A unified agent framework for multidomain autonomous scientific simulations
Project description
FermiLink: Unified Agent Framework for Multidomain Autonomous Scientific Simulations
FermiLink is a unified agent framework for multidomain autonomous scientific simulations. It runs on personal laptops, high-performance computing (HPC) clusters, or even your cellphone. You can interact with it through command-line tools, a web UI with a ChatGPT-like chat interface, or a Telegram bot for remote computing when you travel or sleep.
With an official package channel, FermiLink ships with built-in support for more than 150 scientific packages. You can also use its command-line tools to build a local knowledge base from any local scientific package or simulation pipeline.
Once you describe a goal, FermiLink takes care of the simulations on both workstations and HPC clusters. It is designed to sustain long-running multi-task computational jobs for days or weeks without human intervention.
Quick Start
Install the FermiLink package with pip:
pip install fermilink
To use any FermiLink feature, open a clean directory,
mkdir myproject
cd myproject
fermilink init
Then open any local agent (OpenAI Codex, Claude Code, Gemini CLI, their desktop apps, or VS Code extension, etc) within this directory and ask how to use:
- exec/chat/loop/research/reproduce command line tools
- Web UI
- Telegram remote control
- or a general question regarding how to setup or run simulations with FermiLink
You can also simply type in:
fermilink
for a step-by-step guide regarding how to setup the environment of FermiLink.
Advanced FermiLink command line usage
# 1. Install at least one scientific package knowledge base
fermilink install meep
# 2. set up the agent provider
fermilink agent codex/claude/gemini
# 3. (Optional) initialize default HPC profile at ~/.fermilink/HPC_PROFILE.json
fermilink hpc
# 4.1. Use command-line tool to do autonomous scientific research
fermilink exec/loop/reproduce/research goal.md
# 4.2. Start web UI service for ChatGPT-like experience
fermilink start
# 4.3. Start the gateway for supporting Chatbots via Telegram
export FERMILINK_GATEWAY_TELEGRAM_TOKEN="<token-from-@BotFather>"
export FERMILINK_GATEWAY_TELEGRAM_ALLOW_FROM="<numeric-id-from-@get_telegram_id_smppcenter_bot>"
fermilink gateway
Documentation
Visit the documentation for installation details and usage guide.
In brief, the key design principle of FermiLink is the separation of package knowledge bases from simulation workflows, so that simulation workflows in FermiLink, from figure-level simulations to full-paper-level research on high-performance computing clusters, operate uniformly among supported packages via a four-layer progressive disclosure mechanism.
To accommodate simulations at different scopes, as demonstrated below, FermiLink delivers with three major computational workflows.
- exec mode: Designed for short-duration simulations.
- loop mode: Connects iterative agent reasoning with simulation monitoring for PID and HPC SLURM jobs, thus providing robust support for long-duration simulations on both workstations and HPC clusters.
- research/reproduce mode: Intended for multi-task simulations at the scope of a full research paper.
Citation
If you find FermiLink helpful for your research, please cite the following reference:
- Gang Meng†, Andres Felipe Bocanegra Vargas†, Xinwei Ji†, Federico Garcia-Gaitan, Felipe Reyes-Osorio, Jalil Varela-Manjarres, Yafei Ren, Mohammadhasan Dinpajooh, Branislav K. Nikolić, Tao E. Li. FermiLink: A Unified Agent Framework for Multidomain Autonomous Scientific Simulations. arXiv:2604.03460 (2026).
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file fermilink-0.2.3.tar.gz.
File metadata
- Download URL: fermilink-0.2.3.tar.gz
- Upload date:
- Size: 1.1 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
65cc57d0107c293ef0ae0b115f9ada5ea3fe952c9088602f37adadfd67960b0c
|
|
| MD5 |
5413197a2f5ca7d5b62eb37aa4ea5156
|
|
| BLAKE2b-256 |
6c619d5d1e24253ea02bc2db620d633c4a3783b193c273589920c6c9cfcddd05
|
File details
Details for the file fermilink-0.2.3-py3-none-any.whl.
File metadata
- Download URL: fermilink-0.2.3-py3-none-any.whl
- Upload date:
- Size: 2.3 MB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.12
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
e02ed22a47ee15db5514fb437f549e523ed3b9b7acff75823cd5d92c8bdc6aba
|
|
| MD5 |
77c932a331ce12d36a0ce673aac590bb
|
|
| BLAKE2b-256 |
2429da1a472df6947b0745e329d143f2e37e908c2e737405f8805318a07a4d30
|