qsp-proc 0.1.0

Add your description here

qsp-proc

A QSP phase-finding package implementing the preprocessing pipeline from Skelton 2025.

Features

• Completion methods (4): Wilson/Fejér, BerSün (FFT-based), optimization-based completion, and root-finding/Prony-style completion.
• QSP conventions (4): base conventions, Angle-QSP (Lemma 1), Laurent-QSP (Lemma 3), and G-QSP (Lemma 4).
• Metasolver pipeline: compose approximation, completion, decomposition, and phase extraction into a single workflow.
• Recursive decomposition: carve (F(z)) into ({P_j}) and (E_0) for downstream synthesis.
• Circuit generation: build Qiskit circuits for basic QSP, controlled variants, and LCU constructions.

Installation

Install in editable mode with uv:

uv pip install -e .

Quick Start

Use the preprocessing pipeline with a Jacobi-Anger Laurent approximation:

from qsp_proc import qsp_preprocess
from qsp_proc.polynomials.approximations import jacobi_anger_laurent

target = jacobi_anger_laurent(tau=50, epsilon=1e-10)
result = qsp_preprocess(target=target)
print(result.summary())

CLI Usage

Run the CLI entry script with typical parameters:

uv run python main.py --tau 50 --epsilon 1e-10

Project Structure

src/qsp_proc/
├── polynomials/      # Chebyshev/Laurent polynomial tools and approximations (Jacobi-Anger, Remez).
├── conventions/      # Angle, Laurent, and G-QSP convention mappings and shared interfaces.
├── completion/       # Complementary polynomial solvers (Wilson, BerSün, optimization, rootfinding, Prony).
├── decomposition/    # Recursive carving routines for Laurent matrix polynomial factorization.
├── circuits/         # Qiskit circuit constructors for basic, controlled, and LCU QSP forms.
└── pipeline.py       # End-to-end preprocessing pipeline from target function to phases/circuits.

Documentation

Full docs available at https://bawa27.github.io/qsp-proc/.

References

• Skelton, 2025. Hitchhiker's Guide to QSP Pre-processing. arXiv:2501.05977.
• Berntson and Sunderhauf, 2025. Complementary polynomials in quantum signal processing. arXiv:2406.04246.

License

MIT.