NexaQST: Quantum State Tomography with Physics-Informed Neural Networks

NexaQST is a Quantum State Tomography model built using a Physics-Informed Neural Network (PINN) trained on synthetic 2-qubit experiments. This model leverages quantum mechanical priors such as positivity, Hermiticity, and trace constraints to ensure physically plausible reconstructions of quantum states.

Model Overview

Task: Reconstruct quantum states (density matrices) from tomographic measurement traces
System: Simulated 2-qubit experiments
Architecture: Physics-Informed Neural Network (PINN)
Constraints Embedded:
- Positivity
- Hermiticity
- Trace normalization (Tr(ρ) = 1)

Dataset Generation

Simulation Toolkit: QuTiP
Process:
1. Generated full 2-qubit tomography experiments via simulation
2. Extracted and structured measurement traces
3. Derived backward synthetic traces from known density matrices
4. Created supervised training pairs: (trace → density matrix)
5. Enforced physical constraints directly inside the model during training

Model Input/Output

Input: Vector of tomographic measurement traces
Output: Reconstructed density matrix (complex-valued 4x4 for 2 qubits)

Example Usage

# Pseudocode for usage

from model import NexaQSTModel
import torch

model = NexaQSTModel()
model.load_state_dict(torch.load("nexaqst_model.pt"))
model.eval()

traces = load_measurement_vector("qst_trace.npy")  # shape: (N,)
rho_pred = model.predict_density_matrix(traces)

# rho_pred is a 4x4 complex-valued matrix satisfying physical constraints

Applications

Quantum error correction diagnostics
Quantum system identification
Educational simulation of QST techniques
Physically consistent state estimation for quantum simulations

Licensing & Citation

If used in publications or products, please credit the Nexa Scientific Suite.

🌐 Related Tools

Created by Allan, independent quantum systems architect and ML researcher. Part of the Nexa scientific computing ecosystem.

Allanatrix
/

NexaQST