Qupertino / QuantumStudio

Apple Silicon quantum benchmarking stack with a local simulator, reproducible benchmark pipelines, and a desktop UI.

Qupertino is a three-part project: a self-published technical report, the Qupertino simulator stack, and the dedicated QuantumStudio desktop UI studio. There is no native MLX quantum simulator available today, so Qupertino provides a local simulator layer for QFT, QAOA, VQE, Hamiltonian workflows, and OpenQASM runs.

The performance story is two tiers on the same hardware. The pure-MLX tier expresses every structured gate as MLX array operations; on top of it, an opt-in tier of hand-written Metal shaders (src/mlxq/shaders/, one flag: MLXQ_METAL_KERNELS=1) covers every structured layer family through semantics-preserving fusion detectors. Measured on an M1 Max against Qiskit Aer CPU and PennyLane lightning.qubit, the Metal tier is fastest in all 18 comparison cells — 25-qubit QFT in 59 ms (47× Aer, 95× PennyLane, and faster than MLX's own mx.fft), TFIM Trotter in 0.5 s — and it accelerates 26 of 29 benchmark workloads up to 25× over the pure tier. See the charts and tables below.

Website: https://boltzmannentropy.github.io/QupertinoWEB/
Repository: https://github.com/BoltzmannEntropy/Qupertino
Author: Shlomo Kashani
Technical report: unpublished; PDF and LaTeX source are kept locally and are not distributed in this repository.

How to Cite

Repository (Git / Software Citation)

@software{kashani_qupertino_2026,
  author       = {Shlomo Kashani},
  title        = {Qupertino / QuantumStudio: Apple Silicon Quantum Benchmarking Stack},
  year         = {2026},
  url          = {https://github.com/BoltzmannEntropy/Qupertino},
  note         = {GitHub repository}
}

Technical Report Citation

@techreport{kashani_qupertino_report_2026,
  author       = {Shlomo Kashani},
  title        = {Qupertino: Pure MLX Array Kernels versus Hand-Tuned Metal Shaders for Quantum Circuit Simulation on Apple Silicon},
  year         = {2026},
  institution  = {Self-published technical report (unpublished)},
  url          = {https://github.com/BoltzmannEntropy/Qupertino}
}

Extended Introduction

Qupertino exists to make Apple Silicon quantum benchmarking practical, local-first, and reproducible. The project packages three layers into one workflow:

Research layer: benchmark methodology and framing (see the included technical report).
Simulator layer (mlxq): state-vector and MPS backends on MLX for Apple Silicon, in two performance tiers — pure-MLX structured dispatch and an opt-in hand-written Metal shader tier (src/mlxq/shaders/).
Product layer (QuantumStudio): desktop run orchestration, monitoring, and export.

The core motivation is straightforward: Apple Silicon uses a unified memory model, but there is no default native MLX quantum runtime shipped as a platform quantum simulator. Qupertino fills that gap with a local simulator stack and benchmark harnesses for QFT, QAOA, VQE, QCBM, Grover, Hamiltonian/time-evolution workflows, and OpenQASM runs.

This repository is designed for publication-grade reproducibility:

deterministic CLI run paths
structured CSV/JSON output, raw timing distributions, and run manifests
frozen artifact promotion under assets/benchmarks-frozen/
UI+CLI parity for auditability

Project Structure

src/ Python simulator + benchmark code (mlxq)
bench.sh single/full benchmark launcher
bench_with_logging.sh orchestrated benchmark runs with logging/promotions
bench/runs/ per-run outputs (run_YYYYmmdd_HHMMSS)
quantumstudio/ desktop UI + backend + control scripts
datasets/qasm/local/ OpenQASM local corpus
assets/benchmarks-frozen/ frozen benchmark artifacts and sample bundles

Benchmark Context And Representative Results

Qupertino ships two measured performance tiers. The pure-MLX tier dispatches structured gate classes to specialized MLX kernels (diagonal gates as broadcast phase multiplies, controlled gates as masked half-state updates, SWAP as an axis permutation, runtime fusion of equal-angle ZZ Trotter layers). The Metal shader tier (MLXQ_METAL_KERNELS=1) adds hand-written kernels in src/mlxq/shaders/ for every structured layer family, reached through semantics-preserving fusion detectors and parity-tested against the pure path (251 tests).

Wall time at 25 qubits across four backends

25-qubit wall time, same machine, gate-identical circuits (log scale, lower is better). The Metal shader tier — blue — is fastest in every cell.

Four-way interleaved campaign on M1 Max (two warmups, ten measured repeats per cell, all four backends round-robin inside every repeat, gate-identical circuits, mean seconds at 25q):

Workload @ 25q	Qupertino Metal	Qupertino MLX	Aer CPU	PennyLane lightning
QFT	0.059	0.72	2.80	5.61
Ring-QAOA (6 layers)	0.150	2.07	5.35	6.84
TFIM Trotter (20 steps)	0.495	5.82	17.79	32.95
Phase estimation	0.105	0.91	4.05	6.06
Grover proxy	0.052	1.11	1.21	2.73
GHZ	0.022	0.27	0.69	0.42

The Metal tier is fastest in all 18 comparison cells (15/20/25 qubits), with 25-qubit paired per-repeat ratios of 23–47× over Aer and 19–95× over PennyLane — and its gate-stream QFT (59 ms) beats MLX's own mx.fft primitive (77 ms). The pure-MLX tier is fastest among the CPU-comparable trio on 4 of 6 workloads at 25q (Grover is a statistical tie with Aer; small gate-sparse circuits at 15q favor the CPU baselines). A paired ablation with dispatch disabled (MLXQ_DENSE_ONLY=1) attributes 25–33× to kernel specialization itself. A separate campaign against PennyLane's OpenMP-parallel lightning.kokkos reached the same verdict for the pure tier (fastest in 16/18 cells). Full protocol, t-based CIs, raw artifacts: paper/tqc-acm-2026/evidence_artifacts/.

The Metal shader tier

Seven kernel families in one folder (src/mlxq/shaders/, design + measurements + codex-review record in its README): phase-LUT diagonal layers (ZZ, CZ/CPHASE, grouped weighted couplings), GF(2) affine permutation gathers (any CNOT/X/SWAP run = one pass), fused tensor-product single-qubit layers (uniform and per-qubit-varying, with algebraic window collapse — Grover's H·H·X prologue becomes a single bit-flip gather), radix-4 QFT and Walsh–Hadamard butterflies, basis-conjugated XX/YY Trotter layers, and gate-stream QFT/iQFT stage detectors. Kernel-level at 25q:

Layer @ 25q	Per-gate	MLX fused	Metal kernel
ZZ Trotter layer (24 bonds)	35.9 ms	2.5 ms	1.7 ms
Full QFT (radix-4, 13 launches)	748 ms	255 ms	21.1 ms
RX layer (all 25 qubits)	194 ms	—	20.9 ms
H layer (Walsh radix-4, 7 launches)	698 ms	—	12.7 ms

The shader tier also engages on the OpenQASM import path: the full local QASM corpus (40 executed circuits, QASMBench-derived) matches the pure path to ≤5e-7 everywhere, with structured circuits accelerating (ghz_state_n23 9.1×, cat_state_n22 8.0×, ising_n26 2.2×) and pre-decomposed cx+rz circuits (e.g. QASMBench qft_n18) correctly falling through to the pure path at parity — recognizing decomposed 2-qubit blocks is noted future work. Artifact: paper/tqc-acm-2026/evidence_artifacts/qasm_shader_sweep_20260704/.

Full-suite paired sweep (29 workloads at 25q, pure vs Metal alternating within every repeat): 26 of 29 workloads accelerate, 4.7–25× for those with fusable layer structure (TFIM 2nd order 25×, long-range Ising 24×, Heisenberg 13.8×, variational 14.3×, QCBM 12×, Grover 11.5×). The three 1.0× rows are structural and documented (cross-family Trotter interleave, isolated CZs, strict RY/CNOT alternation).

Metal shader speedup over pure MLX across 29 workloads

Every gate-based workload, Metal shaders vs pure MLX (paired, same session). Blue ≥ 4×, green = VQE (residual is energy evaluation), grey = structurally unfusable and correctly left at parity.

Four Codex CLI review rounds shaped and audited the kernels (archived under paper/tqc-acm-2026/reviews_shaders_v2/), catching real bugs: a wrong radix-4 derivation (retracted by the reviewer itself), a float32/rtol hole that silently dropped small-angle gates, and 1.25e-5 phase drift in 300-term products (fixed with grouped double-precision LUTs).

Full-suite refresh (all 21 benchmark families, `bench.sh --repro`)

The complete batch (one warmup, five measured repeats per qubit point, run bench/runs/run_20260702_131646, summaries frozen under paper/quantics-lncs-2026/evidence_artifacts/full_suite_20260702/) confirms the kernel rewrite across every family. Gains vs the archived pre-rewrite snapshot at 25 qubits:

Family	Archived	Current	Gain
Grover	113.3 s	1.74 s	65×
Random circuit	22.4 s	1.80 s	12×
QCBM (9 layers)	26.3 s	2.35 s	11×
QFT	7.03 s	0.82 s	8.5×
Time evolution	30.1 s	3.93 s	7.7×
Variational circuit	19.2 s	3.29 s	5.8×
GHZ	0.89 s	0.16 s	5.7×
QAOA (ring)	11.1 s	2.70 s	4.1×

New coverage: the Heisenberg/XXZ/random-field/long-range/ladder evolution variants run in 18–33 s at 25q (three Pauli-pair sweeps per Trotter step). Known remaining slow path: the density-matrix steady_state diagnostic (~292 s at its 12-qubit cap) still uses the generic dense path.

Current benchmark runners prefer the project-local .runtime-venv/bin/python when it exists, then .venv/bin/python, then PATH python3. Set PYTHON_BIN=/path/to/python only when you intentionally want to override that default.

Installation (Full)

1) System prerequisites

macOS 13.3+ recommended
Apple Silicon (M1/M2/M3/M4)
Python 3.10+ (3.11 tested heavily)
Optional: Flutter SDK (for UI development builds)

2) Core Python environment

cd /Volumes/SSD4tb/Dropbox/DSS/artifacts/code/QuantumStudioPRJ/QuantumStudioCODE
python3 -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install -e .

pyproject.toml includes core dependencies (mlx, rich, numpy) and optional groups.

3) Backend dependencies (QuantumStudio API)

pip install -r quantumstudio/backend/requirements.txt

4) Runtime path for local commands

export PYTHONPATH=src

5) Flutter UI dependencies (optional, for UI dev)

cd quantumstudio/flutter_app
flutter pub get

Running Benchmarks

Standard full-orchestrated run

cd /Volumes/SSD4tb/Dropbox/DSS/artifacts/code/QuantumStudioPRJ/QuantumStudioCODE
./bench_with_logging.sh

12-qubit smoke test (recommended health check)

cd /Volumes/SSD4tb/Dropbox/DSS/artifacts/code/QuantumStudioPRJ/QuantumStudioCODE
PYTHON_BIN=/Users/sol/.pyenv/shims/python3 ./bench_with_logging.sh --frozen-parity-12

Single-circuit run example

PYTHON_BIN=/Users/sol/.pyenv/shims/python3 ./bench.sh --circuit variational_circuit --simulate-limit 12 --qubits 1,2,5,7,10,11,12

Reproducible timing run

./bench.sh --circuit ghz --simulate-limit 2 --qubits 2 --no-save-plots --warmups 1 --repeats 2

Each benchmark run writes the legacy plotting files plus *_raw_runs.csv, *_raw_runs.json, *_timing_summary.csv, and run_manifest.json. QuantumStudio exposes the same warmup and repeat controls under Advanced Options.

Benchmark Catalog (Detailed)

The benchmark engine accepts the following circuit keys. For each benchmark below:

Use the single-circuit form:
- ./bench.sh --circuit <key> --simulate-limit <N> --qubits 1,2,5,7,10,11,12
Or run coordinated suites with:
- ./bench_with_logging.sh --frozen-parity-12

1) `hamiltonian_simulation`

Purpose: product-formula simulation of spin Hamiltonians.
Typical command:

./bench.sh --circuit hamiltonian_simulation --simulate-limit 12 --qubits 1,2,5,7,10,11,12

2) `time_evolution`

Purpose: time-evolution scaling over qubit count.
Typical command:

./bench.sh --circuit time_evolution --simulate-limit 12 --qubits 1,2,5,7,10,11,12

3) `trotter`

Purpose: Trotterized evolution workload for depth/runtime growth.
Typical command:

./bench.sh --circuit trotter --simulate-limit 12 --qubits 1,2,5,7,10,11,12

4) `steady_state`

Purpose: heavier iterative dynamics workload; often the slowest leg.
Typical command:

./bench.sh --circuit steady_state --simulate-limit 12 --qubits 1,2,5,7,10,11,12

5) `heisenberg`

Purpose: baseline Heisenberg model scaling.
Typical command:

./bench.sh --circuit heisenberg --simulate-limit 12 --qubits 1,2,5,7,10,11,12

6) `heisenberg_xxz`

Purpose: anisotropic XXZ Heisenberg variant.
Typical command:

./bench.sh --circuit heisenberg_xxz --simulate-limit 12 --qubits 1,2,5,7,10,11,12

7) `heisenberg_random_field`

Purpose: disordered Heisenberg workload with random field terms.
Typical command:

./bench.sh --circuit heisenberg_random_field --simulate-limit 12 --qubits 1,2,5,7,10,11,12

8) `tfim`

Purpose: transverse-field Ising model baseline.
Typical command:

./bench.sh --circuit tfim --simulate-limit 12 --qubits 1,2,5,7,10,11,12

9) `tfim_trotter2`

Purpose: second-order Trotter TFIM variant.
Typical command:

./bench.sh --circuit tfim_trotter2 --simulate-limit 12 --qubits 1,2,5,7,10,11,12

10) `tfim_random_field`

Purpose: TFIM with random field perturbations.
Typical command:

./bench.sh --circuit tfim_random_field --simulate-limit 12 --qubits 1,2,5,7,10,11,12

11) `long_range_ising`

Purpose: long-range interaction Ising scaling.
Typical command:

./bench.sh --circuit long_range_ising --simulate-limit 12 --qubits 1,2,5,7,10,11,12

12) `ladder_heisenberg`

Purpose: ladder-geometry Heisenberg workload.
Typical command:

./bench.sh --circuit ladder_heisenberg --simulate-limit 12 --qubits 1,2,5,7,10,11,12

13) `random_circuit`

Purpose: generic random gate-circuit scaling.
Typical command:

./bench.sh --circuit random_circuit --simulate-limit 12 --qubits 1,2,5,7,10,11,12

14) `qcbm`

Purpose: Quantum Circuit Born Machine benchmark family.
Typical command:

./bench.sh --circuit qcbm --simulate-limit 12 --qubits 1,2,5,7,10,11,12

15) `phase_estimation`

Purpose: phase estimation runtime growth.
Typical command:

./bench.sh --circuit phase_estimation --simulate-limit 12 --qubits 1,2,5,7,10,11,12

16) `qft`

Purpose: Quantum Fourier Transform scaling.
Typical command:

./bench.sh --circuit qft --simulate-limit 12 --qubits 1,2,5,7,10,11,12

17) `qaoa`

Purpose: QAOA-style variational optimization workload.
Typical command:

./bench.sh --circuit qaoa --simulate-limit 12 --qubits 1,2,5,7,10,11,12

18) `vqe`

Purpose: VQE-style variational eigensolver workload.
Typical command:

./bench.sh --circuit vqe --simulate-limit 12 --qubits 1,2,5,7,10,11,12

19) `variational_circuit`

Purpose: generic parameterized variational circuit benchmark.
Typical command:

./bench.sh --circuit variational_circuit --simulate-limit 12 --qubits 1,2,5,7,10,11,12

20) `grover`

Purpose: Grover-style amplitude amplification benchmark.
Typical command:

./bench.sh --circuit grover --simulate-limit 12 --qubits 1,2,5,7,10,11,12

21) `ghz`

Purpose: GHZ state generation and scaling.
Typical command:

./bench.sh --circuit ghz --simulate-limit 12 --qubits 1,2,5,7,10,11,12

QASM Suite

Purpose: OpenQASM corpus execution from datasets/qasm/local/.
Typical command:

./bench.sh --qasm-suite --qasm-max-qubits 18 --qasm-timeout-ms 30000

Example & Circuit Gallery

Qupertino ships a broad gallery of runnable examples: 21 parameterized benchmark families (any qubit count) plus a 42-circuit OpenQASM corpus under datasets/qasm/local/. Run any benchmark family with ./bench.sh --circuit <key>; run any QASM circuit through the QASM suite. The full gallery is also published on the website.

A. Parameterized benchmark families (`--circuit <key>`)

Category	Circuit keys
Spin & Hamiltonian dynamics	`hamiltonian_simulation`, `time_evolution`, `trotter`, `steady_state`, `heisenberg`, `heisenberg_xxz`, `heisenberg_random_field`, `tfim`, `tfim_trotter2`, `tfim_random_field`, `long_range_ising`, `ladder_heisenberg`
Variational & QML	`qcbm`, `qaoa`, `vqe`, `variational_circuit`
Core algorithms	`qft`, `phase_estimation`, `grover`, `ghz`, `random_circuit`

B. OpenQASM corpus (`datasets/qasm/local/`, 42 circuits)

Category	Circuit (qubits)
States & entanglement	`bell` (2), `bell_state` (2), `bell_n4` (4), `ghz_state_n23` (23), `cat_state_n22` (22), `wstate_n3` (3), `wstate_n27` (27), `coin_flip` (1), `qrng_n4` (4)
Textbook algorithms	`deutsch_n2` (2), `grover_n2` (2), `grover_2qubit` (2), `qft_n4` (4), `qft_n18` (18), `inverseqft_n4` (4), `qpe_n9` (9), `ipea_n2` (2), `swap_test_n25` (25), `toffoli_n3` (3), `teleport_minimal` (3), `teleportation_n3` (3), `bv_n14` — Bernstein–Vazirani (14), `hs4_n4` — hidden shift (4), `bwt_n21` — binary welded tree (21)
Arithmetic & factoring	`bigadder_n18` (18), `multiplier_n15` (15), `multiply_n13` (13), `qf21_n15` — factor 21 (15), `factor247_n15` — factor 247 (15)
Variational & physics	`vqe_n4` (4), `vqe_n24` (24), `vqe_ising`, `variational_n4` (4), `ising_n10` (10), `ising_n26` (26)
Applied, ML & QEC	`dnn_n16` — quantum DNN (16), `knn_n25` — quantum kNN (25), `sat_n11` — SAT (11), `seca_n11` (11), `qram_n20` — QRAM (20), `qec9xz_n17` — Shor [[9,1,3]] code (17), `advanced_circuit`

Run a QASM circuit:

./bench.sh --qasm-suite --qasm-max-qubits 27 --qasm-timeout-ms 30000

Full Example Catalog (250+)

Beyond the circuit gallery above, Qupertino ships 250 runnable example functions — gate-algebra identities, state preparation, algorithm demonstrations (Bell, GHZ, QFT, Grover, Toffoli, teleportation, QPE), MPS tensor-network parity, OpenQASM execution, and the QuantumStudio backend/MCP API. The complete itemized list is published on the Examples & Catalog page.

Category	Count	Source
Core simulator & gate algebra	149	`src/tests/mlxQCoreTest.py`
Quantum-computing examples (identities, algorithms)	41	`src/tests/mlxQQCExamplesTest.py`
Internal consistency & measurement parity	21	`src/tests/mlxQInternalConsistencyTest.py`, `mlxQMeasurementParityTest.py`
MPS tensor-network backend	12	`src/tests/mlxQMpsBackendTest.py`, `mlxQMpsParamSuiteTest.py`
QML wrapper	5	`src/tests/mlxQQmlWrapperTest.py`
QPE energy estimation	2	`src/tests/mlxQQpeEnergyEstimationTest.py`
Benchmark catalog, protocol & visualization	4	`src/tests/mlxQBenchCatalogTest.py`, `mlxQBenchmarkProtocolTest.py`, `mlxQVisualizationPlotsTest.py`
QuantumStudio backend API & MCP server	17	`quantumstudio/tests/test_backend_api.py`, `test_mcp_server.py`
Total	251

Run them all with ./test.sh, or one module with python3 -m pytest <file> -q.

Sample Benchmark Log (12q Smoke)

[preflight] MLX initialization OK
=== Single-circuit run: variational_circuit (qubits: 1,2,5,7,10,11,12, cap: 12) ===
=== Running variational_circuit (qubits: 1,2,5,7,10,11,12, cap: 12) ===
variational_circuit Scaling Benchmark
Framework: mlx–quantum | Device: apple–silicon–mlx | Backend: mps
Testing qubit counts: 1, 2, 5, 7, 10, 11, 12
variational_circuit    |  1q | gates     8 | wall    3.86 ms
variational_circuit    |  2q | gates    20 | wall   17.51 ms
variational_circuit    |  5q | gates    56 | wall   12.41 ms
variational_circuit    |  7q | gates    80 | wall   16.65 ms
variational_circuit    | 10q | gates   116 | wall   18.76 ms
variational_circuit    | 11q | gates   128 | wall   51.75 ms
variational_circuit    | 12q | gates   140 | wall   38.91 ms

Frozen Assets And Sample Bundles

assets/benchmarks-frozen/ is the reproducibility backbone:

latest/ = current promoted benchmark plots
sample-runs/legacy_25q_logs_2025-10/ = curated historical 24/25q logs
sample-runs/legacy_dmg_stage_bench_snapshot/ = legacy full snapshot (images/csv/json/logs)
sample-runs/legacy_25q_plus_smoke12/ = combined large-qubit + modern 12q smoke artifacts

Recommended workflow:

Run benchmarks (bench.sh / bench_with_logging.sh).
Validate data in bench/runs/<run_id>/.
Promote validated outputs to assets/benchmarks-frozen/latest/.
Keep historical reference bundles immutable under sample-runs/.

Testing (200+ Coverage)

The repository includes broad simulator, algorithm, and backend tests:

src/tests/ + quantumstudio/tests/ currently expose 233+ test functions.
Raw assertion density across test code is well above 200 checks (500+ assert-related lines).
Coverage includes:
- gate algebra and unitary identities
- state preparation and measurement parity
- QFT/QAOA/VQE/QCBM/Grover behavior checks
- MPS backend parity and bond-growth diagnostics
- OpenQASM parser/execution checks
- QuantumStudio backend API tests

Run tests:

./test.sh

Targeted suites:

python3 -m pytest src/tests -q
python3 -m pytest quantumstudio/tests -q

QuantumStudio UI

Start/stop the desktop stack:

cd quantumstudio
./bin/appctl up
./bin/appctl status
./bin/appctl logs backend
./bin/appctl down

Build UI artifacts:

./scripts/build_flutter_app.sh --release
./scripts/build_dmg.sh

Screenshots

Product UI

Benchmark Figures (Paper Workflow)

Licensing

Qupertino / QuantumStudio is free and open source under the MIT License (see LICENSE). The source code and the compiled macOS binaries are both covered by the same MIT terms — use, modify, and redistribute freely, including for commercial purposes. No purchase or license key is required.

Notes

This repo is local-first by design: benchmark execution and artifacts remain on-device.
For web-facing product copy, see https://boltzmannentropy.github.io/QupertinoWEB/.

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
assets		assets
bench		bench
bench_test_unit		bench_test_unit
datasets/qasm/local		datasets/qasm/local
osxQuantum-MLX/datasets/qasm/local		osxQuantum-MLX/datasets/qasm/local
quantumstudio		quantumstudio
scripts		scripts
src		src
tools		tools
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
bench.sh		bench.sh
bench_with_logging.sh		bench_with_logging.sh
pyproject.toml		pyproject.toml
pytest.ini		pytest.ini
test.sh		test.sh

Folders and files

Latest commit

History

Repository files navigation

Qupertino / QuantumStudio

How to Cite

Repository (Git / Software Citation)

Technical Report Citation

Extended Introduction

Project Structure

Benchmark Context And Representative Results

The Metal shader tier

Full-suite refresh (all 21 benchmark families, bench.sh --repro)

Installation (Full)

1) System prerequisites

2) Core Python environment

3) Backend dependencies (QuantumStudio API)

4) Runtime path for local commands

5) Flutter UI dependencies (optional, for UI dev)

Running Benchmarks

Standard full-orchestrated run

12-qubit smoke test (recommended health check)

Single-circuit run example

Reproducible timing run

Benchmark Catalog (Detailed)

1) hamiltonian_simulation

2) time_evolution

3) trotter

4) steady_state

5) heisenberg

6) heisenberg_xxz

7) heisenberg_random_field

8) tfim

9) tfim_trotter2

10) tfim_random_field

11) long_range_ising

12) ladder_heisenberg

13) random_circuit

14) qcbm

15) phase_estimation

16) qft

17) qaoa

18) vqe

19) variational_circuit

20) grover

21) ghz

QASM Suite

Example & Circuit Gallery

A. Parameterized benchmark families (--circuit <key>)

B. OpenQASM corpus (datasets/qasm/local/, 42 circuits)

Full Example Catalog (250+)

Sample Benchmark Log (12q Smoke)

Frozen Assets And Sample Bundles

Testing (200+ Coverage)

QuantumStudio UI

Screenshots

Product UI

Benchmark Figures (Paper Workflow)

Licensing

Notes

About

Topics

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases 1

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Full-suite refresh (all 21 benchmark families, `bench.sh --repro`)

1) `hamiltonian_simulation`

2) `time_evolution`

3) `trotter`

4) `steady_state`

5) `heisenberg`

6) `heisenberg_xxz`

7) `heisenberg_random_field`

8) `tfim`

9) `tfim_trotter2`

10) `tfim_random_field`

11) `long_range_ising`

12) `ladder_heisenberg`

13) `random_circuit`

14) `qcbm`

15) `phase_estimation`

16) `qft`

17) `qaoa`

18) `vqe`

19) `variational_circuit`

20) `grover`

21) `ghz`

A. Parameterized benchmark families (`--circuit <key>`)

B. OpenQASM corpus (`datasets/qasm/local/`, 42 circuits)

Packages