Optimizing Parameters of Quantum Circuits with Sparsity-Inducing Coordinate Descent

Authors: Rudy Raymond, Zichang He

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We provide theoretical analyses and demonstrate experiments showing the effectiveness of Rotolasso to solve instances of combinatorial optimization problems. [...] We show the efficacy of CD-based model selection by experimenting on instances of Max Cut and LABS, two hard problems with promising near-term quantum algorithms.
Researcher Affiliation Industry Rudy Raymond , Zichang He Global Technology Applied Research (GTAR), JPMorgan Chase & Co., New York, NY 10001, USA EMAIL
Pseudocode Yes Coordinate Descent (CD) Evaluate VQC or QAOA at 2m + 1 different rotation values to obtain the tuples {(θj,l, ˆf(θj,l))}2m+1 l=1 . Build a system of linear equations as Eq. (9) for obtaining (c0,j, c1,j, . . . , c2m,j) with classical computers from {(θj,l, ˆf(θj,l))}2m+1 l=1 . Compute θ j = arg minθj ˆf(θj) by classical computers.
Open Source Code No Supplementary Material is available at 10.5281/zenodo.15425252. The paper does not explicitly state that code is provided in the supplementary material or provide a direct link to a code repository.
Open Datasets Yes Experiments demonstrating the role of sparsity-inducing CD in PQCs are conducted using qujax [Duffield et al., 2023] for classical simulation of quantum circuits of both QAOA and VQC for LABS and Max Cut (3-regular graphs) instances. [...] Another hard problem we consider is the so-called low-autocorrelation binary sequences (LABS) [Shaydulin et al., 2024] whose objective is to minimize the sum of the squares of autocorrelations of z { 1, 1}n formulated as min z { 1,1}n Pn 1 k=1 ( Pn k i=1 zizi+k)2.
Dataset Splits No The paper focuses on combinatorial optimization problems (Max Cut and LABS) and evaluates performance on different problem instances and initial parameter settings. It does not describe dataset splits (e.g., train/test/validation) in the context of machine learning. The text refers to 'random initial θ s' and 'random graphs' which relate to initialization and instance generation, not data splitting.
Hardware Specification No Experiments demonstrating the role of sparsity-inducing CD in PQCs are conducted using qujax [Duffield et al., 2023] for classical simulation of quantum circuits of both QAOA and VQC for LABS and Max Cut (3-regular graphs) instances. No specific hardware (e.g., GPU/CPU models, cloud resources) used for these classical simulations is mentioned in the paper.
Software Dependencies No Experiments demonstrating the role of sparsity-inducing CD in PQCs are conducted using qujax [Duffield et al., 2023] for classical simulation of quantum circuits of both QAOA and VQC for LABS and Max Cut (3-regular graphs) instances. The paper mentions 'qujax' but does not specify its version number.
Experiment Setup Yes Following the standard usage of L1-regularized optimizer in conventional machine learning, we iteratively run Rotolasso on a randomly initialized θ(0) with the regularization factor λ initialized to λstart, and update each parameter cyclically to obtain a (sub)optimal θ . The value of λ is then reduced by a factor while utilizing the previously obtained θ as the new initial parameter set. This is repeated until λ becomes very close to 0. [...] Each data point is derived from 5 random initial θ s. [...] Each point is derived from 5 initial θ s and 5 random graphs. [...] Each datapoint is derived from 50 random initial θ.