LLM-SOLVE — PROGRAM
Days: Sunday, 19 July 2026
Sunday, 19 July 2026
09:00-09:50
Talk1
LLM-Solve
Location:
C5.08
09:50-10:30
Session 1 - Explanations with LLMs
LLM-Solve
Location:
C5.08
| 09:50-10:10 |
Towards Human Readable Explanations (abstract) 20 min
1 UCLouvain
|
| 10:10-10:30 |
Large Language Models for Explaining Unsatisfiable Constraint Satisfaction Problems (abstract) 20 min
1 KU Leuven
|
10:30-11:00
Coffee Break
LLM-Solve
Location:
C5.08
11:00-12:40
Session 2 - LLMs for Constraint Model Generation
LLM-Solve
Location:
C5.08
| 11:00-11:20 |
Learn2Zinc: Fine-tuning Small Language Models for Text-to-Model Translation in MiniZinc (abstract) 20 min
1 Fidelity Investments
|
| 11:20-11:40 |
CP-SynC: Multi-Agent Zero-Shot Constraint Modeling in MiniZinc with Synthesized Checkers (abstract) 20 min
1 University of Toronto
|
| 11:40-12:00 |
Using LLMs to Generate CoPTIC Constraint Models (abstract) 20 min
1 University of Reading
|
| 12:00-12:20 |
Solving MaxSAT Problems from Natural Language Descriptions with LLMs and PySAT (abstract) 20 min
1 Artificial Intelligence Research Institute (IIIA), Spanish National Research Council (CSIC), Barcelona, Spain
2 Department of Computer Science, University of Oxford, Oxford, UK
3 Institut de Robòtica i Informàtica Industrial (IRI-CSIC-UPC), Barcelona, Spain
|
| 12:20-12:40 |
ZebraProlog: Solving the Zebra Logic Benchmark1 with the Help of Prolog and CLP(FD) (abstract) 20 min
1 Ensimag
2 Université Côte d'Azur
|
12:40-14:00
Lunch
LLM-Solve
Location:
C5.08
14:00-14:50
Talk 2
LLM-Solve
Location:
C5.08
14:50-15:30
Session 3 - Formal Methods for LLMs
LLM-Solve
Location:
C5.08
| 14:50-15:10 |
Escaping Local Optima in Prompt Optimisation via Failure-Driven Refinement and History-Guided Restarting (abstract) 20 min
1 ITMO University
|
| 15:10-15:30 |
Controlling Agent Behavior with Policy-as-Code via Autoformalization (abstract) 20 min
1 Sondera
|
15:30-16:00
Coffee Break
LLM-Solve
Location:
C5.08
16:00-17:20
Session 4 - LLMs for Improving Solvers and Models
LLM-Solve
Location:
C5.08
| 16:00-16:20 |
Improving Constraint Models with LLM Agents (abstract) 20 min
1 TU Wien
|
| 16:20-16:40 |
LLM-Guided Constraint Model Reformulation for Solving Efficiency (abstract) 20 min
1 KU Leuven
2 University of Western Macedonia
3 University of St Andrews
|
| 16:40-17:00 |
KissatEvolve: Controllable and Scalable Synthesis of SAT Heuristics with Densely Annotated Memory Bank (abstract) 20 min
1 Georgia Institute of Technology
2 Simon Fraser University
|
| 17:00-17:20 |
Can LLMs Build a MaxSAT Solver from Papers? The CoreForge Experience (abstract) 20 min
1 Carnegie Mellon University
|
17:20-17:30
Closing remarks
LLM-Solve
Location:
C5.08