Research – Uncertainty Quantification – Korali

@misc{martin2021koral
title ={Korali: Efficient and Scalable Software Framework for Bayesian Uncertainty Quantification and Stochastic Optimization},
author = {Sergio M. Martin and Daniel Wälchli and Georgios Arampatzis and Athena E. Economides and Petr Karnakov and Petros Koumoutsakos},
year = {2021},
eprint = {2005.13457},
archivePrefix ={arXiv},
primaryClass ={cs.DC}
}

We present Korali, an open-source framework for large-scale Bayesian uncertainty quantification and stochastic optimization. The framework relies on non-intrusive sampling of complex multiphysics models and enables their exploitation for optimization and decision-making. In addition, its distributed sampling engine makes efficient use of massively-parallel architectures while introducing novel fault tolerance and load balancing mechanisms. We demonstrate these features by interfacing Korali with existing high-performance software such as Aphros, Lammps (CPU-based), and Mirheo (GPU-based) and show efficient scaling for up to 512 nodes of the CSCS Piz Daint supercomputer. Finally, we present benchmarks demonstrating that Korali outperforms related state-of-the-art software frameworks.

@inproceedings{10.1145/3394277.3401849,
author = {W\”{a}lchli, Daniel and Martin, Sergio M. and Economides, Athena and Amoudruz, Lucas and Arampatzis, George and Bian, Xin and Koumoutsakos, Petros},
title = {Load Balancing in Large Scale Bayesian Inference},
year = {2020},
isbn = {9781450379939},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3394277.3401849},
doi = {10.1145/3394277.3401849},
booktitle = {Proceedings of the Platform for Advanced Scientific Computing Conference},
articleno = {5},
numpages = {12},
keywords = {High Performance Computing, Erythrocyte Membrane Viscosity, Uncertainty Quantification,
Bayesian Inference, Load Balancing},
location = {Geneva, Switzerland},
series = {PASC ’20
}

We present a novel strategy to improve load balancing for large scale Bayesian inference problems. Load imbalance can be particularly destructive in generation based uncertainty quantification (UQ) methods since all compute nodes in a large-scale allocation have to synchronize after every generation and therefore remain in an idle state until the longest model evaluation finishes. Our strategy relies on the concurrent scheduling of independent Bayesian inference experiments while sharing a group of worker nodes, reducing the destructive effects of workload imbalance in population-based sampling methods.
To demonstrate the efficiency of our method, we infer parameters of a red blood cell (RBC) model. We perform a data-driven calibration of the RBC’s membrane viscosity by applying hierarchical Bayesian inference methods. To this end, we employ a computational model to simulate the relaxation of an initially stretched RBC towards its equilibrium state. The results of this work advance upon the current state of the art towards realistic blood flow simulations by providing inferred parameters for the RBC membrane viscosity.
We show that our strategy achieves a notable reduction in imbalance and significantly improves effective node usage on 512 nodes of the CSCS Piz Daint supercomputer. Our results show that, by enabling multiple independent sampling experiments to run concurrently on a given allocation of supercomputer nodes, our method sustains a high computational efficiency on a large-scale supercomputing setting.

@inproceedings{arampatzis2019,
author = {Arampatzis, Georgios and W\”{a}lchli, Daniel and Weber, Pascal and R\”{a}stas, Henri and Koumoutsakos, Petros},
booktitle = {Proceedings of the Platform for Advanced Scientific Computing Conference on – {PASC} {\textquotesingle}19},
doi = {10.1145/3324989.3325725},
keywords = {Stochastic optimization, constraint handling, covariance matrix adaptation, evolution strategy, pharmacodynamics, viability evolution},
publisher = {{ACM} Press},
title = {(\Μ,{\Lambda})-CCMA-ES for Constrained Optimization with an Application in Pharmacodynamics},
url = {https://drive.google.com/file/d/1gpcJ0z-9SZTMTjon8O6Kkgj1gfBXINdU/view?usp=drive_link},
year = {2019}
}

We present the algorithm CCMA-ES, an extension to CMA-ES, an evolution strategy that has shown to perform well in a broad range of black-box optimization problems. The (μ, λ)-CMA-ES effectively handles nonlinear nonconvex functions but faces difficulties in constrained optimization problems. We introduce viability boundaries to improve the search for an initial point in the valid domain and adapt the covariance matrix using normal approximations to maintain the inequality constraints. Using benchmark problems from 2006 CEC we compare the performance of CCMA-ES with a state of the art optimization algorithm (mViE) showing favorable results. Finally, CCMA-ES is applied to a pharmacodynamics problem describing tumor growth, and we demonstrate that CCMA-ES outperforms mViE in terms of the objective function value and total function evaluations.

@article {Karnakov2020.20313,
author = {Karnakov, Petr and Arampatzis, Georgios and Ki{\v c}i{\’c}, Ivica and Wermelinger, Fabian and W{\”a}lchli, Daniel and Papadimitriou, Costas and Koumoutsakos, Petros},
title = {Data-driven inference of the reproduction number for COVID-19 before and after interventions for 51 European countries},
year = {2020},
doi = {10.4414/smw.2020.20313},
publisher = {Swiss Medical Weekly},
URL = {https://smw.ch/index.php/smw/article/view/2831},
journal = {Swiss Medical Weekly}}
}

The reproduction number is broadly considered as a key
indicator for the spreading of the COVID-19 pandemic. Its
estimated value is a measure of the necessity and, eventually, effectiveness of interventions imposed in various
countries. Here we present an online tool for the data-driven inference and quantification of uncertainties for the
reproduction number, as well as the time points of interventions for 51 European countries. The study relied on
the Bayesian calibration of the SIR model with data from
reported daily infections from these countries. The model fitted the data, for most countries, without individual
tuning of parameters. We also compared the results of
SIR and SEIR models, which give different estimates of
the reproduction number, and provided an analytical relationship between the respective numbers. We deployed
a Bayesian inference framework with efficient sampling
algorithms, to present a publicly available graphical user
interface (https://cse-lab.ethz.ch/coronavirus) that allows
the user to assess and compare predictions for pairs of
European countries. The results quantified the rate of the
disease’s spread before and after interventions, and provided a metric for the effectiveness of non-pharmaceutical
interventions in different countries. They also indicated
how geographic proximity and the times of interventions
affected the progression of the epidemic.

@article {Chatzimanolakis2020.07.20.20157818,
author = {Chatzimanolakis, Michail and Weber, Pascal and Arampatzis, Georgios and W{\”a}lchli, Daniel and Karnakov, Petr and Ki{\v c}i{\’c}, Ivica and Papadimitriou, Costas and Koumoutsakos, Petros},
title = {Optimal Testing Strategy for the Identification of COVID-19 Infections},
elocation-id = {2020.07.20.20157818},
year = {2020},
doi = {10.1101/2020.07.20.20157818},
publisher = {Cold Spring Harbor Laboratory Press},
URL = {https://www.medrxiv.org/content/early/2020/07/27/2020.07.20.20157818},
eprint = {https://www.medrxiv.org/content/early/2020/07/27/2020.07.20.20157818.ful…},
journal = {medRxiv}
}

The systematic identification of infectious, yet unreported, individuals is critical for the containment of the COVID-19 pandemic. We present a strategy for identifying the location, timing and extent of testing that maximizes information gain for such infections. The optimal testing strategy relies on Bayesian experimental design and forecasting epidemic models that account for time dependent interventions. It is applicable at the onset and spreading of the epidemic and can forewarn for a possible recurrence of the disease after relaxation of interventions. We examine its application in Switzerland and show that it can provide timely and systematic guidance for the effective identification of infectious individuals with finite testing resources. The methodology and the open source code are readily adaptable to countries around the world.