LLM-Based Exogenous World Modeling for Dynamic Scheduling

Ni Zhang; Haofeng Yuan; Jianan Zhou; Shiji Song; Zhiguang Cao

cs.AIOct 21, 2026

LLM-Based Exogenous World Modeling for Dynamic Scheduling

Authors: Ni Zhang, Haofeng Yuan, Jianan Zhou, Shiji Song, Zhiguang Cao

Organizations: Singapore Management University · Tsinghua University · Nanyang Technological University

Abstract

Dynamic scheduling requires dispatching online arrivals while anticipating future jobs, as poor decisions can compound downstream costs under nonlinear tardiness objectives. We show that a pretrained large language model (LLM) can serve as the exogenous component of a structured world model, forecasting uncertain future job arrivals. Specifically, we exploit an afterstate decomposition, which separates known deterministic shop transitions from uncertain exogenous job arrivals, enabling the LLM to forecast only the latter through a compact belief representation. Building on this, we develop Afterstate World-Model Planning (AWMP), which combines an evolved, belief-conditioned dispatching strategy with short scenario-based rollouts. A surprise-triggered refresh mechanism queries the LLM only when newly observed arrivals deviate sufficiently from previous forecast, and online residual self-calibration further conditions the LLM on its recent forecast errors to handle nonstationary arrivals. Across diverse arrival distributions, AWMP reduces weighted tardiness by 6-21% on dynamic job-shop and 22-46% on flexible job-shop benchmarks, relative to the best of eight classical priority-dispatching rules.

Figures & tables

Figure 1: Illustration of dynamic scheduling with online job arrivals and dispatching decisions under unknown future arrivals. At the current time $c_{t}$ , the scheduler observes the released jobs and current shop state and selects the next operation to dispatch. Here, $J_{i}$ denotes job $i$ , $J_{i}$ – $O_{j}$ denotes the $j$ -th operation of job $i$ , and “due” indicates the corresponding job due date. Future job arrivals remain unknown.

Figure 2: Overview of our AWMP framework. 1) The dynamic scheduling environment provides the current shop state and the observed arrival history. 2) The afterstate decomposition separates the known, deterministic shop transition from the exogenous job-arrival process. 3) A pretrained LLM forms the exogenous component of the structured world model and, with online residual self-calibration, forecasts a compact early-late arrival belief. 4) The belief-conditioned planner first scores feasible actions, retains the top-

K

candidates, and evaluates them through short scenario-based rollouts. 5) Meta-evolution optimizes planning strategies offline. 6) The best strategy is frozen and evaluated on held-out dynamic JSSP and FJSP benchmarks.

Benchmark	Distribution	Myopic	Rolling	MSA	SAA	EV	Best PDR	AWMP (Ours)
Dynamic JSSP	Pulse train	79931	67930	66509	66568	66178	32923	29238 (-11.2%)
	Late-ramp	70584	59654	58544	58957	58380	34841	27619 (-20.7%)
	Cyclic	82708	70243	69530	69561	68841	34486	32483 (-5.8%)
	Step-up	78631	66590	65719	65885	65388	32686	29630 (-9.4%)
Dynamic FJSP	Pulse train	836	369	384	342	315	212	163 (-23.1%)
	Late-ramp	1543	559	912	591	561	456	258 (-43.4%)

Table 1: Average objective (total weighted tardiness) on dynamic JSSP and FJSP instances under four arrival distributions (lower is better). Best PDR reports the best result among the eight baseline PDRs.

Figure 3: Forecast correlation (a) without and (b) with residual self-calibration, increasing from

r=0.12

r=0.93

Figure 4: (a) Meta-evolution trajectory; (b) seed vs. evolved strategy performance on training and held-out instances.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Empirical arrival-rate profile of each distribution family over the arrival window (

200

seeds,

\beta=1.5

Model-Based Planning

Model-Based Planning uses explicit or learned models of how actions change states to simulate futures, compare plans, and choose actions.

See all 207 papers

DSevolve: Enabling Real-Time Adaptive Scheduling on Dynamic Flexible Job Shop with LLM-Evolved Heuristic Portfolios
Mar 29, 2026XinLei Zhou, Jin Huang, Jie Yang +2Flexible Job Shop Scheduling Real-Time Scheduling
School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, 430074, PR China
Beyond Prediction: Tail-Aware Scheduling for LLM Inference
Jun 16, 2026Yueying Li, Yuanfan Chen, Jiayang Chen +6LLM Serving Inference-Time Optimization
Cornell University, Computer Science Department, NY, USA · Microsoft Azure System Research, WA, USA · Cornell University, Electrical and Computer Engineering Department, NY, USA +2
MAPS: Memory-Aware Predictive Scheduling Framework for Large Language Model Serving
Sep 14, 2026Tiancheng Zhang, Yulin Chen, Yunfeng Zhao +3Disaggregated LLM Serving LLM Inference Scheduling
College of Intelligence and Computing, Tianjin University, Tianjin, China · The International Joint Institute of Tianjin University, Tianjin University, Fuzhou, China · Faculty of Digital Economics and Managements, Tianjin University of Finance and Economics, Tianjin, China

Mar 29, 2026J/K move · Enter open · S save

DSevolve: Enabling Real-Time Adaptive Scheduling on Dynamic Flexible Job Shop with LLM-Evolved Heuristic Portfolios

XinLei Zhou, Jin Huang, Jie Yang, Xinyu Li+1

School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, 430074, PR China

Flexible Job Shop Scheduling Real-Time Scheduling Combinatorial Optimization Automated Heuristic Design Metaheuristic Optimization

Open paper AskPDF arXiv

In dynamic flexible job shops, order arrivals, machine breakdowns, and processing-time deviations continually reshape the scheduling state and the priority trade-offs behind dispatching decisions. Dispatching rules are well suited to this setting because they are fast, interpretable, and easy to deploy, and recent LLM-assisted automatic heuristic design further expands their expressiveness by evolving composite priority functions. The key challenge is to make these evolved rule behaviors state-adaptive without losing the rapid response needed for online rescheduling. This paper proposes a dynamic self-evolutionary framework DSevolve, which separates offline rule-library construction from online state-conditioned rule selection. Offline, an LLM-guided quality-diversity search combines multi-persona seeding, a MAP-Elites behavioral archive, and behavior-guided variation to evolve a library of complementary rules rather than a single elite, and event-level simulation then trains a neural selector to rank the rules by state. Online, after each rescheduling event, a neural selector maps a 22-dimensional state fingerprint to rule scores and dispatches the top-ranked rule within about a second, meeting the response-time requirement after each disruption. Experiments on dynamic instances derived from standard flexible job shop benchmarks show that DSevolve achieves lower mean makespan than individual LLM-evolved rules, classical dispatching rules, and learning-based baselines under a one-active-rule deployment protocol. Trained only on small instances, the selector transfers zero-shot to substantially larger dynamic shops. These results show that state-conditioned selection preserves the speed and interpretability of dispatching rules while improving adaptability.

Abstract

Model-Based Planning

Explore similar work

DSevolve: Enabling Real-Time Adaptive Scheduling on Dynamic Flexible Job Shop with LLM-Evolved Heuristic Portfolios

Beyond Prediction: Tail-Aware Scheduling for LLM Inference

MAPS: Memory-Aware Predictive Scheduling Framework for Large Language Model Serving