{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5Y7AWNYEPJHY5DNPC3CVU27ACF","short_pith_number":"pith:5Y7AWNYE","schema_version":"1.0","canonical_sha256":"ee3e0b37047a4f8e8daf16c55a6be0114d0f963293c590b55b6dd352eaab8d17","source":{"kind":"arxiv","id":"2604.16232","version":2},"attestation_state":"computed","paper":{"title":"Neuro-Symbolic ODE Discovery with Latent Grammar Flow","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Latent Grammar Flow discovers ODEs by placing grammar-based equation representations in a discrete latent space, using a behavioral loss to cluster similar equations, and sampling via a discrete flow model guided by data fit and constraints.","cross_cats":["cs.AI","cs.CE","cs.SC"],"primary_cat":"cs.LG","authors_text":"Eleni Chatzi, Georgios Kissas, Karin Yu","submitted_at":"2026-04-17T16:46:23Z","abstract_excerpt":"Understanding natural and engineered systems often relies on symbolic formulations, such as differential equations, which provide interpretability and transferability beyond black-box models.\n  We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candid"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2604.16232","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-17T16:46:23Z","cross_cats_sorted":["cs.AI","cs.CE","cs.SC"],"title_canon_sha256":"e72b2d2f58748267d1db2aeaf078b0d5fc9003e2115268fc6678a8bc58e08f8b","abstract_canon_sha256":"0ab817126848b42d342b627b52fbed4443f45ddac0c5181d0c0580ecaeb05daf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-15T00:21:18.754261Z","signature_b64":"gA0ZnrUY7wxvCQsPrW5RhZgbZO3GzKhuMKoU1B1prLioLSC9hjMeb0kw/ftOYjJ0wG6v8aCFL5yL2QPQUiLhCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ee3e0b37047a4f8e8daf16c55a6be0114d0f963293c590b55b6dd352eaab8d17","last_reissued_at":"2026-07-15T00:21:18.753364Z","signature_status":"signed_v1","first_computed_at":"2026-07-15T00:21:18.753364Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Neuro-Symbolic ODE Discovery with Latent Grammar Flow","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Latent Grammar Flow discovers ODEs by placing grammar-based equation representations in a discrete latent space, using a behavioral loss to cluster similar equations, and sampling via a discrete flow model guided by data fit and constraints.","cross_cats":["cs.AI","cs.CE","cs.SC"],"primary_cat":"cs.LG","authors_text":"Eleni Chatzi, Georgios Kissas, Karin Yu","submitted_at":"2026-04-17T16:46:23Z","abstract_excerpt":"Understanding natural and engineered systems often relies on symbolic formulations, such as differential equations, which provide interpretability and transferability beyond black-box models.\n  We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candid"},"claims":{"count":3,"items":[{"kind":"strongest_claim","text":"We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a behavioral loss can reliably place semantically similar equations closer in the discrete latent space and that the discrete flow model can efficiently sample equations that both fit data and satisfy domain constraints without exhaustive search.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Latent Grammar Flow discovers ODEs by placing grammar-based equation representations in a discrete latent space, using a behavioral loss to cluster similar equations, and sampling via a discrete flow model guided by data fit and constraints.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"}],"snapshot_sha256":"01ae6c1d39dbc5993bf71d517eab0e6707ff181bb3a9130706a999eaf6fa6237"},"source":{"id":"2604.16232","kind":"arxiv","version":2},"verdict":{"id":"cfedc803-6e6e-49bf-9d6e-dce32bd558dd","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T08:33:51.307611Z","strongest_claim":"We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data.","one_line_summary":"Latent Grammar Flow discovers ODEs by placing grammar-based equation representations in a discrete latent space, using a behavioral loss to cluster similar equations, and sampling via a discrete flow model guided by data fit and constraints.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a behavioral loss can reliably place semantically similar equations closer in the discrete latent space and that the discrete flow model can efficiently sample equations that both fit data and satisfy domain constraints without exhaustive search.","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.16232/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2604.16232","created_at":"2026-07-15T00:21:18.753781+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.16232v2","created_at":"2026-07-15T00:21:18.753781+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.16232","created_at":"2026-07-15T00:21:18.753781+00:00"},{"alias_kind":"pith_short_12","alias_value":"5Y7AWNYEPJHY","created_at":"2026-07-15T00:21:18.753781+00:00"},{"alias_kind":"pith_short_16","alias_value":"5Y7AWNYEPJHY5DNP","created_at":"2026-07-15T00:21:18.753781+00:00"},{"alias_kind":"pith_short_8","alias_value":"5Y7AWNYE","created_at":"2026-07-15T00:21:18.753781+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.02662","citing_title":"Verifier-Guided Model Discovery for Physical Dynamical Systems with Pretrained Symbolic Transformers","ref_index":34,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF","json":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF.json","graph_json":"https://pith.science/api/pith-number/5Y7AWNYEPJHY5DNPC3CVU27ACF/graph.json","events_json":"https://pith.science/api/pith-number/5Y7AWNYEPJHY5DNPC3CVU27ACF/events.json","paper":"https://pith.science/paper/5Y7AWNYE"},"agent_actions":{"view_html":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF","download_json":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF.json","view_paper":"https://pith.science/paper/5Y7AWNYE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.16232&json=true","fetch_graph":"https://pith.science/api/pith-number/5Y7AWNYEPJHY5DNPC3CVU27ACF/graph.json","fetch_events":"https://pith.science/api/pith-number/5Y7AWNYEPJHY5DNPC3CVU27ACF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF/action/storage_attestation","attest_author":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF/action/author_attestation","sign_citation":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF/action/citation_signature","submit_replication":"https://pith.science/pith/5Y7AWNYEPJHY5DNPC3CVU27ACF/action/replication_record"}},"created_at":"2026-07-15T00:21:18.753781+00:00","updated_at":"2026-07-15T00:21:18.753781+00:00"}