New paper: 28TB time-resolved DNS shows turbulence must be treated as 4D phenomenon
jo_brandstetter · x · 2026-10-26
Johannes Brandstetter's team published "Let's Do Turbulence Right: In Space and Time!" (arXiv:2610.32601), using massive time-resolved direct numerical simulations exceeding 28TB per run to assess how numerical choices (grid resolution, arithmetic precision, filtering, rounding) govern the resolved (3+1)-dimensional physics of turbulent flows, and how neural networks represent the recorded motion. Key finding: spatial and temporal assessments can disagree — a record can look convincing in space while its motion changes too quickly in time. Since ML models train on simulated flows, how those flows are computed and recorded shapes what models learn. The paper calls on the ML community to treat (3+1)-dimensional turbulence as a four-dimensional phenomenon.
More from Research
- First-order world-model gradients give minor training speedups over zero-order PG — breadli428 · 2026-10-26
- TACL editor Ehud Reiter: don't submit LLM-written papers, human review will reject them — EhudReiter · 2026-10-26
- DeepGHS previews AnimeCSIP-v3, an artist-style recognition model built on AnimeDINOv3 — bdsqlsz · 2026-10-26
- LeCun co-authored Temporal Straightening: straightening latent paths boosts gradient-based planning success rates — ylecun · 2026-10-26
- Alibaba's four-generation TTS evolution: CosyVoice to Qwen-Audio-3.0-TTS — solyarisoftware · 2026-10-26
- LoopUno: 52-Minute LoRA Training Makes 3B LLM Draft 2 Tokens per Step Losslessly — LucaAmb · 2026-10-26