{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MXGAB4LD53U4P6YUE6XDT5PFLF","short_pith_number":"pith:MXGAB4LD","schema_version":"1.0","canonical_sha256":"65cc00f163eee9c7fb1427ae39f5e5597b8f2bd245e50ea8b0e0485fcf985b4b","source":{"kind":"arxiv","id":"2608.06582","version":1},"attestation_state":"computed","paper":{"title":"CrystalGRPO: Target-Aligned and Coverage-Preserving Reinforcement Learning for Flow-Based Crystal Structure Prediction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hongfei Xue, Kaixiang Su, Qiang Zhu","submitted_at":"2026-08-06T20:53:00Z","abstract_excerpt":"Flow-based generative models can efficiently produce candidate structures for crystal structure prediction (CSP), but their pretrained objectives do not directly optimize downstream target recovery. Reinforcement-learning post-training offers a flexible solution, yet existing approaches rely primarily on energy rewards and coordinate-only stochastic policies. Predicted energy does not identify the reference polymorph, while reward-driven concentration can reduce the candidate coverage required for Top-N recovery. We introduce CrystalGRPO, a CSP-aligned post-training framework that extends exis"},"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":"2608.06582","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-06T20:53:00Z","cross_cats_sorted":[],"title_canon_sha256":"cc917577b04ede8d35e181de8e295490cd127b42ed0f3336e2f852ace91eade2","abstract_canon_sha256":"fa22c3cc2c2d3498b4249cde2bf0da5f75bf7c82ab8074077fa6aeb09247dfe2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-10T01:11:21.371332Z","signature_b64":"flMP77HeFrOVBb2ZB7aQDpNV7f0yh+HtwscErQt2TvfIXlrXjPC8YPV/tVDTOwaPa5PpPmOOUIOgZPbzpayUAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"65cc00f163eee9c7fb1427ae39f5e5597b8f2bd245e50ea8b0e0485fcf985b4b","last_reissued_at":"2026-08-10T01:11:21.369059Z","signature_status":"signed_v1","first_computed_at":"2026-08-10T01:11:21.369059Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CrystalGRPO: Target-Aligned and Coverage-Preserving Reinforcement Learning for Flow-Based Crystal Structure Prediction","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hongfei Xue, Kaixiang Su, Qiang Zhu","submitted_at":"2026-08-06T20:53:00Z","abstract_excerpt":"Flow-based generative models can efficiently produce candidate structures for crystal structure prediction (CSP), but their pretrained objectives do not directly optimize downstream target recovery. Reinforcement-learning post-training offers a flexible solution, yet existing approaches rely primarily on energy rewards and coordinate-only stochastic policies. Predicted energy does not identify the reference polymorph, while reward-driven concentration can reduce the candidate coverage required for Top-N recovery. We introduce CrystalGRPO, a CSP-aligned post-training framework that extends exis"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.06582","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/2608.06582/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":"2608.06582","created_at":"2026-08-10T01:11:21.369956+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.06582v1","created_at":"2026-08-10T01:11:21.369956+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.06582","created_at":"2026-08-10T01:11:21.369956+00:00"},{"alias_kind":"pith_short_12","alias_value":"MXGAB4LD53U4","created_at":"2026-08-10T01:11:21.369956+00:00"},{"alias_kind":"pith_short_16","alias_value":"MXGAB4LD53U4P6YU","created_at":"2026-08-10T01:11:21.369956+00:00"},{"alias_kind":"pith_short_8","alias_value":"MXGAB4LD","created_at":"2026-08-10T01:11:21.369956+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF","json":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF.json","graph_json":"https://pith.science/api/pith-number/MXGAB4LD53U4P6YUE6XDT5PFLF/graph.json","events_json":"https://pith.science/api/pith-number/MXGAB4LD53U4P6YUE6XDT5PFLF/events.json","paper":"https://pith.science/paper/MXGAB4LD"},"agent_actions":{"view_html":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF","download_json":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF.json","view_paper":"https://pith.science/paper/MXGAB4LD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.06582&json=true","fetch_graph":"https://pith.science/api/pith-number/MXGAB4LD53U4P6YUE6XDT5PFLF/graph.json","fetch_events":"https://pith.science/api/pith-number/MXGAB4LD53U4P6YUE6XDT5PFLF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF/action/storage_attestation","attest_author":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF/action/author_attestation","sign_citation":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF/action/citation_signature","submit_replication":"https://pith.science/pith/MXGAB4LD53U4P6YUE6XDT5PFLF/action/replication_record"}},"created_at":"2026-08-10T01:11:21.369956+00:00","updated_at":"2026-08-10T01:11:21.369956+00:00"}