Call for Position Papers: ASCR Workshop on Inverse Methods for Complex Systems under Uncertainty
Due Date: 5:00 PM ET on April 21, 2025
Important Dates
- April 21, 2025: Deadline for position paper submission
- May 9, 2025: Notification of position paper acceptance
- June 10-12, 2025: Workshop (greater Washington, DC area)
- Workshop Website: https://orau.gov/InverseMethods
Motivation
The ability to solve inverse problems – inferring unknown parameters, structures, or states of a system from observed data – is essential for advancing scientific discovery and innovation capabilities for the DOE mission. Basic research needs and challenges are particularly acute in emerging areas such as the interactive, data-driven, modeling and simulation of digital twins; decision support for experiments at DOE scientific user facilities; and for other complex systems and workflows. Inverse problems are at the heart of understanding and controlling complex systems due to factors such as observational data with varying modalities and fidelities, inherent uncertainties in physical measurements and numerical models, and the computational demands of rapid and high-fidelity simulations. The convergence of recent scientific computing trends – scientific machine learning, artificial intelligence, and computing advances such as exascale computing – is creating unprecedented opportunities. These advancements offer the potential to revolutionize how we approach inverse problems to extract actionable insights with the required level of accuracy and computational efficiency. This workshop and the Call for Position Papers are vital steps in bringing together experts to collectively explore and identify the new computational and mathematical directions needed in inverse methods for complex systems under uncertainty.
Invitation
On behalf of the Advanced Scientific Computing Research (ASCR) program in the U.S. Department of Energy (DOE) Office of Science, we are pleased to announce a workshop on the basic research needs for inverse methods for complex systems under uncertainty. We invite community input in the form of two-page position papers that identify key challenges and opportunities for inverse methods for complex systems under uncertainty. These position papers will be used to select workshop participants, define the workshop agenda, and form the basis of a post-workshop report.
The lead author of each selected submission will be invited to participate in the workshop. Authors are not required to have a history of funding by ASCR. By submitting a position paper, authors consent to having their position paper published publicly.
Submission Guidelines
Single or multi-author position papers will be accepted up until 5:00 PM ET April 21 and can be submitted following the instructions on the workshop website. A researcher may only submit one position paper as the corresponding author, but there is no limit on the number of position papers they may co-author. Position papers should aim to improve the community’s shared understanding of the problem space, identify challenging research directions, and help to stimulate discussion. They should not (a) describe the authors’ current or planned research, (b) contain material that should not be disclosed to the public, and (c) recommend specific solutions or discuss narrowly focused research topics.
Position Paper Structure and Format
Position papers should follow the following format:
- Title
- Authors (with affiliations and email addresses)
- Topic: one or more of the following: Optimization algorithms for inverse problems under uncertainty; Probabilistic approaches for solving inverse problems; Inverse problems using incomplete, noisy, or multi-modal data; Uncertainty-aware hybrid modeling for solving inverse problems; Goal-oriented inverse problems; Scalable algorithms for inverse problems.
- Challenge: Identify one or more limitations of current inverse methods when applied to applications of relevance to the DOE.
- Opportunity: Discuss the possibilities for tackling the challenges identified and quantify the impact of effectively addressing these opportunities.
- Innovation: Describe some potential breakthroughs that are essential to facilitate progress that is currently not possible.
- References
- Each position paper must be no more than two pages, in single-column format using 1-inch margins and 11pt or larger font, including figures and references.
Relevant themes and topics
Position papers should be aspirational and advocate for future directions that have the potential to transform inverse methods for complex systems under uncertainty. Successful position papers should draw inspiration from the following themes and topics:
-
Optimization algorithms for inverse problems under uncertainty. New mathematical and computational sciences research aimed at addressing the diverse needs of the various types of inverse problems associated with DOE mission areas. For example, research may address:
a. Multi-objective problems.
b. Mixed-integer and bi-level problems.
c. Randomized and deterministic methods, including derivative-free, inexact gradient, and non-smooth methods.
d. Robust and risk-averse approaches.
e. Computationally efficient AI/ML training.
-
Probabilistic approaches for solving inverse problems. Methods to solve inverse problems that use probabilistic representations to account for the uncertainty in both the data and the model. For example, research could address:
a. Sampling methods for Bayesian inference, e.g., Markov Chain Monte Carlo.
b. Variational inference and generative modeling.
c. Alternative approaches to Bayesian inference.
-
Inverse problems using incomplete, noisy, or multi-modal data. Methods that address the challenges of limited and noisy data while maximizing the information extracted from all available data. For example, research could address:
a. Inverse methods for small/sparse data.
b. Multi-modal or multi-source data fusion.
c. Inverse methods for unlabeled data.
d. Physics- and data-driven regularization/priors.
-
Uncertainty-aware hybrid modeling for solving inverse problems. Methods for synergistically combining multiple models to accelerate the solution of inverse problems while also quantifying and/or ensuring robustness to model errors. For example, research could address:
a. Inverse methods to deal with model-form uncertainties, e.g., arising from incomplete physics and numerical discretizations.
b. Surrogate models, e.g., operator learning, meta networks, to accelerate the solution to inverse problems.
c. Multi-fidelity inverse methods.
-
Goal-oriented inverse problems. Methods that solve inverse problems for specific objectives beyond inferring system parameters. For example, research could address:
a. Methods for conditioning predictions, rather than parameters, on observations.
b. Data assimilation on actionable time scales.
c. Optimal experimental design.
d. Digital Twins for control and decision support.
-
Scalable algorithms for inverse problems. Novel approaches that can address scalability challenges faced when solving inverse problems. For example, research could address:
a. Data analysis across experimental facilities.
b. High-dimensional data and parameter spaces.
c. Large-scale experimental data analysis.
d. Algorithms for emerging architectures, near real-time, or edge computing.
e. Structure-exploiting intrusive methods.
Selection
Submissions will be reviewed by the workshop’s organizing committee based on overall quality, relevance to the themes and topics, potential to stimulate constructive discussion, and ability to contribute to the post-workshop report. Preference will be given to unique perspectives that are well-articulated and highlight potentially transformative research directions.
Workshop Organizers
Co-Chairs
- Jeffrey Donatelli, Lawrence Berkeley National Laboratory
- John Jakeman, Sandia National Laboratories
- Michael Shields, John Hopkins University
Organizing Committee
- Anne Gelb, Dartmouth College
- Felix Hermann, Georgia Institute of Technology
- Sanket Jantre, Brookhaven National Laboratory
- Jeffrey Larson, Argonne National Laboratory
- Juliane Mueller, National Renewable Energy Laboratory
- Assad Oberai, University of Southern California
- Cosmin Petra, Lawrence Livermore National Laboratory
- Brendt Wohlberg, Los Alamos National Laboratory
- Guannan Zhang, Oak Ridge National Laboratory
DOE Program Managers
- Steven Lee, Department of Energy, Advanced Scientific Computing Research
- Bill Spotz, Department of Energy, Advanced Scientific Computing Research
Please direct inquiries about meeting logistics to Todd Munson, Todd.Munson@science.doe.gov
###
|
