Physics-based numerical simulation codes are widely used in science and engineering to model complex systems that would be infeasible to study otherwise. While such codes may provide the highest-fidelity representation of system behavior, they are often so slow to run that insight into the system is limited. Trying to understand the effects of inputs on outputs by conducting an exhaustive grid-based sweep over the input parameter space is simply too time-consuming. An alternative approach called “directed exploration” (see figure) has been developed to harvest information from numerical simulators more efficiently. The basic idea is to employ active learning and supervised machine learning to choose cleverly at each step which simulation trials to run next based on the results of previous trials.

Illustration of the Directed Exploration approach in an asteroid collision application. The central image shows the Ida-Dactyl asteroid pair observed serendipitously by the Galileo spacecraft. Planetary scientists are interested in understanding how such systems form and more generally in how asteroid families form. Physics-based numerical simulations offer a means to gain insight into such systems; however, the simulations are so slow to run that a directed exploration strategy is required.
SIM_EXPLORE is a new computer program that uses directed exploration to explore efficiently complex systems represented by numerical simulations. The software sequentially identifies and runs simulation trials that it believes will be most informative given the results of previous trials. The results of new trials are incorporated into the software’s model of the system behavior. The updated model is then used to pick the next round of new trials. This process, implemented as a closed-loop system wrapped around existing simulation code, provides a means to improve the speed and efficiency with which a set of simulations can yield scientifically useful results.

The software focuses on the case in which the feedback from the simulation trials is binary-valued, i.e., the learner is only informed of the success or failure of the simulation trial to produce a desired output. The software offers a number of choices for the supervised learning algorithm (the method used to model the system behavior given the results so far) and a number of choices for the active learning strategy (the method used to choose which new simulation trials to run given the current behavior model). The software also makes use of the LEGION distributed computing framework to leverage the power of a set of compute nodes. The approach has been demonstrated on a planetary science application in which numerical simulations are used to study the formation of asteroid families.

This work was done by Michael Burl and Esther Wang of Caltech, and Brian Enke and William J. Merline of SWRI for NASA’s Jet Propulsion Laboratory.

This software is available for commercial licensing. Please contact Dan Broderick at This email address is being protected from spambots. You need JavaScript enabled to view it.. NPO-47919

Topics:
Artificial intelligence (AI) Computer software and hardware Education and training Machine learning Mathematical models People and personalities Simulation and modeling Simulators Software
This Brief includes a Technical Support Package (TSP).
Document cover
SIM_EXPLORE: Software for Directed Exploration of Complex Systems

(reference NPO-47919) is currently available for download from the TSP library.

Don't have an account?

Magazine cover
Software Tech Briefs Magazine

This article first appeared in the September, 2013 issue of Software Tech Briefs Magazine (Vol. 37 No. 9).

Read more articles from this issue here.

Read more articles from the archives here.

Overview

The document outlines the SIM_EXPLORE software developed by NASA's Jet Propulsion Laboratory (JPL) under the Applied Information Systems Research (AISR) program. This software is designed for the directed exploration of complex systems, particularly in scenarios where traditional methods of study are impractical due to constraints such as cost, safety, or the inability to replicate conditions in a laboratory.

The primary challenge addressed by SIM_EXPLORE is the long run-times associated with physics-based numerical simulations, which can hinder exhaustive exploration of input parameter landscapes. To overcome this, the software employs state-of-the-art supervised machine learning techniques combined with active learning strategies. The key innovation is using the simulator as an oracle to generate labeled training data (input-output pairs), which allows for the development of predictive models of the systems being studied.

Active learning is then utilized to determine which simulation trials would yield the most valuable new data, thereby optimizing the exploration process. This approach enables researchers to efficiently navigate the complex parameter space and gain insights into the behavior of systems that are otherwise difficult to study directly.

The document also acknowledges contributions from various individuals and references several related works that support the development and application of the SIM_EXPLORE software. It highlights the importance of collaboration in advancing knowledge discovery from simulations and emphasizes the potential of this software to enhance understanding in fields such as planetary science, particularly in studying asteroid formation and dynamics.

Overall, the SIM_EXPLORE software represents a significant advancement in the ability to model and explore complex systems, providing a framework that can be applied to various scientific and engineering challenges. The document serves as a technical support package, offering insights into the software's capabilities and the broader implications of its use in research and technology development.