PyTran
A hybrid Python–Fortran language runtime for AI/ML, HPC, and quantum–classical computation.
PyTran augments Python with deterministic, Fortran-class numerical execution, optional GPU acceleration, and first-class support for hybrid quantum workflows — without forcing researchers to abandon Python.
Why PyTran Exists
Modern AI, scientific computing, and quantum research live in Python — but performance still lives elsewhere.
PyTran closes this gap without forcing teams to rewrite entire systems in C++, CUDA, or bespoke toolchains.
- Prototype algorithms in Python
- Annotate performance-critical regions
- Compile those regions into optimized Fortran or GPU kernels
- Orchestrate classical and quantum execution in one workflow
Python, Not Replaced
PyTran preserves Python’s readability and ecosystem while removing the performance ceiling that limits large-scale simulations and training loops.
Fortran Where It Matters
Numerically intensive paths are lowered into Fortran-class execution, enabling predictable performance and cache-efficient computation.
Explicit GPU Acceleration
GPU usage is opt-in and explicit. PyTran avoids hidden magic and gives developers control over parallel execution.
Quantum-Classical Native
Quantum circuits are treated as first-class constructs alongside classical kernels — enabling hybrid algorithms such as VQE and QAOA.
Install (Linux)
PyTran ships as a single release tarball with a works-on-first-run bootstrap installer. No system Python is modified.
curl -L -o pytran.tar.gz "https://github.com/amitb-quantum/pytran-platform/releases/latest/download/pytran-0.1.0.tar.gz"
tar -xzf pytran.tar.gz
cd pytran-0.1.0
./bootstrap.sh
Optional GPU acceleration:
./bootstrap.sh --extras gpuAmit Brahmbhatt
Founder & Lead Architect, Quantum-Clarity LLC