Active Learning for Saddle Point Calculation

The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state (TS). The traditional methods need to evaluate the gradients of the energy function at a very large number o...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Gu, Shuting, Wang, Hongqiao, Zhou, Xiang
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Gu, Shuting
Wang, Hongqiao
Zhou, Xiang
description The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state (TS). The traditional methods need to evaluate the gradients of the energy function at a very large number of locations. To reduce the number of expensive computations of the true gradients, we propose an active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states. SP is detected by the GAD applied to the GPR surrogate for the gradient vector and the Hessian matrix. Our key ingredient for efficiency improvements is an active learning method which sequentially designs the most informative locations and takes evaluations of the original model at these locations to train GPR. We formulate this active learning task as the optimal experimental design problem and propose a very efficient sample-based sub-optimal criterion to construct the optimal locations. We show that the new method significantly decreases the required number of energy or force evaluations of the original model.
doi_str_mv 10.48550/arxiv.2108.04698
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2108_04698</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2108_04698</sourcerecordid><originalsourceid>FETCH-LOGICAL-a678-12b53ebeca428ef07ae853104c855a284a298ca3cd2709f12f8e318474f447f33</originalsourceid><addsrcrecordid>eNotzs1qwkAUQOHZuCjaB-jKeQCTzs-Nc7OUYG0hoKD7cJ3ckYGYyDQV-_a2tquzO3xCvGiVAxaFeqV0i9fcaIW5gmWJT2Kx8mO8sqyZUh_7kwxDkntq247lboj9KCvq_FdHYxz6mZgE6j75-b9TcXhbH6r3rN5uPqpVndHSYabNsbB8ZE9gkINyxFhYrcD_GMggkCnRk_WtcaoM2gRkqxEcBAAXrJ2K-d_2wW0uKZ4pfTe_7ObBtne3mTuf</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Active Learning for Saddle Point Calculation</title><source>arXiv.org</source><creator>Gu, Shuting ; Wang, Hongqiao ; Zhou, Xiang</creator><creatorcontrib>Gu, Shuting ; Wang, Hongqiao ; Zhou, Xiang</creatorcontrib><description>The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state (TS). The traditional methods need to evaluate the gradients of the energy function at a very large number of locations. To reduce the number of expensive computations of the true gradients, we propose an active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states. SP is detected by the GAD applied to the GPR surrogate for the gradient vector and the Hessian matrix. Our key ingredient for efficiency improvements is an active learning method which sequentially designs the most informative locations and takes evaluations of the original model at these locations to train GPR. We formulate this active learning task as the optimal experimental design problem and propose a very efficient sample-based sub-optimal criterion to construct the optimal locations. We show that the new method significantly decreases the required number of energy or force evaluations of the original model.</description><identifier>DOI: 10.48550/arxiv.2108.04698</identifier><language>eng</language><subject>Computer Science - Learning ; Statistics - Machine Learning</subject><creationdate>2021-08</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2108.04698$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2108.04698$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Gu, Shuting</creatorcontrib><creatorcontrib>Wang, Hongqiao</creatorcontrib><creatorcontrib>Zhou, Xiang</creatorcontrib><title>Active Learning for Saddle Point Calculation</title><description>The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state (TS). The traditional methods need to evaluate the gradients of the energy function at a very large number of locations. To reduce the number of expensive computations of the true gradients, we propose an active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states. SP is detected by the GAD applied to the GPR surrogate for the gradient vector and the Hessian matrix. Our key ingredient for efficiency improvements is an active learning method which sequentially designs the most informative locations and takes evaluations of the original model at these locations to train GPR. We formulate this active learning task as the optimal experimental design problem and propose a very efficient sample-based sub-optimal criterion to construct the optimal locations. We show that the new method significantly decreases the required number of energy or force evaluations of the original model.</description><subject>Computer Science - Learning</subject><subject>Statistics - Machine Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotzs1qwkAUQOHZuCjaB-jKeQCTzs-Nc7OUYG0hoKD7cJ3ckYGYyDQV-_a2tquzO3xCvGiVAxaFeqV0i9fcaIW5gmWJT2Kx8mO8sqyZUh_7kwxDkntq247lboj9KCvq_FdHYxz6mZgE6j75-b9TcXhbH6r3rN5uPqpVndHSYabNsbB8ZE9gkINyxFhYrcD_GMggkCnRk_WtcaoM2gRkqxEcBAAXrJ2K-d_2wW0uKZ4pfTe_7ObBtne3mTuf</recordid><startdate>20210810</startdate><enddate>20210810</enddate><creator>Gu, Shuting</creator><creator>Wang, Hongqiao</creator><creator>Zhou, Xiang</creator><scope>AKY</scope><scope>EPD</scope><scope>GOX</scope></search><sort><creationdate>20210810</creationdate><title>Active Learning for Saddle Point Calculation</title><author>Gu, Shuting ; Wang, Hongqiao ; Zhou, Xiang</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a678-12b53ebeca428ef07ae853104c855a284a298ca3cd2709f12f8e318474f447f33</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Computer Science - Learning</topic><topic>Statistics - Machine Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Gu, Shuting</creatorcontrib><creatorcontrib>Wang, Hongqiao</creatorcontrib><creatorcontrib>Zhou, Xiang</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv Statistics</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Gu, Shuting</au><au>Wang, Hongqiao</au><au>Zhou, Xiang</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Active Learning for Saddle Point Calculation</atitle><date>2021-08-10</date><risdate>2021</risdate><abstract>The saddle point (SP) calculation is a grand challenge for computationally intensive energy function in computational chemistry area, where the saddle point may represent the transition state (TS). The traditional methods need to evaluate the gradients of the energy function at a very large number of locations. To reduce the number of expensive computations of the true gradients, we propose an active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states. SP is detected by the GAD applied to the GPR surrogate for the gradient vector and the Hessian matrix. Our key ingredient for efficiency improvements is an active learning method which sequentially designs the most informative locations and takes evaluations of the original model at these locations to train GPR. We formulate this active learning task as the optimal experimental design problem and propose a very efficient sample-based sub-optimal criterion to construct the optimal locations. We show that the new method significantly decreases the required number of energy or force evaluations of the original model.</abstract><doi>10.48550/arxiv.2108.04698</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2108.04698
ispartof
issn
language eng
recordid cdi_arxiv_primary_2108_04698
source arXiv.org
subjects Computer Science - Learning
Statistics - Machine Learning
title Active Learning for Saddle Point Calculation
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-03T20%3A27%3A26IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Active%20Learning%20for%20Saddle%20Point%20Calculation&rft.au=Gu,%20Shuting&rft.date=2021-08-10&rft_id=info:doi/10.48550/arxiv.2108.04698&rft_dat=%3Carxiv_GOX%3E2108_04698%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true