{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2602-08538","title":"Trajectory Stitching for Solving Inverse Problems with Flow-Based Models","arxiv_id":"2602.08538","date":"2026-02-09","proceeding":"ICML","authors":["Alexander Denker","Moshe Eliasof","Zeljko Kereta","Carola-Bibiane Schönlieb"],"abstract":"Flow-based generative models have emerged as powerful priors for solving inverse problems. One option is to directly optimize the initial latent code (noise), such that the flow output solves the inverse problem. However, this requires backpropagating through the entire generative trajectory, incurring high memory costs and numerical instability. We propose MS-Flow, which represents the trajectory as a sequence of intermediate latent states rather than a single initial code. By enforcing the flow dynamics locally and coupling segments through trajectory-matching penalties, MS-Flow alternates between updating intermediate latent states and enforcing consistency with observed data. This reduces memory consumption while improving reconstruction quality. We demonstrate the effectiveness of MS-Flow over existing methods on image recovery and inverse problems, including inpainting, super-resolution, and computed tomography.","url_abs":"https://arxiv.org/abs/2602.08538","url_pdf":"https://arxiv.org/pdf/2602.08538","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2602.08538","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2602.08538"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/annegnx/PnP-Flow","reach":null}],"summary":{"ran_draft_wrong":1,"unverified":7},"by_repo_kind":{"found_in_text":{"samples":8,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":8,"samples":[{"code_sha256_prefix":"0ee1a2c94acafeca","entry":"group_norm","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/models.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/models.py","link_basis":"plan_row","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"0ee1a2c94acafeca"}},{"code_sha256_prefix":"014ed7df4050e8a4","entry":"calculate_activation_statistics","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/fid_score.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/fid_score.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"014ed7df4050e8a4"}},{"code_sha256_prefix":"0def50a351111624","entry":"calculate_frechet_distance","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/fid_score.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/fid_score.py","link_basis":"plan_row","language":"python","status":"unverified","verification_level":0,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"0def50a351111624"}},{"code_sha256_prefix":"3b05acd66d4cf394","entry":"custom_collate","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/dataloaders.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/dataloaders.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3b05acd66d4cf394"}},{"code_sha256_prefix":"64fff7f072e05415","entry":"downsample","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/models.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/models.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"64fff7f072e05415"}},{"code_sha256_prefix":"f10c698bdf0953d7","entry":"get_activations","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/fid_score.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/fid_score.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f10c698bdf0953d7"}},{"code_sha256_prefix":"42e5d1476a6ac5cb","entry":"to_nn_parameter","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/methods/pnp_diff.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/methods/pnp_diff.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"42e5d1476a6ac5cb"}},{"code_sha256_prefix":"77e6e60b8a8a69da","entry":"upsample","repo":"annegnx/PnP-Flow","repo_kind":"found_in_text","path":"pnpflow/models.py","file_url":"https://github.com/annegnx/PnP-Flow/blob/HEAD/pnpflow/models.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"77e6e60b8a8a69da"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}