{"id":{"repo_id":"cornell","oai_identifier":"oai:ecommons.cornell.edu:1813/116015"},"canonical_url":"https://search.ndltd.org/etd/cornell/oai:ecommons.cornell.edu:1813/116015","repository":{"repo_id":"cornell","name":"Cornell University","base_url":"https://ecommons.cornell.edu/server/oai/request"},"display":{"title":"METRIC AND TAME GEOMETRY IN OPTIMIZATION","abstract":"Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example.","abstract_html":"Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example.","abstract_has_math":false,"creators":["Tian, Tonghua"],"institution":"Cornell University","degree_name":"Ph. D., Operations Research and Information Engineering","degree_level":"Doctor of Philosophy","degree_discipline":"Operations Research and Information Engineering","degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":["Lewis, Adrian"],"committee_members":["Davis, Damek","Scheinberg, Katya"],"year":2024,"date_issued":"2024-05","date_published":"2024-05","updated_at":"2026-11-19T03:11:32Z","subjects":[],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.7298/7m96-ey91"],"render_values":[{"text":"https://doi.org/10.7298/7m96-ey91","href":"https://doi.org/10.7298/7m96-ey91","code":true}]},{"key":"dc:identifier.bibid","label":"Dc Identifier Bibid","values":["16575526"],"render_values":[{"text":"16575526","href":null,"code":true}]},{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["ProQuest Submission ID: 14207","ProQuest Publication ID: 31237872"],"render_values":[{"text":"ProQuest Submission ID: 14207","href":null,"code":true},{"text":"ProQuest Publication ID: 31237872","href":null,"code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/1813/116015","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/1813/116015","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://ecommons.cornell.edu/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Aecommons.cornell.edu%3A1813%2F116015","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.chair","label":"Chair","values":["Lewis, Adrian"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Davis, Damek","Scheinberg, Katya"]},{"key":"dc:creator","label":"Author","values":["Tian, Tonghua"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2024-11-05T19:47:24Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2024-11-05T19:47:24Z"]},{"key":"dc:date.issued","label":"Date","values":["2024-05"]},{"key":"dc:relation.uri","label":"Dc Relation URI","values":["https://newcatalog.library.cornell.edu/catalog/16575526"]},{"key":"dc:type","label":"Dc Type","values":["dissertation or thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Operations Research and Information Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Doctor of Philosophy"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Ph. D., Operations Research and Information Engineering"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Cornell University"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.bibid","label":"Dc Identifier Bibid","values":["16575526"]},{"key":"dc:identifier.doi","label":"DOI","values":["https://doi.org/10.7298/7m96-ey91"]},{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["ProQuest Submission ID: 14207","ProQuest Publication ID: 31237872"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1813/116015"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"cerif:openaire","label":"Cerif Openaire","values":["<Publication xmlns=\"https://www.openaire.eu/cerif-profile/1.1/\" id=\"bf02c6c1-7885-457c-ae55-edcf1840b3dd\"> <Type xmlns=\"https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types\">http://purl.org/coar/resource_type/c_1843</Type> <Language>en</Language> <Title>METRIC AND TAME GEOMETRY IN OPTIMIZATION</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2024-05</PublicationDate> <DOI>https://doi.org/10.7298/7m96-ey91</DOI> <Authors> <Author> <DisplayName>Tian, Tonghua</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <OrgUnit /> </Publisher> </Publishers> <Abstract>Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example.</Abstract> <Access xmlns=\"http://purl.org/coar/access_right\" > </Access> </Publication>"]},{"key":"cris:virtual.author","label":"Cris Virtual Author","values":["Tian, Tonghua"]},{"key":"cris:virtualsource.collection","label":"Cris Virtualsource Collection","values":["5893a6ea-7af3-41d7-abc6-04bcd26ab5df"]},{"key":"dc:description","label":"Description","values":["198 pages"]},{"key":"dc:description.abstract","label":"Abstract","values":["Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example."]},{"key":"dc:description.collection","label":"Dc Description Collection","values":["Cornell Theses and Dissertations"]},{"key":"dc:format.mimetype","label":"Dc Format Mimetype","values":["application/pdf"]},{"key":"dcterms:license","label":"Dcterms License","values":["https://hdl.handle.net/1813/59810.2"]},{"key":"dc:title","label":"Title","values":["METRIC AND TAME GEOMETRY IN OPTIMIZATION"]},{"key":"dspace:entity.type","label":"Dspace Entity Type","values":["Publication"]},{"key":"others:access_status","label":"Others Access Status","values":["open.access"]}]}],"canonical_facts":{"cerif:openaire":["<Publication xmlns=\"https://www.openaire.eu/cerif-profile/1.1/\" id=\"bf02c6c1-7885-457c-ae55-edcf1840b3dd\"> <Type xmlns=\"https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types\">http://purl.org/coar/resource_type/c_1843</Type> <Language>en</Language> <Title>METRIC AND TAME GEOMETRY IN OPTIMIZATION</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2024-05</PublicationDate> <DOI>https://doi.org/10.7298/7m96-ey91</DOI> <Authors> <Author> <DisplayName>Tian, Tonghua</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <OrgUnit /> </Publisher> </Publishers> <Abstract>Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example.</Abstract> <Access xmlns=\"http://purl.org/coar/access_right\" > </Access> </Publication>"],"cris:virtual.author":["Tian, Tonghua"],"cris:virtualsource.collection":["5893a6ea-7af3-41d7-abc6-04bcd26ab5df"],"dc:contributor.chair":["Lewis, Adrian"],"dc:contributor.committeemember":["Davis, Damek","Scheinberg, Katya"],"dc:creator":["Tian, Tonghua"],"dc:date.accessioned":["2024-11-05T19:47:24Z"],"dc:date.available":["2024-11-05T19:47:24Z"],"dc:date.issued":["2024-05"],"dc:description":["198 pages"],"dc:description.abstract":["Structure in optimization is traditionally studied through variational analysis. Structural results usually require convexity or other regularity conditions. However, contemporary optimization involves problems that lie beyond this traditional scope, especially in machine learning, both in Euclidean space and in non-Euclidean metric spaces. This thesis pursues theories for such problems that reveal underlying geometric structure and aid algorithm design and analysis. A central philosophy in the thesis is to highlight the slope as a tool for capturing first-order information of objectives. This engenders a subgradient-free development: subgradients are indirect in depicting structure and do not extend naturally to non-Euclidean settings. The development focuses on two ideas traditionally studied through subgradients. The first idea is identification: the phenomenon of a set being identified by certain convergent sequences in finite time. By re-interpreting identification via the slope, the thesis extends the idea from Euclidean space to general metric spaces, and establishes results on objective growth rates and necessary conditions for optimization on metric spaces. An extension of identification behavior from discrete-time sequences to continuous-time trajectories is also included. The second idea is the Kurdyka-Łojasiewicz (KL) inequality: a popular tool for convergence analysis in optimization. Defining the KL inequality via the slope reveals its connection with identification, and more importantly its impact on complexity analysis of first-order methods in metric spaces. The prevalence of identifiability and of objectives satisfying the KL inequality is established in the literature from semialgebraic (or, more generally, tame) geometry. Based on the same technique, the thesis presents two other results. The first characterizes the structure of conservative gradient fields, a notion of generalized gradients especially useful in analyzing deep learning methods. The second concerns the generic prevalence of partly smooth set-valued mappings, a type of structure known to be associated with identifiable sets. Furthermore, the thesis gives a short slope-based proof for the fundamental fact that semialgebraic functions possess the KL property. Finally, to unify and relax traditional structural assumptions such as convexity, smoothness, and their compositions, the thesis proposes a new type of implicit structure — smooth approximate convexity. The thesis shows the prevalence of such structure and demonstrates its applicability with a Riemannian optimization example."],"dc:description.collection":["Cornell Theses and Dissertations"],"dc:format.mimetype":["application/pdf"],"dc:identifier.bibid":["16575526"],"dc:identifier.doi":["https://doi.org/10.7298/7m96-ey91"],"dc:identifier.other":["ProQuest Submission ID: 14207","ProQuest Publication ID: 31237872"],"dc:identifier.uri":["https://hdl.handle.net/1813/116015"],"dc:language.iso":["en"],"dc:relation.uri":["https://newcatalog.library.cornell.edu/catalog/16575526"],"dcterms:license":["https://hdl.handle.net/1813/59810.2"],"dc:title":["METRIC AND TAME GEOMETRY IN OPTIMIZATION"],"dc:type":["dissertation or thesis"],"dspace:entity.type":["Publication"],"others:access_status":["open.access"],"thesis:degree_discipline":["Operations Research and Information Engineering"],"thesis:degree_level":["Doctor of Philosophy"],"thesis:degree_name":["Ph. D., Operations Research and Information Engineering"],"thesis:institution_name":["Cornell University"]},"updated_at":"2026-11-19T03:11:32Z"}