Random Circuits for Distributed Quantum Computing

Code accompanying the paper "On the Distortion of Partitioning Performance by Random Quantum Circuits"

This repository contains the partitioning pipeline, analysis scripts, and statistical tests used to evaluate how well different hypergraph partitioning strategies distribute quantum circuits across multi-QPU networks — and whether the choice of strategy matters more for real circuits than for random or procedurally generated ones.

---

Note this README was generated by Claude

---

Repository structure

File	Description
main.py	Partitioning pipeline — reads .qasm circuits, maps them to hypergraphs, and runs a partitioning strategy across QPU counts k = 2…10
analysis.py	Generates all paper figures (violin plots, heatmaps, scatter plots, normalised-cut lines) from partitioning_results.csv
mannwhitney_test.py	Pairwise Mann-Whitney U tests comparing cost distributions across circuit origin categories
twoqubit_density.py	Computes two-qubit gate density per circuit origin
zoltan_phg_partition.c	C subprocess wrapper for Zoltan's Parallel HyperGraph (PHG) partitioner
partitioning_results.csv	Pre-computed partitioning results used in the paper

---

Circuit origins

Circuits are placed in ./Circuits/ following the naming convention:

monolithic__{origin}__{identifier}__{n_qubits}.qasm

The five circuit origins studied are:

Origin tag	Category	Description
mqt	Real	MQT Bench application circuits
quipper	Real	Quipper library circuits
randomqiskit	Random	Randomly generated circuits via Qiskit
randomfromgraphqiskit	Random	Random circuits sampled from random graphs via Qiskit
qgen	Generated	Procedurally generated circuits (QGen)

---

Partitioning model

Quantum circuits are modelled as hypergraphs:

• Vertices = qubits
• Hyperedges = multi-qubit gates (single-qubit gates are ignored)

Communication cost = number of cut hyperedges (gates whose qubits span more than one QPU).

The pipeline partitions circuits for every valid k in {2, …, 10} subject to a minimum of 5 qubits per QPU.

---

Partitioning strategies

The following strategies are implemented in main.py (some are commented out and can be swapped in):

Strategy	Key	Notes
StocG (active)	StocG	Randomised iterated greedy with local improvement and a wall-clock cap
Zoltan PHG	zoltan_phg	Zoltan's Parallel HyperGraph partitioner via C subprocess
Greedy	greedy	Single-pass plurality greedy
FM	fm	k-way Fiduccia-Mattheyses iterative improvement
EA	ea	Evolutionary / genetic algorithm
Random	random	Uniform random assignment (baseline)
KaHyPar	kahypar_default	KaHyPar (requires separate install)

All improvements in the analysis are measured relative to the random baseline:

improvement (%) = (random_cut − strategy_cut) / max(random_cut, 1) × 100

---

Getting started

Requirements: Python 3.10+, plus the packages listed below.

python -m venv venv
source venv/bin/activate   # or venv/bin/activate.fish
pip install qiskit tqdm numpy pandas matplotlib scipy

To compile the Zoltan PHG wrapper (requires Zoltan and MPI headers):

# Example — adjust include/lib paths for your system
mpicc zoltan_phg_partition.c -o zoltan_phg_partition \
    -I/path/to/zoltan/include -L/path/to/zoltan/lib -lzoltan -lm

Run the partitioning pipeline (appends to partitioning_results.csv, resumes automatically if interrupted):

python main.py

Generate figures (saved as SVG to ./analysis_figures/):

python analysis.py

Run statistical tests:

python mannwhitney_test.py
python twoqubit_density.py

---

Output

partitioning_results.csv columns:

Column	Description
partitioning_strategy	Strategy name
circuit_partitioned	Circuit filename stem
origin_of_circuit	Origin tag (e.g. mqt, randomqiskit)
circuit_identifier	Algorithm/benchmark name
n_qubits_circuit	Number of qubits
n_qpus	Number of QPUs (k)
coms_cost	Cut hyperedges (communication cost)

---

Citation

If you use this code or data, please cite the accompanying paper:

@inproceedings{grageragarces2026distortion,
  title     = {On the Distortion of Partitioning Performance by Random Quantum Circuits},
  author    = {Gragera Garces, Maria},
  booktitle = {Proceedings of the Workshop on Distributed Quantum Information and Computing (DisQIC)
               at the 46th IEEE International Conference on Distributed Computing Systems (ICDCS 2026)},
  year      = {2026},
  month     = {June},
  address   = {Seoul, South Korea},
}