Tight Lower Bounds for Block-Structured Integer Programs
We study fundamental block-structured integer programs called tree-fold and multi-stage IPs. Tree-fold IPs admit a constraint matrix with independent blocks linked together by few constraints in a recursive pattern; and transposing their constraint matrix yields multi-stage IPs. The state-of-the-art...
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Artikel |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2402.17290 http://arxiv.org/abs/2402.17290 |
| Tags: |
Keine Tags, Fügen Sie das erste Tag hinzu!
|
| _version_ | 1879175405325254656 |
|---|---|
| author | Hunkenschröder, Christoph Klein, Kim-Manuel Koutecký, Martin Lassota, Alexandra Levin, Asaf |
| author_browse | Hunkenschröder, Christoph Klein, Kim-Manuel Koutecký, Martin Lassota, Alexandra Levin, Asaf |
| author_facet | Hunkenschröder, Christoph Klein, Kim-Manuel Koutecký, Martin Lassota, Alexandra Levin, Asaf |
| author_sort | Hunkenschröder, Christoph |
| building | arXiv |
| collection | Acceso abierto |
| description | We study fundamental block-structured integer programs called tree-fold and multi-stage IPs. Tree-fold IPs admit a constraint matrix with independent blocks linked together by few constraints in a recursive pattern; and transposing their constraint matrix yields multi-stage IPs. The state-of-the-art algorithms to solve these IPs have an exponential gap in their running times, making it natural to ask whether this gap is inherent. We answer this question affirmative. Assuming the Exponential Time Hypothesis, we prove lower bounds showing that the exponential difference is necessary, and that the known algorithms are near optimal. Moreover, we prove unconditional lower bounds on the norms of the Graver basis, a fundamental building block of all known algorithms to solve these IPs. This shows that none of the current approaches can be improved beyond this bound. |
| first_indexed | 2026-10-16T00:28:35Z |
| format | Article |
| fullrecord | <oai_dc:dc xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ https://www.openarchives.org/OAI/2.0/oai_dc.xsd"><identifier>arxiv-2402.17290</identifier><datestamp>2024-02-28</datestamp>
<dc:title>Tight Lower Bounds for Block-Structured Integer Programs</dc:title>
<dc:creator>Hunkenschröder, Christoph</dc:creator>
<dc:creator>Klein, Kim-Manuel</dc:creator>
<dc:creator>Koutecký, Martin</dc:creator>
<dc:creator>Lassota, Alexandra</dc:creator>
<dc:creator>Levin, Asaf</dc:creator>
<dc:subject>Computational Complexity</dc:subject>
<dc:description>We study fundamental block-structured integer programs called tree-fold and multi-stage IPs. Tree-fold IPs admit a constraint matrix with independent blocks linked together by few constraints in a recursive pattern; and transposing their constraint matrix yields multi-stage IPs. The state-of-the-art algorithms to solve these IPs have an exponential gap in their running times, making it natural to ask whether this gap is inherent. We answer this question affirmative. Assuming the Exponential Time Hypothesis, we prove lower bounds showing that the exponential difference is necessary, and that the known algorithms are near optimal. Moreover, we prove unconditional lower bounds on the norms of the Graver basis, a fundamental building block of all known algorithms to solve these IPs. This shows that none of the current approaches can be improved beyond this bound. </dc:description>
<dc:date>2024-02-27</dc:date>
<dc:type>text</dc:type>
<dc:identifier>https://arxiv.org/abs/2402.17290</dc:identifier>
<dc:identifier>http://arxiv.org/abs/2402.17290</dc:identifier>
</oai_dc:dc> |
| id | arxiv-2402.17290 |
| institution | arXiv |
| last_indexed | 2026-10-16T00:28:35Z |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| record_format | oai_dc |
| spelling | arxiv-2402.172902024-02-28 Tight Lower Bounds for Block-Structured Integer Programs Hunkenschröder, Christoph Klein, Kim-Manuel Koutecký, Martin Lassota, Alexandra Levin, Asaf Computational Complexity We study fundamental block-structured integer programs called tree-fold and multi-stage IPs. Tree-fold IPs admit a constraint matrix with independent blocks linked together by few constraints in a recursive pattern; and transposing their constraint matrix yields multi-stage IPs. The state-of-the-art algorithms to solve these IPs have an exponential gap in their running times, making it natural to ask whether this gap is inherent. We answer this question affirmative. Assuming the Exponential Time Hypothesis, we prove lower bounds showing that the exponential difference is necessary, and that the known algorithms are near optimal. Moreover, we prove unconditional lower bounds on the norms of the Graver basis, a fundamental building block of all known algorithms to solve these IPs. This shows that none of the current approaches can be improved beyond this bound. 2024-02-27 text https://arxiv.org/abs/2402.17290 http://arxiv.org/abs/2402.17290 |
| spellingShingle | Computational Complexity Hunkenschröder, Christoph Klein, Kim-Manuel Koutecký, Martin Lassota, Alexandra Levin, Asaf Tight Lower Bounds for Block-Structured Integer Programs |
| title | Tight Lower Bounds for Block-Structured Integer Programs |
| title_full | Tight Lower Bounds for Block-Structured Integer Programs |
| title_fullStr | Tight Lower Bounds for Block-Structured Integer Programs |
| title_full_unstemmed | Tight Lower Bounds for Block-Structured Integer Programs |
| title_short | Tight Lower Bounds for Block-Structured Integer Programs |
| title_sort | tight lower bounds for block structured integer programs |
| topic | Computational Complexity |
| url | https://arxiv.org/abs/2402.17290 http://arxiv.org/abs/2402.17290 |
| work_keys_str_mv | AT hunkenschroderchristoph tightlowerboundsforblockstructuredintegerprograms AT kleinkimmanuel tightlowerboundsforblockstructuredintegerprograms AT kouteckymartin tightlowerboundsforblockstructuredintegerprograms AT lassotaalexandra tightlowerboundsforblockstructuredintegerprograms AT levinasaf tightlowerboundsforblockstructuredintegerprograms |