Differences

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

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
highlights:2026:mpc_fno [2026/04/09 13:37] – Didier Barradas Bautistahighlights:2026:mpc_fno [2026/04/12 06:59] (current) – Didier Barradas Bautista
Line 10: Line 10:
  
 <WRAP center> <WRAP center>
-{{:wiki:highlights:2026:PPI_IterCR_v1.png?600|}}+{{:wiki:highlights:2025:fno_mpc_01.gif?nolink }}
 </WRAP> </WRAP>
  
Line 21: Line 21:
  
 <WRAP center round box 100%> {{:wiki:kaust-seeds.png?nolink&0x30}} <fs:26px>** KAUST PI on Project**</fs> <WRAP center round box 100%> {{:wiki:kaust-seeds.png?nolink&0x30}} <fs:26px>** KAUST PI on Project**</fs>
 +
 +
 +
 +{{:icon:twbs:link:w-25:h-auto:person.svg?nolink&}} William Roberts \\
 +{{:icon:twbs:link:w-25:h-auto:envelope-at.svg?nolink&}} william.roberts@kaust.edu.sa \\
 +
 +</WRAP>
 +
 +
 +<WRAP center round box 100%> {{:wiki:kaust-seeds.png?nolink&0x30}} <fs:26px>** KAUST RS on Project**</fs>
  
 {{:icon:twbs:link:w-25:h-auto:person.svg?nolink&}} Paolo Guida  \\ {{:icon:twbs:link:w-25:h-auto:person.svg?nolink&}} Paolo Guida  \\
Line 33: Line 43:
 ===== Research Breakthrough ===== ===== Research Breakthrough =====
 <WRAP> <WRAP>
-Protein–protein docking generates thousands of possible interaction models—but only a few are correct. In our latest collaborative work, we introduce Iter-CONSRANK, an iterative scoring algorithm that filters out incorrect models and enriches the ensemble with correct ones. +    * Bridging the Real-Time Gap: Developed an AI framework that replaces slow, traditional physics simulations with Fourier Neural Operators, enabling the first real-time control of complex multiphase flows. 
-  * 📊 Tested on two challenging datasets, Iter-CONSRANK increased the fraction of correct models by up to 8× for medium-difficulty targets and outperformed over 150 scoring functions in ranking accuracy. +    * Precision AI Decision-Making: Integrated Bayesian Optimization with neural surrogates to precisely track liquid levels in bubble columns, overcoming the mathematical "noise" that typically breaks standard control algorithms. 
-  * 🎯 The method is available for use in pre-processing docking ensembles or as an independent scoring tool. +    * Accelerating Industrial Innovation: Reduced the computational cost of multiphase flow management by orders of magnitude, providing a scalable foundation for smarter, more efficient chemical reactors and energy systems.
-  * 🔬 This work was led in collaboration with Luigi Cavallo and our partners at the University of Naples “Parthenope”+
 </WRAP>  </WRAP> 
      
  
  
-<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> <WRAP center>
-{{:wiki:highlights:2026:Graphical_abstract.gif|}}+{{:wiki:highlights:2025:fno_mpc_02.png|}}
 </WRAP> </WRAP>
  
Line 53: Line 61:
 </WRAP> /* end column */ </WRAP> /* end column */
 <fs 18>**🎯 KVL's Contribution:**</fs> <fs 18>**🎯 KVL's Contribution:**</fs>
-KVL's visualization scientist **Didier Barradas-Bautista** contributed significantly to the development and validation of the Iter-CONSRANK algorithm through:+KVL's visualization scientist **Didier Barradas-Bautista** contributed significantly to the implementation of algorithm through:
  
-  * Design and implementation of contact-based clustering to enhance scoring accuracy +  * Updating the code to efficientiyl use the GPU memory 
-  * Generation of the 3K-BM5up benchmark dataset using multiple docking tools +  * refactoring the code to output efficientyl the results 
-  * Optimization of iteration parameters and performance evaluation across difficulty categories +
-  * Comparative benchmarking against 157 scoring functions, demonstrating top performance +
-  * Public release of the Iter-CONSRANK software for community use +
-  * Active participation in CAPRI scoring rounds, validating the method in blind tests+
  
 </WRAP> /* end group */ </WRAP> /* end group */