Skip to main content
OSTI.GOVU.S. Department of Energy Office of Scientific and Technical Information
U.S. Department of Energy
Office of Scientific and Technical Information
 

aphBO-2GP-3B: a budgeted asynchronous parallel multi-acquisition functions for constrained Bayesian optimization on high-performing computing architecture

Journal Article · · Structural and Multidisciplinary Optimization
 [1];  [2];  [3];  [3];  [3];  [4]
  1. Georgia Institute of Technology, Atlanta, GA (United States); GIW Industries Inc., Grovetown, GA (United States); Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
  2. Xilinx Inc., San Jose, CA (United States)
  3. GIW Industries Inc., Grovetown, GA (United States)
  4. Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
+ Show Author Affiliations

High-fidelity complex engineering simulations are often predictive, but also computationally expensive and often require substantial computational efforts. The mitigation of computational burden is usually enabled through parallelism in high-performance cluster (HPC) architecture. Optimization problems associated with these applications is a challenging problem due to the high computational cost of the high-fidelity simulations. In this paper, an asynchronous parallel constrained Bayesian optimization method is proposed to efficiently solve the computationally expensive simulation-based optimization problems on the HPC platform, with a budgeted computational resource, where the maximum number of simulations is a constant. The advantage of this method are three-fold. Firstly, the efficiency of the Bayesian optimization is improved, where multiple input locations are evaluated parallel in an asynchronous manner to accelerate the optimization convergence with respect to physical runtime. This efficiency feature is further improved so that when each of the inputs is finished, another input is queried without waiting for the whole batch to complete. Second, the proposed method can handle both known and unknown constraints. Third, the proposed method samples several acquisition functions based on their rewards using a modified GP-Hedge scheme. The proposed framework is termed aphBO-2GP-3B, which means asynchronous parallel hedge Bayesian optimization with two Gaussian processes and three batches. The numerical performance of the proposed framework aphBO-2GP-3B is comprehensively benchmarked using 16 numerical examples, compared against other 6 parallel Bayesian optimization variants and 1 parallel Monte Carlo as a baseline, and demonstrated using two real-world high-fidelity expensive industrial applications. The first engineering application is based on finite element analysis (FEA) and the second one is based on computational fluid dynamics (CFD) simulations.

Research Organization:
Sandia National Laboratories (SNL-NM), Albuquerque, NM (United States)
Sponsoring Organization:
USDOE National Nuclear Security Administration (NNSA)
Grant/Contract Number:
NA0003525
OSTI ID:
1834333
Alternate ID(s):
OSTI ID: 1882855
Report Number(s):
SAND2020-3391J; 684908
Journal Information:
Structural and Multidisciplinary Optimization, Journal Name: Structural and Multidisciplinary Optimization Journal Issue: 4 Vol. 65; ISSN 1615-147X
Publisher:
SpringerCopyright Statement
Country of Publication:
United States
Language:
English

References (52)

On bayesian methods for seeking the extremum book January 1975
Fast Computation of the Multi-Points Expected Improvement with Applications in Batch Selection book January 2013
Constrained efficient global optimization with support vector machines journal January 2012
Active Reward Learning conference July 2014
DiceKriging , DiceOptim : Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization journal January 2012
Leveraging the nugget parameter for efficient Gaussian process modeling journal February 2018
Gaussian Processes in Machine Learning book January 2004
Parallel Gaussian Process Optimization with Upper Confidence Bound and Pure Exploration book January 2013
On Bayesian Methods for Seeking the Extremum book January 1975
The Bayesian approach to global optimization book January 2005
A robust optimization approach based on multi-fidelity metamodel journal August 2017
Constrained mixed-integer Gaussian mixture Bayesian optimization and its applications in designing fractal and auxetic metamaterials journal January 2019
Gaussian process optimization with failures: classification and convergence proof journal July 2020
Constitutive equations for hot-working of metals journal January 1985
Blind Kriging: Implementation and performance analysis journal July 2012
pBO-2GP-3B: A batch parallel known/unknown constrained Bayesian optimization with feasibility classification and its applications in computational fluid dynamics journal April 2019
WearGP: A computationally efficient machine learning framework for local erosive wear predictions via nodal Gaussian processes journal March 2019
Efficient Global Optimization of Expensive Black-Box Functions journal January 1998
Random Forests journal January 2001
Least Squares Support Vector Machine Classifiers journal June 1999
Deep learning journal May 2015
Modeling an Augmented Lagrangian for Blackbox Constrained Optimization journal January 2016
On parallel policies for ranking and selection problems journal October 2017
Infill sampling criteria for surrogate-based optimization with constraint handling journal October 2012
Optimization Under Unknown Constraints book October 2011
Support vector machines journal July 1998
Board-Level Reliability of 3D through Glass via Filters During Thermal Cycling conference May 2016
Warpage and Reliability Challenges for Stacked Silicon Interconnect Technology in Large Packages conference May 2018
Budgeted Batch Bayesian Optimization conference December 2016
Taking the Human Out of the Loop: A Review of Bayesian Optimization journal January 2016
Amdahl's Law in the Multicore Era journal July 2008
Use of Birefringence to Determine Redistribution Layer Stresses to Create Design Guidelines to Prevent Glass Cracking journal September 2017
Information-Theoretic Regret Bounds for Gaussian Process Optimization in the Bandit Setting journal May 2012
A New Method of Locating the Maximum Point of an Arbitrary Multipeak Curve in the Presence of Noise journal March 1964
An Efficient First-Principles Saddle Point Searching Method Based on Distributed Kriging Metamodels journal September 2017
Globally Approximate Gaussian Processes for Big Data With Application to Data-Driven Metamaterials Design journal September 2019
sMF-BO-2CoGP: A Sequential Multi-Fidelity Constrained Bayesian Optimization Framework for Design Applications journal April 2020
Scalable Adaptive Batch Sampling in Simulation-Based Design With Heteroscedastic Noise journal December 2020
sBF-BO-2CoGP: A Sequential Bi-Fidelity Constrained Bayesian Optimization for Design Applications
  • Tran, Anh; Wildey, Tim; McCann, Scott
  • ASME 2019 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, Volume 1: 39th Computers and Information in Engineering Conference https://doi.org/10.1115/DETC2019-97986
conference November 2019
Solid-Liquid Flow-Induced Erosion Prediction in Three-Dimensional Pump Casing conference July 2010
The Correlated Knowledge Gradient for Simulation Optimization of Continuous Parameters using Gaussian Process Regression journal July 2011
Multidimensional binary search trees used for associative searching journal September 1975
Completely Derandomized Self-Adaptation in Evolution Strategies journal June 2001
Reducing the Time Complexity of the Derandomized Evolution Strategy with Covariance Matrix Adaptation (CMA-ES) journal March 2003
Constrained Bayesian Optimization with Noisy Experiments journal June 2019
The Knowledge-Gradient Policy for Correlated Normal Beliefs journal November 2009
Stochastic Kriging for Simulation Metamodeling journal April 2010
Active Reward Learning conference July 2014
On the Likelihood that One Unknown Probability Exceeds Another in View of the Evidence of Two Samples journal December 1933
A one-equation turbulence model for aerodynamic flows conference February 2013
Multi-class AdaBoost journal January 2009
Modeling an Augmented Lagrangian for Blackbox Constrained Optimization collection January 2015

Figures / Tables (12)

Figure 1(p. 13)figure Figure 1
Figure 2(p. 16)figure Figure 2
Table 1(p. 17)table Table 1
Figure 3(p. 18)figure Figure 3
Table 2(p. 18)table Table 2
Figure 4(p. 19)figure Figure 4
Figure 5(p. 20)figure Figure 5
Figure 6(p. 20)figure Figure 6
Figure 7(p. 22)figure Figure 7
Figure 8(p. 22)figure Figure 8
Figure 10(p. 23)figure Figure 10
Figure 9(p. 23)figure Figure 9

Similar Records

Related Subjects

97 MATHEMATICS AND COMPUTING
Bayesian optimization
Gaussian process
asynchronously parallel
known/unknown constrained
multi-acquisition