{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:ER353KTMMJGZRI4O4KXT5BQ2X3","short_pith_number":"pith:ER353KTM","schema_version":"1.0","canonical_sha256":"2477ddaa6c624d98a38ee2af3e861abeee5a668eb450234e4d9d0872b8715ee7","source":{"kind":"arxiv","id":"2007.14527","version":1},"attestation_state":"computed","paper":{"title":"When and why PINNs fail to train: A neural tangent kernel perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA","math.NA","stat.ML"],"primary_cat":"cs.LG","authors_text":"Paris Perdikaris, Sifan Wang, Xinling Yu","submitted_at":"2020-07-28T23:44:56Z","abstract_excerpt":"Physics-informed neural networks (PINNs) have lately received great attention thanks to their flexibility in tackling a wide range of forward and inverse problems involving partial differential equations. However, despite their noticeable empirical success, little is known about how such constrained neural networks behave during their training via gradient descent. More importantly, even less is known about why such models sometimes fail to train at all. In this work, we aim to investigate these questions through the lens of the Neural Tangent Kernel (NTK); a kernel that captures the behavior "},"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":"2007.14527","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-07-28T23:44:56Z","cross_cats_sorted":["cs.NA","math.NA","stat.ML"],"title_canon_sha256":"263f27c2733734c4765a2128a036f288484b0fe3c03a024e0b15ab00eaf88747","abstract_canon_sha256":"870cb48e2f30af118cdda0e613fd0946562c3c499e48562e3454ef48ca5c5398"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:23:16.478662Z","signature_b64":"SFzMR6pwhi3ItQtjge8J9vSxLT0nhW5UKQde1/iTAQ9VKwjK/Dm+dlf894x6NoMmXsGGKJjyqnF0fbfOAEz2DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2477ddaa6c624d98a38ee2af3e861abeee5a668eb450234e4d9d0872b8715ee7","last_reissued_at":"2026-07-05T01:23:16.478191Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:23:16.478191Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When and why PINNs fail to train: A neural tangent kernel perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.NA","math.NA","stat.ML"],"primary_cat":"cs.LG","authors_text":"Paris Perdikaris, Sifan Wang, Xinling Yu","submitted_at":"2020-07-28T23:44:56Z","abstract_excerpt":"Physics-informed neural networks (PINNs) have lately received great attention thanks to their flexibility in tackling a wide range of forward and inverse problems involving partial differential equations. However, despite their noticeable empirical success, little is known about how such constrained neural networks behave during their training via gradient descent. More importantly, even less is known about why such models sometimes fail to train at all. In this work, we aim to investigate these questions through the lens of the Neural Tangent Kernel (NTK); a kernel that captures the behavior "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2007.14527","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2007.14527/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":"2007.14527","created_at":"2026-07-05T01:23:16.478247+00:00"},{"alias_kind":"arxiv_version","alias_value":"2007.14527v1","created_at":"2026-07-05T01:23:16.478247+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2007.14527","created_at":"2026-07-05T01:23:16.478247+00:00"},{"alias_kind":"pith_short_12","alias_value":"ER353KTMMJGZ","created_at":"2026-07-05T01:23:16.478247+00:00"},{"alias_kind":"pith_short_16","alias_value":"ER353KTMMJGZRI4O","created_at":"2026-07-05T01:23:16.478247+00:00"},{"alias_kind":"pith_short_8","alias_value":"ER353KTM","created_at":"2026-07-05T01:23:16.478247+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19368","citing_title":"Neural Architectures as Functional Priors in Physics-Informed Control Problems","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02335","citing_title":"Neural Spectral Element Methods for stiff multiphysics PDEs with electrochemical transport benchmarks","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2212.08989","citing_title":"Deep learning applied to computational mechanics: A comprehensive review, state of the art, and the classics","ref_index":288,"is_internal_anchor":false},{"citing_arxiv_id":"2308.13222","citing_title":"Bayesian Reasoning for Physics Informed Neural Networks","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27052","citing_title":"Man, Machine, and Mathematics","ref_index":97,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3","json":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3.json","graph_json":"https://pith.science/api/pith-number/ER353KTMMJGZRI4O4KXT5BQ2X3/graph.json","events_json":"https://pith.science/api/pith-number/ER353KTMMJGZRI4O4KXT5BQ2X3/events.json","paper":"https://pith.science/paper/ER353KTM"},"agent_actions":{"view_html":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3","download_json":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3.json","view_paper":"https://pith.science/paper/ER353KTM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2007.14527&json=true","fetch_graph":"https://pith.science/api/pith-number/ER353KTMMJGZRI4O4KXT5BQ2X3/graph.json","fetch_events":"https://pith.science/api/pith-number/ER353KTMMJGZRI4O4KXT5BQ2X3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3/action/storage_attestation","attest_author":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3/action/author_attestation","sign_citation":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3/action/citation_signature","submit_replication":"https://pith.science/pith/ER353KTMMJGZRI4O4KXT5BQ2X3/action/replication_record"}},"created_at":"2026-07-05T01:23:16.478247+00:00","updated_at":"2026-07-05T01:23:16.478247+00:00"}