{"id":"cd316861-5371-48fd-8802-d1c84dce9f23","arxiv_id":"2508.06736","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":4.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":2,"one_line_summary":"ParBalans uses parallel multi-armed bandit search to speed up MIP solving and competes with Gurobi on hard benchmarks.","lead":"ParBalans is a parallel extension of Balans, a multi-armed bandit based adaptive large neighborhood search for mixed-integer programming. It aims to match or beat the commercial solver Gurobi on hard optimization problems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central empirical claim of competitive performance over Gurobi is unsupported by the abstract; full experimental methodology is the load-bearing unknown.","rationale":"The reader identified two load-bearing assumptions: fair comparison and scaling of the multi-armed bandit. My concern focuses on the former—specifically, that the abstract provides no evidence to support the comparative claim. This is the most fundamental concern because the paper's contribution is empirical; without experimental controls, any claim of competitiveness is unsubstantiated. The scaling issue is a separate technical concern that may or may not be addressed in the full text. I do not see an internal logical flaw from the abstract alone, but the lack of data is enough to keep the verdict at UNVERDICTED. The verdict should remain unchanged because the reader already made the correct call: insufficient information for evaluation. I set agreement to 'partial' because while we both point to the experimental comparison, the reader also mentioned scaling, which I did not focus on. My concrete test is to inspect the full experimental section to verify fairness and statistical rigor, which would settle the concern.","tokens_in":621,"tokens_out":3220,"duration_ms":39247,"concrete_test":"Obtain the full paper and examine Section 4 (Experiments). Check that the comparison with Gurobi is performed under identical wall-clock time limits and hardware, that the benchmarks include standard hard MIP instances (e.g., MIPLIB 2017, or at least a diverse set), that multiple runs with different random seeds are reported, and that the measure of 'competitive' is clearly defined and statistically supported. If these details are present and sound, the concern is resolved; if any control is missing or ambiguous, the claim weakens.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that ParBalans exhibits competitive performance compared to Gurobi, especially on hard benchmarks. This is an empirical statement, yet the abstract provides no quantitative evidence: no performance metric (e.g., primal integral, solution time), no benchmark set, no resource limits, no hardware description, and no statistical tests. The claim's validity rests entirely on the full paper's experimental design. If that design is unfair or poorly controlled—for instance, if Gurobi is not given the same time limit or thread count, or if instances are selected to favor ALNS—the central claim collapses. This is not an internal inconsistency but a severe gap in verifiability. The reader's UNVERDICTED verdict correctly recognizes that the claim cannot be assessed without the full text. The most load-bearing concern is thus the absence of any experimental details in the abstract, which makes the core contribution unfalsifiable from the available evidence.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces ParBalans, a parallel extension of the recently proposed Balans algorithm, which is a multi-armed bandit-based adaptive large neighborhood search for mixed-integer programming (MIP). ParBalans is claimed to exploit both solver-level and algorithmic-level parallelism, and the abstract states that experimental results demonstrate competitive performance against the state-of-the-art commercial solver Gurobi, particularly on hard optimization benchmarks. No additional technical details, equations, experimental design, or quantitative results are provided; the manuscript currently consists only of the abstract.","tokens_in":859,"tokens_out":1507,"duration_ms":19268,"significance":"If the central claim is substantiated, ParBalans would offer a meaningful contribution: a parallelizable, open (presumably) heuristic that can compete with a leading commercial MIP solver on difficult instances. The MAB-based adaptive control and the dual-level parallelism are interesting ideas. However, because the abstract provides no quantitative evidence, the significance cannot be assessed. The potential is real, but the manuscript as presented does not support it.","major_comments":[{"comment":"The central claim—'competitive performance compared to Gurobi, particularly on hard optimization benchmarks'—is entirely unsupported by quantitative data. No performance metric (e.g., primal integral, time to best solution, gap), no benchmark set, no computational budget, no hardware description, and no statistical tests are reported. This is the load-bearing claim of the paper; without these details, the contribution is unverifiable. A full experimental section is required to support it.","section":"Abstract"},{"comment":"The claim that ParBalans 'leverages both solver-level and algorithmic-level parallelism' is vague. The abstract does not state how many parallel workers were used, how the multi-armed bandit exploration-exploitation trade-off is preserved under parallelization, or whether additional workers produce meaningful speedups rather than duplicating identical search effort. The scaling behavior is central to the parallelization contribution and needs explicit evidence.","section":"Abstract"},{"comment":"The experimental comparison to Gurobi is mentioned without any specification of fairness controls: equal time limits, thread counts, or termination criteria. Since Balans is the authors' own prior work, transparency about the experimental protocol is especially important to rule out selection or configuration bias. The abstract gives no basis to judge whether the comparison is fair or generalizable.","section":"Abstract"}],"minor_comments":[{"comment":"The phrase 'solver-level and algorithmic-level parallelism' would benefit from a brief definition, as these terms are not standard.","section":"Abstract"},{"comment":"The abstract does not mention any public availability of the code or benchmark data, which would be valuable for reproducibility.","section":"Abstract"}],"recommendation":"uncertain","confidential_remarks":"The manuscript is abstract-only, so I cannot perform a substantive technical review. The central empirical claim is plausible but unsupported. I recommend requesting the full manuscript before making a final decision; if the experiments are rigorous and the scaling results hold, this may be a solid contribution. If the full paper is unavailable, the editor should consider whether an abstract-only submission is appropriate for review."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, this is an abstract-only read, so take this as provisional. The contribution is straightforward: ParBalans takes Balans, a recently proposed bandit-guided adaptive large neighborhood search for MIPs, and parallelizes it at both solver and algorithmic levels. That is genuinely useful—the modular design does suggest parallel workers should explore different configurations—and the authors are honest that the parallel potential had not been examined. The writing is clear and the motivation is fine.\n\nThe soft spot is the one you'd expect: the central claim, competitive performance with Gurobi on hard benchmarks, has zero quantitative support in the abstract. No metrics, no instances, no resource limits, no hardware. That makes the claim unfalsifiable from what is shown, and the whole contribution hangs on the experimental design. If the full paper gives Gurobi equal time and threads and uses standard benchmarks, this could be a solid incremental result. If not, it is noise. I cannot check any of that here.\n\nThe novelty is incremental—parallelizing an existing method—but that is not a flaw; it is a reasonable engineering contribution. Self-citation is not an issue given Balans is the direct base.\n\nBottom line: worth a serious referee, because the question—whether a bandit-controlled ALNS can be parallelized enough to rival Gurobi—is legitimate, and the method is specified well enough to be testable. But the abstract alone would not convince me to cite it. I would want the full experimental section and, ideally, code.\n\nRecommend to the editor: get the full text before deciding, and if the experiments check out, send it to review.","headline":"ParBalans is a sensible incremental parallelization of Balans; the abstract's Gurobi-competitiveness claim needs full experimental details before it can be assessed.","tokens_in":1269,"tokens_out":1439,"would_cite":false,"duration_ms":15895,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["90C10","90C59","68W10"],"pacs":[],"model":"deepseek-v4-flash","headline":"ParBalans shows that parallelizing a bandit-guided large neighborhood search makes it competitive with the commercial solver Gurobi on hard mixed-integer programming instances.","keywords":["mixed-integer programming","large neighborhood search","multi-armed bandits","parallel optimization","adaptive search","Gurobi","combinatorial optimization"],"falsifier":"Run ParBalans and Gurobi on a held-out set of hard MIP instances with identical wall-clock time and core counts; if Gurobi solves more instances or reaches lower objective gaps on most of them, the competitiveness claim fails. Also measure speedup as worker count increases: if solution time stops improving or worsens beyond a small number of workers, the bandit control is not scaling.","tokens_in":577,"feed_emoji":"⚙️","tokens_out":3788,"duration_ms":41983,"temperature":0.7,"pith_summary":"Solving mixed-integer programming problems is computationally expensive, and parallelism is one route to faster solutions. This paper introduces ParBalans, a parallel version of the Balans algorithm, which uses multi-armed bandits to adaptively steer large neighborhood search for MIPs. ParBalans combines solver-level parallelism, running multiple search processes at once, with algorithm-level parallelism, letting the bandit controller explore diverse parameter settings across workers. The paper's claim is that ParBalans is competitive with Gurobi, especially on hard optimization benchmarks. A sympathetic reader would care because it suggests a parallel adaptive search method can rival a commercial standard on difficult instances.","feed_headline":"Parallel bandit search matches Gurobi on hard MIP instances","feed_subtitle":"ParBalans spreads bandit-controlled local search across workers and holds its own against a top commercial solver.","key_machinery":"The central mechanism is the multi-armed bandit controller inherited from Balans: a learning policy that treats each large-neighborhood-search parameter configuration as an arm and adaptively focuses computational effort on configurations that have performed well. ParBalans places this controller on top of two parallelization layers, multiple solver-level searches running in parallel and algorithmic-level diversity across parameter configurations, so that exploration and exploitation happen both across configurations and across workers. The bandit's exploration-exploitation balance is what keeps parallel workers from duplicating identical effort and what the paper credits for competitive per","core_discovery":"On the paper's own terms, the discovery is that Balans's modular design, where a multi-armed bandit chooses among large-neighborhood-search parameter configurations, can be extended into ParBalans by adding two levels of parallelism: solver-level parallelism, in which independent MIP searches run concurrently, and algorithm-level parallelism, in which the bandit explores a wider set of configurations across workers. The paper reports that this parallel extension performs competitively against Gurobi, with the strongest results on hard benchmarks. In other words, the bandit-guided adaptive neighborhood search, once parallelized, does not lose its search quality and can match a state-of-the-ar","pith_inferences":["The reward signal used by the bandit, presumably how quickly a configuration improves the incumbent solution, is likely the main lever for scaling to many workers; different reward definitions could change parallel efficiency more than adding cores. The abstract does not test this.","The competitive results on hard benchmarks may not extend to easier instances, where Gurobi's presolve and cutting planes could dominate; if so, ParBalans is best deployed as a hard-instance supplement rather than a general replacement.","A natural testable extension is to vary the number of workers and measure speedup curves; if the bandit does not adjust exploration as workers increase, speedup should plateau or reverse.","The same bandit-plus-parallel-large-neighborhood-search recipe could transfer to other combinatorial optimization domains, such as constraint programming or SAT, where neighborhood search and configuration selection are natural."],"forward_implications":["If ParBalans' results hold, parallel adaptive large neighborhood search can serve as a strong alternative to commercial MIP solvers on hard instances.","The two-level parallel structure means adding workers can translate into broader configuration coverage rather than duplicated search, making speedups more scalable.","The bandit controller's adaptive behavior lets the method shift effort to promising configurations without human tuning, which is valuable when instance difficulty varies.","Competitive performance on hard benchmarks suggests the method is most useful precisely where exact solvers struggle.","The modular architecture means further algorithmic improvements to Balans can be inherited by ParBalans with little additional engineering."],"supporting_citations":[],"fun_headline_variants":["ParBalans: parallel bandit search matches Gurobi on hardest benchmarks","Bandit-based parallel search matches Gurobi on hard MIP instances","Two-level parallel bandit search competitive with Gurobi","ParBalans: bandit-guided parallel search rivals Gurobi","ParBalans: parallel MIP search holds its own vs Gurobi"],"cache_read_input_tokens":2816,"weakest_assumption_plain":"The load-bearing premise is that the reported comparison with Gurobi is fair and representative, with equal time and hardware budgets and no cherry-picked instances, and that the bandit's exploration-exploitation tradeoff still works when many parallel workers are added.","fun_headline_variants_meta":{"raw":{"variants":["ParBalans: parallel bandit search matches Gurobi on hardest benchmarks","Bandit-based parallel search matches Gurobi on hard MIP instances","Two-level parallel bandit search competitive with Gurobi","ParBalans: bandit-guided parallel search rivals Gurobi","ParBalans: parallel MIP search holds its own vs Gurobi"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000766,"raw_usage":{"total_tokens":3186,"prompt_tokens":648,"completion_tokens":2538,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":392,"completion_tokens_details":{"reasoning_tokens":2444}},"tokens_in":392,"tokens_out":2538,"duration_ms":18707,"temperature":1.0,"reasoning_tokens":2444,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T22:32:36.940213+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run ParBalans and Gurobi on a held-out set of hard MIP instances with identical wall-clock time and core counts; if Gurobi solves more instances or reaches lower objective gaps on most of them, the competitiveness claim fails. Also measure speedup as worker count increases: if solution time stops improving or worsens beyond a small number of workers, the bandit control is not scaling.","supporting_citations":[],"review_version":1}