The Polynomial Iteration Complexity for Variance Exploding Diffusion Models: Elucidating SDE and ODE Samplers

Ruofeng Yang, Bo Jiang, Shuai Li
Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, PMLR 258:730-738, 2025.

Abstract

Recently, variance exploding (VE) diffusion models have achieved state-of-the-art (SOTA) performance in two implementations: (1) the SDE-based implementation and (2) the probability flow ODE (PFODE) implementation. However, only a few works analyze the iteration complexity of VE-based models, and most focus on SDE-based implementation with strong assumptions. In this work, we prove the first polynomial iteration complexity under the realistic bounded support assumption for these two implementations. For the SDE-based implementation, we explain why the current SOTA VE-based model performs better than previous VE models. After that, we provide an improved result under the linear subspace data assumption and explain the great performance of VE models under the manifold data. For the PFODE-based implementation, the current results depend exponentially on problem parameters. Inspired by the previous predictor-corrector analysis framework, we propose the PFODE-Corrector algorithm and prove the polynomial complexity for the basic algorithm with uniform stepsize. After that, we show that VE-based models are more suitable for large stepsize and propose an exponential-decay stepsize version algorithm to improve the results.

Cite this Paper

BibTeX
@InProceedings{pmlr-v258-yang25a, title = {The Polynomial Iteration Complexity for Variance Exploding Diffusion Models: Elucidating SDE and ODE Samplers}, author = {Yang, Ruofeng and Jiang, Bo and Li, Shuai}, booktitle = {Proceedings of The 28th International Conference on Artificial Intelligence and Statistics}, pages = {730--738}, year = {2025}, editor = {Li, Yingzhen and Mandt, Stephan and Agrawal, Shipra and Khan, Emtiyaz}, volume = {258}, series = {Proceedings of Machine Learning Research}, month = {03--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v258/main/assets/yang25a/yang25a.pdf}, url = {https://proceedings.mlr.press/v258/yang25a.html}, abstract = {Recently, variance exploding (VE) diffusion models have achieved state-of-the-art (SOTA) performance in two implementations: (1) the SDE-based implementation and (2) the probability flow ODE (PFODE) implementation. However, only a few works analyze the iteration complexity of VE-based models, and most focus on SDE-based implementation with strong assumptions. In this work, we prove the first polynomial iteration complexity under the realistic bounded support assumption for these two implementations. For the SDE-based implementation, we explain why the current SOTA VE-based model performs better than previous VE models. After that, we provide an improved result under the linear subspace data assumption and explain the great performance of VE models under the manifold data. For the PFODE-based implementation, the current results depend exponentially on problem parameters. Inspired by the previous predictor-corrector analysis framework, we propose the PFODE-Corrector algorithm and prove the polynomial complexity for the basic algorithm with uniform stepsize. After that, we show that VE-based models are more suitable for large stepsize and propose an exponential-decay stepsize version algorithm to improve the results.} }
Endnote
%0 Conference Paper %T The Polynomial Iteration Complexity for Variance Exploding Diffusion Models: Elucidating SDE and ODE Samplers %A Ruofeng Yang %A Bo Jiang %A Shuai Li %B Proceedings of The 28th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2025 %E Yingzhen Li %E Stephan Mandt %E Shipra Agrawal %E Emtiyaz Khan %F pmlr-v258-yang25a %I PMLR %P 730--738 %U https://proceedings.mlr.press/v258/yang25a.html %V 258 %X Recently, variance exploding (VE) diffusion models have achieved state-of-the-art (SOTA) performance in two implementations: (1) the SDE-based implementation and (2) the probability flow ODE (PFODE) implementation. However, only a few works analyze the iteration complexity of VE-based models, and most focus on SDE-based implementation with strong assumptions. In this work, we prove the first polynomial iteration complexity under the realistic bounded support assumption for these two implementations. For the SDE-based implementation, we explain why the current SOTA VE-based model performs better than previous VE models. After that, we provide an improved result under the linear subspace data assumption and explain the great performance of VE models under the manifold data. For the PFODE-based implementation, the current results depend exponentially on problem parameters. Inspired by the previous predictor-corrector analysis framework, we propose the PFODE-Corrector algorithm and prove the polynomial complexity for the basic algorithm with uniform stepsize. After that, we show that VE-based models are more suitable for large stepsize and propose an exponential-decay stepsize version algorithm to improve the results.
APA
Yang, R., Jiang, B. & Li, S.. (2025). The Polynomial Iteration Complexity for Variance Exploding Diffusion Models: Elucidating SDE and ODE Samplers. Proceedings of The 28th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 258:730-738 Available from https://proceedings.mlr.press/v258/yang25a.html.

Related Material