell's equations as in Minkowski space-time, (3) it eliminates the need for dark matter and dark energy in the Friedmann equations, and (4) it successfully predicts the expansion and morphology of spiral galaxies in agreement with observations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:22:54 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Vavrycuk",
"Vaclav",
""
]
] |
2506.19100 | Alexander Clow A.Clow | Ben Cameron and Alexander Clow | On Gy\'arf\'as' Path-Colour Problem | 30 pages, 5 figures | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | In their 1997 paper titled ``Fruit Salad", Gy\'{a}rf\'{a}s posed the following conjecture: there exists a constant $k$ such that if each path of a graph spans a $3$-colourable subgraph, then the graph is $k$-colourable. It is noted that $k=4$ might suffice. Let $r(G)$ be the maximum chromatic number of any subgraph $H$ of $G$ where $H$ is spanned by a path. The only progress on this conjecture comes from Randerath and Schiermeyer in 2002, who proved that if $G$ is an $n$ vertex graph, then $\chi(G) \leq r(G)\log_{\frac{8}{7}}(n)$.
We prove that for all natural numbers $r$, there exists a graph $G$ with $r(G)\leq r$ and $\chi(G)\geq \lfloor\frac{3r}{2}\rfloor -1$. Hence, for all constants $k$ there exists a graph with $\chi - r > k$. Our proof is constructive.
We also study this problem in graphs with a forbidden induced subgraph. We show that if $G$ is $K_{1,t}$-free, for $t\geq 4$, then $\chi(G) \leq (t-1)(r(G)+\binom{t-1}{2}-3)$. If $G$ is claw-free, then we prove $\chi(G) \leq 2r(G)$. Additionally, the graphs $G$ where every induced subgraph $G'$ of $G$ satisfy $\chi(G') = r(G')$ are considered. We call such graphs path-perfect, as this class generalizes perfect graphs. We prove that if $H$ is a forest with at most $4$ vertices other than the claw, then every $H$-free graph $G$ has $\chi(G) \leq r(G)+1$. We also prove that if $H$ is additionally not isomorphic to $2K_2$ or $K_2+2K_1$, then all $H$-free graphs are path-perfect. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:23:07 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Cameron",
"Ben",
""
],
[
"Clow",
"Alexander",
""
]
] |
2506.19101 | Vladyslav Naboka Yurievich | Vladislav Naboka, Yuri Sinyukov, Musfer Adzhymambetov, Hanna Zbroszczyk | Interferometry Radii at RHIC BES Energies within the integrated HydroKinetic Model | 9 pages, 3 figures | null | null | null | hep-ph nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study presents a correlation femtoscopy analysis of relativistic heavy-ion collisions at RHIC Beam Energy Scan (BES) energies using the extended integrated HydroKinetic Model (iHKMe). The model combines initial conditions from UrQMD transport simulations, a gradual pre-equilibrium relaxation stage, viscous hydrodynamic evolution, and a hadronic cascade, providing a comprehensive description of the system's full dynamical evolution. Special emphasis is placed on the impact of extended nuclear overlap times and partial thermalization, which are characteristic of intermediate collision energies (sqrt(s_NN) = 7.7-39 GeV). The corresponding correlation femtoscopy scales, or interferometry (HBT) radii, are calculated across a range of collision energies and compared with experimental data to assess the sensitivity of femtoscopic observables to the equation of state (EoS). Two EoS scenarios are explored. The model demonstrates good agreement with experimental femtoscopy measurements, particularly at lower BES energies. These results enhance our understanding of the space-time structure of the particle-emitting source and provide valuable insight into the nature of the QCD phase transition in the intermediate energy regime. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:30:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Naboka",
"Vladislav",
""
],
[
"Sinyukov",
"Yuri",
""
],
[
"Adzhymambetov",
"Musfer",
""
],
[
"Zbroszczyk",
"Hanna",
""
]
] |
2506.19102 | Bharat Sharma | Maedeh Rahimitouranposhti, Bharat Sharma, Mustafa Can Camur, Olufemi A. Omitaomu and Xueping Li | Investigating Resiliency of Transportation Network Under Targeted and Potential Climate Change Disruptions | null | null | null | null | stat.AP physics.soc-ph | http://creativecommons.org/licenses/by/4.0/ | Ensuring robustness and resilience in intermodal transportation systems is essential for the continuity and reliability of global logistics. These systems are vulnerable to various disruptions, including natural disasters and technical failures. Despite significant research on freight transportation resilience, investigating the robustness of the system after targeted and climate-change driven disruption remains a crucial challenge. Drawing on network science methodologies, this study models the interdependencies within the rail and water transport networks and simulates different disruption scenarios to evaluate system responses. We use the data from the US Department of Energy Volpe Center for network topology and tonnage projections. The proposed framework quantifies deliberate, stochastic, and climate driven infrastructure failure, using higher resolution downscaled multiple Earth System Models simulations from Coupled Model Intercomparison Project Phase version 6. We show that the disruptions of a few nodes could have a larger impact on the total tonnage of freight transport than on network topology. For example, the removal of targeted 20 nodes can bring the total tonnage carrying capacity to 30 percent with about 75 percent of the rail freight network intact. This research advances the theoretical understanding of transportation resilience and provides practical applications for infrastructure managers and policymakers. By implementing these strategies, stakeholders and policymakers can better prepare for and respond to unexpected disruptions, ensuring sustained operational efficiency in the transportation networks. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:33:58 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Rahimitouranposhti",
"Maedeh",
""
],
[
"Sharma",
"Bharat",
""
],
[
"Camur",
"Mustafa Can",
""
],
[
"Omitaomu",
"Olufemi A.",
""
],
[
"Li",
"Xueping",
""
]
] |
2506.19103 | Ilia Beletskii | Ilia Beletskii, Andrey Kuznetsov, Aibek Alanov | Inverse-and-Edit: Effective and Fast Image Editing by Cycle Consistency Models | The code of our method is available on GitHub at https://github.com/ControlGenAI/Inverse-and-Edit | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent advances in image editing with diffusion models have achieved impressive results, offering fine-grained control over the generation process. However, these methods are computationally intensive because of their iterative nature. While distilled diffusion models enable faster inference, their editing capabilities remain limited, primarily because of poor inversion quality. High-fidelity inversion and reconstruction are essential for precise image editing, as they preserve the structural and semantic integrity of the source image. In this work, we propose a novel framework that enhances image inversion using consistency models, enabling high-quality editing in just four steps. Our method introduces a cycle-consistency optimization strategy that significantly improves reconstruction accuracy and enables a controllable trade-off between editability and content preservation. We achieve state-of-the-art performance across various image editing tasks and datasets, demonstrating that our method matches or surpasses full-step diffusion models while being substantially more efficient. The code of our method is available on GitHub at https://github.com/ControlGenAI/Inverse-and-Edit. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:34:43 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Beletskii",
"Ilia",
""
],
[
"Kuznetsov",
"Andrey",
""
],
[
"Alanov",
"Aibek",
""
]
] |
2506.19104 | Daan Huybrechs | Daan Huybrechs | On the algorithmic construction of deep ReLU networks | null | null | null | null | cs.LG cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is difficult to describe in mathematical terms what a neural network trained on data represents. On the other hand, there is a growing mathematical understanding of what neural networks are in principle capable of representing. Feedforward neural networks using the ReLU activation function represent continuous and piecewise linear functions and can approximate many others. The study of their expressivity addresses the question: which ones? Contributing to the available answers, we take the perspective of a neural network as an algorithm. In this analogy, a neural network is programmed constructively, rather than trained from data. An interesting example is a sorting algorithm: we explicitly construct a neural network that sorts its inputs exactly, not approximately, and that, in a sense, has optimal computational complexity if the input dimension is large. Such constructed networks may have several billion parameters. We construct and analyze several other examples, both existing and new. We find that, in these examples, neural networks as algorithms are typically recursive and parallel. Compared to conventional algorithms, ReLU networks are restricted by having to be continuous. Moreover, the depth of recursion is limited by the depth of the network, with deep networks having superior properties over shallow ones. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:35:52 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Huybrechs",
"Daan",
""
]
] |
2506.19105 | Elias Gabriel Minian | Agust\'in Nicol\'as Barreto and Elias Gabriel Minian | Weakly $G$-slim complexes and the non-positive immersion property for generalized Wirtinger presentations | 11 pages | null | null | null | math.GR math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate weakly $G$-slim complexes, a more flexible variant of Helfer and Wise's slim complexes, which can be defined on any regular $G$-covering. We prove that if a $2$-complex $X$ associated to a group presentation is weakly $H$-slim for some regular $H$-covering, and $\pi_1X$ is left-orderable, then $X$ is slim and, in particular, it has non-positive immersions. We apply this result to study conditions on generalized Wirtinger presentations that guarantee the non-positive immersion property for their associated $2$-complexes. This class of presentations includes classical Wirtinger presentations of (perhaps high-dimensional) knots in spheres, some classes of Adian presentations and LOTs presentations. On the one hand, our results provide a wide variety of examples of presentation complexes with the non-positive immersion property via a condition that can be checked with a simple algorithm. On the other hand, as an application of these methods, we derive an extension of a result by Wise on Adian presentations and prove a stronger formulation of a result of Howie on LOT presentations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:36:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Barreto",
"Agustín Nicolás",
""
],
[
"Minian",
"Elias Gabriel",
""
]
] |
2506.19106 | Pekka Ruusuvuori | Umair Khan, Jouni H\"ark\"onen, Marjukka Friman, Leena Latonen, Teijo Kuopio and Pekka Ruusuvuori | Staining normalization in histopathology: Method benchmarking using multicenter dataset | 18 pages, 9 figures | null | null | null | eess.IV cs.CV q-bio.TO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Hematoxylin and Eosin (H&E) has been the gold standard in tissue analysis for decades, however, tissue specimens stained in different laboratories vary, often significantly, in appearance. This variation poses a challenge for both pathologists' and AI-based downstream analysis. Minimizing stain variation computationally is an active area of research. To further investigate this problem, we collected a unique multi-center tissue image dataset, wherein tissue samples from colon, kidney, and skin tissue blocks were distributed to 66 different labs for routine H&E staining. To isolate staining variation, other factors affecting the tissue appearance were kept constant. Further, we used this tissue image dataset to compare the performance of eight different stain normalization methods, including four traditional methods, namely, histogram matching, Macenko, Vahadane, and Reinhard normalization, and two deep learning-based methods namely CycleGAN and Pixp2pix, both with two variants each. We used both quantitative and qualitative evaluation to assess the performance of these methods. The dataset's inter-laboratory staining variation could also guide strategies to improve model generalizability through varied training data | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:37:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Khan",
"Umair",
""
],
[
"Härkönen",
"Jouni",
""
],
[
"Friman",
"Marjukka",
""
],
[
"Latonen",
"Leena",
""
],
[
"Kuopio",
"Teijo",
""
],
[
"Ruusuvuori",
"Pekka",
""
]
] |
2506.19107 | Ruiwei Xiao | Ruiwei Xiao, Xinying Hou, Runlong Ye, Majeed Kazemitabaar, Nicholas Diana, Michael Liut, John Stamper | Improving Student-AI Interaction Through Pedagogical Prompting: An Example in Computer Science Education | Under review for Computer & Education: Artificial Intelligence. Journal policy allows submitting as preprint | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by/4.0/ | With the proliferation of large language model (LLM) applications since 2022, their use in education has sparked both excitement and concern. Recent studies consistently highlight students' (mis)use of LLMs can hinder learning outcomes. This work aims to teach students how to effectively prompt LLMs to improve their learning. We first proposed pedagogical prompting, a theoretically-grounded new concept to elicit learning-oriented responses from LLMs. To move from concept design to a proof-of-concept learning intervention in real educational settings, we selected early undergraduate CS education (CS1/CS2) as the example context. We began with a formative survey study with instructors (N=36) teaching early-stage undergraduate-level CS courses to inform the instructional design based on classroom needs. Based on their insights, we designed and developed a learning intervention through an interactive system with scenario-based instruction to train pedagogical prompting skills. Finally, we evaluated its instructional effectiveness through a user study with CS novice students (N=22) using pre/post-tests. Through mixed methods analyses, our results indicate significant improvements in learners' LLM-based pedagogical help-seeking skills, along with positive attitudes toward the system and increased willingness to use pedagogical prompts in the future. Our contributions include (1) a theoretical framework of pedagogical prompting; (2) empirical insights into current instructor attitudes toward pedagogical prompting; and (3) a learning intervention design with an interactive learning tool and scenario-based instruction leading to promising results on teaching LLM-based help-seeking. Our approach is scalable for broader implementation in classrooms and has the potential to be integrated into tools like ChatGPT as an on-boarding experience to encourage learning-oriented use of generative AI. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:39:17 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Xiao",
"Ruiwei",
""
],
[
"Hou",
"Xinying",
""
],
[
"Ye",
"Runlong",
""
],
[
"Kazemitabaar",
"Majeed",
""
],
[
"Diana",
"Nicholas",
""
],
[
"Liut",
"Michael",
""
],
[
"Stamper",
"John",
""
]
] |
2506.19108 | Darius Afchar | Darius Afchar, Gabriel Meseguer-Brocal, Kamil Akesbi and Romain Hennequin | A Fourier Explanation of AI-music Artifacts | Accepted at ISMIR 2025 | null | null | null | cs.SD | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The rapid rise of generative AI has transformed music creation, with millions of users engaging in AI-generated music. Despite its popularity, concerns regarding copyright infringement, job displacement, and ethical implications have led to growing scrutiny and legal challenges. In parallel, AI-detection services have emerged, yet these systems remain largely opaque and privately controlled, mirroring the very issues they aim to address. This paper explores the fundamental properties of synthetic content and how it can be detected. Specifically, we analyze deconvolution modules commonly used in generative models and mathematically prove that their outputs exhibit systematic frequency artifacts -- manifesting as small yet distinctive spectral peaks. This phenomenon, related to the well-known checkerboard artifact, is shown to be inherent to a chosen model architecture rather than a consequence of training data or model weights. We validate our theoretical findings through extensive experiments on open-source models, as well as commercial AI-music generators such as Suno and Udio. We use these insights to propose a simple and interpretable detection criterion for AI-generated music. Despite its simplicity, our method achieves detection accuracy on par with deep learning-based approaches, surpassing 99% accuracy on several scenarios. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:39:23 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Afchar",
"Darius",
""
],
[
"Meseguer-Brocal",
"Gabriel",
""
],
[
"Akesbi",
"Kamil",
""
],
[
"Hennequin",
"Romain",
""
]
] |
2506.19109 | Valerii Gakh | Valerii Gakh, Hayretdin Bahsi | Enhancing Security in LLM Applications: A Performance Evaluation of Early Detection Systems | 18 pages, 8 tables, 7 figures | null | null | null | cs.CR cs.AI | http://creativecommons.org/licenses/by/4.0/ | Prompt injection threatens novel applications that emerge from adapting LLMs for various user tasks. The newly developed LLM-based software applications become more ubiquitous and diverse. However, the threat of prompt injection attacks undermines the security of these systems as the mitigation and defenses against them, proposed so far, are insufficient. We investigated the capabilities of early prompt injection detection systems, focusing specifically on the detection performance of techniques implemented in various open-source solutions. These solutions are supposed to detect certain types of prompt injection attacks, including the prompt leak. In prompt leakage attacks, an attacker maliciously manipulates the LLM into outputting its system instructions, violating the system's confidentiality. Our study presents analyzes of distinct prompt leakage detection techniques, and a comparative analysis of several detection solutions, which implement those techniques. We identify the strengths and weaknesses of these techniques and elaborate on their optimal configuration and usage in high-stake deployments. In one of the first studies on existing prompt leak detection solutions, we compared the performances of LLM Guard, Vigil, and Rebuff. We concluded that the implementations of canary word checks in Vigil and Rebuff were not effective at detecting prompt leak attacks, and we proposed improvements for them. We also found an evasion weakness in Rebuff's secondary model-based technique and proposed a mitigation. Then, the result of the comparison of LLM Guard, Vigil, and Rebuff at their peak performance revealed that Vigil is optimal for cases when minimal false positive rate is required, and Rebuff is the most optimal for average needs. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:39:43 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Gakh",
"Valerii",
""
],
[
"Bahsi",
"Hayretdin",
""
]
] |
2506.19110 | Sara Congia | Sara Congia, Massimo Borghi, Emanuele Brusaschi, Federico Andrea Sabattoli, Houssein El Dirani, Laurene Youssef, Camille Petit-Etienne, Erwine Pargon, Corrado Sciancalepore, Marco Liscidini, Johan Rothman, S\'egol\`ene Olivier, Matteo Galli, Daniele Bajoni | Generation of hyperentangled photon pairs in the time and frequency domain on a silicon photonic chip | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Multi-dimensional entangled photon states represent an important resource in quantum communication networks. Specifically, hyperentangled states presenting simultaneous entanglement in several degrees of freedom (DoF), stand out for their noise resilience and information capacity. In this work, we demonstrate the generation of hyperentangled photon pairs in the time and frequency-bin domain by spontaneous four-wave mixing from the coherent driving of two integrated Silicon microresonators. We demonstrate entanglement in each DoF by proving the violation of the Clauser Horne Shimony Holt (CHSH) inequality by more than 27 standard deviations (STDs) in each reduced space. Genuine hyperentanglement is then assessed from the negativity of an hyperentanglement witness, which is verified by more than 60 STDs. These results mark, to the best of our knowledge, the first demonstration of time-frequency bin hyperentanglement in an integrated silicon photonic device. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:40:29 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Congia",
"Sara",
""
],
[
"Borghi",
"Massimo",
""
],
[
"Brusaschi",
"Emanuele",
""
],
[
"Sabattoli",
"Federico Andrea",
""
],
[
"Dirani",
"Houssein El",
""
],
[
"Youssef",
"Laurene",
""
],
[
"Petit-Etienne",
"Camille",
""
],
[
"Pargon",
"Erwine",
""
],
[
"Sciancalepore",
"Corrado",
""
],
[
"Liscidini",
"Marco",
""
],
[
"Rothman",
"Johan",
""
],
[
"Olivier",
"Ségolène",
""
],
[
"Galli",
"Matteo",
""
],
[
"Bajoni",
"Daniele",
""
]
] |
2506.19111 | Mathias Marchal | Mathias Marchal, Evangelos Dimopoulos, Kasper Spiegelhauer, Nikolaos Chatzaras, Marco Saldutti, Kresten Yvind, Yi Yu, Jesper M{\o}rk | Carrier Transport in Electrically-Driven Photonic Crystal Membrane Lasers | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | We model carrier transport in photonic crystal lasers with lateral current injection through two-dimensional (2D) finite-volume simulations. Though such lasers can achieve ultra-low threshold currents, leakage paths reduce the carrier injection efficiency. The design is evaluated through its performance in terms of injection efficiency, internal quantum efficiency, and IV characteristics. Our model predicts the presence of unconventional leakage paths, explaining experimental observations of low injection efficiencies and enhanced spontaneous recombination at doping interfaces. Carrier leakage paths arise due to insufficient injection of holes into the active region, leading to an electric field that increases the energy barrier for electrons, reducing the injection efficiency. The profile of the p-doped region is shown to play a critical role in achieving a high electrical injection efficiency and low-threshold lasing. The model is an important step towards modelling and optimizing properties of 2D photonic crystal membrane lasers. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:40:41 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Marchal",
"Mathias",
""
],
[
"Dimopoulos",
"Evangelos",
""
],
[
"Spiegelhauer",
"Kasper",
""
],
[
"Chatzaras",
"Nikolaos",
""
],
[
"Saldutti",
"Marco",
""
],
[
"Yvind",
"Kresten",
""
],
[
"Yu",
"Yi",
""
],
[
"Mørk",
"Jesper",
""
]
] |
2506.19112 | Yizhar Or | Rom Levy, Ari Dantus, Zitao Yu and Yizhar Or | Analysis and experiments of the dissipative Twistcar: direction reversal and asymptotic approximations | null | null | null | null | cs.RO math.DS | http://creativecommons.org/licenses/by/4.0/ | Underactuated wheeled vehicles are commonly studied as nonholonomic systems with periodic actuation. Twistcar is a classical example inspired by a riding toy, which has been analyzed using a planar model of a dynamical system with nonholonomic constraints. Most of the previous analyses did not account for energy dissipation due to friction. In this work, we study a theoretical two-link model of the Twistcar while incorporating dissipation due to rolling resistance. We obtain asymptotic expressions for the system's small-amplitude steady-state periodic dynamics, which reveals the possibility of reversing the direction of motion upon varying the geometric and mass properties of the vehicle. Next, we design and construct a robotic prototype of the Twistcar whose center-of-mass position can be shifted by adding and removing a massive block, enabling demonstration of the Twistcar's direction reversal phenomenon. We also conduct parameter fitting for the frictional resistance in order to improve agreement with experiments. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:41:08 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Levy",
"Rom",
""
],
[
"Dantus",
"Ari",
""
],
[
"Yu",
"Zitao",
""
],
[
"Or",
"Yizhar",
""
]
] |
2506.19113 | Ramaravind Kommiya Mothilal | Ramaravind K. Mothilal and Joanna Roy and Syed Ishtiaque Ahmed and Shion Guha | Human-Aligned Faithfulness in Toxicity Explanations of LLMs | 21 pages, 5 figures, 7 tables | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The discourse around toxicity and LLMs in NLP largely revolves around detection tasks. This work shifts the focus to evaluating LLMs' reasoning about toxicity -- from their explanations that justify a stance -- to enhance their trustworthiness in downstream tasks. Despite extensive research on explainability, it is not straightforward to adopt existing methods to evaluate free-form toxicity explanation due to their over-reliance on input text perturbations, among other challenges. To account for these, we propose a novel, theoretically-grounded multi-dimensional criterion, Human-Aligned Faithfulness (HAF), that measures the extent to which LLMs' free-form toxicity explanations align with those of a rational human under ideal conditions. We develop six metrics, based on uncertainty quantification, to comprehensively evaluate \haf of LLMs' toxicity explanations with no human involvement, and highlight how "non-ideal" the explanations are. We conduct several experiments on three Llama models (of size up to 70B) and an 8B Ministral model on five diverse toxicity datasets. Our results show that while LLMs generate plausible explanations to simple prompts, their reasoning about toxicity breaks down when prompted about the nuanced relations between the complete set of reasons, the individual reasons, and their toxicity stances, resulting in inconsistent and nonsensical responses. We open-source our code and LLM-generated explanations at https://github.com/uofthcdslab/HAF. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:41:45 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Mothilal",
"Ramaravind K.",
""
],
[
"Roy",
"Joanna",
""
],
[
"Ahmed",
"Syed Ishtiaque",
""
],
[
"Guha",
"Shion",
""
]
] |
2506.19114 | Ashwin Bhat | Ashwin Bhat and Michael Dymond | Fast repetitivity in non-rectifiable Delone sets | null | null | null | null | math.MG cs.DM math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a construction of non-rectifiable, repetitive Delone sets in every Euclidean space $\mathbb{R}^d$ with $d \geq 2$. We further obtain a close to optimal repetitivity function for such sets. The proof is based on the process of encoding a non-realisable density in a Delone set, due to Burago and Kleiner. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:42:11 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bhat",
"Ashwin",
""
],
[
"Dymond",
"Michael",
""
]
] |
2506.19115 | Sebastian Angermund | Sebastian Angermund | A Two-Operator Calculus for Arithmetic-Progression Paths in the Collatz Graph | null | null | null | null | math.GM | http://creativecommons.org/licenses/by/4.0/ | A recast of the standard residue-class analysis of the 3x+1 (Collatz) map in terms of two elementary operators on arithmetic progressions. The resulting calculus (i) splits any progression into its even and odd subsequences in a single step, (ii) gives a closed formula for every set of seeds that realises a prescribed parity word, (iii) yields a one line affine invariant that forbids trajectories consisting of infinitely many odd moves, and (iv) reduces the non-trivial-cycle problem to a pair of linear congruences. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:43:08 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Angermund",
"Sebastian",
""
]
] |
2506.19116 | Edit Matyus | \'Ad\'am Marg\'ocsy, Bal\'azs R\'acsai, P\'eter Jeszenszki, Edit M\'atyus | Rovibrational computations for the He$_2$ a $^3\Sigma_\mathrm{u}^+$ state including non-adiabatic, relativistic, and QED corrections | null | null | null | null | physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A potential energy curve (PEC) accurate to a fraction of 1 ppm ($1:10^6$) is computed for the $^3\Sigma_\mathrm{u}^+$ state of He$_2$ endowed with relativistic and QED corrections. The nuclear Schr\"odinger equation is solved on this PEC with diagonal Born-Oppenheimer and non-adiabatic mass corrections to obtain highly accurate rotational-vibrational levels. The computed rovibrational intervals and fine-structure splittings, spanning over several orders of magnitude in energy, are found to be in remarkable agreement with available high-resolution spectroscopy data. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:44:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Margócsy",
"Ádám",
""
],
[
"Rácsai",
"Balázs",
""
],
[
"Jeszenszki",
"Péter",
""
],
[
"Mátyus",
"Edit",
""
]
] |
2506.19117 | Christina Ourania Tze | Christina Ourania Tze, Daniel Dauner, Yiyi Liao, Dzmitry Tsishkou, Andreas Geiger | PrITTI: Primitive-based Generation of Controllable and Editable 3D Semantic Scenes | Project page: https://raniatze.github.io/pritti/ | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Large-scale 3D semantic scene generation has predominantly relied on voxel-based representations, which are memory-intensive, bound by fixed resolutions, and challenging to edit. In contrast, primitives represent semantic entities using compact, coarse 3D structures that are easy to manipulate and compose, making them an ideal representation for this task. In this paper, we introduce PrITTI, a latent diffusion-based framework that leverages primitives as the main foundational elements for generating compositional, controllable, and editable 3D semantic scene layouts. Our method adopts a hybrid representation, modeling ground surfaces in a rasterized format while encoding objects as vectorized 3D primitives. This decomposition is also reflected in a structured latent representation that enables flexible scene manipulation of ground and object components. To overcome the orientation ambiguities in conventional encoding methods, we introduce a stable Cholesky-based parameterization that jointly encodes object size and orientation. Experiments on the KITTI-360 dataset show that PrITTI outperforms a voxel-based baseline in generation quality, while reducing memory requirements by up to $3\times$. In addition, PrITTI enables direct instance-level manipulation of objects in the scene and supports a range of downstream applications, including scene inpainting, outpainting, and photo-realistic street-view synthesis. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:47:18 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Tze",
"Christina Ourania",
""
],
[
"Dauner",
"Daniel",
""
],
[
"Liao",
"Yiyi",
""
],
[
"Tsishkou",
"Dzmitry",
""
],
[
"Geiger",
"Andreas",
""
]
] |
2506.19119 | Bharat Sharma | Bharat Sharma, Jitendra Kumar, Auroop R. Ganguly, and Forrest M. Hoffman | Carbon Cycle Extremes Accelerate Weakening of the Land Carbon Sink in the Late 21st Century | null | null | null | null | stat.AP | http://creativecommons.org/licenses/by/4.0/ | Increasing surface temperature could lead to enhanced evaporation, reduced soil moisture availability, and more frequent droughts and heat waves. The spatiotemporal co-occurrence of such effects further drives extreme anomalies in vegetation productivity and net land carbon storage. However, the impacts of climate change on extremes in net biospheric production (NBP) over longer time periods are unknown. Using the percentile threshold on the probability distribution curve of NBP anomalies, we computed negative and positive extremes in NBP. Here we show that due to climate warming, about 88% of global regions will experience a larger magnitude of negative NBP extremes than positive NBP extremes toward the end of 2100, which accelerate the weakening of the land carbon sink. Our analysis indicates the frequency of negative extremes associated with declines in biospheric productivity was larger than positive extremes, especially in the tropics. While the overall impact of warming at high latitudes is expected to increase plant productivity and carbon uptake, high-temperature anomalies increasingly induce negative NBP extremes toward the end of the 21st century. Using regression analysis, we found soil moisture anomalies to be the most dominant individual driver of NBP extremes. The compound effect of hot, dry, and fire caused extremes at more than 50% of the total grid cells. The larger proportion of negative NBP extremes raises a concern about whether the Earth is capable of increasing vegetation production with growing human population and rising demand for plant material for food, fiber, fuel, and building materials. The increasing proportion of negative NBP extremes highlights the consequences of not only reduction in total carbon uptake capacity but also of conversion of land to a carbon source. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:48:56 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Sharma",
"Bharat",
""
],
[
"Kumar",
"Jitendra",
""
],
[
"Ganguly",
"Auroop R.",
""
],
[
"Hoffman",
"Forrest M.",
""
]
] |
2506.19120 | Edit Matyus | Edit M\'atyus, \'Ad\'am Marg\'ocsy | Rovibrational computations for He$_2^+$ X$^+\ \Sigma_\mathrm{u}^+$ including non-adiabatic, relativistic and QED corrections | null | null | null | null | physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We report the potential energy curve, the diagonal Born-Oppenheimer, non-adiabatic mass, relativistic, and leading-order QED corrections for the ground electronic state of the helium dimer cation; the higher-order QED and finite-nuclear size effects are also estimated. The computations are carried out with an improved error control and over a much broader configuration range compared to earlier work [D. Ferenc, V. I. Korobov, and E. M\'atyus, Phys. Rev. Lett. 125, 213001 (2020)]. As a result, all rovibrational bound states are reported with an estimated accuracy of 0.005 cm$^{-1}$. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:49:17 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Mátyus",
"Edit",
""
],
[
"Margócsy",
"Ádám",
""
]
] |
2506.19121 | Christopher Agia | Christopher Agia, Rohan Sinha, Jingyun Yang, Rika Antonova, Marco Pavone, Haruki Nishimura, Masha Itkina, Jeannette Bohg | CUPID: Curating Data your Robot Loves with Influence Functions | Project page: https://cupid-curation.github.io. 28 pages, 15 figures | null | null | null | cs.RO cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In robot imitation learning, policy performance is tightly coupled with the quality and composition of the demonstration data. Yet, developing a precise understanding of how individual demonstrations contribute to downstream outcomes - such as closed-loop task success or failure - remains a persistent challenge. We propose CUPID, a robot data curation method based on a novel influence function-theoretic formulation for imitation learning policies. Given a set of evaluation rollouts, CUPID estimates the influence of each training demonstration on the policy's expected return. This enables ranking and selection of demonstrations according to their impact on the policy's closed-loop performance. We use CUPID to curate data by 1) filtering out training demonstrations that harm policy performance and 2) subselecting newly collected trajectories that will most improve the policy. Extensive simulated and hardware experiments show that our approach consistently identifies which data drives test-time performance. For example, training with less than 33% of curated data can yield state-of-the-art diffusion policies on the simulated RoboMimic benchmark, with similar gains observed in hardware. Furthermore, hardware experiments show that our method can identify robust strategies under distribution shift, isolate spurious correlations, and even enhance the post-training of generalist robot policies. Additional materials are made available at: https://cupid-curation.github.io. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:49:34 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Agia",
"Christopher",
""
],
[
"Sinha",
"Rohan",
""
],
[
"Yang",
"Jingyun",
""
],
[
"Antonova",
"Rika",
""
],
[
"Pavone",
"Marco",
""
],
[
"Nishimura",
"Haruki",
""
],
[
"Itkina",
"Masha",
""
],
[
"Bohg",
"Jeannette",
""
]
] |
2506.19122 | Ricardo Rodriguez-Pedraza | Ricardo Rodriguez, Nam Nguyen, Elizabeth Behrman, Andrew C. Y. Li and James Steck | Existence of a robust optimal control process for efficient measurements in a two-qubit system | 13 pages, 2 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Using a Hamiltonian of physical interest, we show the existence of a unitary transformation that drives the initial state of a two-qubit system to a designated final state. We implement an application of the above framework by finding an optimal control process that enables us to determine the entanglement of the two-qubit system with a single measurement. We also show that this process is robust to environmental noise. Our results provide mathematical justification for our earlier computational experiments. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:50:24 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Rodriguez",
"Ricardo",
""
],
[
"Nguyen",
"Nam",
""
],
[
"Behrman",
"Elizabeth",
""
],
[
"Li",
"Andrew C. Y.",
""
],
[
"Steck",
"James",
""
]
] |
2506.19123 | Jacopo Borga | Arka Adhikari, Jacopo Borga, Thomas Budzinski, William Da Silva, Delphin S\'enizergues | The longest increasing subsequence of Brownian separable permutons | Comments are welcome! | null | null | null | math.PR math.CO | http://creativecommons.org/licenses/by/4.0/ | We establish a scaling limit result for the length $\operatorname{LIS}(\sigma_n)$ of the longest increasing subsequence of a permutation $\sigma_n$ of size $n$ sampled from the Brownian separable permuton $\boldsymbol{\mu}_p$ of parameter $p\in(0,1)$, which is the universal limit of pattern-avoiding permutations. Specifically, we prove that \[\frac{\operatorname{LIS}(\sigma_n)}{n^\alpha}\;\underset{n\to\infty}{\overset{\mathrm{a.s.}}{\longrightarrow}}\; X,\] where $\alpha=\alpha(p)$ is the unique solution in the interval $(1/2,1)$ to the equation \[\frac{1}{4^{\frac{1}{2\alpha}}\sqrt{\pi}}\,\frac{\Gamma\big(\tfrac{1}{2}-\tfrac{1}{2\alpha}\big)}{\Gamma\big(1-\tfrac{1}{2\alpha}\big)}=\frac{p}{p-1},\] and $X=X(p)$ is a non-deterministic and a.s. positive and finite random variable, which is a measurable function of the Brownian separable permuton. Notably, the exponent $\alpha(p)$ is an increasing continuous function of $p$ with $\alpha(0^+)=1/2$, $\alpha(1^-)=1$ and $\alpha(1/2)\approx0.815226$, which corresponds to the permuton limit of uniform separable permutations. We prove analogous results for the size of the largest clique of a graph sampled from the Brownian cographon of parameter $p\in(0,1)$. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:51:11 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Adhikari",
"Arka",
""
],
[
"Borga",
"Jacopo",
""
],
[
"Budzinski",
"Thomas",
""
],
[
"Da Silva",
"William",
""
],
[
"Sénizergues",
"Delphin",
""
]
] |
2506.19124 | Abdus Salam Sarkar Dr | Abdus Salam Sarkar | Future prospect of anisotropic 2D tin sulfide (SnS) for emerging electronic and quantum device applications | null | null | null | null | cond-mat.mes-hall physics.app-ph physics.optics quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The family of anisotropic two-dimensional (2D) emerging materials is rapidly evolving due to their low crystal symmetry and in-plane structural anisotropy. Among these, 2D tin sulfide (SnS) has gained significant attention because of its distinctive crystalline symmetry and the resulting extraordinary anisotropic physical properties. This perspective explores recent developments in anisotropic 2D SnS. In particular, it highlights advances in isolating high-quality SnS monolayers (1L-SnS) and in applying advanced techniques for anisotropic characterization. The discussion continues with an overview of the anisotropic optical properties of SnS, followed by recent progress in emerging electronic device applications, including energy storage, neuromorphic (synaptic) systems, and quantum technologies. In addition to presenting significant research findings on SnS, this perspective outlines current limitations and discusses emerging opportunities and future prospects for its application in quantum devices. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:51:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Sarkar",
"Abdus Salam",
""
]
] |
2506.19125 | Parikshit Ram | Kenneth L. Clarkson, Lior Horesh, Takuya Ito, Charlotte Park, Parikshit Ram | Finding Clustering Algorithms in the Transformer Architecture | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | The invention of the transformer architecture has revolutionized Artificial Intelligence (AI), yielding unprecedented success in areas such as natural language processing, computer vision, and multimodal reasoning. Despite these advances, it is unclear whether transformers are able to learn and implement precise algorithms. Here, we demonstrate that transformers can exactly implement a fundamental and widely used algorithm for $k$-means clustering: Lloyd's algorithm. First, we theoretically prove the existence of such a transformer architecture, which we term the $k$-means transformer, that exactly implements Lloyd's algorithm for $k$-means clustering using the standard ingredients of modern transformers: attention and residual connections. Next, we numerically implement this transformer and demonstrate in experiments the exact correspondence between our architecture and Lloyd's algorithm, providing a fully neural implementation of $k$-means clustering. Finally, we demonstrate that interpretable alterations (e.g., incorporating layer normalizations or multilayer perceptrons) to this architecture yields diverse and novel variants of clustering algorithms, such as soft $k$-means, spherical $k$-means, trimmed $k$-means, and more. Collectively, our findings demonstrate how transformer mechanisms can precisely map onto algorithmic procedures, offering a clear and interpretable perspective on implementing precise algorithms in transformers. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:52:01 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Clarkson",
"Kenneth L.",
""
],
[
"Horesh",
"Lior",
""
],
[
"Ito",
"Takuya",
""
],
[
"Park",
"Charlotte",
""
],
[
"Ram",
"Parikshit",
""
]
] |
2506.19126 | Aaron Abrams | Aaron Abrams | Upper Chromatic Numbers: An Update | This is a survey written in 2000 about upper chromatic numbers | Geombinatorics, vol. 10 no. 1 (2000), pp. 4-11 | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This is a survey written in 2000 about upper chromatic numbers | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:55:56 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Abrams",
"Aaron",
""
]
] |
2506.19127 | Gordon Semenoff | Duncan MacIntyre and Gordon W. Semenoff | Entropy from scattering in weakly interacting systems | null | null | null | null | quant-ph cond-mat.other hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | Perturbation theory is used to investigate the evolution of the von Neumann entropy of a subsystem of a bipartite quantum system in the course of a gedanken scattering experiment. We find surprisingly simple criteria for the initial state and the scattering matrix that guarantee that the subsystem entropy increases. The class of states that meet these criteria are more correlated than simple product states of the subsystems. They form a subclass of the set of all separable states, and they can therefore be assembled by classical processes alone. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 20:55:58 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"MacIntyre",
"Duncan",
""
],
[
"Semenoff",
"Gordon W.",
""
]
] |
2506.19128 | Tucker Paron | Tucker Paron | Longitudinal analysis of heart rate variability as it pertains to anxiety and readiness | 16 pages, 10 figures | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The aim of this study is to explore the relationship between lifestyle choices, subjective experiences and objective biometric data in a single individual. The participant, at the time a male in his twenties, used the EliteHRV app to perform Heart Rate Variability Readings across twenty-six months accompanied by logs about the previous days activity as well as current emotional and physical state. The study will use a mixed-methods approach to analyze the data, including quantitative analysis of the biometric data and correlation analysis between the biometric data and subjective experience tags. Qualitative analysis of the daily logs will also be conducted to gain a deeper understanding of the participant's experiences and to identify keywords, people, or ideas that affect biometric output. The results of this study will provide insights into the relationship between subjective and objective measures, and the potential benefits or drawbacks of certain lifestyle choices and ways of thinking. The findings could have implications for the development of wearable-based personalized interventions for improving mental health and well-being. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:01:00 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Paron",
"Tucker",
""
]
] |
2506.19129 | Andrea Bovo | Andrea Bovo, Alessandro Milazzo | Global regularity of the value function in a stopper vs. singular-controller game | 25 pages | null | null | null | math.OC math.AP math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study a class of zero-sum stochastic games between a stopper and a singular-controller, previously considered in [Bovo and De Angelis (2025)]. The underlying singularly-controlled dynamics takes values in $\mathcal{O}\subseteq\mathbb{R}$. The problem is set on a finite time-horizon and is connected to a parabolic variational inequality of min-max type with spatial-derivative and obstacle constraints.
We show that the value function of the problem is of class $C^1$ in the whole domain $[0,T)\times\mathcal{O}$ and that the second-order spatial derivative and the second-order mixed derivative are continuous everywhere except for a (potential) jump across a non-decreasing curve (the stopping boundary of the game). The latter discontinuity is a natural consequence of the partial differential equation associated to the problem. Beyond its intrinsic analytical value, such a regularity for the value function is a stepping stone for further exploring the structure and properties of the free-boundaries of the stochastic game, which in turn determine the optimal strategies of the players. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:01:19 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bovo",
"Andrea",
""
],
[
"Milazzo",
"Alessandro",
""
]
] |
2506.19130 | Blair Davey | Blair Davey | A frequency function approach to quantitative unique continuation for elliptic equations | 22 pages | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the quantitative unique continuation properties of solutions to second order elliptic equations with bounded lower order terms. In particular, we establish quantitative forms of the strong unique continuation property for solutions to generalized Schr\"odinger equations of the form $- \nabla \cdot (A \nabla u) + W \cdot \nabla u + V u = 0$, where we assume that $A$ is bounded, elliptic, symmetric, and Lipschitz, while $W$ and $V$ belong to $L^\infty$. We also study the global unique continuation properties of solutions to these equations, establishing results that are related to Landis' conjecture concerning the optimal rate of decay at infinity. Versions of the theorems in this article have been previously proved using Carleman estimates, but here we present novel proof techniques that rely on frequency functions. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:02:10 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Davey",
"Blair",
""
]
] |
2506.19131 | Edit Matyus | P\'eter Jeszenszki, P\'eter Holl\'osy, \'Ad\'am Marg\'ocsy, Edit M\'atyus | Spin-dependent terms of the Breit-Pauli Hamiltonian evaluated with explicitly correlated Gaussian basis set for molecular computations | null | null | null | null | physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work compiles the spin-dependent leading-order relativistic and quantum electrodynamical corrections to the electronic structure of atoms and molecules within the non-relativistic quantum electrodynamics framework. We report the computation of perturbative corrections using an explicitly correlated Gaussian basis set, which allows high-precision computations for few-electron systems. In addition to numerical tests for triplet Be, triplet H$_2$, and triplet H$_3^+$ states and comparison with no-pair Dirac-Coulomb-Breit Hamiltonian energies, numerical results are reported for electronically excited states of the triplet helium dimer, He$_2$, for which the present implementation delivers high-precision magnetic coupling curves necessary for a quantitative understanding of the fine structure of its high-resolution rovibronic spectrum. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:03:53 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Jeszenszki",
"Péter",
""
],
[
"Hollósy",
"Péter",
""
],
[
"Margócsy",
"Ádám",
""
],
[
"Mátyus",
"Edit",
""
]
] |
2506.19132 | Simeon Mistakidis | F. Bristy, G. A. Bougas, G. C. Katsimiga, S. I. Mistakidis | Localization and splitting of a quantum droplet by immersing a heavy impurity | 13 pages, 8 figures | null | null | null | cond-mat.quant-gas physics.atom-ph quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We unravel the existence and nonequilibrium response of one-dimensional harmonically trapped droplet configurations in the presence of a central repulsive or attractive potential well mimicking the effect of a heavy impurity. For fixed negative chemical potentials, it is shown that droplets fragment into two for increasing repulsive potential heights, a process that occurs faster for larger widths. However, atoms from the droplet accumulate at the attractive potential, especially for wider ones, leading to a deformed droplet and eventually to the termination of the solution. Linearization analysis yields the underlying excitation spectrum which dictates stability and the behavior of the ensuing collective modes. Quenches in the potential height are used to demonstrate dynamical fragmentation of the droplet for repulsive potentials as well as self-evaporation along with droplet localization and eventual relaxation for longer evolution times in the case of attractive potential wells. The many-body character of the dynamics is explicated by evaluating the participating single-particle eigenstates manifesting the superposition nature of the droplet state. Our results should be detectable by current ultracold atom experiments and may inspire engineered droplet dynamics with the aid of external potentials. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:06:07 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bristy",
"F.",
""
],
[
"Bougas",
"G. A.",
""
],
[
"Katsimiga",
"G. C.",
""
],
[
"Mistakidis",
"S. I.",
""
]
] |
2506.19133 | Andreas Bjerregaard | Andreas Bjerregaard and S{\o}ren Hauberg and Anders Krogh | Riemannian generative decoder | GenBio ICML 2025 (Proceedings of the Workshop on Generative AI for Biology at the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025) | null | null | null | cs.LG q-bio.QM stat.ML | http://creativecommons.org/licenses/by/4.0/ | Riemannian representation learning typically relies on approximating densities on chosen manifolds. This involves optimizing difficult objectives, potentially harming models. To completely circumvent this issue, we introduce the Riemannian generative decoder which finds manifold-valued maximum likelihood latents with a Riemannian optimizer while training a decoder network. By discarding the encoder, we vastly simplify the manifold constraint compared to current approaches which often only handle few specific manifolds. We validate our approach on three case studies -- a synthetic branching diffusion process, human migrations inferred from mitochondrial DNA, and cells undergoing a cell division cycle -- each showing that learned representations respect the prescribed geometry and capture intrinsic non-Euclidean structure. Our method requires only a decoder, is compatible with existing architectures, and yields interpretable latent spaces aligned with data geometry. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:06:13 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bjerregaard",
"Andreas",
""
],
[
"Hauberg",
"Søren",
""
],
[
"Krogh",
"Anders",
""
]
] |
2506.19135 | Juan Manuel Burgos Mieres | J. M. Burgos | McGehee blowup for Lagrangian systems and instability of equilibria | 52 pages, 5 figures | null | null | null | math.DS math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We prove that total instability is a generic phenomenon in the real analytic class of electromagnetic Lagrangian systems under a weak magnetism hypothesis. The main object in the proof is an adaptation of the McGehee blowup for these systems. Together with this result, new criteria for total instability are introduced for both generic and non-generic cases. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:09:42 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Burgos",
"J. M.",
""
]
] |
2506.19136 | Cyrill B\"osch | Cyrill B\"osch, Geoffrey Roeder, Marc Serra-Garcia, Ryan P. Adams | Local Learning Rules for Out-of-Equilibrium Physical Generative Models | 6 pages, 2 figures | null | null | null | cs.LG cond-mat.mes-hall cs.ET cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the out-of-equilibrium driving protocol of score-based generative models (SGMs) can be learned via a local learning rule. The gradient with respect to the parameters of the driving protocol are computed directly from force measurements or from observed system dynamics. As a demonstration, we implement an SGM in a network of driven, nonlinear, overdamped oscillators coupled to a thermal bath. We first apply it to the problem of sampling from a mixture of two Gaussians in 2D. Finally, we train a network of 10x10 oscillators to sample images of 0s and 1s from the MNIST dataset. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:11:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bösch",
"Cyrill",
""
],
[
"Roeder",
"Geoffrey",
""
],
[
"Serra-Garcia",
"Marc",
""
],
[
"Adams",
"Ryan P.",
""
]
] |
2506.19137 | Mohamed Amazioug PhD | Hamza Harraf, M'bark Amghar, Wiam Kaydi, Mohamed Amazioug and Rachid Ahl Laamara | Extracted work as measure of entanglement in optomechanics | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we investigate quantum entanglement and work extraction in two distinct optomechanical systems. The first system consists of two spatially separated Fabry-P\'erot cavities driven by squeezed light in the resolved-sideband regime, while the second system comprises a laser field incident on a vibrating mirror. We analyze the entanglement dynamics between optical and mechanical modes, as well as quantum correlations in a mixed optomechanical bipartite system (optic-optic mode) mediated by radiation pressure. Using logarithmic negativity as a measure, we quantify the entanglement evolution in both optic-optic and mirror-mirror bipartite subsystems. Furthermore, we show how three distinct types of extractable work vary with system parameters and examine their relationship with entanglement. Our results highlight the relationship between quantum correlations and extracted work in optomechanical system, offering insights for quantum information processing and energy transfer in hybrid systems. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 22:57:12 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Harraf",
"Hamza",
""
],
[
"Amghar",
"M'bark",
""
],
[
"Kaydi",
"Wiam",
""
],
[
"Amazioug",
"Mohamed",
""
],
[
"Laamara",
"Rachid Ahl",
""
]
] |
2506.19138 | Moh Kamalul Wafi | Moh Kamalul Wafi, Katherin Indriawati, Bambang L. Widjiantoro | Model Reference Adaptive Control of Networked Systems with State and Input Delays | This is the extended version of http://doi.org/10.11591/ijece.v14i5.pp5055-5063 | null | 10.11591/ijece.v14i5.pp5055-5063 | null | eess.SY cs.SY | http://creativecommons.org/licenses/by-sa/4.0/ | Adaptive control strategies have progressively advanced to accommodate increasingly uncertain, delayed, and interconnected systems. This paper addresses the model reference adaptive control (MRAC) of networked, heterogeneous, and unknown dynamical agents subject to both state and input delays. The objective is to ensure that all follower agents asymptotically track the trajectory of a stable leader system, despite system uncertainties and communication constraints. Two communication topologies are considered, full connectivity between each agent and the leader, and partial connectivity wherein agents rely on both neighboring peers and the leader. The agent-to-agent and agent-to-leader interactions are encoded using a Laplacian-like matrix and a diagonal model-weighting matrix, respectively. To compensate for the delays, a predictor-based control structure and an auxiliary dynamic system are proposed. The control framework includes distributed adaptive parameter laws derived via Lyapunov-based analysis, ensuring convergence of the augmented tracking error. Stability conditions are established through a carefully constructed Lyapunov Krasovskii functional, under minimal assumptions on connectivity and excitation. Numerical simulations of both network structures validate the proposed method, demonstrating that exact leader tracking is achieved under appropriately designed learning rates and initializations. This work lays a foundation for future studies on fault-resilient distributed adaptive control incorporating data-driven or reinforcement learning techniques. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:17:28 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Wafi",
"Moh Kamalul",
""
],
[
"Indriawati",
"Katherin",
""
],
[
"Widjiantoro",
"Bambang L.",
""
]
] |
2506.19139 | Lukas Radl | Lukas Radl, Felix Windisch, Thomas Deixelberger, Jozef Hladky, Michael Steiner, Dieter Schmalstieg, Markus Steinberger | SOF: Sorted Opacity Fields for Fast Unbounded Surface Reconstruction | null | null | null | null | cs.GR cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Recent advances in 3D Gaussian representations have significantly improved the quality and efficiency of image-based scene reconstruction. Their explicit nature facilitates real-time rendering and fast optimization, yet extracting accurate surfaces - particularly in large-scale, unbounded environments - remains a difficult task. Many existing methods rely on approximate depth estimates and global sorting heuristics, which can introduce artifacts and limit the fidelity of the reconstructed mesh. In this paper, we present Sorted Opacity Fields (SOF), a method designed to recover detailed surfaces from 3D Gaussians with both speed and precision. Our approach improves upon prior work by introducing hierarchical resorting and a robust formulation of Gaussian depth, which better aligns with the level-set. To enhance mesh quality, we incorporate a level-set regularizer operating on the opacity field and introduce losses that encourage geometrically-consistent primitive shapes. In addition, we develop a parallelized Marching Tetrahedra algorithm tailored to our opacity formulation, reducing meshing time by up to an order of magnitude. As demonstrated by our quantitative evaluation, SOF achieves higher reconstruction accuracy while cutting total processing time by more than a factor of three. These results mark a step forward in turning efficient Gaussian-based rendering into equally efficient geometry extraction. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:20:52 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Radl",
"Lukas",
""
],
[
"Windisch",
"Felix",
""
],
[
"Deixelberger",
"Thomas",
""
],
[
"Hladky",
"Jozef",
""
],
[
"Steiner",
"Michael",
""
],
[
"Schmalstieg",
"Dieter",
""
],
[
"Steinberger",
"Markus",
""
]
] |
2506.19140 | Barry Wang | Barry Wang, Avi Schwarzschild, Alexander Robey, Ali Payani, Charles Fleming, Mingjie Sun, Daphne Ippolito | Command-V: Pasting LLM Behaviors via Activation Profiles | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Retrofitting large language models (LLMs) with new behaviors typically requires full finetuning or distillation-costly steps that must be repeated for every architecture. In this work, we introduce Command-V, a backpropagation-free behavior transfer method that copies an existing residual activation adapter from a donor model and pastes its effect into a recipient model. Command-V profiles layer activations on a small prompt set, derives linear converters between corresponding layers, and applies the donor intervention in the recipient's activation space. This process does not require access to the original training data and needs minimal compute. In three case studies-safety-refusal enhancement, jailbreak facilitation, and automatic chain-of-thought reasoning--Command-V matches or exceeds the performance of direct finetuning while using orders of magnitude less compute. Our code and data are accessible at https://github.com/GithuBarry/Command-V/. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:21:49 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Wang",
"Barry",
""
],
[
"Schwarzschild",
"Avi",
""
],
[
"Robey",
"Alexander",
""
],
[
"Payani",
"Ali",
""
],
[
"Fleming",
"Charles",
""
],
[
"Sun",
"Mingjie",
""
],
[
"Ippolito",
"Daphne",
""
]
] |
2506.19141 | Bruno Aristimunha | Bruno Aristimunha, Dung Truong, Pierre Guetschel, Seyed Yahya Shirazi, Isabelle Guyon, Alexandre R. Franco, Michael P. Milham, Aviv Dotan, Scott Makeig, Alexandre Gramfort, Jean-Remi King, Marie-Constance Corsi, Pedro A. Vald\'es-Sosa, Amit Majumdar, Alan Evans, Terrence J Sejnowski, Oren Shriki, Sylvain Chevallier, Arnaud Delorme | EEG Foundation Challenge: From Cross-Task to Cross-Subject EEG Decoding | Approved at Neurips Competition track. webpage: https://eeg2025.github.io/ | The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS), 2025 | null | null | eess.SP cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Current electroencephalogram (EEG) decoding models are typically trained on small numbers of subjects performing a single task. Here, we introduce a large-scale, code-submission-based competition comprising two challenges. First, the Transfer Challenge asks participants to build and test a model that can zero-shot decode new tasks and new subjects from their EEG data. Second, the Psychopathology factor prediction Challenge asks participants to infer subject measures of mental health from EEG data. For this, we use an unprecedented, multi-terabyte dataset of high-density EEG signals (128 channels) recorded from over 3,000 child to young adult subjects engaged in multiple active and passive tasks. We provide several tunable neural network baselines for each of these two challenges, including a simple network and demographic-based regression models. Developing models that generalise across tasks and individuals will pave the way for ML network architectures capable of adapting to EEG data collected from diverse tasks and individuals. Similarly, predicting mental health-relevant personality trait values from EEG might identify objective biomarkers useful for clinical diagnosis and design of personalised treatment for psychological conditions. Ultimately, the advances spurred by this challenge could contribute to the development of computational psychiatry and useful neurotechnology, and contribute to breakthroughs in both fundamental neuroscience and applied clinical research. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:25:19 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Aristimunha",
"Bruno",
""
],
[
"Truong",
"Dung",
""
],
[
"Guetschel",
"Pierre",
""
],
[
"Shirazi",
"Seyed Yahya",
""
],
[
"Guyon",
"Isabelle",
""
],
[
"Franco",
"Alexandre R.",
""
],
[
"Milham",
"Michael P.",
""
],
[
"Dotan",
"Aviv",
""
],
[
"Makeig",
"Scott",
""
],
[
"Gramfort",
"Alexandre",
""
],
[
"King",
"Jean-Remi",
""
],
[
"Corsi",
"Marie-Constance",
""
],
[
"Valdés-Sosa",
"Pedro A.",
""
],
[
"Majumdar",
"Amit",
""
],
[
"Evans",
"Alan",
""
],
[
"Sejnowski",
"Terrence J",
""
],
[
"Shriki",
"Oren",
""
],
[
"Chevallier",
"Sylvain",
""
],
[
"Delorme",
"Arnaud",
""
]
] |
2506.19142 | Yubai Yuan | Yubai Yuan, Siyu Huang, Abdul Basit Adeel | Inferring Diffusion Structures of Heterogeneous Network Cascade | null | null | null | null | cs.SI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Network cascade refers to diffusion processes in which outcome changes within part of an interconnected population trigger a sequence of changes across the entire network. These cascades are governed by underlying diffusion networks, which are often latent. Inferring such networks is critical for understanding cascade pathways, uncovering Granger causality of interaction mechanisms among individuals, and enabling tasks such as forecasting or maximizing information propagation. In this project, we propose a novel double mixture directed graph model for inferring multi-layer diffusion networks from cascade data. The proposed model represents cascade pathways as a mixture of diffusion networks across different layers, effectively capturing the strong heterogeneity present in real-world cascades. Additionally, the model imposes layer-specific structural constraints, enabling diffusion networks at different layers to capture complementary cascading patterns at the population level. A key advantage of our model is its convex formulation, which allows us to establish both statistical and computational guarantees for the resulting diffusion network estimates. We conduct extensive simulation studies to demonstrate the model's performance in recovering diverse diffusion structures. Finally, we apply the proposed method to analyze cascades of research topics in the social sciences across U.S. universities, revealing the underlying diffusion networks of research topic propagation among institutions. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:26:19 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Yuan",
"Yubai",
""
],
[
"Huang",
"Siyu",
""
],
[
"Adeel",
"Abdul Basit",
""
]
] |
2506.19144 | Kyeongwon Lee | Kyeongwon Lee, Lizhen Lin, Jaewoo Park, and Seonghyun Jeong | Posterior Contraction for Sparse Neural Networks in Besov Spaces with Intrinsic Dimensionality | null | null | null | null | stat.ML cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work establishes that sparse Bayesian neural networks achieve optimal posterior contraction rates over anisotropic Besov spaces and their hierarchical compositions. These structures reflect the intrinsic dimensionality of the underlying function, thereby mitigating the curse of dimensionality. Our analysis shows that Bayesian neural networks equipped with either sparse or continuous shrinkage priors attain the optimal rates which are dependent on the intrinsic dimension of the true structures. Moreover, we show that these priors enable rate adaptation, allowing the posterior to contract at the optimal rate even when the smoothness level of the true function is unknown. The proposed framework accommodates a broad class of functions, including additive and multiplicative Besov functions as special cases. These results advance the theoretical foundations of Bayesian neural networks and provide rigorous justification for their practical effectiveness in high-dimensional, structured estimation problems. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:29:40 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Lee",
"Kyeongwon",
""
],
[
"Lin",
"Lizhen",
""
],
[
"Park",
"Jaewoo",
""
],
[
"Jeong",
"Seonghyun",
""
]
] |
2506.19145 | Krzysztof Grygiel | K. Chmielewski, K. Grygiel, K. Bartkiewicz | Stability Analysis of Three Coupled Kerr Oscillators: Implications for Quantum Computing | 15 pages, 7 figures | null | null | null | nlin.CD quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the classical dynamics of optical nonlinear Kerr couplers, focusing on their potential relevance to quantum computing applications. The system consists of three Kerr-type nonlinear oscillators arranged in two configurations: a triangular arrangement, where each oscillator is coupled to the others, and a sandwich arrangement, where only the middle oscillator interacts with the two outer ones. The system is driven by an external periodic field and includes dissipative processes. Its evolution is governed by six non-autonomous differential equations derived from a Kerr Hamiltonian with nonlinear coupling terms. We show that even for identical Kerr media, the interplay between nonlinear couplings and mismatched fundamental and pump frequencies leads to rich and complex dynamics, including the emergence of multiple stable attractors. These attractors are highly sensitive to both the coupling configuration and initial conditions. A key contribution of this work is a detailed stability analysis based on the numerical calculation of Lyapunov exponents, which reveals transitions from regular to chaotic dynamics as damping is reduced. We identify critical damping thresholds for the onset of chaos and characterize phenomena such as chaotic beats. These results offer insights for potential experimental realizations and are directly relevant to emerging quantum technologies, where Kerr parametric oscillators play a central role in quantum gates, error correction protocols, and quantum neural network architectures. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:31:17 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Chmielewski",
"K.",
""
],
[
"Grygiel",
"K.",
""
],
[
"Bartkiewicz",
"K.",
""
]
] |
2506.19146 | Mehmet Fatih Ozkan | Mehmet Fatih Ozkan, Samuel Filgueira da Silva, Faissal El Idrissi, Prashanth Ramesh and Marcello Canova | Optimal Design of Experiment for Electrochemical Parameter Identification of Li-ion Battery via Deep Reinforcement Learning | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Accurate parameter estimation in electrochemical battery models is essential for monitoring and assessing the performance of lithium-ion batteries (LiBs). This paper presents a novel approach that combines deep reinforcement learning (DRL) with an optimal experimental design (OED) framework to identify key electrochemical parameters of LiB cell models. The proposed method utilizes the twin delayed deep deterministic policy gradient (TD3) algorithm to optimize input excitation, thereby increasing the sensitivity of the system response to electrochemical parameters. The performance of this DRL-based approach is evaluated against a nonlinear model predictive control (NMPC) method and conventional tests. Results indicate that the DRL-based method provides superior information content, reflected in higher Fisher information (FI) values and lower parameter estimation errors compared to the NMPC design and conventional test practices. Additionally, the DRL approach offers a substantial reduction in experimental time and computational resources. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:32:38 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Ozkan",
"Mehmet Fatih",
""
],
[
"da Silva",
"Samuel Filgueira",
""
],
[
"Idrissi",
"Faissal El",
""
],
[
"Ramesh",
"Prashanth",
""
],
[
"Canova",
"Marcello",
""
]
] |
2506.19147 | James Hanson | James E. Hanson | Indiscernible extraction at small large cardinals from a higher-arity stability notion | 16 pages, 2 figures | null | null | null | math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a higher-arity stability notion defined in terms of $k$-splitting, a higher-arity generalization of splitting. We show that theories with bounded $k$-splitting have improved indiscernible extraction at $k$-ineffable cardinals, and we give a non-trivial example of a theory with bounded $k$-splitting but unbounded $(k-1)$-splitting for each odd $k > 1$. We also show that bounded $k$-splitting implies $\mathrm{NFOP}_k$, a higher arity stability notion introduced by Terry and Wolf. We then use our indiscernible extraction result together with a construction of Kaplan and Shelah to give a strong counterexample to the converse: an $\mathrm{NIP}$ theory with unbounded $k$-splitting for every $k$. Finally, as a thematically related but technically independent result, we show that treelessness implies $\mathrm{NFOP}_2$, sharpening a result of Kaplan, Ramsey, and Simon. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:33:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Hanson",
"James E.",
""
]
] |
2506.19148 | Linquan Ma | Raymond Heitmann and Linquan Ma | On complete integral closedness of the $p$-adic completion of absolute integral closure | 15 pages, comments welcome | null | null | null | math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fix a prime $p$ and let $(R,\mathfrak{m})$ be a Noetherian complete local domain of mixed characteristic $(0,p)$ with fraction field $K$. Let $R^+$ denote the absolute integral closure of $R$, which is the integral closure of $R$ in an algebraic closure $\overline{K}$ of $K$. The first author has shown that $\widehat{R^+}$, the $p$-adic completion of $R^+$, is an integral domain. In this paper, we prove that $\widehat{R^+}$ is completely integrally closed in $\widehat{R^+}\otimes_{R^+}\overline{K}$, but $\widehat{R^+}$ is not completely integrally closed in its own fraction field when $\dim(R)\geq 2$. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:37:46 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Heitmann",
"Raymond",
""
],
[
"Ma",
"Linquan",
""
]
] |
2506.19149 | Peter Borg | Karl Bartolo, Peter Borg, Dayle Scicluna | Solution to a problem on isolation of $3$-vertex paths | 12 pages | null | null | null | math.CO cs.DM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The $3$-path isolation number of a connected $n$-vertex graph $G$, denoted by $\iota(G,P_3)$, is the size of a smallest subset $D$ of the vertex set of $G$ such that the closed neighbourhood $N[D]$ of $D$ in $G$ intersects each $3$-vertex path of $G$, meaning that no two edges of $G-N[D]$ intersect. Zhang and Wu proved that $\iota(G,P_3) \leq 2n/7$ unless $G$ is a $3$-path or a $3$-cycle or a $6$-cycle. The bound is attained by infinitely many graphs having induced $6$-cycles. Huang, Zhang and Jin proved that if $G$ has no $6$-cycles, or $G$ has no induced $5$-cycles and no induced $6$-cycles, then $\iota(G, P_3) \leq n/4$ unless $G$ is a $3$-path or a $3$-cycle or a $7$-cycle or an $11$-cycle. They asked if the bound still holds asymptotically for connected graphs having no induced $6$-cycles. More precisely, taking $f(n)$ to be the maximum value of $\iota(G,P_3)$ over all connected $n$-vertex graphs $G$ having no induced $6$-cycles, their question is whether $\limsup_{n \to\infty}\frac{f(n)}{n} = \frac{1}{4}$. We verify this by proving that $f(n) = \left \lfloor (n+1)/4 \right \rfloor$. The proof hinges on further proving that if $G$ is such a graph and $\iota(G, P_3) = (n+1)/4$, then $\iota(G-v, P_3) < \iota(G, P_3)$ for each vertex $v$ of $G$. This new idea promises to be of further use. We also prove that if the maximum degree of such a graph $G$ is at least $5$, then $\iota(G,P_3) \leq n/4$. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:38:22 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Bartolo",
"Karl",
""
],
[
"Borg",
"Peter",
""
],
[
"Scicluna",
"Dayle",
""
]
] |
2506.19150 | Martin Mootz | Martin Mootz and Yong-Xin Yao | Efficient Berry Phase Calculation via Adaptive Variational Quantum Computing Approach | null | null | null | null | quant-ph cond-mat.str-el physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a quantum computing approach for efficiently calculating the Berry phase in topological Hamiltonians. Our method leverages cyclic adiabatic evolution of the Hamiltonian and employs adaptive variational quantum algorithms for state preparation and evolution, optimizing circuit efficiency while maintaining high accuracy. We benchmark our approach on dimerized Fermi-Hubbard chains with four sites, demonstrating precise Berry phase simulations in both noninteracting and interacting regimes. Our results show that circuit depths reach up to 106 layers for noninteracting systems and increase to 279 layers for interacting systems due to added complexity. Additionally, we demonstrate the robustness of our scheme across a wide range of parameters governing adiabatic evolution and variational algorithm. These findings highlight the potential of adaptive variational quantum algorithms for advancing quantum simulations of topological materials and computing geometric phases in strongly correlated systems. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:38:55 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Mootz",
"Martin",
""
],
[
"Yao",
"Yong-Xin",
""
]
] |
2506.19151 | Aaron Abrams | Aaron Abrams and Peter Johnson | Yet Another Species of Forbidden-distances Chromatic Number | This 2001 paper introduces a new type of chromatic number for point sets | Geombinatorics, vol. 10 no. 3 (2001), pp. 89-95 | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This 2001 paper introduces a new type of chromatic number for point sets. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:40:02 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Abrams",
"Aaron",
""
],
[
"Johnson",
"Peter",
""
]
] |
2506.19152 | Anne-Cathrine Dott | A.-C. Dott, J. Saur, S. Schlegel, D.F. Strobel, K. de Kleer, I. de Pater | Observed Latitudinal, Longitudinal and Temporal Variability of Io's Atmosphere Simulated by a Purely Sublimation Driven Atmosphere | 34 pages, 12 figures | null | null | null | astro-ph.EP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | How much of Io's SO$_2$ atmosphere is driven by volcanic outgasing or sublimation of SO$_2$ surface frost is a question with a considerable history. We develop a time dependent surface temperature model including thermal inertia and the exact celestial geometry to model the radiation driven global structure and temporal evolution of Io's atmosphere. We show that many observations can be explained by assuming a purely sublimation driven atmosphere. We find that a thermal diffusivity $\alpha=2.41\times10^{-7}$ m$^2$s$^{-1}$ yields an averaged atmospheric SO$_2$ column density decreasing by more than one order of magnitude from the equator to the poles in accordance with the observed spatial variations of Io's column densities. Our model produces a strong day-night-asymmetry with modeled column density variations of almost two orders of magnitude at the equator as well as a sub-anti-Jovian hemisphere asymmetry, with maximum dayside column densities of $3.7\times10^{16}$ cm$^{-2}$ for the sub-Jovian and $8.5\times10^{16}$ cm$^{-2}$ for the anti-Jovian hemisphere. Both are consistent with the observed temporal and large-scale longitudinal variation of Io's atmosphere. We find that the diurnal variations of the surface temperature affect the subsurface structure up to a depth of 0.6m. Furthermore, we quantify seasonal effects with Io having a northern summer close to perihelion and a northern winter close to aphelion. Finally, we found that at Io's anomalous warm polar regions a conductive heat flux of at least 1.2 Wm$^{-2}$ is necessary to reach surface temperatures consistent with observations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:42:24 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Dott",
"A. -C.",
""
],
[
"Saur",
"J.",
""
],
[
"Schlegel",
"S.",
""
],
[
"Strobel",
"D. F.",
""
],
[
"de Kleer",
"K.",
""
],
[
"de Pater",
"I.",
""
]
] |
2506.19153 | Krzysztof Fona{\l} | Krzysztof Fonal | Dataset of Yul Contracts to Support Solidity Compiler Research | 4 pages | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | The YulCode dataset presents a comprehensive collection of 348,840 Yul-based smart contract instances, comprising approximately 135,013 unique contracts. These contracts were generated through the compilation of Solidity source files that have been deployed on the Ethereum mainnet, making the dataset directly representative of real-world decentralized applications. YulCode provides a rich foundation for a variety of research and development tasks, including but not limited to machine learning applications, formal verification, optimization analysis, and software engineering tool evaluation in the context of low-level smart contract code. To the best of our knowledge at the time of writing, YulCode is the first and only publicly available dataset that focuses specifically on Yul, an intermediate language designed for the Ethereum Virtual Machine (EVM). As such, it fills a critical gap in the current ecosystem of smart contract datasets and opens new avenues for research and tooling aimed at low-level contract analysis and generation. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:43:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Fonal",
"Krzysztof",
""
]
] |
2506.19154 | Mahdi Falaki | Mahdi Falaki, Maria A. Amer | Lightweight RGB-T Tracking with Mobile Vision Transformers | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Single-modality object tracking (e.g., RGB-only) encounters difficulties in challenging imaging conditions, such as low illumination and adverse weather conditions. To solve this, multimodal tracking (e.g., RGB-T models) aims to leverage complementary data such as thermal infrared features. While recent Vision Transformer-based multimodal trackers achieve strong performance, they are often computationally expensive due to large model sizes. In this work, we propose a novel lightweight RGB-T tracking algorithm based on Mobile Vision Transformers (MobileViT). Our tracker introduces a progressive fusion framework that jointly learns intra-modal and inter-modal interactions between the template and search regions using separable attention. This design produces effective feature representations that support more accurate target localization while achieving a small model size and fast inference speed. Compared to state-of-the-art efficient multimodal trackers, our model achieves comparable accuracy while offering significantly lower parameter counts (less than 4 million) and the fastest GPU inference speed of 122 frames per second. This paper is the first to propose a tracker using Mobile Vision Transformers for RGB-T tracking and multimodal tracking at large. Tracker code and model weights will be made publicly available upon acceptance. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:46:22 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Falaki",
"Mahdi",
""
],
[
"Amer",
"Maria A.",
""
]
] |
2506.19155 | Jasone Ram\'irez-Ayerbe | Jasone Ram\'irez-Ayerbe and Emma Frejinger | Relative Explanations for Contextual Problems with Endogenous Uncertainty: An Application to Competitive Facility Location | null | null | null | null | math.OC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this paper, we consider contextual stochastic optimization problems subject to endogenous uncertainty, where the decisions affect the underlying distributions. To implement such decisions in practice, it is crucial to ensure that their outcomes are interpretable and trustworthy. To this end, we compute relative counterfactual explanations, providing practitioners with concrete changes in the contextual covariates required for a solution to satisfy specific constraints. Whereas relative explanations have been introduced in prior literature, to the best of our knowledge, this is the first work focused on problems with binary decision variables and subject to endogenous uncertainty. We propose a methodology that uses Wasserstein distance as regularization and to compute a lower bound. It leads to a drastic reduction in computation times, compared to the unregularized counterpart. We illustrate the method using a choice-based competitive facility location problem, and present numerical experiments that demonstrate its ability to efficiently compute sparse and interpretable explanations. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:46:41 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Ramírez-Ayerbe",
"Jasone",
""
],
[
"Frejinger",
"Emma",
""
]
] |
2506.19156 | Ath\'ena\"is Vaginay | Hans-J\"org Schurr and Ath\'ena\"is Vaginay | Looking for Signs: Reasoning About FOBNNs Using SAT | Author version, accepted at CMSB25 | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | First-Order Boolean Networks with Non-deterministic updates (FOBNN) compute a boolean transition graph representing the absence and presence of species over time. The utility of FOBNNs has been justified by their theoretical soundness with respect to the Euler simulation of the differential equations. However, we lack practical means to work with FOBNNs and an empirical evaluation of their properties. We present a sound and efficient reduction of the first-order FOBNN transition relation to a propositional logic formula. This makes it possible to use modern SAT solvers to reason on the full transition graph, even for large models. We use this encoding to assess the feasibility and efficiency of practical reasoning with FOBNNs. To do so, we focus on the computation of fixed points. We also compare the transition graphs obtained via FOBNNs to those computed by the classic boolean semantics of reaction networks. Overall, our encoding opens new directions for the analysis of FOBNNs and deepens the understanding of their relationship with reaction networks. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:46:44 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Schurr",
"Hans-Jörg",
""
],
[
"Vaginay",
"Athénaïs",
""
]
] |
2506.19157 | Stephen Jon Quiton | Stephen Jon Quiton, Juan D. F. Pottecher, Xin Xing, Martin Head-Gordon, Lin Lin | Optimized auxiliary functions for robust mitigation of finite-size errors in periodic hybrid density functional theory | 44 pages, 19 figures (including supporting information) | null | null | null | physics.comp-ph physics.chem-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When calculating properties of periodic systems at the thermodynamic limit (TDL), the dominant source of finite size error (FSE) arises from the long-range Coulomb interaction, and can manifest as a slowly converging quadrature error when approximating an integral in the reciprocal space by a finite sum. The singularity subtraction (SS) method offers a systematic approach for reducing this quadrature error and thus the FSE. In this work, we first investigate the performance of the SS method in the simplest setting, aiming at reducing the FSE in exact exchange calculations by subtracting the Coulomb contribution with a single, adjustable Gaussian auxiliary function. We demonstrate that a simple fitting method can robustly estimate the optimal Gaussian width and leads to rapid convergence toward the TDL. Furthermore, we suggest new forms of the auxiliary function, whose optimal parameters could also be determined through least-squares fitting. For a range of semiconductors and insulators, the proposed auxiliary functions achieve robust, millihartree-level accuracy in hybrid density functional theory calculations, including cases with sparse k-meshes and large basis sets. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:51:08 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Quiton",
"Stephen Jon",
""
],
[
"Pottecher",
"Juan D. F.",
""
],
[
"Xing",
"Xin",
""
],
[
"Head-Gordon",
"Martin",
""
],
[
"Lin",
"Lin",
""
]
] |
2506.19158 | John W. Moffat | John. W. Moffat, Ethan. J. Thompson | Embedding $\mathrm{SL}(2,\mathbb{C})/\mathbb{Z}_2$ in Complex Riemannian Geometry | 8 pages | null | null | null | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We present a unified framework demonstrating how the spinor complex Lorentz group SL(2,C)/Z\_2 is realized as a canonical subgroup within a four-dimensional complex Riemannian manifold. Building on the complex, holomorphic metric extension and contour-integration regularization of classical singularities, we show that promoting the metric to a complex-valued tensor on a complex 4-fold enlarges the frame bundle to SO(4,C). Its spin double cover factorizes as a product of two independent SL(2,C) factors modulo a shared Z\_2, and selecting one Weyl factor recovers the familiar SL(2,C)/Z\_2 spin cover of the Lorentz group. By explicitly extending metric components and connection forms into the complex domain, using contour deformations to avoid coordinate-singular loci, we exhibit how left- and right-handed Weyl spinors transform under separate SL(2,C) factors, and how modding out a common Z\_2 reproduces the standard spin-Lorentz structure. In particular, the same contour-integration techniques that yield singularity-free Schwarzschild and Kerr solutions via a holomorphic radial coordinate also furnish a holomorphic spin bundle in which chiral spin representations live without imposing exotic matter or modifying the Einstein equations. This embedding clarifies the geometric origin of chirality, enables holomorphic factorization of curvature and connection forms, and provides a foundation for constructing holomorphic field theories and complexified gravity backgrounds. Our results indicate that extending the Lorentz group into a complex Riemannian setting not only recovers SL(2,C)/Z\_2 in a canonical fashion, but also establishes a geometric arena for studying chiral fermions, Bogomol'nyi-Prasad-Sommerfield (BPS) instantons, and potential quantum-gravity corrections within a rigorously defined complex manifold. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:51:18 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Moffat",
"John. W.",
""
],
[
"Thompson",
"Ethan. J.",
""
]
] |
2506.19159 | Yun Tang | Yun Tang, Eesung Kim, Vijendra Raj Apsingekar | Enhanced Hybrid Transducer and Attention Encoder Decoder with Text Data | Accepted by Interspeech2025 | null | null | null | cs.CL cs.SD eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A joint speech and text optimization method is proposed for hybrid transducer and attention-based encoder decoder (TAED) modeling to leverage large amounts of text corpus and enhance ASR accuracy. The joint TAED (J-TAED) is trained with both speech and text input modalities together, while it only takes speech data as input during inference. The trained model can unify the internal representations from different modalities, and be further extended to text-based domain adaptation. It can effectively alleviate data scarcity for mismatch domain tasks since no speech data is required. Our experiments show J-TAED successfully integrates speech and linguistic information into one model, and reduce the WER by 5.8 ~12.8% on the Librispeech dataset. The model is also evaluated on two out-of-domain datasets: one is finance and another is named entity focused. The text-based domain adaptation brings 15.3% and 17.8% WER reduction on those two datasets respectively. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:51:39 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Tang",
"Yun",
""
],
[
"Kim",
"Eesung",
""
],
[
"Apsingekar",
"Vijendra Raj",
""
]
] |
2506.19160 | Seyyed Ali Emami | Mohammad Narimani, Seyyed Ali Emami | AgenticControl: An Automated Control Design Framework Using Large Language Models | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Traditional control system design, reliant on expert knowledge and precise models, struggles with complex, nonlinear, or uncertain dynamics. This paper introduces AgenticControl, a novel multi-agent framework that automates controller design using coordinated Large Language Model (LLM) agents. Through structured JSON communication, these agents handle tasks including controller selection, scenario design, parameter optimization, performance evaluation, and decision-making. Through an actor-critic optimization approach, the system iteratively improves performance while progressing through scenarios of increasing complexity to ensure robustness under nominal conditions, measurement noise, actuator disturbances, and parametric uncertainties. Key innovations include structured multi-agent collaboration, robust optimization mechanisms, and real-time adaptability via in-context learning. Validated across four diverse control systems, namely, DC Motor Position control, Ball and Beam, Inverted Pendulum, and Double Inverted Pendulum, the framework achieves competitive performance against classical methods. Its Full State Feedback solution closely matches Linear Quadratic Regulator (LQR) results, while the designed PID controller significantly outperforming MATLAB's PIDTuner, reducing PID tracking error by 55% through adaptive parameter exploration. A comparative study of five LLM models reveals distinct optimization profiles, with DeepSeek achieving the fastest convergence. This work demonstrates the potential of LLM-driven control design, paving the way for advanced techniques like model predictive control and reinforcement learning. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:53:05 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Narimani",
"Mohammad",
""
],
[
"Emami",
"Seyyed Ali",
""
]
] |
2506.19161 | John W. Moffat | John. W. Moffat, Ethan. J. Thompson | Holomorphic Unified Field Theory of Gravity and the Standard Model | 13 pages | null | null | null | physics.gen-ph gr-qc hep-ph hep-th | http://creativecommons.org/licenses/by/4.0/ | We present a holomorphic framework in which gravity, gauge interactions, and their couplings to charges and currents emerge from a single geometric action on a four-complex-dimensional manifold. The Hermitian metric yields on the real slice $y^\mu = 0$, a real symmetric metric $g_{(\mu \nu)}$ giving the vacuum Einstein equations, and an antisymmetric part $g_{[\mu \nu]}$ that reproduces Maxwell's equations with sources. A single holomorphic gauge connection for $G_{\text{GUT}}$, such as $SU(5)$ or $SO(10)$, encodes all gauge sectors; its Bianchi identities give homogeneous Yang--Mills equations, and variation imposes $\nabla_\mu F^{\mu \nu}_A = J^\nu_A$. Chiral fermions arise from a holomorphic Dirac Lagrangian and couple minimally to all gauge fields, reproducing the Standard Model spectrum. Anomaly cancellation follows from holomorphic gauge invariance. A holomorphic adjoint Higgs breaks $G_{\text{GUT}} \rightarrow SU(3) \times SU(2) \times U(1)$ with unified coupling, and a second Higgs breaks electroweak symmetry, generating $W^\pm$, $Z$, and fermion masses. Below the unification scale, couplings run by standard renormalization-group flow. This construction unifies Einstein gravity, Yang--Mills theory, electromagnetism, and chiral fermions into a single classical geometric framework, and admits quantization via a holomorphic path integral that reproduces standard Feynman rules. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 21:59:56 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Moffat",
"John. W.",
""
],
[
"Thompson",
"Ethan. J.",
""
]
] |
2506.19162 | Margaret Verrico | Margaret E. Verrico, K. Decker French, Katherine Suess, Tanay Agrawal, Lauranne Lanz, Yuanze Luo, Pallavi Patil, Kate Rowlands, Margaret Shepherd, and Maya Skarbinski | Modeling Star Formation Histories of Changing-Look AGN Host Galaxies with Prospector | Submitted to ApJ | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by-sa/4.0/ | Changing-look active galactic nuclei, or CL-AGN, are AGN which appear to transition between Seyfert Type 1 and 2 over periods of months to years. Several mechanisms to trigger these transitions have been proposed, but we have yet to conclusively determine their cause. Recent studies suggest CL-AGN are hosted primarily in galaxies which are shutting down star formation (Dodd et al. 2021; Liu et al. 2021, Wang et al. 2023), which may indicate a link between galaxy quenching and changing look events. We use Prospector stellar population synthesis software (Leja et al. 2017; Johnson & Leja 2017; Johnson et al. 2021) to model non-parametric star formation histories for 39 CL-AGN host galaxies. We find that $43^{+13}_{-12}\%$ of our gold sample CL-AGN at z < 0.15 are star forming, while $29^{+13}_{-10}\%$ fall in the Green Valley of the stellar mass-sSFR diagram. At z > 0.15, $57^{+13}_{-18}\%$ of CL-AGN in the gold sample are star-forming and $29^{+19}_{-14}\%$ are in the Green Valley. CL-AGN hosts have similar star formation properties to the host galaxies of Seyfert 1 and 2 AGN at z < 0.15 and to Seyfert 2 AGN at z > 0.15. We find no statistically significant differences in the star formation properties of turn-on and turn-off CL-AGN. We also find no evidence for rapid quenching in the Green Valley CL-AGN. We conclude that CL-AGN state transitions are not associated with the formation history of CL-AGN host galaxies on large spatial scales, implying CL-AGN state transitions may instead result from nuclear-scale or accretion disk effects. | [
{
"version": "v1",
"created": "Mon, 23 Jun 2025 22:02:04 GMT"
}
] | 2025-06-25T00:00:00 | [
[
"Verrico",
"Margaret E.",
""
],
[
"French",
"K. Decker",
""
],
[
"Suess",
"Katherine",
""
],
[
"Agrawal",
"Tanay",
""
],
[
"Lanz",
"Lauranne",
""
],
[
"Luo",
"Yuanze",
""
],
[
"Patil",
"Pallavi",
""
],
[
"Rowlands",
"Kate",
""
],
[
"Shepherd",
"Margaret",
""
],
[
"Skarbinski",
"Maya",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.