quasar

Code for "Quantum simulation of mutation-selection dynamics reveals a structural obstruction to quantum advantage" (submitted).

The Crow-Kimura and Eigen mutation-selection models map onto transverse-field Ising chains evolving in imaginary time. Mutation rate is the transverse field, per-locus fitness the longitudinal field, epistasis the ZZ coupling, and the quasispecies is the Perron eigenvector of the generator.

This repository implements that mapping as a quantum circuit and compares two quantum routes, variational imaginary-time evolution and QSVT eigenstate filtering, with three classical baselines: Wright-Fisher sampling, an exact solver for the landscape classes that admit one, and DMRG on matrix-product states.

Install

Everything that writes a record runs inside the Docker image:

docker build -t quasar:v1 .

For development and the fast tests:

make setup

Usage

make test      # unit and regression tests
make gates     # every check, inside the image, about nine hours

Single work packages:

python scripts/run_all_gates.py --list
python scripts/run_all_gates.py --wp wp_r

The comparison sweep and its scoring:

python scripts/sweep_runner.py --wp 7 --grid full --methods classical --workers 8
python scripts/sweep_runner.py --wp 7 --grid full --methods quantum
python scripts/score_g7.py

scripts/rescore_hardware.py reproduces the hardware result from the raw counts. scripts/make_manifest.py writes and verifies a SHA-256 manifest of a data deposit.

Layout

quasarstack/analytic     closed-form solutions, exact diagonalisation
quasarstack/hamiltonian  generator to Pauli operators
quasarstack/circuit      Trotterised propagator
quasarstack/ite          variational and Motta imaginary-time evolution
quasarstack/qsvt         block encoding, phase factors, eigenstate filter
quasarstack/spectral     gap, conditioning, order parameter
quasarstack/classical    landscapes, Wright-Fisher, exact solver, DMRG
quasarstack/scoring      cosine, total variation, bootstrap
quasarstack/backends     noise models, IBM Quantum submission
quasarstack/io           record schema, conventions, storage

experiments/             one script per check, each writing a JSON record
scripts/                 sweep runner, scoring, hardware rescoring, manifests
results/                 records
docs/                    settings and derivations

Results

Records are in results/, one JSON file per check, and docs/protocol.md lists the settings. No quantum advantage appears at the sizes compared, up to 12 loci: the matrix-product baseline matches exact diagonalisation on every instance, so no group meets the advantage criterion. The hardware run reaches 4 loci on an IBM Heron r2 device.

Requirements

Python 3.12, Qiskit 2.5.1 and quimb 1.14.0, with the pinned set in environment.lock.txt. IBM Quantum credentials are needed only for the hardware run.

Development note

This codebase was written with AI-assisted tooling. All results come from the pipeline in this repository and reproduce from a clean checkout; the authors are responsible for correctness.

Citation

Code and data are archived at Zenodo: https://doi.org/10.5281/zenodo.22828425

@software{quasar,
  author  = {Mahaboob Ali, Anees Ahmed and Nelson, Everette Jacob Remington and Delhibabu, Radhakrishnan},
  title   = {quasar: quantum simulation of mutation-selection dynamics},
  year    = {2026},
  doi     = {10.5281/zenodo.22828425},
  url     = {https://github.com/ahmedanees-m/quasar}
}

Licence

Apache-2.0 for code, CC-BY-4.0 for data and records. See LICENSE.