PathWise: Planning through World Model for Automated Heuristic Design via Self-Evolving LLMs

arXiv 26.1 2026 31.6 method, application

TLDR

PathWise uses a world model agent for heuristic rollouts in a multi-agent LLM framework to automate heuristic design for combinatorial optimization.

Reasoning

The paper introduces a novel multi-agent reasoning framework with a world model for state-aware planning, which is a strength. However, the abstract lacks details on the world model's implementation and scalability, and the evaluation is limited to combinatorial optimization problems.

Read-first score

Read-first score 31.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 19.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=evaluation,experiment

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Topical relevance 29%

27.1

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Citation impact 18%

10.7

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.1065087

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 73.

Keyword Scores

world model

generative world model

world dynamics prediction

model-based reinforcement learning world model

world simulator

interactive world model

video world model

Deep Analysis

Innovations

Formulating heuristic generation as a sequential decision process over an entailment graph that serves as a compact, stateful memory of the search trajectory
Multi-agent reasoning framework with a policy agent for planning, a world model agent for heuristic rollouts, and critic agents for routed reflections
Shifting LLM-based automated heuristic design from trial-and-error evolution to state-aware planning through reasoning

Methodology

PathWise employs a multi-agent framework where a policy agent plans evolutionary actions, a world model agent generates heuristic rollouts conditioned on those actions, and critic agents provide routed reflections summarizing lessons from prior steps. The system operates over an entailment graph that acts as a stateful memory, enabling the reuse or avoidance of derivation information across generations. This approach formulates heuristic generation as a sequential decision process, moving beyond fixed evolutionary rules and static prompt templates.

Key Results

Experiments across diverse combinatorial optimization problems show that PathWise converges faster to better heuristics, generalizes across different LLM backbones, and scales to larger problem sizes.

Resources

Paper