{"id":87829,"updated":"2025-01-21T17:00:57.524180+00:00","links":{},"created":"2025-01-18T23:38:25.601298+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00087829","sets":["6164:6165:6242:6971"]},"path":["6971"],"owner":"11","recid":"87829","title":["特異値計算アルゴリズムdqds法およびmdLVs法のための新しいシフト戦略"],"pubdate":{"attribute_name":"公開日","attribute_value":"2013-01-08"},"_buckets":{"deposit":"2756331c-5aeb-4c00-a58f-12d57b526668"},"_deposit":{"id":"87829","pid":{"type":"depid","value":"87829","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"特異値計算アルゴリズムdqds法およびmdLVs法のための新しいシフト戦略","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"特異値計算アルゴリズムdqds法およびmdLVs法のための新しいシフト戦略"},{"subitem_title":"New Shift Strategy in the dqds Algorithm and the mdLVs Algorithm for Singular Values","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"数値計算","subitem_subject_scheme":"Other"}]},"item_type_id":"18","publish_date":"2013-01-08","item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_18_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"奈良女子大学"},{"subitem_text_value":"京都大学"},{"subitem_text_value":"京都大学"},{"subitem_text_value":"京都大学"},{"subitem_text_value":"京都府立大学"},{"subitem_text_value":"京都大学"}]},"item_18_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Nara Women's University","subitem_text_language":"en"},{"subitem_text_value":"Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"Kyoto University","subitem_text_language":"en"},{"subitem_text_value":"Kyoto Prefectural University","subitem_text_language":"en"},{"subitem_text_value":"Kyoto University","subitem_text_language":"en"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/87829/files/IPSJ-HPCS2013001.pdf"},"date":[{"dateType":"Available","dateValue":"2015-01-08"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-HPCS2013001.pdf","filesize":[{"value":"762.2 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"330","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"330","billingrole":"14"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"48fdedd9-3661-4bb2-a57e-98317eb3d4f3","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2013 by the Information Processing Society of Japan"}]},"item_18_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"髙田, 雅美"},{"creatorName":"石上, 裕之"},{"creatorName":"木村, 欣司"},{"creatorName":"山下, 巧"},{"creatorName":"岩﨑, 雅史"},{"creatorName":"中村, 佳正"}],"nameIdentifiers":[{}]}]},"item_18_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Masami, Takata","creatorNameLang":"en"},{"creatorName":"Hiroyuki, Ishigami","creatorNameLang":"en"},{"creatorName":"Kinji, Kimura","creatorNameLang":"en"},{"creatorName":"Takumi, Yamashita","creatorNameLang":"en"},{"creatorName":"Masashi, Iwasaki","creatorNameLang":"en"},{"creatorName":"Yoshimasa, Nakamura","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_5794","resourcetype":"conference paper"}]},"item_18_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"本論文では，新しいシフト戦略としてAlgebraicシフトを提案することによって，特異値計算アルゴリズムdqds法およびm2dLVs法の計算速度と相対精度を改善する．従来，LAPACKに実装されたdqds法では，経験則に基づくaggressiveシフトを採用している．従来のmdLVs法では，Johnsonシフトを採用している．aggressiveシフトは高速に計算されるが，無誤差の計算でも過大なシフトを与える可能性がある．Johnsonシフトについては，その計算において平方根を多用するため，速度の面できわめて不十分である．そこで，本論文では，dqds法およびm2dLVs法の両方に対して，数学理論に基づくAlgebraicシフトを導入する．このシフト戦略の有効性を確認するために，従来のシフト戦略との比較を行う．","subitem_description_type":"Other"}]},"item_18_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"In this paper, we propose a new shift strategy ”Algebraic shift”, which improves computational time and relative errors in the dqds and m2dLVs algorithms for singular values. The dqds algorithm in LAPACK has adopted the aggressive shift, which is based on a rule of heuristics. The original mdLVs algorithm adopts the Johnson shift. While the aggressive shift is quickly computed, it may give a too large shift even in exact computation. Since the Johnson shift needs a lot of square root, the computational time is weakness. Therefore, we introduce the Algebraic shift, which has the proof based on mathematical theory, to the dqds and m2dLVs algorithms. To evaluate the shift strategy, we compare with existing shift strategy.","subitem_description_type":"Other"}]},"item_18_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"10","bibliographic_titles":[{"bibliographic_title":"ハイパフォーマンスコンピューティングと計算科学シンポジウム論文集"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2013-01-08","bibliographicIssueDateType":"Issued"},"bibliographicVolumeNumber":"2013"}]},"relation_version_is_last":true,"weko_creator_id":"11"}}