Project Description
The Tachyon Resilient Modeling project develops the theory, algorithms, and software needed to make large-scale computational models resilient — able to produce trustworthy predictions even when inputs are noisy, components fail, or the system being modeled drifts away from its assumptions.
Modern scientific and engineering models are increasingly assembled from many coupled components running across heterogeneous, failure-prone computing environments. Our work spans uncertainty quantification, fault-tolerant numerical methods, and reproducible modeling workflows so that results remain valid and explainable as scale and complexity grow. We release the resulting tools, datasets, and source code openly so the broader research community can reproduce and build on our findings.
This site collects the project's people, publications, and released research artifacts. Content marked as a placeholder should be replaced with the project's specific details before public launch.
Research Focus Areas
Uncertainty Quantification
Rigorous, scalable propagation of uncertainty through coupled models so predictions come with honest, calibrated error bars.
Fault-Tolerant Methods
Numerical algorithms and workflows that detect, absorb, and recover from hardware, data, and component failures at scale.
Reproducible Workflows
Provenance-tracked, portable modeling pipelines that produce the same results across platforms and over time.
Adaptive Modeling
Models that self-monitor and adjust fidelity as conditions and data availability change during a run.
Open Benchmarks
Curated datasets and reference problems for evaluating resilience and reproducibility across the community.
High-Performance Computing
Implementations tuned for HPC and heterogeneous accelerators with an emphasis on scalability and portability.