Irregularly Sampled Time Series Interpolation for Binary Evolution Simulations Using Dynamic Time Warping
Irregularly Sampled Time Series Interpolation for Binary Evolution Simulations Using Dynamic Time Warping
Authors: Ugur Demir, Philipp M. Srivastava, Aggelos Katsaggelos, Vicky Kalogera, S. Tapia, Manuel Ballester, Shamal Lalvani, Patrick Koller, Jeff J. Andrews, Seth Gossage, Max M. Briel, Elizabeth Teng Date: 2026-04-15 Paper ID: openalex:2604.13604
Summary
Binary stellar evolution simulations are computationally expensive, necessitating efficient interpolation for stellar population synthesis. This paper introduces a novel approach for interpolating irregularly sampled binary stellar tracks using Dynamic Time Warping to perform track alignment and iterative averaging. By computing a shared warping path across physical parameters, the method preserves causal relationships and essential physical laws. The proposed technique significantly outperforms traditional interpolation methods, offering a more efficient way to generate accurate binary population samples for astrophysical research.
Key Contributions
- Proposes a novel track alignment and iterative averaging method using Dynamic Time Warping to handle irregularities and discontinuities in binary evolution simulation data.
- Enables the generation of accurate binary population samples by computing a shared warping path that maintains essential physical relationships like the Stefan-Boltzmann law.
- Demonstrates consistent performance improvements over traditional linear interpolation baselines across multiple binary configuration datasets.
Open Questions & Future Work
Key Concepts
- dtw-multi-parameter-alignment: A method for aligning irregularly sampled, multi-parameter time series by computing a shared warping path across all variables to preserve internal correlations and physical laws during averaging.
Archivist Review
The approved concept generalizes the alignment technique to emphasize its utility across multi-parameter physical systems rather than just the stellar evolution domain. The approved open question addresses the structural limitation of static neighbor selection in interpolation tasks, which is a common bottleneck in scientific machine learning for physical emulators.
Approved Concepts
- Dynamic Time Warping for Multi-Parameter Alignment: It provides a principled way to interpolate across high-dimensional, irregularly sampled trajectories where temporal consistency is critical to preserving causal physics, offering a robust alternative to standard time-agnostic interpolation.
Approved Open Questions
- Morphology-Aware Neighbor Selection: This is a fundamental limitation in surrogate modeling and emulation for expensive scientific simulations, where sparse data often forces the inclusion of unrepresentative ‘neighboring’ samples.
Rejected Candidates
- [concept] Dynamic Time Warping for Stellar Track Alignment (
dynamic-time-warping-stellar-track-alignment) - other: Renamed and generalized to focus on the multi-parameter alignment mechanism rather than the specific application (stellar evolution).
Links
Metadata & Links
- url
- https://arxiv.org/abs/2604.13604
- paper_id
- 2604.13604
- paper_source
- openalex
- domain
- time-series
- tags
-
- architectures
-
- datasets
-
- concept_slugs
- dtw-multi-parameter-alignment
- dataset_slugs
-
- skill
- TimeSeriesSkill
- created_at
- 2026-04-18T05:34:41Z