AI Paper Atlas

Research map / Time series / Causal and generative time series

PINFDiT: Energy-Based Physics-Informed Diffusion Transformers for General-purpose Time Series Tasks

Defu Cao, Wen Ye, Yizhou Zhang, Sam Griesemer, Yan Liu

ICLR 2026 · Time series › Causal and generative time series · approach: Diffusion models for time series

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add physics-guided Langevin injection (ablation)PINFDiT · AdvectionRMSE: 0.0052 -> 0.0039Improved
add physics-guided Langevin injection (ablation)PINFDiT · BurgersRMSE: 0.0136 -> 0.0133Improved
add physics-guided Langevin injection (ablation)PINFDiT · ERA5 2m temperature (6h)ACC: 0.985 -> 0.987Improved
add physics-guided Langevin injection (ablation)PINFDiT · Solar zero-shot forecastingCRPSsum: 0.445 -> 0.424Improved
add block masking (ablation)PINFDiT · Solar zero-shot forecastingCRPSsum: 0.862 -> 0.424Improved
add random masking (ablation)PINFDiT · Solar zero-shot forecastingCRPSsum: 0.465 -> 0.424Improved
add stride masking (ablation)PINFDiT · Electricity zero-shot forecastingCRPSsum: 0.037 -> 0.030Improved
use direct embedding instead of patch tokens (ablation)PINFDiT · Solar zero-shot forecastingCRPSsum: 0.874 -> 0.424Improved
use AdaLN conditioning instead of feature addition (ablation)PINFDiT · Solar zero-shot forecastingCRPSsum: 0.677 -> 0.424Improved
use AdaLN conditioning instead of cross attention (ablation)PINFDiT · Electricity zero-shot forecastingCRPSsum: 0.077 -> 0.030Improved
inject incorrect Burgers PDE instead of correct Advection PDE (ablation)PINFDiT · AdvectionRMSE: 0.0052 -> 0.0061Hurt
apply fluid physics constraints to non-physical data (Burgers PDE) (ablation)PINFDiT · NASDAQMAE: 0.516 -> 0.537Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy