{"id":"3b5f571f-dbd5-4b2c-a5ce-0eef1e16fb63","arxiv_id":"2607.29228","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"SOMA's and DE/rand/1/bin's candidate generation is shown to be an exactly linear proposal operator whose closed-form moments guide new, competitive optimizer variants.","lead":"This paper shows that the candidate-generation step in two common optimization algorithms, SOMA and differential evolution, can be written as a simple linear operation on the current positions, and derives exact formulas for the statistical spread of the candidates. This gives researchers a principled way to tune the algorithms' exploratory behavior and to design new variants that perform competitively on benchmark problems.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Canonical SOMA mask semantics: Eq. (13) and the benchmark baseline assume independently regenerated PRT masks, conflicting with the classic single-PRT-vector-per-migration SOMA; the 'canonical' label needs verification or relabeling.","rationale":"The mathematical core of the paper is verified by exact enumeration and appears sound. The reader's weakest_assumption correctly identified the procedural definitions of canonical SOMA and DE as the load-bearing assumptions. My stress-test focuses on the one part of that assumption that is not merely a harmless scope restriction: the mask-regeneration semantics. The paper's own Section 2.1 leaves open whether masks are reused or regenerated, yet Eq. (13), the path-level coverage claims, and the Canonical-SOMA baseline all assume regeneration. If standard SOMA actually reuses a single PRT vector per migration, the paper mislabels a variant as canonical and its coordinate-coverage contribution does not transfer. This is a verifiable, empirical and terminological issue rather than a mathematical error; it warrants a conditional acceptance pending a check of the reference implementation or a relabeling of the analyzed variant.","tokens_in":24994,"tokens_out":19014,"duration_ms":205429,"concrete_test":"Run the D=5, 500D BBOB configuration with Canonical-SOMA modified to reuse a single PRT vector across all path points within each migration, keeping all other settings from Table C.9 unchanged. Compare empirical per-path coordinate coverage to Eq. (13) and final-error ranks to Table 4. If coverage is approximately p rather than 1-(1-p)^J and the ranks shift materially, the paper must either relabel its analyzed variant as independently-masked SOMA or derive reuse-aware coverage formulas.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Section 2.1 explicitly notes that some SOMA formulations reuse one PRT vector across the whole migration path while others regenerate it at each path point, but the paper does not settle which semantics is 'canonical.' The path-level coverage formulas in Eq. (13) are derived only for independent mask regeneration at J path points, and Appendix A.2 confirms that reuse changes the result drastically: with D=30 and p=0.2, regeneration gives all-coordinate coverage 0.706 at J=20, while reuse leaves coverage approximately p, essentially zero for all-coordinate coverage. The experimental 'Canonical-SOMA' baseline in Table C.9 is explicitly configured with masks 'regenerated independently at every proposal point.' If the standard SOMA implementation uses one PRT vector per migration, as in the classic formulation, then the paper's 'canonical SOMA' is really an independently-masked variant. This does not invalidate the fixed-t linear representation (Eq. 4), the leader-relative rescaling (Eq. 7), or the per-path moments (Eqs. 9-10), because those hold for any mask realization. But it puts the 'canonical' label and the coordinate-coverage contribution on weaker footing, and the reported canonical baseline may not match the usual algorithm.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper proposes an operator–selection factorization for SOMA and DE/rand/1/bin, separating raw candidate generation from boundary repair and fitness-based selection. For SOMA, the fixed-mask proposal z = x + tΠ(L−x) is written as an exact linear map on an augmented state (Eq. 4), reducing in leader-relative coordinates to coordinate-wise rescaling by (1−t) (Eq. 7). Under independent Bernoulli perturbation masks, the paper derives closed forms for the proposal mean, covariance, expected squared step length, expected squared leader distance, active dimensionality, and path-level coordinate coverage (Eqs. 9–13). For DE/rand/1/bin, it derives finite-population donor moments with an exact sampling-without-replacement covariance factor, and forced-coordinate binomial crossover moments (Appendices A.3–A.4). The analysis motivates three new SOMA variants: GC-SOMA, RA-SOMA, and an adaptive composite iL-SHOMA-RA, benchmarked on the full noiseless BBOB suite at D∈{5,10,20} and budgets 500D/2500D. The empirical results show improvements over canonical SOMA and mixed but competitive performance against iL-SHADE and SciPy-DE.","tokens_in":106,"tokens_out":10005,"duration_ms":203510,"significance":"If accepted, the paper provides a clean, reusable decomposition of two widely used optimizers. Its strongest assets are the parameter-free exact identities, the finite-population DE correction factor (1 + 2F²M/(M−1)), and the exact forced-coordinate crossover covariance with negative off-diagonal entries; all are verified by exact enumeration to floating-point precision, and the code is made available. The boundary-repair and selection-distortion studies in Appendix A.5 are honest about the scope of the raw-proposal analysis. The paper also makes a credible design point: proposal-level geometry can be converted into algorithmic controls, even though the composite iL-SHOMA-RA is correctly labeled exploratory. The central derivation is sound; the main risks are terminological overreach and benchmark confounds rather than mathematical error.","major_comments":[{"comment":"","section":"Section 2.1; Eq. (13); Table C.9"},{"comment":"","section":"Section 5.1; Eq. (21); Appendix A.4"},{"comment":"","section":"Section 6.1; Table 4; Table C.9"}],"minor_comments":[{"comment":"","section":"Figure 1"},{"comment":"","section":"Section 5.1, Eq. (21)"},{"comment":"","section":"Section 6.2, Table 5"},{"comment":"","section":"Appendix A.5"}],"recommendation":"major_revision","confidential_remarks":"The mathematical core is sound and well verified; I would not reject on technical grounds. The main risk is overclaiming the word \"canonical\" for SOMA, especially because the paper itself documents two mask semantics that behave very differently, and the baseline uses the regenerated-mask variant. The DE main-text covariance also needs to distinguish the independent-mask model from exact forced-coordinate crossover. Finally, the benchmark claims would be strengthened by a matched canonical baseline. These are fixable within the manuscript's scope, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this paper is a genuinely useful, carefully scoped analysis, not a flashy empirical win. The core contribution is the exact linear augmented-state representation of the SOMA proposal (Eq. 4) and the finite-population DE moments, including the forced-coordinate crossover covariance. I checked the algebra against the paper and against exact enumeration; both hold. The derivations are parameter-free, and the appendices honestly quantify how boundary repair and fitness-based selection distort the raw proposal geometry. Credit where due: the paper ships code and data, verifies its formulas to floating-point precision, and does not oversell the unde rived parts.\n\nThe stress-test concern about the 'canonical SOMA' label is real but minor. The paper explicitly acknowledges in Section 2.1 that SOMA variants differ on whether the PRT mask is reused or regenerated, and it clearly states its own procedural assumption. So the math is on solid ground. The weak spot is the experimental baseline: calling the independently-masked variant 'Canonical-SOMA' in Table C.9 may mislead readers who know the classic single-PRT-vector-per-migration formulation. The paper should either verify which convention is dominant in the standard SOMA literature or rename the baseline to something like 'SOMA with regenerated masks.' That is a presentational fix, not a load-bearing flaw.\n\nThe only other soft spot worth naming is the iL-SHOMA-RA composite: it bundles rotation-aware masking, success-history adaptation, and population-size reduction with no ablation. The authors say this openly and treat it as an exploratory method. That is honest, but it does limit what the benchmark results can tell us about individual mechanisms.\n\nWho is this for? Anyone who teaches or analyzes SOMA or DE and wants a clean, teachable account of candidate-generation geometry. The closed forms for active dimensionality, coordinate coverage, and the finite-population covariance factor are directly usable. The benchmark results are secondary; the analytical core is the reason to read it.\n\nMy verdict: the paper deserves a serious referee. It would benefit from a terminology pass on 'canonical' and a more explicit statement that the empirical claims depend on the chosen mask semantics, but the mathematical results are reproducible and the limitations are honestly stated. I would send it out.","headline":"Solid, carefully scoped analysis of SOMA/DE proposal geometry with verified closed-form moments; the main caveat is a labeling issue around the 'canonical' SOMA mask assumption.","tokens_in":25781,"tokens_out":1087,"would_cite":true,"duration_ms":14455,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["68T20","90C59"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that the candidate-generation step of two classic population optimizers—SOMA and Differential Evolution—is, conditional on the sampled state and random masks, an exact linear or affine transformation, and derives closed-for","keywords":["SOMA","Differential Evolution","linear proposal operators","stochastic search geometry","Bernoulli masks","finite-population moments","crossover covariance","BBOB benchmark"],"falsifier":"Run canonical SOMA with one fixed perturbation mask reused across the entire path; if per-coordinate coverage does not rise with path length J as 1−(1−p)^J but stays close to p, then the independent-regeneration assumption behind Eq. (13) fails. Similarly, sampling DE donors with replacement would change the finite-population covariance factor, providing a direct check of the mutation-moment formula.","tokens_in":24904,"feed_emoji":"📐","tokens_out":4656,"duration_ms":48615,"temperature":0.7,"pith_summary":"Population-based optimizers are usually treated as nonlinear black boxes because selection, replacement, and rank-based decisions are discontinuous. This paper tries to establish that the hidden variation stage is not: for canonical SOMA, each candidate proposal is an exact linear map of an augmented migrant–leader state, and relative to the leader it is simply a coordinate-wise rescaling. Treating the perturbation mask as independent Bernoulli random variables yields closed-form formulas for the proposal mean, covariance, expected step length, expected squared leader distance, active dimensionality, and coordinate coverage; for DE/rand/1/bin the paper derives matching finite-population moments including the extra covariance induced by forced-coordinate crossover. If these results hold, parameter setting (PRT, path parameter, F, CR) can be replaced by direct geometric targeting, and algorithm design can be guided by proposal-level quantities. The paper verifies the identities numerically and shows that repair and selection, while unavoidable, are separate transformations that can substantially distort the raw proposals.","feed_headline":"SOMA candidate steps are exact linear maps","feed_subtitle":"Closed-form means, covariances, and coverage formulas turn SOMA's PRT and path length into designable geometry.","key_machinery":"The load-bearing object is the augmented migrant–leader state s = (x,L) in R^{2D} and the block matrix M(t,Π) = [[I−tΠ, tΠ],[0,I]], which turns the proposal into an exact linear map; in leader-relative coordinates it collapses to the diagonal rescaling I−tΠ. For the stochastic analysis, the Bernoulli perturbation mask Π = diag(π_d) with π_d ~ Bernoulli(p) is the random operator that produces all the closed-form moments. For DE, the analogous machinery is the augmented donor/target state and the forced-coordinate binomial crossover mask, whose marginal activation probability becomes q⋆ = q + (1−q)/D and whose covariance has off-diagonal entries −(1−q)²/D².","core_discovery":"The central claim is that the nonlinearity of a full optimization iteration lives in selection, target construction, boundary repair, and adaptation, not in the variation operator. Once the migrant, leader, path parameter, and perturbation mask are fixed, the SOMA proposal z = x + tΠ(L−x) is exactly linear in the augmented state (x,L) through the block matrix M(t,Π), and relative to the leader it reduces to a coordinate-wise rescaling by (1−t) on active coordinates. Under independent Bernoulli masks, the paper derives closed forms: E[z] = x + pt(L−x), covariance p(1−p)t² diag(r²), normalized expected squared leader distance 1−pt(2−t), active dimensionality Dp, and per-coordinate coverage 1−(","pith_inferences":["If raw proposal moments are exact, then selection and repair are the only remaining sources of nonlinearity; one could try to characterize fitness-based selection as an operator on the proposal covariance, estimated from acceptance rates, to close the gap between raw and selected distributions.","The coverage formula suggests a testable design rule: choose path length J and PRT so that 1−(1−p)^J covers a desired fraction of coordinates, turning exploration budgets into explicit geometric targets.","Rotation-aware masking in a population-covariance basis is promising in low dimensions, but in higher dimensions the empirical basis may be noisy; shrinkage, low-rank, or rank-one-updated bases are natural extensions to test.","The DE/SOMA correspondence via masked affine difference operators hints at a unified analysis of other strategies, such as current-to-best or all-to-random, through the same augmented-state machinery."],"forward_implications":["SOMA's PRT, path parameter, and path length become designable geometric quantities: expected active dimensionality, covariance, step length, and coordinate coverage are all computable in closed form.","The same separation applies to DE/rand/1/bin: finite-population mutation moments and forced-coordinate crossover moments fully describe the raw trial distribution, including the cross-coordinate covariance that the forced coordinate introduces.","Because the proposal operator is objective-independent, the derived geometry holds for any fitness function before repair and selection, providing a common basis for comparing masking and covariance across different algorithms.","Operator-guided variants—geometry-controlled, rotation-aware, and a composite iSOMA extension—improve canonical SOMA on the noiseless benchmark, with rotation-aware masking strongest at low dimension and high budget.","Boundary repair and fitness-based selection can substantially change raw proposal means and covariances, so the exact results describe the variation stage only; predicting full optimizer behavior requires modeling repair and selection as well."],"fun_headline_variants":["SOMA proposal geometry is exactly linear, not heuristic","Linearization of SOMA and DE yields designable search steps","Closed-form means and covariances for SOMA's random steps","Operator factorization splits SOMA into variation and selection","Geometry-aware SOMA variants rival DE on BBOB suite"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The closed-form results describe the raw proposal stage under exactly the stated procedural variants—independent regenerated Bernoulli masks and donor sampling without replacement—and they do not by themselves predict the full algorithm because boundary repair and fitness-based selection can substantially reshape the proposal distribution.","fun_headline_variants_meta":{"raw":{"variants":["SOMA proposal geometry is exactly linear, not heuristic","Linearization of SOMA and DE yields designable search steps","Closed-form means and covariances for SOMA's random steps","Operator factorization splits SOMA into variation and selection","Geometry-aware SOMA variants rival DE on BBOB suite"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1219,"prompt_tokens":832,"completion_tokens":387,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":576,"completion_tokens_details":{"reasoning_tokens":305}},"tokens_in":576,"tokens_out":387,"duration_ms":4554,"temperature":1.0,"reasoning_tokens":305,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T11:12:15.208750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run canonical SOMA with one fixed perturbation mask reused across the entire path; if per-coordinate coverage does not rise with path length J as 1−(1−p)^J but stays close to p, then the independent-regeneration assumption behind Eq. (13) fails. Similarly, sampling DE donors with replacement would change the finite-population covariance factor, providing a direct check of the mutation-moment formula.","supporting_citations":[],"review_version":1}