{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:DFNHQBQFWPBQDUKLOU25IOCEIG","short_pith_number":"pith:DFNHQBQF","canonical_record":{"source":{"id":"2605.05179","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-06T17:46:12Z","cross_cats_sorted":["cond-mat.dis-nn","stat.ML"],"title_canon_sha256":"e83d27cd0f0820321d5c976495742f2ea233ec3502cc7ea1a34c9f6db17548e4","abstract_canon_sha256":"bfa6caa99c84bfde05d21320502be24c830a92578b16dd19e8fc560f1eb10b03"},"schema_version":"1.0"},"canonical_sha256":"195a780605b3c301d14b7535d4384441a92e5447f7ca4f02f98abb3e4c54a2f9","source":{"kind":"arxiv","id":"2605.05179","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.05179","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"arxiv_version","alias_value":"2605.05179v2","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.05179","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_12","alias_value":"DFNHQBQFWPBQ","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_16","alias_value":"DFNHQBQFWPBQDUKL","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_8","alias_value":"DFNHQBQF","created_at":"2026-05-20T00:00:40Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:DFNHQBQFWPBQDUKLOU25IOCEIG","target":"record","payload":{"canonical_record":{"source":{"id":"2605.05179","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-06T17:46:12Z","cross_cats_sorted":["cond-mat.dis-nn","stat.ML"],"title_canon_sha256":"e83d27cd0f0820321d5c976495742f2ea233ec3502cc7ea1a34c9f6db17548e4","abstract_canon_sha256":"bfa6caa99c84bfde05d21320502be24c830a92578b16dd19e8fc560f1eb10b03"},"schema_version":"1.0"},"canonical_sha256":"195a780605b3c301d14b7535d4384441a92e5447f7ca4f02f98abb3e4c54a2f9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:40.856368Z","signature_b64":"DxjnqUbPkAE354cwwlGY0GljskQlhc7r5ey/DyG7ENYeBzcxjpBzlyzhWy90NQFP6jYKmwsKbcIAxqHKfXUzCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"195a780605b3c301d14b7535d4384441a92e5447f7ca4f02f98abb3e4c54a2f9","last_reissued_at":"2026-05-20T00:00:40.855454Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:40.855454Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.05179","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rf51Gss7AoDEb6u3W4H4C/BHwkuPLRQ8Ea/r9CC96uWri1szgC0+Gza+xCk9yIkcF9EPxKEzEQ+x0fQpj58CAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T05:59:03.962840Z"},"content_sha256":"41391a69b59a2d3525f9c15575d7994dd24ea0c077b7f35da3b1bbbdb632b6a2","schema_version":"1.0","event_id":"sha256:41391a69b59a2d3525f9c15575d7994dd24ea0c077b7f35da3b1bbbdb632b6a2"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:DFNHQBQFWPBQDUKLOU25IOCEIG","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Estimating the expected output of wide random MLPs more efficiently than sampling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations.","cross_cats":["cond-mat.dis-nn","stat.ML"],"primary_cat":"cs.LG","authors_text":"George Robinson, Jacob Hilton, Michael Winer, Paul Christiano, Victor Lecomte, Wilson Wu","submitted_at":"2026-05-06T17:46:12Z","abstract_excerpt":"By far the most common way to estimate an expected loss in machine learning is to draw samples, compute the loss on each one, and take the empirical average. However, sampling is not necessarily optimal. Given an MLP at initialization, we show how to estimate its expected output over Gaussian inputs without running samples through the network at all. Instead, we produce approximate representations of the distributions of activations at each layer, leveraging tools such as cumulants and Hermite expansions. We show both theoretically and empirically that for sufficiently wide networks, our estim"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We show both theoretically and empirically that for sufficiently wide networks, our estimator achieves a target mean squared error using substantially fewer FLOPs than Monte Carlo sampling.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The claim depends on the networks being sufficiently wide for the cumulant and Hermite approximations of activation distributions to remain accurate; the abstract qualifies the result with this width condition but does not specify the scaling or error bounds that would make the assumption hold for a given target MSE.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"For sufficiently wide random MLPs, cumulant and Hermite approximations of layer-wise activation distributions yield expected outputs at lower computational cost than Monte Carlo sampling, with good performance on rare-event probabilities.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"8fbdc12f9e2f11b2ae3245988eb697520e8bc8cff9024f19661764e1079ad485"},"source":{"id":"2605.05179","kind":"arxiv","version":2},"verdict":{"id":"3f5036cb-150c-4f26-92bc-2f2d99e0220e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T17:23:29.366184Z","strongest_claim":"We show both theoretically and empirically that for sufficiently wide networks, our estimator achieves a target mean squared error using substantially fewer FLOPs than Monte Carlo sampling.","one_line_summary":"For sufficiently wide random MLPs, cumulant and Hermite approximations of layer-wise activation distributions yield expected outputs at lower computational cost than Monte Carlo sampling, with good performance on rare-event probabilities.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The claim depends on the networks being sufficiently wide for the cumulant and Hermite approximations of activation distributions to remain accurate; the abstract qualifies the result with this width condition but does not specify the scaling or error bounds that would make the assumption hold for a given target MSE.","pith_extraction_headline":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.05179/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T21:01:19.755676Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T13:47:15.455270Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"49e50c907f5a4c3eb5b3444b04868498f96d774b5a4543761575b018a7947899"},"references":{"count":26,"sample":[{"doi":"10.1090/mcom/4127","year":2017,"title":"URLhttps://www.alignment.org/blog/competing-with-sampling/. J. Pennington, S. S. Schoenholz, and S. Ganguli. Resurrecting the sigmoid in deep learning through dynamical isometry: theory and practice. ","work_id":"bbe1c6ec-a544-413a-bd96-43457aec3c48","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Our algorithms perform poorly at low width, especially when the number of hidden layers L is large, and can even getworsewith the maximum cumulant order K in this setting","work_id":"a38cc0bf-8753-4c65-8017-8740ac38c481","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":", ϕ(Z n) from the cumu- lants of Z1,","work_id":"326620d1-2904-45cc-9a5e-5af38de3dc78","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"When the maximum cumulant order K is odd, the basic version of our algorithm tracks the full trace of the (K+ 1) th-order cumulant tensor, as defined in Section 4.1. The ablated version of our algorit","work_id":"47e14565-b591-4f2d-9102-4389025e65a3","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"However, it performs very poorly once the probability drops below 1/(number of samples), since the squared error is very large in the unlikely event that there is a positive sample","work_id":"797124c4-6ea5-49c0-8d47-9a6097a2b1b5","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":26,"snapshot_sha256":"16351e9c24b405a91dabf3e8e1f610a1ffed6be350c9f1881397def877737547","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"02acf1ba0f4b5e8bdd915eaed5562e5480b37ce13f886fee3f5e3523e21a0f1c"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"3f5036cb-150c-4f26-92bc-2f2d99e0220e"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/4+HgMvKMAxKGg8XWcKpJnovL8uEIElMlXvpIycV2eDXsqOeJ9Dt+oxoZ1UKpd2DNqSRSxVoGeZWE8nXhXBHBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-28T05:59:03.963550Z"},"content_sha256":"9f20b4b76f9fd1a90cf6e933830e694a1b9018a61bc1a6cb0b7a464c510ed00a","schema_version":"1.0","event_id":"sha256:9f20b4b76f9fd1a90cf6e933830e694a1b9018a61bc1a6cb0b7a464c510ed00a"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/bundle.json","state_url":"https://pith.science/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-28T05:59:03Z","links":{"resolver":"https://pith.science/pith/DFNHQBQFWPBQDUKLOU25IOCEIG","bundle":"https://pith.science/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/bundle.json","state":"https://pith.science/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/state.json","well_known_bundle":"https://pith.science/.well-known/pith/DFNHQBQFWPBQDUKLOU25IOCEIG/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:DFNHQBQFWPBQDUKLOU25IOCEIG","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"bfa6caa99c84bfde05d21320502be24c830a92578b16dd19e8fc560f1eb10b03","cross_cats_sorted":["cond-mat.dis-nn","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-06T17:46:12Z","title_canon_sha256":"e83d27cd0f0820321d5c976495742f2ea233ec3502cc7ea1a34c9f6db17548e4"},"schema_version":"1.0","source":{"id":"2605.05179","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.05179","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"arxiv_version","alias_value":"2605.05179v2","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.05179","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_12","alias_value":"DFNHQBQFWPBQ","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_16","alias_value":"DFNHQBQFWPBQDUKL","created_at":"2026-05-20T00:00:40Z"},{"alias_kind":"pith_short_8","alias_value":"DFNHQBQF","created_at":"2026-05-20T00:00:40Z"}],"graph_snapshots":[{"event_id":"sha256:9f20b4b76f9fd1a90cf6e933830e694a1b9018a61bc1a6cb0b7a464c510ed00a","target":"graph","created_at":"2026-05-20T00:00:40Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We show both theoretically and empirically that for sufficiently wide networks, our estimator achieves a target mean squared error using substantially fewer FLOPs than Monte Carlo sampling."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The claim depends on the networks being sufficiently wide for the cumulant and Hermite approximations of activation distributions to remain accurate; the abstract qualifies the result with this width condition but does not specify the scaling or error bounds that would make the assumption hold for a given target MSE."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"For sufficiently wide random MLPs, cumulant and Hermite approximations of layer-wise activation distributions yield expected outputs at lower computational cost than Monte Carlo sampling, with good performance on rare-event probabilities."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations."}],"snapshot_sha256":"8fbdc12f9e2f11b2ae3245988eb697520e8bc8cff9024f19661764e1079ad485"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"02acf1ba0f4b5e8bdd915eaed5562e5480b37ce13f886fee3f5e3523e21a0f1c"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T21:01:19.755676Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T13:47:15.455270Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.05179/integrity.json","findings":[],"snapshot_sha256":"49e50c907f5a4c3eb5b3444b04868498f96d774b5a4543761575b018a7947899","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"By far the most common way to estimate an expected loss in machine learning is to draw samples, compute the loss on each one, and take the empirical average. However, sampling is not necessarily optimal. Given an MLP at initialization, we show how to estimate its expected output over Gaussian inputs without running samples through the network at all. Instead, we produce approximate representations of the distributions of activations at each layer, leveraging tools such as cumulants and Hermite expansions. We show both theoretically and empirically that for sufficiently wide networks, our estim","authors_text":"George Robinson, Jacob Hilton, Michael Winer, Paul Christiano, Victor Lecomte, Wilson Wu","cross_cats":["cond-mat.dis-nn","stat.ML"],"headline":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-06T17:46:12Z","title":"Estimating the expected output of wide random MLPs more efficiently than sampling"},"references":{"count":26,"internal_anchors":0,"resolved_work":26,"sample":[{"cited_arxiv_id":"","doi":"10.1090/mcom/4127","is_internal_anchor":false,"ref_index":1,"title":"URLhttps://www.alignment.org/blog/competing-with-sampling/. J. Pennington, S. S. Schoenholz, and S. Ganguli. Resurrecting the sigmoid in deep learning through dynamical isometry: theory and practice. ","work_id":"bbe1c6ec-a544-413a-bd96-43457aec3c48","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Our algorithms perform poorly at low width, especially when the number of hidden layers L is large, and can even getworsewith the maximum cumulant order K in this setting","work_id":"a38cc0bf-8753-4c65-8017-8740ac38c481","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":", ϕ(Z n) from the cumu- lants of Z1,","work_id":"326620d1-2904-45cc-9a5e-5af38de3dc78","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"When the maximum cumulant order K is odd, the basic version of our algorithm tracks the full trace of the (K+ 1) th-order cumulant tensor, as defined in Section 4.1. The ablated version of our algorit","work_id":"47e14565-b591-4f2d-9102-4389025e65a3","year":2016},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"However, it performs very poorly once the probability drops below 1/(number of samples), since the squared error is very large in the unlikely event that there is a positive sample","work_id":"797124c4-6ea5-49c0-8d47-9a6097a2b1b5","year":null}],"snapshot_sha256":"16351e9c24b405a91dabf3e8e1f610a1ffed6be350c9f1881397def877737547"},"source":{"id":"2605.05179","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-19T17:23:29.366184Z","id":"3f5036cb-150c-4f26-92bc-2f2d99e0220e","model_set":{"reader":"grok-4.3"},"one_line_summary":"For sufficiently wide random MLPs, cumulant and Hermite approximations of layer-wise activation distributions yield expected outputs at lower computational cost than Monte Carlo sampling, with good performance on rare-event probabilities.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"The expected output of wide random MLPs can be estimated without sampling using layer-wise cumulant and Hermite approximations.","strongest_claim":"We show both theoretically and empirically that for sufficiently wide networks, our estimator achieves a target mean squared error using substantially fewer FLOPs than Monte Carlo sampling.","weakest_assumption":"The claim depends on the networks being sufficiently wide for the cumulant and Hermite approximations of activation distributions to remain accurate; the abstract qualifies the result with this width condition but does not specify the scaling or error bounds that would make the assumption hold for a given target MSE."}},"verdict_id":"3f5036cb-150c-4f26-92bc-2f2d99e0220e"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:41391a69b59a2d3525f9c15575d7994dd24ea0c077b7f35da3b1bbbdb632b6a2","target":"record","created_at":"2026-05-20T00:00:40Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"bfa6caa99c84bfde05d21320502be24c830a92578b16dd19e8fc560f1eb10b03","cross_cats_sorted":["cond-mat.dis-nn","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-06T17:46:12Z","title_canon_sha256":"e83d27cd0f0820321d5c976495742f2ea233ec3502cc7ea1a34c9f6db17548e4"},"schema_version":"1.0","source":{"id":"2605.05179","kind":"arxiv","version":2}},"canonical_sha256":"195a780605b3c301d14b7535d4384441a92e5447f7ca4f02f98abb3e4c54a2f9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"195a780605b3c301d14b7535d4384441a92e5447f7ca4f02f98abb3e4c54a2f9","first_computed_at":"2026-05-20T00:00:40.855454Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:40.855454Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"DxjnqUbPkAE354cwwlGY0GljskQlhc7r5ey/DyG7ENYeBzcxjpBzlyzhWy90NQFP6jYKmwsKbcIAxqHKfXUzCg==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:40.856368Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.05179","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:41391a69b59a2d3525f9c15575d7994dd24ea0c077b7f35da3b1bbbdb632b6a2","sha256:9f20b4b76f9fd1a90cf6e933830e694a1b9018a61bc1a6cb0b7a464c510ed00a"],"state_sha256":"70fb016449dea665a99ede0037889719cf0d701119508ef7dea81467077a0eb8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"TrzYL+hmOi1HxHXgN6BSeLZ02y4TGceURwBykofkQnuJiqvq8lJRtm1yTZoYgJlXwkWNMwL871RRR7vCrCZXCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-28T05:59:03.966393Z","bundle_sha256":"d84887753ee58d07af474c76348b3fb9b68c78a825c63815c4dff79fb821ed61"}}