AutoSND: From Execution Evidence to Structural Policies for Automated Network Dismantling Heuristic Discovery

Changjun Fan; Yufan Deng; Zhiguang Cao; Zhijing Hu

REVIEW 4 major objections 3 minor 35 references

AutoSND: From Execution Evidence to Structural Policies for Automated Network Dismantling Heuristic Discovery

T0 review · 4 major / 3 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Machine-found heuristics beat hand-built rules on 12 real networks

desk verdict AutoSND's Stage II structural-policy induction is a real step forward for LLM-based heuristic discovery, but the paper's headline comparison rests on a single search seed on one synthetic proxy, so the reliability claim is not yet demonstrated. read the letter →

arxiv 2608.03653 v1 pith:XLC4HEJR submitted 2026-08-04 cs.AI

Zhijing Hu , Changjun Fan , Yufan Deng , Zhiguang Cao This is my paper

classification cs.AI

keywords networkdismantlingautomaticheuristicdesignlargelanguagemodelsstructuralpolicyinductiontreesearchcomplexnetworksprogramsynthesis

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

AutoSND is a three-stage automated search that treats a complete network dismantling program—not just a node-scoring rule—as the object to discover. It first runs a broad tree search from a simple degree-based heuristic, archiving quality, runtime, execution states, and code structures; then it compiles these records into an explicit structural policy (reuse, avoid, bound) that governs local signals, neighborhood access, and update ranges; finally it continues tree search under that policy. The two frozen candidates, AutoSND-Q and AutoSND-S, achieve the lowest mean accumulated normalized connectivity among 18 methods completing all 12 real networks, finish all 12 within seconds, and also return valid removal sequences on million-node networks. If true, this shows execution evidence can be converted into reusable structural guidance, turning LLM-based heuristic design from isolated candidate screening into interpretable, transferable program discovery.

What carries the argument

The central object is the complete removal-sequence heuristic h: G ↦ π, combined with the Execution-to-Evidence Structural Policy Induction. Stage II uses LLM-Struct to give each archived candidate a unified description z_i over a vocabulary of structural features (degree, frontier, weak_tie, two_hop, boundary, redundancy, phase, heap), then forms contrastive evidence C1 = P ∪ N (positive versus poor-quality/slow/failed candidates) and statistics T1, and compiles them into an explicit policy Π = (A+, A-, B): what structures to reuse, what to avoid, and how to bound neighborhood access and state updates. Π is embedded in the Stage III generation prompt, so subsequent search is constrained by

What would settle it

Re-run the full AutoSND search with the proxy replaced by an Erdős–Rényi or Watts–Strogatz 500-node graph on the same budget and seed, freeze the resulting Q/S candidates, and evaluate them on the 12 main graphs. If the frozen candidates drop below 12/12 coverage or their mean 100×ANC(GCC) rises above the best native baseline (about 12.5), then the transfer from the power-law proxy does not follow as described.

Watch

Extended reading notes

Core claim

AutoSND claims that a three-stage LLM-driven tree search can discover complete executable network dismantling heuristics whose quality and runtime are competitive with—and on the tested networks better than—expert-designed and learning-based baselines. Stage II is the key step: every archived candidate is annotated with a unified structural description, aligned with its ANC@30%, runtime, and execution state, and compiled into an explicit structural policy Π = (A+, A-, B) that says what to reuse, what to avoid, and how to bound neighborhood access and state updates. Stage III continues generation conditioned on this policy and selects quality- and speed-prioritized endpoints from the quality-

Load-bearing premise

The search is run and the final candidates are frozen using one 500-node synthetic power-law graph as the only proxy; the reported quality and coverage assume that structural patterns learned on this single graph transfer to real networks of different sizes, densities, and formation mechanisms.

Editorial extensions

If this is right

Complete heuristics, not scoring functions: the discovered objects output a full removal sequence, so they can be used directly in robustness and vulnerability analysis.
Structural policies make search interpretable: Stage II turns one run's execution evidence into explicit reuse/avoid/bound rules, preventing costly whole-graph rescans from recurring in later generations.
Transfer across scales: candidates frozen from a 500-node synthetic power-law proxy produce valid complete sequences on 12 real networks and on million-node networks, suggesting the induced structure is not overfit to proxy size.
A reusable pattern: the final programs support a testable heuristic-design principle—residual degree backbone plus bounded local corrections plus restricted update scope—that can seed future manual or automatic designs.
Cross-task feasibility: the paper reports that the same three-stage process transfers to influence maximization, improving quality and runtime over compared controls, though this is presented as a limited feasibility result.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The explicit policy induction could be decoupled from tree search and used to distill a corpus of previously run candidate programs into design rules for other algorithm-discovery settings; the paper tests this only for network dismantling and one influence-maximization transfer.
The comparison depends on a single 500-node synthetic proxy; searching on a small portfolio of proxies spanning power-law, Erdős–Rényi, Watts–Strogatz, and stochastic block model graphs would test the transfer claim more directly than the current single-proxy design.
The final score formula resembles classical dismantling ideas (degree, CI, CoreHD) but with explicit caps on access and updates; ablating the caps alone would quantify how much of the gain comes from bounded computation versus the local signals themselves.
A practical test is to use the induced policy Π as a prompt-level prior for zero-shot generation without tree search, comparing validity and quality to see whether the policy itself, rather than the search process, carries the discovered knowledge.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The genuinely new thing here is Stage II: compiling execution records into explicit structural policies that condition later generation. That is a real step beyond scalar fitness or language reflection, and the paper does a lot right. The three-stage tree search with archived execution evidence, LLM-Struct annotations, contrastive positive/negative evidence, and a compiled "reuse, avoid, bound" constraint is a solid mechanism. The discovered AutoSND-Q/S are complete executable heuristics, not just scoring functions, and the evaluation is broad: 12 real networks, 3 large real networks, component ablations, lineage analysis, a CoreHD-root robustness check, a model-swap experiment, and an influence-maximization transfer. The candidates themselves are competitive and fast, and the paper ships code and a reproduction manifest with hashes, which earns real trust.

The soft spot is precisely what the stress-test note flags: the central claim of "better search performance and stability" is underdetermined. All endpoints in Table 1 come from a single search run with seed 42 on Powerlaw_500. There are no repeated-search statistics, and the ablations mostly use the same proxy and seed. The CoreHD-root runs with seeds 42 and 43 give some additional evidence, but the improvement over CoreHD is only 1.2% on Q, so it does not substitute for repeated full-pipeline runs. Without multi-seed and ideally multi-proxy searches, the Table 1 margins could be a favorable draw. That does not invalidate the existence result—the paper shows a discovery path exists—but it does not support the reliability claim as stated. Also, several AHD baselines are local reproductions rather than original implementations, so the exact head-to-head margins should be read with care, though the appendix documents the reproductions unusually well.

Who is this for? Researchers in LLM-based heuristic discovery and network dismantling. The mechanism is worth discussing and the paper deserves a serious referee. The revision should ask for multi-seed runs (at least three to five seeds) and confidence intervals for the 12-network means before the comparative claims are treated as definitive.

Recommendation: send it to peer review with that request. The core idea is sound and the reported reproducibility artifacts make the work worth engaging with.

Referee Report

4 major / 3 minor

Summary. The paper proposes AutoSND, a three-stage framework for automatic heuristic discovery for network dismantling. Stage I performs a broad LLM-driven tree search from an HDA root, archiving candidate code, execution metrics, and runtime. Stage II applies LLM-Struct to every archived candidate, aligns structures with quality/runtime/execution states, and compiles a structural policy (reuse/avoid/bound) from positive and negative evidence. Stage III performs policy-conditioned tree search with four branch seed types and selects quality- and speed-prioritized endpoints (AutoSND-Q/S) from a Pareto frontier. The main experiments evaluate the two frozen candidates on 12 real networks and 3 large networks, reporting lower mean ANC(GCC) than 18 baselines, near-100% candidate validity, complete coverage, and low runtime. Ablations test the contribution of Stage II and structural-family diversity, and a component-level analysis interprets the final programs as residual-degree backbones with bounded local corrections.

Significance. If the empirical claims hold, the paper makes a useful contribution to LLM-based automatic heuristic design by converting execution evidence into structural-level constraints, a step beyond scalar-fitness or language-reflection feedback. The strengths are substantial: the code is released; the large-network runs are strict complete-sequence verifications with hashes and independently recomputed ANC; and the component ablations in Appendix D.1 directly test which code structures matter. The lineage analysis in Section 5.3.2 and the structural-policy tracing in Appendix E provide unusually detailed provenance. The main risk is that the central comparative claim is established from a single search trajectory, which I detail below.

major comments (4)

[§5.1.4, §A.3, §E.1] The entire comparative evaluation rests on one stochastic execution: search seed 42 on the single 500-node Powerlaw_500 proxy. The frozen endpoints AutoSND-Q (cd9e3818033d) and AutoSND-S (0ade8d3405c2) are selected from that run. Temperature-0.2 LLM generation is stochastic, so the reported mean ANC, coverage, and validity are outcomes of a single draw, not estimates of typical behavior. The DeepSeek-v4-flash and CoreHD-root experiments use the same seed/proxy, so they do not address seed sensitivity. The abstract's 'search performance and stability' claim is therefore underdetermined. Please run multiple search seeds (e.g., 5) and at least one additional proxy family (ER/WS/SBM), and report the distribution of external metrics.
[Table 1, §5.2.1] The 12-graph means for AutoSND-Q/S (10.99/10.93) are close to Clade-AHD (11.19) and ERA (11.12), yet no confidence intervals, standard errors, or significance tests are given. The claim that AutoSND is 'better' is based on point estimates over 12 graphs, some of which are near-ties. The average rank comparison (3.50 vs 4.92) also lacks a measure of variability. Please provide per-graph paired differences, standard errors, or a paired test against the strongest baselines.
[§5.1.3, §C.2] All AHD baselines are local reproductions under a unified interface, not original implementations. The validity, coverage, and runtime comparisons may be influenced by reproduction choices. For example, LLM4CN completes 9/12 and HSEvo 9/12; these may not reflect the published methods' behavior. Unless the reproductions are validated against known outputs or the limitation is explicitly stated, the 'AutoSND vs. published methods' comparison is not a strict comparison. Please add a fidelity check or treat the comparison as 'AutoSND vs. local re-implementations' throughout.
[§5.3.1, Fig. 4] The ablation interpretation is partially underspecified. The w/o-Stage-II variant has slightly lower mean ANC on the 8 common graphs (the text says 'slightly lower') but falls to 8/12 coverage. The conclusion 'Stage II mainly supports cross-network executability' rests on coverage differences over a small, non-random set of 8 graphs; no test is given. Similarly, the w/o-family variant changes both seed diversity and seed selection procedure, so it does not isolate the structural-family restriction. Please clarify what is held fixed and add error bars or tests for the ablation comparisons.

minor comments (3)

[§3.1, Eqs. (2)-(7)] The notation for complete-sequence ANC in Eq. (2) uses N both as node count and as the number of removed nodes, while the search metric in Eq. (7) uses K. A brief sentence distinguishing the full-sequence metric from the search-time ANC@30% would improve readability.
[Fig. 3 caption] The figure caption states 'The red solid line denotes AutoSND-S' but does not label the curves for ERA, FunSearch, Clade-AHD, MCTS-AHD, and AlphaEvolve. Please add a legend or explicit color mapping.
[§C.1, Table 6] The appendix states that PH and hamster have identical undirected simple edge sets and that Collaboration and GrQC are isomorphic, yet both pairs appear as distinct rows in Table 6. Clarify why both names are retained and whether the main-test label differs from the archive label only by provenance.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported heuristics are frozen on a separate search proxy before external evaluation, and the structural policy is an empirical summary tested by ablations.

full rationale

AutoSND's derivation chain is self-contained rather than circular. Stage I performs tree search on a single 500-node synthetic Powerlaw_500 proxy; Stage II induces a structural policy by contrasting successful and failed candidates from that archive; Stage III continues search on the same proxy and freezes two endpoints (AutoSND-Q cd9e3818033d, AutoSND-S 0ade8d3405c2) from the ANC@30%-runtime frontier (Appendix A.3, Eq. 38). All main-text quality claims are then obtained by replaying these frozen candidates on 12 real networks and 3 large real networks that were not used during search (Sections 5.1.1 and 5.2). The evaluation metric for the headline comparison, complete-sequence 100×ANC(GCC), is not the same as the search-time ANC@30%(GCC); there is no equation in which the reported improvement is a fitted parameter renamed as a prediction. The structural-policy contribution is tested rather than assumed: the w/o-Stage-II ablation and w/o-family ablation (Fig. 4, Table 19) show that removing the policy degrades coverage or validity, and the CoreHD-root rerun (Appendix D.5) and DeepSeek-v4-flash rerun (Appendix G) probe sensitivity to initialization and generator. The FINDER citation is a baseline and data source, not a load-bearing uniqueness or ansatz argument, and the FINDER curves are reproduced from released files with an explicit offset correction (Appendix D.4). The only substantive weakness is that the entire comparative evidence rests on one seed and one proxy graph, but that is a statistical robustness limitation, not a circularity: it does not make the external results true by construction. Therefore no circular step meets the evidentiary bar of Eq. X = Eq. Y or fitted-input-called-prediction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The search is empirical and LLM-driven; there is no closed-form derivation. The central claims rest on hand-tuned hyperparameters, a single-proxy transfer assumption, and the reliability of LLM-generated code and annotations.

free parameters (7)

Stage I/III quality-runtime weights (alpha, beta) = 0.80/0.20
Hand-chosen; balances ANC@30% and runtime during parent selection
Parent exploration bonus lambda1, lambda3 = 0.10
Hand-chosen to encourage less-expanded parents
Speed-branch quality gate r_S = 0.55
Hand-chosen threshold on normalized quality rank for S selection
Branch seed quotas K_Q, K_S, K_B, K_R = 8, 6, 6, 4
Hand-chosen seed counts for Q/S/B/R branches
Stage III branch expansion allocation = 60/60/50/30
Fixed 200 Stage III expansions split across branches
Initial policy caps neighbor/two-hop = 64/128
BasePolicy default caps before LLM proposals refine them
Search horizon = 30% of nodes
ANC@30% used as proxy for full-sequence quality during search

assumptions (5)

domain assumption A single 500-node Powerlaw_500 graph is a sufficient proxy for discovering heuristics that generalize to real networks
Section 5.1.1 uses only this graph in search; external performance assumes transfer.
domain assumption ANC@30% on the proxy is a faithful proxy for full-sequence ANC(GCC)
Eq. (7) in Sec 4.1; candidates are selected on this metric, final results use full ANC.
domain assumption LLM-Struct's structural annotations faithfully describe candidate code
Section 4.2.1; policy compilation depends on z_i.
domain assumption Local reproductions of AHD baselines preserve the mechanisms of the original methods
Appendix C.2; comparisons assume faithful baselines.
domain assumption LLM code generation (gpt-5.5) can produce executable candidates with sufficient frequency
Entire search depends on LLM producing valid code; validity rates vary (99% main, 79% DeepSeek appendix).

how reviews work

0 comments

Cite this review

Pith. "Pith review of AutoSND: From Execution Evidence to Structural Policies for Automated Network Dismantling Heuristic Discovery." pith.science (2026). https://pith.science/paper/XLC4HEJR

@misc{pith2026260803653,
  author       = {Pith},
  title        = {Pith review of: AutoSND: From Execution Evidence to Structural Policies for Automated Network Dismantling Heuristic Discovery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XLC4HEJR}},
  note         = {Machine review of arXiv:2608.03653}
}

read the original abstract

Network dismantling is fundamental to analyzing the robustness and vulnerability of complex systems, yet practical heuristics must balance effectiveness and computational efficiency, and are usually designed manually by researchers. Existing large language model based automatic heuristic design methods can generate and screen candidates, yet they have difficulty further transforming candidate quality or failure states during execution into structural-level guid- ance for subsequent generation. We propose AutoSND, a three stage tree search framework for complete network dismantling pro- grams. Stage I broadly explores from simple heuristics and archives execution evidence. Stage II compiles candidate records into struc- tural policies concerning local signals, neighborhood access, and state update ranges. Stage III continues tree search conditioned on these policies and obtains the final quality prioritized and speed prioritized candidates, AutoSND-Q/S. Experiments on 12 real world networks and 3 large real world networks show that AutoSND achieves better search performance and stability and discovers more competitive and structurally interpretable network disman- tling programs. The final candidates form an interpretable structure that uses residual degree as the backbone, adjusts node order with bounded local signals, and restricts the state update range. Code is available at https://github.com/MirrorNew/AutoSND.

Figures

Figures reproduced from arXiv: 2608.03653 by the authors.

**Figure 1.** Two key challenges faced by LLM driven search for discovering network dismantling heuristics: exposing candidate [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** The three stage sequential workflow of AutoSND. AutoSND searches a network dismantling heuristic that can generate [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

**Figure 3.** Stagewise search progress of AutoSND-S and AHD methods. The left panel shows the best observed ANC@30% (GCC), [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Structural-guidance ablation of AutoSND-Q/S. Left: Stage I/III validity, Q/S main-12 coverage, and mean ANC (GCC) [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

**Figure 5.** Figure 5: Complete sequence quality on three large real [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

**Figure 6.** Figure 6: Stagewise candidate distribution and AutoSND-S lineage on [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

35 extracted references · 18 canonical work pages

[1]

Eser Aygün et al. 2026. An AI System to Help Scientists Write Expert-Level Empirical Software.Nature654 (2026), 909–916. arXiv:2509.06503. doi:10.1038/ s41586-026-10658-6

arXiv 2026
[2]

Albert-László Barabási and Réka Albert. 1999. Emergence of Scaling in Random Networks.Science286, 5439 (1999), 509–512. doi:10.1126/science.286.5439.509

doi:10.1126/science.286.5439.509 1999
[3]

Alfredo Braunstein, Luca Dall’Asta, Guilhem Semerjian, and Lenka Zdeborová

work page
[4]

Chentong Chen, Mengyuan Zhong, Ye Fan, Jialong Shi, and Jianyong Sun. 2026. HiFo-Prompt: Prompting with Hindsight and Foresight for LLM-Based Auto- matic Heuristic Design. InInternational Conference on Learning Representations. OpenReview: imSLzfZ6av, arXiv:2508.13333

arXiv 2026
[5]

Pham Vu Tuan Dat, Long Doan, and Huynh Thi Thanh Binh. 2025. HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. doi:10.1609/aaai.v39i25.34898

doi:10.1609/aaai.v39i25.34898 2025
[6]

Davis and Yifan Hu

Timothy A. Davis and Yifan Hu. 2011. The University of Florida Sparse Matrix Collection.ACM Trans. Math. Software38, 1 (2011), 1:1–1:25. doi:10.1145/2049662. 2049663

work page doi:10.1145/2049662 2011
[7]

Paul Erdős and Alfréd Rényi. 1959. On Random Graphs I.Publicationes Mathe- maticae Debrecen6 (1959), 290–297. doi:10.5486/PMD.1959.6.3-4.12

doi:10.5486/pmd.1959.6.3-4.12 1959
[8]

Changjun Fan, Li Zeng, Yizhou Sun, and Yang-Yu Liu. 2020. Finding Key Players in Complex Networks through Deep Reinforcement Learning.Nature Machine Intelligence2 (2020), 317–324. doi:10.1038/s42256-020-0177-2

doi:10.1038/s42256-020-0177-2 2020

Show all 35 references

[9]

Marco Grassia, Manlio De Domenico, and Giuseppe Mangioni. 2021. Machine Learning Dismantling and Early-Warning Signals of Disintegration in Complex Systems.Nature Communications12 (2021). doi:10.1038/s41467-021-25485-8

2021 doi
[10]

Roger Guimerà, Leon Danon, Albert Díaz-Guilera, Francesc Giralt, and Alex Are- nas. 2003. Self-Similar Community Structure in a Network of Human Interactions. Physical Review E68 (2003), 065103. doi:10.1103/PhysRevE.68.065103

2003 doi
[11]

Chungu Guo, Liangwei Yang, Xiao Chen, Duanbing Chen, Hui Gao, and Jing Ma. 2020. Influential Nodes Identification in Complex Networks via Information Entropy.Entropy22, 2 (2020), 242. doi:10.3390/e22020242

2020 doi
[12]

Holland, Kathryn Blackmond Laskey, and Samuel Leinhardt

Paul W. Holland, Kathryn Blackmond Laskey, and Samuel Leinhardt. 1983. Stochastic Blockmodels: First Steps.Social Networks5, 2 (1983), 109–137. doi:10.1016/0378-8733(83)90021-7

1983 doi
[13]

Petter Holme, Beom Jun Kim, Chang No Yoon, and Seung Kee Han. 2002. Attack Vulnerability of Complex Networks.Physical Review E65 (2002), 056109. doi:10. 1103/PhysRevE.65.056109

2002
[14]

Jérôme Kunegis. 2013. KONECT—The Koblenz Network Collection. InProceedings of the 22nd International Conference on World Wide Web Companion. 1343–1350. doi:10.1145/2487788.2488173

2013
[15]

Kezhao Lai, Yutao Lai, and Hai-Lin Liu. 2026. Beyond the Node: Clade-Level Selection for Efficient MCTS in Automatic Heuristic Design.arXiv preprint arXiv:2602.00549(2026). doi:10.48550/arXiv.2602.00549

2026 doi
[16]

Jure Leskovec and Andrej Krevl. 2014. SNAP Datasets: Stanford Large Network Dataset Collection. https://snap.stanford.edu/data/

2014
[17]

Fei Liu, Xialiang Tong, Mingxuan Yuan, Xi Lin, Fu Luo, Zhenkun Wang, Zhichao Lu, and Qingfu Zhang. 2024. Evolution of Heuristics: Towards Efficient Automatic Algorithm Design Using Large Language Model. InInternational Conference on Machine Learning. OpenReview: BwAkaxqiLB, ar...

2024 arXiv
[18]

Jinzhu Mao, Dongyun Zou, Li Sheng, Siyi Liu, Chen Gao, Yue Wang, and Yong Li
[19]

Michael Molloy and Bruce Reed. 1995. A Critical Point for Random Graphs with a Given Degree Sequence.Random Structures & Algorithms6, 2–3 (1995), 161–180. doi:10.1002/rsa.3240060204

1995 doi
[20]

Flaviano Morone and Hernán A. Makse. 2015. Influence Maximization in Complex Networks through Optimal Percolation.Nature524 (2015), 65–68. doi:10.1038/ nature14604

2015
[21]

Salomon Mugisha and Hai-Jun Zhou. 2016. Identifying Optimal Targets of Network Attack by Belief Propagation.Physical Review E94 (2016), 012305. doi:10.1103/PhysRevE.94.012305

2016 doi
[22]

Alexander Novikov et al. 2025. AlphaEvolve: A Coding Agent for Scientific and Algorithmic Discovery.arXiv preprint arXiv:2506.13131(2025). doi:10.48550/ arXiv.2506.13131

2025 doi arXiv
[23]

Liming Pan, Cheng Shi, and Ivan Dokmanić. 2022. Neural Link Prediction with Walk Pooling. InInternational Conference on Learning Representations. OpenRe- view: CCu6RcUMwK0, arXiv:2110.04375. doi:10.48550/arXiv.2110.04375

2022 doi arXiv
[24]

Xiao-Long Ren, Niels Gleinig, Dirk Helbing, and Nino Antulov-Fantulin. 2019. Generalized Network Dismantling.Proceedings of the National Academy of Sciences116, 14 (2019), 6554–6559. doi:10.1073/pnas.1806108116

2019 doi
[25]

Bernardino Romera-Paredes et al. 2024. Mathematical Discoveries from Program Search with Large Language Models.Nature625 (2024), 468–475. doi:10.1038/ s41586-023-06924-6

2024
[26]

Christian von Mering et al. 2002. Comparative Assessment of Large-Scale Data Sets of Protein–Protein Interactions.Nature417 (2002), 399–403. doi:10.1038/ nature750

2002
[27]

Small- World

Duncan J. Watts and Steven H. Strogatz. 1998. Collective Dynamics of “Small- World” Networks.Nature393 (1998), 440–442. doi:10.1038/30918

1998 doi
[28]

Haoran Ye, Jiarui Wang, Zhiguang Cao, Federico Berto, Chuanbo Hua, Haeyeon Kim, Jinkyoo Park, and Guojie Song. 2024. ReEvo: Large Language Models as Hyper-Heuristics with Reflective Evolution. InAdvances in Neural Information Processing Systems. doi:10.48550/arXiv.2402.01145

2024 doi arXiv
[29]

Lenka Zdeborová, Pan Zhang, and Hai-Jun Zhou. 2016. Fast and Simple Decycling and Dismantling of Networks.Scientific Reports6 (2016), 37954. doi:10.1038/ srep37954

2016
[30]

Jiazheng Zhang and Bang Wang. 2022. Dismantling Complex Networks by a Neural Model Trained from Tiny Networks. InProceedings of the 31st ACM International Conference on Information and Knowledge Management. doi:10.1145/ 3511808.3557290

2022
[31]

Yu Zheng, Jingtao Ding, Depeng Jin, Jianxi Gao, and Yong Li. 2025. Advancing Network Resilience Theories with Symbolized Reinforcement Learning.arXiv preprint arXiv:2507.08827(2025)

2025 arXiv
[32]

Please design an evo- lutionary algorithm scheme for solving the network dismantling problem

Zhi Zheng, Zhuoliang Xie, Zhenkun Wang, and Bryan Hooi. 2025. Monte Carlo Tree Search for Comprehensive Exploration in LLM-Based Automatic Heuristic Design. InInternational Conference on Machine Learning. OpenReview: Do1OdZzYHr, arXiv:2501.08603. A Search Evaluation, Candidate...

2025 arXiv
[35]

Broad Tree Search, Execution- to-Evidence Structural Policy Induction, and Policy-Conditioned Tree Search

Every curve contains𝑁 rows, is monotonically nonincreasing, and ends at 0. The environment is Windows 11, Python 3.12.9, and NetworkX 3.4.2 on a machine with an Intel Core Ultra 9 275HX and 64 GB memory. The FINDER comparison uses only the without reinsertion curves correspond...

2027
[2016]

doi:10.1073/pnas.1605083113

Network Dismantling.Proceedings of the National Academy of Sciences113, 44 (2016), 12368–12373. doi:10.1073/pnas.1605083113

2016 doi
[2024]

arXiv preprint arXiv:2403.03962(2024)

Identify Critical Nodes in Complex Network with Large Language Models. arXiv preprint arXiv:2403.03962(2024). doi:10.48550/arXiv.2403.03962

2024 doi arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Eser Aygün et al. 2026. An AI System to Help Scientists Write Expert-Level Empirical Software.Nature654 (2026), 909–916. arXiv:2509.06503. doi:10.1038/ s41586-026-10658-6

arXiv 2026

[2] [2]

Albert-László Barabási and Réka Albert. 1999. Emergence of Scaling in Random Networks.Science286, 5439 (1999), 509–512. doi:10.1126/science.286.5439.509

doi:10.1126/science.286.5439.509 1999

[3] [3]

Alfredo Braunstein, Luca Dall’Asta, Guilhem Semerjian, and Lenka Zdeborová

work page

[4] [4]

Chentong Chen, Mengyuan Zhong, Ye Fan, Jialong Shi, and Jianyong Sun. 2026. HiFo-Prompt: Prompting with Hindsight and Foresight for LLM-Based Auto- matic Heuristic Design. InInternational Conference on Learning Representations. OpenReview: imSLzfZ6av, arXiv:2508.13333

arXiv 2026

[5] [5]

Pham Vu Tuan Dat, Long Doan, and Huynh Thi Thanh Binh. 2025. HSEvo: Elevating Automatic Heuristic Design with Diversity-Driven Harmony Search and Genetic Algorithm Using LLMs. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. doi:10.1609/aaai.v39i25.34898

doi:10.1609/aaai.v39i25.34898 2025

[6] [6]

Davis and Yifan Hu

Timothy A. Davis and Yifan Hu. 2011. The University of Florida Sparse Matrix Collection.ACM Trans. Math. Software38, 1 (2011), 1:1–1:25. doi:10.1145/2049662. 2049663

work page doi:10.1145/2049662 2011

[7] [7]

Paul Erdős and Alfréd Rényi. 1959. On Random Graphs I.Publicationes Mathe- maticae Debrecen6 (1959), 290–297. doi:10.5486/PMD.1959.6.3-4.12

doi:10.5486/pmd.1959.6.3-4.12 1959

[8] [8]

Changjun Fan, Li Zeng, Yizhou Sun, and Yang-Yu Liu. 2020. Finding Key Players in Complex Networks through Deep Reinforcement Learning.Nature Machine Intelligence2 (2020), 317–324. doi:10.1038/s42256-020-0177-2

doi:10.1038/s42256-020-0177-2 2020