Differences

This shows you the differences between two versions of the page.

--- highlights:2026:mpc_fno [2026/04/09 13:56] – Didier Barradas Bautista
+++ highlights:2026:mpc_fno [2026/04/12 06:59] (current) – Didier Barradas Bautista
@@ Line 10: / Line 10: @@
 <WRAP center>
-{{:wiki:highlights:2025:fno_mpc_01.gif?600|}}
+{{:wiki:highlights:2025:fno_mpc_01.gif?nolink }}
 </WRAP>
@@ Line 50: / Line 50: @@
-<WRAP notice>
+<WRAP>
-<fs 18>**🔬 Research Visualization**</fs>
+<fs 18>**🔬 Graphical Representation of Computational Acceleration**</fs>
-The graphical abstract below illustrates the iterative filtering strategy used to enrich correct protein–protein docking models. The paper entitled "Increasing the Fraction of Correct Solutions in Ensembles of Protein-Protein Docking Models by an Iterative Consensus Algorithm" is available for download and review. [[https://onlinelibrary.wiley.com/doi/pdf/10.1002/pro.70314|here]].
+ The acceleration plot highlights why surrogate-based control is practical for real-time multiphase systems: by replacing repeated CFD evaluations with a trained Fourier Neural Operator, the control pipeline achieves much lower latency while preserving predictive quality. In our benchmarks, the optimized training workflow (GPU-resident preloading plus CuPy preprocessing) reached up to 3.09x speedup per epoch, and inference throughput reached about 19,711 to 24,508 samples per second (0.041 to 0.051 ms per sample). This speed gain is critical because MPC requires many horizon evaluations at each control step, and faster model rollouts directly translate into faster optimization and more responsive closed-loop control. You can access the paper[[https://doi.org/10.48550/arXiv.2603.25308|here]].
-<WRAP info>
 <WRAP center>
 {{:wiki:highlights:2025:fno_mpc_02.png|}}