Comparison of Computational Speed between Landau-Lif-shitz- Gilbert and Hybrid-Monte-Carlo Micromagnetics

The Landau-Lifshitz-Gilbert (LLG) equation is widely utilized to analyze the magnetic properties in micromagnetics [1]. However, the calculation of micromagnetics with Landau-Lifshitz-Gilbert equations is time-consuming, so the FDM-FFT method was brought up to speed up the simulation in large scale...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2018 IEEE International Magnetics Conference (INTERMAG) s. 1
Hlavní autoři: Song, J., Zhao, Z., Wei, D.
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 01.04.2018
Témata:
ISSN:2150-4601
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract The Landau-Lifshitz-Gilbert (LLG) equation is widely utilized to analyze the magnetic properties in micromagnetics [1]. However, the calculation of micromagnetics with Landau-Lifshitz-Gilbert equations is time-consuming, so the FDM-FFT method was brought up to speed up the simulation in large scale [2]. Recently, we have developed a new micromagnetic method based on the Hybrid Monte Carlo (HMC) algorithm, which can calculate M-H loops and domains at finite temperature below Curie point [3]. Some simulation results have been put forward to confirm the validity of the HMC algorithm [4]-[6]. Furthermore, we also wish to focus on the computational efficiency to compare HMC micromagnetics with traditional LLG micromagnetics. Comparison of computational speed has been made of these two methods in various micromagnetic scales in this work. In LLG or HMC micromagnetics, a regular mesh with cell size 5 \times 5 \times 5 nm 3 is utilized. Other important parameters are: anisotropy energy constant K(0 \mathrm {K}) = 1 \times 10 ^{5}\mathrm {J}/ \mathrm {m}^{3}, saturation magnetization at zero temperatur e M_{s}(0 \mathrm {K}) = 7.98 \times 10 ^{5}\mathrm {A} /\mathrm {m}, exchange constant A^{\ast }= 1.0 \times 10 ^{-11}\mathrm {J} /\text{m}. In LLG micromagnetics, the iterations can go endless unless the spin error \vert \Delta\vert ( the spin maximum deviation between two simulation steps) is set to be a limited number. To ensure the stability and accuracy, \vert \Delta\vert is set as 10 ^{-6} in LLG algorithm. In HMC micromagnetics, the Monte Carlo time t is not real time. Based on the feature of the leap-frog algorithm used for iteration of the Hamilton equations, in HMC micromagnetics, the spin error \vert \Delta\vert stands for the spin deviation of two adjacent trajectories. To compare the computational speed of the LLG and the HMC micromagnetic methods, the spin error \vert \Delta\vert in HMC MuMag is also set as 10 ^{-6}. The M-H loops are same under these two micromagnetic algorithms. Fig. 1 shows the computational time to reach \vert \Delta \vert \sim 10 ^{-6} at each external magnetic field (\mathrm {H}_{ext}), with 16*8*16 micromagnetic cells. The computational time for Hybrid Monte Carlo MuMag keeps around 59 seconds for different \mathrm {H}_{ext}; while for Laudau-Lifshitz-Gilbert MuMag, the simulation time is typically several hundred of seconds and shows a sharp rising near coercivity to about 1904 seconds, which is different from HMC MuMag. In Fig. 2, the computational speed of the two micromagnetic methods is presented. The horizontal axis is {log}_{2}(N_{cell}), and the vertical axis is the total computation time for a loop in Fig. 1 (in seconds). The blue line is the speed of the LLG MuMag. In the Ref. 7, the LLG algorithm has a running time proportional to the number of particles N, which yields {O(Nlog}_{2}N) computational time. We have compared the computational time of LLG algorithm with {O(Nlog}_{2}N). The triangle-dot line is the computation time vs. {log}_{2}(N_{cell}), which shows great agreement with {O(Nlog}_{2}N)( black line in Fig. 2). The red dotted line shows the computation time consumed by the HMC algorithm, which show great coincidence with O(N) in the insert figure. From the comparison, the HMC algorithm has a significant advantage for the computational speed. When the number of micromagnetic cells rise to 32*32*32, LLG algorithm spends 16 times more than HMC algorithm.
AbstractList The Landau-Lifshitz-Gilbert (LLG) equation is widely utilized to analyze the magnetic properties in micromagnetics [1]. However, the calculation of micromagnetics with Landau-Lifshitz-Gilbert equations is time-consuming, so the FDM-FFT method was brought up to speed up the simulation in large scale [2]. Recently, we have developed a new micromagnetic method based on the Hybrid Monte Carlo (HMC) algorithm, which can calculate M-H loops and domains at finite temperature below Curie point [3]. Some simulation results have been put forward to confirm the validity of the HMC algorithm [4]-[6]. Furthermore, we also wish to focus on the computational efficiency to compare HMC micromagnetics with traditional LLG micromagnetics. Comparison of computational speed has been made of these two methods in various micromagnetic scales in this work. In LLG or HMC micromagnetics, a regular mesh with cell size 5 \times 5 \times 5 nm 3 is utilized. Other important parameters are: anisotropy energy constant K(0 \mathrm {K}) = 1 \times 10 ^{5}\mathrm {J}/ \mathrm {m}^{3}, saturation magnetization at zero temperatur e M_{s}(0 \mathrm {K}) = 7.98 \times 10 ^{5}\mathrm {A} /\mathrm {m}, exchange constant A^{\ast }= 1.0 \times 10 ^{-11}\mathrm {J} /\text{m}. In LLG micromagnetics, the iterations can go endless unless the spin error \vert \Delta\vert ( the spin maximum deviation between two simulation steps) is set to be a limited number. To ensure the stability and accuracy, \vert \Delta\vert is set as 10 ^{-6} in LLG algorithm. In HMC micromagnetics, the Monte Carlo time t is not real time. Based on the feature of the leap-frog algorithm used for iteration of the Hamilton equations, in HMC micromagnetics, the spin error \vert \Delta\vert stands for the spin deviation of two adjacent trajectories. To compare the computational speed of the LLG and the HMC micromagnetic methods, the spin error \vert \Delta\vert in HMC MuMag is also set as 10 ^{-6}. The M-H loops are same under these two micromagnetic algorithms. Fig. 1 shows the computational time to reach \vert \Delta \vert \sim 10 ^{-6} at each external magnetic field (\mathrm {H}_{ext}), with 16*8*16 micromagnetic cells. The computational time for Hybrid Monte Carlo MuMag keeps around 59 seconds for different \mathrm {H}_{ext}; while for Laudau-Lifshitz-Gilbert MuMag, the simulation time is typically several hundred of seconds and shows a sharp rising near coercivity to about 1904 seconds, which is different from HMC MuMag. In Fig. 2, the computational speed of the two micromagnetic methods is presented. The horizontal axis is {log}_{2}(N_{cell}), and the vertical axis is the total computation time for a loop in Fig. 1 (in seconds). The blue line is the speed of the LLG MuMag. In the Ref. 7, the LLG algorithm has a running time proportional to the number of particles N, which yields {O(Nlog}_{2}N) computational time. We have compared the computational time of LLG algorithm with {O(Nlog}_{2}N). The triangle-dot line is the computation time vs. {log}_{2}(N_{cell}), which shows great agreement with {O(Nlog}_{2}N)( black line in Fig. 2). The red dotted line shows the computation time consumed by the HMC algorithm, which show great coincidence with O(N) in the insert figure. From the comparison, the HMC algorithm has a significant advantage for the computational speed. When the number of micromagnetic cells rise to 32*32*32, LLG algorithm spends 16 times more than HMC algorithm.
Author Zhao, Z.
Wei, D.
Song, J.
Author_xml – sequence: 1
  givenname: J.
  surname: Song
  fullname: Song, J.
  organization: School of Materials Science and Engineering, Tsinghua University, Beijing, China
– sequence: 2
  givenname: Z.
  surname: Zhao
  fullname: Zhao, Z.
  organization: School of Materials Science and Engineering, Tsinghua University, Beijing, China
– sequence: 3
  givenname: D.
  surname: Wei
  fullname: Wei, D.
  organization: School of Materials Science and Engineering, Tsinghua University, Beijing, China
BookMark eNotkM1OAjEYRavRRECegE1foNgf2uksyUSBZNCFuCad9qvWDC3plBh8ejWyujk5yVncMbqJKQJCM0bnjNH6YfO82y5Xc06ZnmtJtRL1FRozKbRSCy7razTiTFKyUJTdoekwfFJKudJCczpCoUmHo8lhSBEnj__oVEwJKZoevx4BHO6gfAFE3JrozIm0wZPhI5Rvgleh7yAX_Cvw-tzl4Mg2xQKkMblPeBtsTgfzHqEEO9yjW2_6AaaXnaC3p8ddsybty2rTLFsSGJWFMMe4BMu5tLo21nte2QqsNQwqrkAo5o2TlZU1eO4WnYVOc-284AIUaCEmaPbfDQCwP-ZwMPm8vzwjfgD3XFvq
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/INTMAG.2018.8508639
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
EISBN 1538664259
9781538664254
EISSN 2150-4601
EndPage 1
ExternalDocumentID 8508639
Genre orig-research
GroupedDBID 6IE
6IF
6IL
6IN
AAJGR
AAWTH
ABLEC
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
IEGSK
OCL
RIE
RIL
ID FETCH-LOGICAL-i105t-1d125ec225c89acff27c7ecca1e726e361fad57c59ef2d4bceb828df323e6e833
IEDL.DBID RIE
ISICitedReferencesCount 0
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000455204800418&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:51:53 EDT 2025
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i105t-1d125ec225c89acff27c7ecca1e726e361fad57c59ef2d4bceb828df323e6e833
PageCount 1
ParticipantIDs ieee_primary_8508639
PublicationCentury 2000
PublicationDate 2018-April
PublicationDateYYYYMMDD 2018-04-01
PublicationDate_xml – month: 04
  year: 2018
  text: 2018-April
PublicationDecade 2010
PublicationTitle 2018 IEEE International Magnetics Conference (INTERMAG)
PublicationTitleAbbrev INTMAG
PublicationYear 2018
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0002683820
ssj0000331248
Score 1.650886
Snippet The Landau-Lifshitz-Gilbert (LLG) equation is widely utilized to analyze the magnetic properties in micromagnetics [1]. However, the calculation of...
SourceID ieee
SourceType Publisher
StartPage 1
SubjectTerms Conferences
Magnetics
Title Comparison of Computational Speed between Landau-Lif-shitz- Gilbert and Hybrid-Monte-Carlo Micromagnetics
URI https://ieeexplore.ieee.org/document/8508639
WOSCitedRecordID wos000455204800418&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1bS8MwFD5swwd98bKJd_Lgo9nappfkUYbbhG0MnLC3keaiBWlH1wn6603aUhV88a1pSSg5lO_k9DvfB3BLhe9JwgJMHOljXzCGuXVNDczQIJKOZcBLs4loPqerFVu04K7phVFKleQz1beX5b98mYmdLZUNqMkmDKK2oR1FYdWr1dRTHEIMVNFm7IWUGHSrhYZchw0e58vZ_diyuWi_XumXpUqJKKPD_73LEfS-W_PQogGdY2ip9AQOfqgKdiEZNt6CKNOo8m2oa37oaWNmopqdhaa2jrDD00Tj7WtSfGI0TqzoVYHMAzT5sO1ceGYFrPCQ528ZmpX8Pf6S2t7HbQ-eRw_L4QTXjgo4MXlUgV1p8hklzDcsKONCay8SkQ2iqyIvVCR0NZdBJAKmtCf9WKjYnMikJh5RoaKEnEInzVJ1Boj4gghBYs59x9dhEHtmIclEzIkUjMfn0LWbtt5Uohnrer8u_r59Cfs2LhUl5go6Rb5T17An3otkm9-Ukf4CJFOrXg
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1bS8MwFD7MKagvXjbxbh58NNva9JZHGe6CbRk4YW8jzUUL0o6tE_TXm3SlKvjiW9OSUHIo38npd74P4Dbgji0IdTHpCQc7nFLMjGuqq4cakVQiXFaaTfhxHMxmdNKAu7oXRkpZks9kx1yW__JFztemVNYNdDahEXULto1zVtWtVVdUeoRosArqse0FRONbJTVk9Wh3HE-j-6HhcwWdaq1fpiolpgwO_vc2h9D-bs5Dkxp2jqAhs2PY_6Er2IK0X7sLolyhjXNDVfVDTws9E1X8LBSaSsIah6nCq9e0-MRomBrZqwLpB2j0YRq6cGQkrHCfLd9yFJUMPvaSme7HVRueBw_T_ghXngo41ZlUgS2hMxrJ9VfMA8q4UrbPfRNGS_q2J4lnKSZcn7tUKls4CZeJPpMJRWwiPRkQcgLNLM_kKSDicMI5SRhzeo7y3MTWCwnKE0YEpyw5g5bZtPliI5sxr_br_O_bN7A7mkbhPBzHjxewZ2K0IchcQrNYruUV7PD3Il0tr8uofwE8pa6n
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=2018+IEEE+International+Magnetics+Conference+%28INTERMAG%29&rft.atitle=Comparison+of+Computational+Speed+between+Landau-Lif-shitz-+Gilbert+and+Hybrid-Monte-Carlo+Micromagnetics&rft.au=Song%2C+J.&rft.au=Zhao%2C+Z.&rft.au=Wei%2C+D.&rft.date=2018-04-01&rft.pub=IEEE&rft.eissn=2150-4601&rft.spage=1&rft.epage=1&rft_id=info:doi/10.1109%2FINTMAG.2018.8508639&rft.externalDocID=8508639