Nebulous

Where uncertainty becomes innovation.

ECLIPSE: Evolutionary Computation Library for Instrumentation Prototyping in Scientific Engineering

ECLIPSE is a general framework for the evolutionary design of hardware in space-science domains. It provides a modular architecture consisting of Individuals, which encode hardware designs using physically constrained representations; Evaluators, which translate simulation outputs into fitness measures; and Evolvers, which implement EC algorithms.

ECLIPSE Framework Overview

Goal & Motivation

Designing space-science hardware is a complex, expensive, and highly iterative process. Traditional engineering workflows rely on expert-driven design followed by extensive simulation and refinement cycles, which can be time-consuming and limit the exploration of unconventional geometries. Evolutionary computation (EC) offers an appealing alternative: by automatically exploring large, high-dimensional design spaces, EC can discover high-performing and sometimes unintuitive solutions that human designers may overlook.

However, applying evolutionary methods to new scientific hardware remains challenging because most scientific simulation tools were not originally developed with evolutionary workflows in mind. They are often optimized for accuracy rather than high-throughput evaluation. ECLIPSE provides the integration layer needed to couple evolutionary search with real scientific modeling environments in a consistent, extensible, and domain-aware manner.

Reusable Software Infrastructure

The ECLIPSE framework underpins all Nebulous projects, providing shared evolutionary algorithms, geometry encodings, solver interfaces, and orchestration tools. A key research theme is extending ECLIPSE so that new classes of instruments can be brought into the framework with minimal additional effort. By separating representation, evaluation, and evolutionary logic, the framework enables interdisciplinary teams to explore unconventional hardware geometries while utilizing well-established scientific modeling tools.

Results & Updates

Who are we?

Contributing Team

Dr. Julie Rolla

Dr. Julie Rolla

Principal Investigator (PI), Director
Astrophysicist / Earth Scientist
NASA Jet Propulsion Laboratory

Dr. Emily Dolson

Dr. Emily Dolson

Deputy Director
Assistant Professor, CSE
Michigan State University

Dr. Charles Ofria

Dr. Charles Ofria

Professor, Computer Science & Engineering
Michigan State University

Max Foreback

Max Foreback

Lead Developer
PhD Candidate, Michigan State University

Dr. Vincent Ragusa

Postdoc
Michigan State University

Evan Imata

Evan Imata

Postbacc
UC Berkeley

Christina Shao

Postbacc
UC Berkeley

Evan Imata

Joey Wagner

Undergrad
Michigan State University