{"id":"arxiv:2604.19341","title":"Structured Scaling of AI Discovery Across Diverse Scientific Domains","date":"2026-04-21","section":"Harness","quality":0.65138,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/","source_url":"https://arxiv.org/abs/2604.19341","text_available":true,"authors":["Haotian Ye","Haowei Lin","Jingyi Tang","Yizhen Luo"],"outgoing":[{"source":"arxiv:2604.19341","target":"arxiv:2001.08361","evidence":"id","id":"arxiv:2001.08361","title":"Scaling Laws for Neural Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/82302458a6d03bc051b2e4fdb8277893a81d4bd6a029969045df6a8380904eac/"},{"source":"arxiv:2604.19341","target":"arxiv:2203.11171","evidence":"title","id":"arxiv:2203.11171","title":"Self-Consistency Improves Chain of Thought Reasoning in Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/02a15b098d96a94d463bb470f59a34723abc34e59196f35c539f9796d8f3c3df/"},{"source":"arxiv:2604.19341","target":"arxiv:2210.10760","evidence":"title","id":"arxiv:2210.10760","title":"Scaling Laws for Reward Model Overoptimization","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/fd55acb4c34a2366d8f9c1fc0672a20be0ac4da8643118d0cbf627e891346d2d/"},{"source":"arxiv:2604.19341","target":"arxiv:2303.17651","evidence":"title","id":"arxiv:2303.17651","title":"Self-Refine: Iterative Refinement with Self-Feedback","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/accef2157d87cf33be93cb3a9af246528b0a3742a9e2e700a194956fac8f5cf4/"},{"source":"arxiv:2604.19341","target":"arxiv:2402.03300","evidence":"id","id":"arxiv:2402.03300","title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/636fb871904bc06f742cfdea069796f9e9665f95bd58dc63260eaff5154ef985/"},{"source":"arxiv:2604.19341","target":"arxiv:2410.07095","evidence":"id","id":"arxiv:2410.07095","title":"MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/f0a0c6321ec079098a36ad19b3900d77197a3af1f6bbf6eee8a083e5eb03d42e/"},{"source":"arxiv:2604.19341","target":"arxiv:2501.19393","evidence":"title","id":"arxiv:2501.19393","title":"s1: Simple test-time scaling","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/2a88818f7605b71cd0160831faebd51b3274c24ffd29626b4d519390b4bffffe/"},{"source":"arxiv:2604.19341","target":"arxiv:2505.21444","evidence":"id","id":"arxiv:2505.21444","title":"Can Large Reasoning Models Self-Train?","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/206224ca50098e0bdc230140722feec107cfc90c1cef2ed8d3ed52355ef56bca/"},{"source":"arxiv:2604.19341","target":"arxiv:2507.19457","evidence":"id","id":"arxiv:2507.19457","title":"GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c1ef72d96fa3292590a6a3d112ab509840081b59aaa771cf42b8b0e8ef62ba08/"},{"source":"arxiv:2604.19341","target":"arxiv:2601.16175","evidence":"id","id":"arxiv:2601.16175","title":"Learning to Discover at Test Time","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/5d0b45cbbc72945765e689cbfb7ac7e4e92dfbb764861b27682ac0744ef05b70/"},{"source":"arxiv:2604.19341","target":"doi:10.1038/s41586-023-06924-6","evidence":"id","id":"doi:10.1038/s41586-023-06924-6","title":"Mathematical discoveries from program search with large language models","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/70a6b427d42fd28c9d538648e148a97db2fb2614c7764f3231e53f4bebdce887/"},{"source":"arxiv:2604.19341","target":"url:sakana.ai/shinka-evolve","evidence":"id","id":"url:sakana.ai/shinka-evolve","title":"ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/a13eace17f3407cb1634b579b2738f15f1e035897a8f1842305198c6d2bf35e1/"}],"incoming":[{"source":"arxiv:2605.13821","target":"arxiv:2604.19341","evidence":"id","id":"arxiv:2605.13821","title":"Harnessing Agentic Evolution","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/"},{"source":"arxiv:2609.14858","target":"arxiv:2604.19341","evidence":"id","id":"arxiv:2609.14858","title":"Dream-RSI: Recursive Self-Improvement through Evolving Worlds","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/4c9ce361df4d9a5f89f3dc0a70a71e795eb66f2c4e3c262f73f04b2d117c8ea1/"}],"text_parts":[{"start":0,"end":15628,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-1.html"},{"start":15628,"end":31592,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html"},{"start":31592,"end":47458,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-3.html"},{"start":47458,"end":62964,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html"},{"start":62964,"end":77722,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html"},{"start":77722,"end":93604,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html"},{"start":93604,"end":109590,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html"},{"start":109590,"end":125360,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html"},{"start":125360,"end":140742,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html"},{"start":140742,"end":156284,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html"},{"start":156284,"end":171691,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html"},{"start":171691,"end":175457,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-12.html"}],"contents":[{"title":"Introduction","offset":2053,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-1.html#text"},{"title":"Overview of SimpleTES","offset":10547,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-1.html#text"},{"title":"SimpleTES discovers effective algorithms to reduce quantum compilation overhead","offset":15543,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-1.html#text"},{"title":"SimpleTES improves optimizers for fuel-efficient deep-space trajectories","offset":20401,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html#text"},{"title":"SimpleTES discovers better scientific algorithms","offset":23896,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html#text"},{"title":"SimpleTES discovers a geometry-aware lasso solver that surpasses glmnet .","offset":25108,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html#text"},{"title":"SimpleTES discovers new state-of-the-art predictor for Zebrafish whole-brain neural-activity.","offset":28508,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html#text"},{"title":"SimpleTES improves AI foundations","offset":31553,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-2.html#text"},{"title":"GPU kernel optimization.","offset":32522,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-3.html#text"},{"title":"Scaling-law discovery.","offset":35899,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-3.html#text"},{"title":"Post-training the LLM for SimpleTES leads to new mathematical discoveries","offset":38409,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-3.html#text"},{"title":"Discussion","offset":42438,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-3.html#text"},{"title":"Code availability","offset":48806,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Supplementary Information","offset":49208,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Contents","offset":49238,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"A Authors","offset":49252,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"B Related Work","offset":49913,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"B.1 Existing Evaluation-Driven Discovery Methods","offset":49933,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"AlphaEvolve [ 100 ] .","offset":51071,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"OpenEvolve [ 6 ] .","offset":52219,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"ShinkaEvolve [ 73 ] .","offset":53249,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"ThetaEvolve [ 155 ] .","offset":54515,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"TTT-Discover [ 165 ] .","offset":55757,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"AdaEvolve [ 23 ] .","offset":57006,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"EvoX [ 82 ] .","offset":58236,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"GEPA [ 2 ] .","offset":59430,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Summary.","offset":60688,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"B.2 Self-evolving AI","offset":61680,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Reward-based Self-Evolution.","offset":62188,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Imitation & Demonstration Learning.","offset":62921,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-4.html#text"},{"title":"Population-based & Evolutionary Methods.","offset":63545,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"B.3 LLM for Scientific Discovery","offset":64854,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"LLM based scientific discovery systems.","offset":65441,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Scientific benchmarks for LLMs.","offset":66092,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Important results proposed by LLM systems.","offset":66873,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"B.4 Test-Time Scaling","offset":68123,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Parallel Scaling.","offset":68845,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Sequential Scaling.","offset":69508,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Hybrid Scaling.","offset":70151,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Internal Scaling.","offset":70829,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"C Theoretical Modeling of SimpleTES","offset":71717,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Definition .2 (Multidimensional Problem Refinement Trajectory) .","offset":73467,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Number of global independent trials C C .","offset":76003,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Theorem .3 .","offset":76532,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Proof.","offset":77298,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-5.html#text"},{"title":"Remark.","offset":81533,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"Local sample size K K .","offset":83228,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"D Method Analysis","offset":86106,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"D.1 Experiments on the Scaling Behavior of SimpleTES","offset":86754,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"Scalability of global width ( C C ) and refinement depth ( L L ).","offset":87386,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"Scalability of local sample size ( K K ) and refinement depth ( L L ).","offset":89568,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"Scalability of the total evaluation budget ( N N ).","offset":91028,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"D.2 Experimental Details on Post-Training","offset":91941,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-6.html#text"},{"title":"Training improves the trajectory scores on the training tasks.","offset":94444,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Training enables generalization to the held-out tasks.","offset":95313,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Training unlocks stronger solutions that vanilla models fail to uncover.","offset":96661,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"D.3 From Golden Metrics to Surrogate: Hacking Analysis","offset":97246,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Hacking in math construction tasks.","offset":97860,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Hacking in kernel optimization tasks.","offset":99202,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Discussion.","offset":104753,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"E Variants of Proposal Constructor Φ \\Phi","offset":105599,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"Balance policy.","offset":105646,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"LLM-Elite policy.","offset":107436,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-7.html#text"},{"title":"F Ablation Studies","offset":109590,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"F.1 Ablations on Different Designs of Φ \\Phi","offset":109787,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"F.2 Ablation on Reflection and Failure Patterns","offset":112993,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"F.3 Efficiency Analysis: Trajectory-level Pruning","offset":115038,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"G Wall-clock Time Efficiency","offset":118624,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"H Quantum Compilation Tasks","offset":119450,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"H.1 Problem definitions","offset":119832,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"Superconducting qubit routing.","offset":119862,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"Zoned neutral-atom compilation.","offset":121120,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"H.2 Experimental scaffolds and evaluation setup","offset":123544,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"Superconducting routing setup.","offset":123598,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"Zoned neutral-atom setup.","offset":125327,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-8.html#text"},{"title":"H.3 Discovered programs","offset":127075,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"Discovered superconducting algorithm.","offset":127105,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"Discovered zoned neutral-atom algorithm.","offset":127940,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"H.4 Discovered superconducting routing policy using Gemini","offset":130052,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I Astrodynamics trajectory design","offset":131618,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I.1 Task formulation and program interface","offset":132131,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I.2 Independent trajectory evaluator","offset":133587,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I.3 Benchmark instances","offset":137831,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I.4 Construction of historical-sequence references","offset":140031,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-9.html#text"},{"title":"I.5 How evolution changes mission-design search","offset":141451,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"J GPU Kernel Optimization","offset":144776,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Task definitions and benchmark settings.","offset":144807,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"TriMul evaluation alignment.","offset":145560,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Cross-device performance generalization.","offset":146420,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"K Scaling Law Discovery","offset":146923,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Overview.","offset":146952,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Experimental setting.","offset":149142,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Results analysis.","offset":151313,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Case study.","offset":152323,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"L Whole-brain neural-activity forecast on ZAPBench","offset":156180,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Problem definition and benchmark design.","offset":156236,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-10.html#text"},{"title":"Forecasting task and data splits.","offset":157511,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Evaluation and search protocol.","offset":158414,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"M Single-Cell RNA-Seq Denoising","offset":159151,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"N Lasso Regularization Path","offset":162134,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Overview.","offset":162167,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Experimental setting.","offset":162864,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Results analysis.","offset":164169,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Case study.","offset":164828,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"O Mathematical Scientific Discovery Tasks","offset":165487,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"O.1 Extremal analysis","offset":165988,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Erdős minimum-overlap problem.","offset":166306,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Autocorrelation Inequalities.","offset":167941,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"O.2 Combinatorial construction","offset":169741,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Sum–Difference Problem.","offset":170115,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Circle Packing in a Unit Square.","offset":171478,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-11.html#text"},{"title":"Hadamard maximum-determinant problem of order 29 29 .","offset":172901,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-12.html#text"},{"title":"Instructions for reporting errors","offset":174493,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/part-12.html#text"}],"text_code_points":175457,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/source.md"}