AshiteshSingh

AshiteshSingh / Tpu-Accelerated-Quantum-JAX

Public

High-performance, differentiable quantum state-vector & tensor network simulator in 100% pure JAX (no classical framework overhead). Accelerated on NVIDIA GPUs and Google Cloud TPU v6e-64/v5e VM clusters up to 40 qubits! Supported by Google's TPU Research Cloud (TRC) program.

github.comAshiteshSinghTpu-Accelerated-Quantum-JAX
11
2
85% credibility
Found May 27, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A high-performance quantum state-vector simulator built purely in JAX that runs on NVIDIA GPUs and Google Cloud TPU clusters. It implements variational quantum algorithms (VQE, VQC, QAOA), quantum noise simulations, Shor's algorithm, Grover's search, and large-scale qubit scaling benchmarks. The project targets researchers studying quantum machine learning, molecular simulation, and quantum algorithm performance on classical hardware.

How It Works

1
🔬 Hear about a powerful quantum simulator

You discover a project that lets you run quantum physics experiments on your home computer's graphics card or Google's super-fast AI chips.

2
⚙️ Set up the quantum research tools

You install the software on your computer and it automatically connects to whichever processing hardware you have available—your GPU or Google's TPU cluster.

3
🎓 Learn quantum physics through experiments

You run your first experiment: teaching a quantum system to prepare a special 'GHZ' entangled state, watching as the computer learns to create quantum weirdness step by step.

4
🤖 Train a quantum classifier to solve puzzles

You train a quantum circuit to solve the XOR problem—a classic puzzle that regular computers struggle with—using a technique called variational quantum classification.

5
⚛️ Find the energy of a hydrogen molecule

You run a calculation that finds the exact ground-state energy of a hydrogen molecule, achieving 'chemical accuracy'—the gold standard that chemists use for molecular simulations.

6
📈 Solve graph optimization problems with quantum circuits

You use the QAOA algorithm to find the best way to cut a weighted graph into two parts, exploring different circuit depths to see how close you can get to the perfect solution.

7
Choose your research path
🌊
Study quantum noise effects

You simulate how real-world imperfections decay quantum states over time using Monte Carlo trajectory methods.

📊
Scale up to 33+ qubits

You push the simulation to 33 qubits distributed across a 16-chip TPU cluster, handling massive state vectors that would never fit in regular memory.

✅ Get your research results and visualizations

You receive beautiful plots, CSV data, and JSON summaries of all your quantum experiments, ready to share with colleagues or use in publications.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Tpu-Accelerated-Quantum-JAX?

This is a research-grade quantum state-vector simulator built entirely in Python using JAX. Instead of wrapping existing quantum frameworks, the author wrote everything from scratch to squeeze out maximum hardware performance. The simulator runs circuits on NVIDIA GPUs via CUDA and scales across Google Cloud TPU clusters with up to 64 chips. You get full state-vector simulation for circuits with 40 qubits, automatic differentiation for gradient-based training, and distributed memory sharding that partitions massive state vectors across physical devices. It includes implementations of VQE, QAOA, Grover's search, Shor's algorithm, GHZ state preparation, variational classifiers, and noise simulation.

Why is it gaining traction?

The hook is raw performance through pure JAX compilation. By avoiding heavyweight frameworks like Qiskit or PennyLane, every quantum operation compiles into a single XLA kernel that runs directly in GPU or TPU memory without CPU overhead. Reverse-mode autodifferentiation comes native in JAX, so computing gradients for variational circuits is a single function call rather than the parameter-shift approximations other frameworks require. The TPU Research Cloud backing gives legitimate access to hardware that would otherwise cost thousands per hour.

Who should use this?

Quantum computing researchers running variational algorithms who need differentiability and scale beyond what consumer GPUs handle. Machine learning researchers exploring quantum-classical hybrid models. Anyone with TPU access who wants to simulate circuits beyond 30 qubits without building the infrastructure themselves. Not suitable for beginners or production quantum hardware programming.

Verdict

A technically impressive project with a credibility score of 0.85%, but the 11 stars and sparse documentation suggest early-stage research code. The performance claims are compelling, but you should verify benchmarks on your own hardware before committing. Worth exploring if you have TPU access and need large-scale differentiable quantum simulation.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.