Maximal Independent Set – Quantum Computing Dataset

aqora / Maximal Independent Set

Public

About dataset version

Maximum Independent Set (MIS) — Neutral-Atom Benchmark Problem Instances

Overview

This dataset provides a Parquet-formatted collection of Maximum Independent Set (MIS) problem instances used to benchmark large-scale neutral atom quantum processors. The problems originate from the study:

Andrea B. Rava, Kristel Michielsen, J. A. Montañez-Barrera,
Benchmarking neutral atom-based quantum processors at scale,
arXiv:2511.22967 (2025).

Each instance encodes an MIS optimization problem as an Ising Hamiltonian, suitable for quantum adiabatic algorithms (QAA), QAOA-style protocols, and analog neutral-atom implementations. The original .ising files have been preserved verbatim and reorganized into a single Parquet file for scalable analysis, benchmarking, and machine-learning workflows.

Scientific Background

The Maximum Independent Set (MIS) problem asks for the largest subset of vertices in a graph such that no two selected vertices share an edge. MIS is NP-hard and is widely used as a benchmark for combinatorial optimization on quantum hardware.

In this dataset, MIS instances are defined on unit-disk graphs (UDGs) and, in particular, diagonal-connected unit-disk grid graphs (DUGGs). These graphs naturally map to neutral-atom Rydberg platforms, where the Rydberg blockade enforces the MIS constraint that adjacent vertices cannot both be occupied.

Each .ising file specifies the diagonal and off-diagonal terms of the Ising cost Hamiltonian:

C(\vec{x}) = \sum_i x_i - \lambda_0 \sum_{(i,j)\in E} x_i x_j

These instances were used to benchmark quantum adiabatic algorithms (QAA) and quantum approximate optimization algorithms (QAOA) on real neutral-atom quantum processors such as QuEra Aquila and Pasqal Fresnel, as well as large-scale classical emulators.

Dataset Contents

Rows: 72 MIS problem instances
One row per problem: Each row corresponds to one .ising file
Problem sizes: From small grids up to large unit-disk graphs
Graph family: Unit-disk grid graphs with diagonal connectivity
Transformations:
No reformulation, optimization, or normalization
Only serialization from text-based .ising format to Parquet

The full original Ising specification is retained for reproducibility and downstream verification.

Parquet Schema

Column	Type	Description
filename	String	Original .ising filename
num_qubits_header	Int32	Number of variables declared
offset	Float64	Constant offset in the objective
qubit_map	Map[String, Int32]	Symbolic variable to index mapping
metadata	Struct	Grid size, seed, and filling parameters
bias_terms	List[Struct]	Diagonal Ising terms
couplers	List[Struct]	Pairwise interactions
ising_text	String	Full original file content

Loading Examples

Polars


import polars as pl

from aqora_cli.pyarrow import dataset

df = pl.scan_pyarrow_dataset(

    dataset("aqora/maximal-independent-set", "v0.0.0")

).collect()

print(df.head())

Pandas


import pandas as pd

df = pd.read_parquet(

    "aqora://aqora/maximal-independent-set/v0.0.0"

)

print(df.head())

Citation

If you use this dataset, please cite the original work:

Andrea B. Rava, Kristel Michielsen, J. A. Montañez-Barrera,

“Benchmarking neutral atom-based quantum processors at scale,”

arXiv:2511.22967 (2025).

https://arxiv.org/abs/2511.22967