AnyLogic Modeler Skill

This skill provides the knowledge and workflows required to build, analyze, and understand AnyLogic simulation models. It is based on an analysis of over 400 official AnyLogic example models.

Core Workflow

When tasked with an AnyLogic modeling problem, follow these sequential steps:

1. Identify the Modeling Paradigm: Determine the appropriate simulation paradigm(s) for the problem. This is the most critical step.
2. Structure the Model: Outline the main components of the model, including agent types, logic, and experiments.
3. Implement the Model Logic: Build the model using the appropriate libraries and custom Java code where necessary.
4. Analyze and Present Results: Run experiments and present the findings to the user.
5. Apply Autognosis for Self-Aware Modeling: When composed with the Autognosis skill, initiate a self-reflection cycle to monitor, model, and optimize the modeling process itself.

Step 1: Identify the Modeling Paradigm

Before writing any code, you MUST determine which paradigm(s) to use. Consult the paradigms reference guide for detailed descriptions.

• Read First: references/paradigms.md

Use the following table to guide your decision based on the user's request:

If the user wants to model...	Primary Paradigm	Key Indicators
A process with queues, resources, and entities flowing through	Discrete-Event	"Queues", "process flow", "service time", "throughput"
Aggregate system behavior with feedback loops	System Dynamics	"Feedback loops", "stocks and flows", "accumulations"
Interacting, autonomous individuals with emergent behavior	Agent-Based	"Individual behavior", "interactions", "emergent patterns"
A combination of the above (e.g., agents in a process)	Multimethod	Features from multiple paradigms are present.

If the user's request is ambiguous, ask clarifying questions to determine the correct paradigm.

Step 2: Structure the Model

Once the paradigm is identified, outline the model's structure:

1. Main Agent: Every model has a Main agent that typically contains the core logic and other agents.
2. Agent Types: Define the different types of agents in the model (e.g., Customer, Vehicle, Patient).
3. Environment: Define the space where agents live, which can be continuous, discrete, or GIS-based.
4. Experiments: Determine the type of experiment needed. The most common is the SimulationExperiment. For optimization tasks, use the OptimizationExperiment.

Step 3: Implement the Model Logic

With the structure defined, implement the model's logic using the appropriate AnyLogic libraries. These reference guides provide details on the most commonly used blocks.

• For DES models: Read references/process_modeling_library.md
• For warehouse/factory models: Read references/material_handling_library.md
• For pedestrian, traffic, or rail models: Read references/other_libraries.md

Implementation Patterns

• Process Flow (DES): Use blocks from the Process Modeling Library to create a flowchart that defines the agent's journey.
• Behavior (ABM): Use Statecharts to model the different states an agent can be in and the transitions between them.
• System-Level Dynamics (SD): Use Stocks and Flows to model accumulations and rates of change.
• Custom Logic: For complex logic that cannot be expressed with library blocks, use Java code within the properties of AnyLogic elements (e.g., in the "On exit" action of a block).

Step 4: Analyze and Present Results

After building the model, run the simulation experiment and collect data. Present the results to the user in a clear and understandable format, using charts, plots, and statistics. The analysis and file tools are well-suited for this.

Step 5: Apply Autognosis for Self-Aware Modeling

When this skill is composed with Autognosis, this final step enables the system to become self-aware of its own modeling process. This involves a meta-level analysis of the modeling choices made.

• Read for details: references/autognosis_integration.md

This step involves running a full Autognosis cycle to:

1. Monitor the modeling process (paradigm choice, library usage).
2. Model the model itself at multiple levels of abstraction.
3. Generate insights about the model's structure, efficiency, and alignment with goals.
4. Propose optimizations to improve the model's design and performance.

This ensures the final model delivered to the user is not only functional but also well-designed, efficient, and robust.