Open-source code for the Spatial Optimization Model for Land Use (SSFLA-MLAS)

The open-source content is the source code of a land use spatial optimization model (SSFLA-MLAS) that couples the spatial leapfrog algorithm with multi-level multi-agent systems. The related paper is titled "Multi-objective synergy and hierarchical decision-making in land use optimization: Coupling bionic algorithm with multi-level agent". Since this research is still in the submission stage, the following is a summary of the research for ease of understanding.Land use optimization is a crucial method to reduce carbon emission and promote sustainable development. However, existing optimization models exhibit limitations in addressing complex human-land conflicts: (1) Local-scale models fail to incorporate spatial non-stationarity constraints and multi-level agent interactions, resulting in inadequate representation of heterogeneous decision-making. (2) Global-scale methods struggle with the endogenization of geographical rules, hindering effective translation of numerical solutions into spatial configurations. (3) The insufficient spatial-scale compatibility between global and local optimizations, coupled with shallow model integration and the absence of multi-scale feedback mechanisms, collectively constrain their application in supporting low-carbon-oriented territorial spatial management. Therefore, this study constructs a multi-objective optimization scenario under low-carbon development, introducing a new coupling model (SSFLA-MLAS). It designs a multi-level agent system (MLAS) where mapping and interaction coexist, and creates a spatialized frog leaping algorithm (SSFLA) by reconfiguring heuristic operators. The coupling of multi-models is achieved through an internal and external nested approach. The results, using Huangpi District as the experimental area, demonstrate that: (1) SSFLA-MLAS outperforms single algorithms in effectiveness and efficiency, achieving multi-objective benefits 3.4%, 32.5% and 26.5% greater than SSFLA, MLAS and MAS. (2) MLAS improves the comprehensive landscape pattern index by 26.4% over MAS while aligning spatial layouts with development strategies. (3) SSFLA-MLAS demonstrates remarkable optimization results, improving economic, social and ecological objectives by 5.01%, 2.87% and 0.74%, while enhancing carbon sinks by 1.38%. This research provides both a methodological framework for achieving sustainable development goals and novel insights into coupling global optimization with local simulation for modeling.

本开源内容为土地利用空间优化模型（SSFLA-MLAS）的源代码，该模型将空间跳蛙算法（spatial leapfrog algorithm）与多层次多智能体系统（multi-level multi-agent systems）进行耦合。相关研究论文题为《土地利用优化中的多目标协同与层级决策：仿生算法与多层次智能体的耦合》。鉴于本研究仍处于投稿阶段，以下为便于理解的研究概要。土地利用优化是降低碳排放、推动可持续发展的关键手段。然而现有优化模型在应对复杂人地冲突时存在诸多局限：(1) 局地尺度模型未能纳入空间非平稳性约束与多层次智能体交互，导致对异质性决策的刻画不足；(2) 全局尺度方法难以实现地理规则的内生化，阻碍了数值解向空间配置的有效转化；(3) 全局与局部优化间的空间尺度适配性不足，加之模型集成程度较浅且缺乏多尺度反馈机制，共同制约了其在支撑低碳导向的国土空间治理中的应用。因此，本研究构建了低碳发展视角下的多目标优化场景，提出新型耦合模型SSFLA-MLAS。设计了兼具映射与交互特性的多层次智能体系统（MLAS，multi-level agent system），并通过重构启发式算子（heuristic operators）构建了空间化蛙跳算法（SSFLA，spatialized frog leaping algorithm）。通过内外嵌套的方式实现多模型耦合。以黄陂区为实验区域的研究结果表明：(1) SSFLA-MLAS的有效性与效率均优于单一算法，其多目标收益较SSFLA、MLAS及MAS分别高出3.4%、32.5%与26.5%；(2) MLAS较MAS将综合景观格局指数提升26.4%，同时使空间布局契合发展战略；(3) SSFLA-MLAS展现出优异的优化效果，使经济、社会与生态目标分别提升5.01%、2.87%与0.74%，同时增强碳汇（carbon sinks）能力1.38%。本研究既为实现可持续发展目标提供了方法论框架，也为全局优化与局部模拟的耦合建模提供了全新视角。