MASPRM: Multi-Agent System Process Reward Model
arXiv preprint arXiv:2510.24803 (2025)
Publications
Grouped by research program, newest first. Google Scholar.
Agent reasoning and adaptation
Automated algorithm discovery and learned search
LLM-Guided Evolution of Symbolic Acceleration Inequality Families for MILP
Transactions on Machine Learning Research (2026); arXiv:2508.11850
COAgents: Multi-Agent Framework to Learn and Navigate Routing Problems Search Space
LION 20: Learning and Intelligent Optimization Conference (2026)
Latent Heuristic Search: Continuous Optimization for Automated Algorithm Design
LION 20: Learning and Intelligent Optimization Conference (2026)
Language, modeling, and evaluation
Discarded or Unreachable? Diagnosing and Repairing Spatial Relation Failure in Vision–Language Similarity
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027)
Separates two causes of near-chance left/right scoring in contrastive vision-language models, an access failure and a pooling failure, and repairs both with a zero-parameter spatial readout from patch features that leaves non-relational rankings unchanged.
Evaluating LLM Reasoning in the Operations Research Domain with ORQA
Proceedings of the AAAI Conference on Artificial Intelligence (2025)
SmartAPS: Tool-Augmented LLMs for Operations Management
AAAI 2025 Bridge on AI and ORMS; arXiv:2507.17927 (2025)
Optimization Modeling and Verification from Problem Specifications Using a Multi-Agent Multi-Stage LLM Framework
INFOR: Information Systems and Operational Research (2024)
Towards Human-Aligned Evaluation for Linear Programming Word Problems
LREC-COLING 2024
LaTeX2Solver: A Hierarchical Semantic Parsing of LaTeX Document into Code for an Assistive Optimization Modeling Application
ACL 2023, System Demonstrations
NL4Opt Competition: Formulating Optimization Problems Based on Their Natural Language Descriptions
NeurIPS 2022 Competition Track, PMLR (2023)
Current work
Work under anonymous review is listed by topic only. Titles and links are added as decisions and preprints appear.
- Agent reasoning and adaptation: pre-execution action selection, memory for frozen models, reusable multi-agent designs, and compute allocation for embodied agents. under review
- Agent reasoning and adaptation: HERMES, a memory architecture for multi-agent systems. in progress
- Language, modeling, and evaluation: ambiguity-aware modeling benchmarks. under review
- Language, modeling, and evaluation: an evaluation harness for coding agents that modify a production optimization solver. in progress
Optimization methods and decision systems
A Branch-and-Price Algorithm for Fast and Equitable Last-Mile Relief Aid Distribution
European Journal of Operational Research (2025)
Discrete Optimization of Min-Max Violation and Its Applications Across Computational Sciences
arXiv preprint arXiv:2508.13437 (2025)
Bayan Algorithm: Detecting Communities in Networks Through Exact and Approximate Optimization of Modularity
Physical Review E (2024)
Analyzing Modularity Maximization in Approximation, Heuristic, and Graph Neural Network Algorithms for Community Detection
Journal of Computational Science (2024)
Heuristic Modularity Maximization Algorithms for Community Detection Rarely Return an Optimal Partition or Anything Similar
International Conference on Computational Science, ICCS (2023)
Two Dimensional Guillotine Cutting Stock and Scheduling Problem in Printing Industry
Computers & Operations Research (2022)
Inequity-Averse Shelter Location for Disaster Preparedness
IISE Transactions (2019)
Patents
Methods and Systems for Generating Advanced Planning Tools for Operations Analysis and Management
US Patent App. US20250363435A1 (2025)
Methods and Systems for Specification Extraction and Model Building Using Language Models
WO Patent WO2025043698A1 (2025)
Methods and Systems for Explainable Template Retrieval for Optimization Modeling
WO Patent WO2024174195A1 (2024)
Systems and Methods for Towards Human-Aligned Evaluation for Auto-Formulating Optimization Modeling with Large Language Models
US Patent App. US20250131059A1 (2025)