Support

_images/gekko_support.png

Bugs

GEKKO is in its early stages and still under active development. If you identify a bug in the package, please submit an issue on GitHub.

If you would like to contribute to GEKKO development, please make pull requests on the GitHub repo.

Questions

Further clarification regarding GEKKO, please visit Stack Overflow with questions tagged gekko for human-generated questions and answers. Generative AI is built into Gekko with the support module. Queries are sent to a web service and a text response is returned. The web service searches a list of hundreds of related Gekko questions and answers to add question context and receive a more relevant answer.

from gekko import support
a = support.agent()
a.ask("Can you optimize the Rosenbrock function?")

Create a support agent with support.agent() and then ask any questions. Prior questions and answers are stored as context for follow-up questions. User questions are not stored for LLM training. The Gekko AI Assistant web service may be unavailable or slow down during periods of high use.

A local Gekko AI Assistant can be run with Retrieval Augmented Generation (RAG) and choice of LLM on a local ollama server. Advanced Process Solutions, LLC (APS) is not responsible in any way for the Gekko AI Assistant accuracy, completeness, performance, timeliness, reliability, content, upgrades, or availability of any information and/or software received as a result of the use of the service. See APS terms and conditions.

Advanced Debugging

The files read/written by GEKKO are in the folder indicated in m._path. Information on debugging these files is available on APMonitor.

Generative AI

Generative AI can also help with basic and advanced questions. Most Large Language Models (LLMs) have knowledge of the Gekko syntax and are able to help with generating prototype applications, suggesting performance improvements, answering questions that would be found in the documentation, designing applications, and working through errors. A context aware OpenAI GPT model is available for Gekko GenAI Support.

_images/gekko_gpt.png

Gekko Skill for AI Tools

A Gekko AI skill package is available in the root of the GEKKO repository as gekko-skill.zip. The skill gives an AI coding assistant additional instructions, examples, and troubleshooting guidance for building, debugging, and running GEKKO models. It is designed to supplement the official GEKKO documentation, APMonitor examples, and the GEKKO source code.

To use the skill in an AI tool:

  1. Download gekko-skill.zip from GitHub. Use the Download raw file button when viewing the file on GitHub.

  2. Add the zip file, or the extracted skill folder, to an AI tool that supports skills, project knowledge, custom instructions, or repository context.

  3. If the tool supports a skills directory, extract the archive so that SKILL.md is inside a folder such as skills/gekko/. Keep the included docs/ and examples/ folders with it.

  4. If the tool supports only file uploads, upload the zip file or extracted folder and tell the assistant to read SKILL.md before answering GEKKO questions.

  5. Start each request with the modeling goal and relevant code, data shape, solver output, or error message. Ask the AI tool to use the Gekko skill when proposing a model formulation, debugging a failed solve, or selecting GEKKO options.

Example prompts for an AI assistant with the skill loaded:

  1. Use the Gekko skill to formulate this optimization problem as a GEKKO model. Include variable declarations, equations, objective function, solver options, and debugging checks.

  2. Use the Gekko skill to debug this failed solve. Explain whether the issue is degrees of freedom, initialization, scaling, solver selection, infeasibility, or an invalid GEKKO expression.

  3. Use the Gekko skill to convert this dynamic system into an MPC application with MV, CV, IMODE, time grid, tuning parameters, and a minimal runnable example.