{"id":"1ee0f922-e03a-48a4-9e8d-ed56a9c03698","arxiv_id":"2607.22710","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"The variational free-energy / ELBO identity is the unique exact decomposition of log Z that is additively separable with a strict residual; alpha- and Renyi-divergence objectives can only ever give bounds.","lead":"This paper proves a uniqueness theorem: the standard split of a log-partition or log-evidence quantity into a computable objective plus an error term is forced under natural conditions, leaving no room for alternative divergences such as Rényi or alpha to give an exact identity. For variational Bayes and mean-field theory, this means the freedom in the method lives in the choice of trial distribution, not in the choice of objective.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The weakest point is the unresolved status of Hypothesis (R): the proof needs it, but the paper neither proves it necessary nor redundant, so the abstract's claim that counterexamples show each hypothesis is needed overstates what is shown.","rationale":"I went through the proof step by step. Step 1 reduces to the residual pair, Step 2 gives scale invariance, Step 3 correctly produces the measurable homomorphism and then the linear functional, Step 4 identifies the forms of A' and B', and Step 5's perturbation argument forces m_Q=0. No algebraic error appears. The theorem is valid conditional on (R). The reader's weakest assumption identifies (R), and I agree it is the least secure premise: it is exactly the assumption that rules out nonmeasurable additive solutions, and the paper explicitly leaves open whether (P) would do the job alone. The abstract's 'Counterexamples show that each hypothesis is needed' is stronger than the body, which for (R) offers only proof-mechanism failure (Remark 6.5) and an open question (Question 6.6). This is a real but contained issue: it affects the claimed minimality of the hypotheses and the boundary of the uniqueness result, not the validity of Theorem 1.1 under its stated assumptions. Therefore no change to the ACCEPT verdict is needed; the concrete test above would settle the open point.","tokens_in":963,"tokens_out":1016,"duration_ms":157064,"concrete_test":"Settle Question 6.6 in the minimal case X={1,2}. Parameterize Q=(q,1-q), pi=(r,1-r), and let phi:R^2 -> R be additive with phi(1,1)=0. Without (R), any candidate residual is Delta(q,r)=q log(q/r)+(1-q)log((1-q)/(1-r))+phi(log(r/q),log((1-r)/(1-q))). Determine whether the condition Delta(q,r) >= 0 for all q,r in (0,1), with equality iff q=r, forces phi ≡ 0. This is a single functional-inequality check: if phi ≡ 0 is forced, (R) is redundant; if some nonzero phi survives, (R) is necessary and supplies the missing counterexample. Either answer resolves the abstract's 'each hypothesis is needed' claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 1.1 is proved correctly conditional on (S), (R), and (P); I checked the argument step by step. The only essential use of (R) is in Step 3 (Section 5, Eqs. (6)-(7)), where measurability promotes the group homomorphism chi_Q:(P,.) -> (R,+) to a linear functional <m_Q, log .>. Without (R), Hamel-basis additive maps survive; Step 4 would give Delta(Q,pi)=D(Q||pi)+phi_Q(log(pi/Q)) with phi_Q:R^X -> R additive and phi_Q(1)=0. The paper's Remark 6.5 and Question 6.6 leave open whether the nonnegativity clause of (P) automatically kills such phi_Q. If it does, (R) is redundant and the abstract's claim that counterexamples show each hypothesis is needed is false; if it does not, the theorem is false without (R), so the uniqueness result again does not have the minimal hypotheses the abstract advertises. Either way, the unsettled status of (R), not an internal proof error, is the load-bearing concern. This does not invalidate Theorem 1.1 as stated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a uniqueness theorem for exact decompositions of log Z(p, ℓ) into a computable variational objective and a residual divergence. On a finite alphabet, it shows that if (G, Δ) satisfies log Z = -G + Δ, with G additively separable as A(Q,p)+B(Q,ℓ), with ℓ ↦ B(Q,ℓ) Lebesgue measurable for each Q, and with Δ ≥ 0 vanishing exactly at Q = π_{p,ℓ}, then G must be the variational free energy F(Q;p,ℓ)=D(Q∥p)-E_Q[log ℓ] and Δ must be D(Q∥π), up to a Q-dependent gauge a(Q) that cancels in G. The proof reduces the hypotheses to a measurable group homomorphism χ_Q:(P,·)->(R,+) and eliminates it by combining linear representation with a second-order expansion of D(Q∥π_ε) and the nonnegativity of Δ. The paper also gives examples showing that the separability and strictness hypotheses are needed and that (D) alone is vacuous; the status of the measurability hypothesis (R) is left explicitly open.","tokens_in":10055,"tokens_out":11287,"duration_ms":97600,"significance":"If accepted, this is a satisfying structural result: the ELBO/Gibbs–Bogoliubov identity is the unique exact separable decomposition with a strict residual depending on the model only through the Gibbs measure. The proof is short, self-contained, and I find it correct conditional on (R). The corollary that α- and Rényi-divergence objectives cannot be arranged to satisfy such an exact identity under (S) and (P) is a useful, falsifiable consequence. The paper's main advertised caveat is the unresolved status of (R), which affects the sharpness claim but not the truth of Theorem 1.1 as stated.","major_comments":[{"comment":"The abstract and the opening of Section 6 claim that counterexamples show each hypothesis is needed, but no counterexample for (R) is provided. Example 6.3 shows (S) is needed and Example 6.4 shows (P) is needed; Remark 6.5 and Question 6.6 explicitly leave open whether a nonmeasurable Hamel-basis homomorphism can be completed to a decomposition satisfying (P). Thus the necessity/redundancy of (R) is unresolved, and the advertised statement that all hypotheses are necessary is stronger than what is proved. Please revise the abstract and Section 6 to say, e.g., that (S) and (P) are shown necessary, while (R) is used essentially in the proof but its optimality remains open.","section":"Abstract; Section 6; Remark 6.5; Question 6.6"}],"minor_comments":[{"comment":"The extension beyond finite X is described as expected but 'not carried out'. This is an honest limitation, but the sentence 'We expect no obstruction' should be labeled as a conjecture to avoid being read as a result.","section":"Section 7, Remark 7.3"},{"comment":"The range assertion for p' ↦ Σ_x tv(x)p'(x) is correct for finite X, but the proof is terse; adding one sentence on concentrating mass near the coordinates attaining the min and max would improve readability.","section":"Section 5, Step 3"},{"comment":"In the concrete case X={1,2}, m=(1,-1), the displayed expression Δ(Q,π)=D(Q∥π)+log(π_1/π_2) is negative whenever π_1<π_2 and Q=π; consider saying 'for Q=π with π_1<π_2' explicitly, since otherwise the sentence is a bit compressed.","section":"Section 6, Example 6.4"}],"recommendation":"minor_revision","confidential_remarks":"The core theorem is sound and the paper is well written. The only substantive problem is the overstatement about the status of (R); this can be fixed by a careful wording change. I do not see grounds for rejection or for requiring new mathematics in this revision, although resolving Question 6.6 would strengthen a future version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this paper proves something real and new. The prior literature characterizes divergences or inference rules; this is the first result I know that characterizes which functionals admit an exact decomposition of log Z as a computable part plus a strict divergence in the Gibbs measure. Theorem 1.1 and Corollary 4.2 are the payload, and they are worth having. I traced the proof in Section 5 and it is sound. Step 3's ratio-realization argument is valid, the measurable-homomorphism step is standard Aczel, and Step 5's perturbation correctly forces m_Q = 0. The gauge freedom for A and B is handled honestly, and the examples in Section 6 do real work. I also give credit for explicitly flagging the open questions: Question 6.6 and Remarks 6.7 and 7.3 are not buried.\n\nThe soft spot is the status of (R), and the abstract oversells it. The paper says \"counterexamples show that each hypothesis is needed,\" but that is only true for (S) and (P). For (R), Section 6 does not give a counterexample; it shows that dropping (R) lets Hamel-basis homomorphisms survive, and then asks in Question 6.6 whether (P) would kill them automatically. The theorem as stated is proved correctly, so this is not a flaw in the main result. But the minimality claim is stronger than what is established. If (R) turns out to be redundant, the theorem is a three-hypothesis theorem; if not, the extra hypothesis is doing essential work and the necessity story should say so. Either way, the abstract needs adjustment, and the body should acknowledge that the necessity of (R) is unresolved, not demonstrated by counterexample.\n\nEverything else checks out. The distinction between characterizing divergences and characterizing decompositions is well drawn, and the Discussion's separation from convex duality is fair. The finite-state restriction is reasonable for a mechanism proof, and the authors say they expect no obstruction in the infinite case without claiming to have checked it.\n\nWho should read this: anyone working on variational inference, mean-field theory, or the foundations of the ELBO. It is not a computational paper; it is a structural theorem. It deserves a serious referee, and with a minor revision on the abstract and a clearer statement about (R) I would accept it. Send it to peer review.","headline":"A genuinely new characterization of the ELBO/Gibbs-Bogoliubov identity, with a clean proof; the only real blemish is an abstract that overstates what the necessity examples show about hypothesis (R).","tokens_in":710,"tokens_out":919,"would_cite":true,"duration_ms":25673,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B03","94A17","62F15","39B22"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that any exact decomposition of log Z into a separable objective minus a nonnegative gap vanishing only at the Gibbs measure must be the variational free energy with the Kullback-Leibler divergence.","keywords":["variational free energy","Gibbs-Bogoliubov inequality","evidence lower bound","relative entropy","uniqueness theorem","functional equations","alpha-divergence","decomposition"],"falsifier":"Construct a decomposition satisfying (D), (S), and (P) using a nonmeasurable additive map φ: R^n → R with φ(1)=0, giving G ≠ F and Δ ≠ D(·||π); that would refute the conclusion if (R) is dropped. Alternatively, a proof that any such φ must be unbounded below on the hypersurface {h: ⟨Q,e^h⟩=1} would show (R) is redundant and strengthen the theorem.","tokens_in":9619,"feed_emoji":"🧮","tokens_out":4786,"duration_ms":42576,"temperature":0.7,"pith_summary":"This paper asks how much freedom exists in the identity log Z = -F(Q) + D(Q||π) that underlies the Gibbs–Bogoliubov inequality and variational inference. It proves that if a functional G is additively separable into a reference-measure term and a weight term, and the residual Δ is nonnegative and vanishes exactly at the Gibbs measure π, then G must be the variational free energy F(Q;p,ℓ)=D(Q||p)-E_Q[logℓ] and Δ must be the Kullback-Leibler divergence D(Q||π). The proof reduces the assumptions to a measurable group homomorphism and shows only the standard split survives. The significance is that objectives built from alpha- or Rényi divergences can provide bounds on log Z but can never yield an exact identity of this form; the ELBO identity is structural.","feed_headline":"Only KL divergence gives an exact variational split","feed_subtitle":"Under separability and strictness, the ELBO identity is unique — alpha and Rényi divergences can only bound log Z.","key_machinery":"The central device is the group homomorphism χ_Q: (P,·)→(R,+) defined by χ_Q(v)=B'(Q,ℓv)-B'(Q,ℓ), which the regularity assumption (R) promotes to a linear functional <m_Q, log v>. The proof then uses scale invariance (giving Σ m_Q = 0), the fact that the residual depends only on the Gibbs measure, and the second-order vanishing of relative entropy near Q = π, to force m_Q = 0 and hence G = F, Δ = D(·||π).","core_discovery":"Theorem 1.1: Let (G,Δ) satisfy log Z = -G + Δ for every trial Q and model (p,ℓ), with G additively separable (G = A(Q,p)+B(Q,ℓ), B measurable in ℓ) and Δ ≥ 0 with equality iff Q = π. Then G must be the variational free energy F(Q;p,ℓ) = D(Q||p) - E_Q[logℓ] and Δ(Q,π) = D(Q||π). The split A/B is unique up to a Q-dependent constant a(Q) that cancels in G. Equivalently, no divergence other than the Kullback-Leibler divergence can serve as the residual in an exact, additively separable decomposition of log Z.","pith_inferences":["If the open regularity question is resolved negatively (i.e., nonmeasurable homomorphisms can satisfy all other hypotheses), then the theorem's uniqueness would fail only in nonconstructive, Hamel-basis territory, leaving all computable applications untouched.","One can read the theorem as a foundational justification for why the ELBO, rather than any alpha-divergence objective, is the canonical exact variational identity; this could motivate reconsidering variational methods that abandon exact decompositions.","The gauge freedom suggests that constant offsets in log-likelihoods or baseline energies are absorbed by the Q-dependent split, which may clarify why additive constants in energy functions are harmless in practice.","Because the theorem characterizes decompositions rather than divergences, it could pair with axiomatic characterizations of relative entropy to give a two-sided picture: KL is the unique divergence and the unique exact-decomposition residual."],"forward_implications":["Objectives based on α- or Rényi divergences cannot be arranged into an exact identity of the form log Z = -G + Δ with Δ a strict divergence and G separable; the absence of an identity is structural.","In mean-field theory and variational inference, the functional to optimize is forced once one demands an exact decomposition; the only modelling freedom is the trial class.","The additive split A(Q,p)+B(Q,ℓ) is ambiguous by a Q-dependent gauge that cancels in G; only the sum is pinned down.","The bound and the identity are not two grades of the same thing: KL divergence delivers both, alternative divergences deliver only bounds.","The proof mechanism suggests the result extends beyond finite state spaces with bounded densities and a suitable topology, though this is not carried out."],"fun_headline_variants":["KL divergence is the only exact residual in variational splits","Exact log Z split forces KL: no other divergence works","Uniqueness: only KL divergence gives exact ELBO split","Only KL is exact: variational decomposition unique"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The most fragile premise is the regularity assumption (R), that the weight-dependent term B(Q,ℓ) is Lebesgue measurable in ℓ; without it, unmeasurable additive functions of the log-weights could in principle break the conclusion, and the paper leaves open whether strictness alone would exclude them.","fun_headline_variants_meta":{"raw":{"variants":["KL divergence is the only exact residual in variational splits","Exact log Z split forces KL: no other divergence works","Uniqueness: only KL divergence gives exact ELBO split","Only KL is exact: variational decomposition unique"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000662,"raw_usage":{"total_tokens":2924,"prompt_tokens":868,"completion_tokens":2056,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":612,"completion_tokens_details":{"reasoning_tokens":1991}},"tokens_in":612,"tokens_out":2056,"duration_ms":12647,"temperature":1.0,"reasoning_tokens":1991,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:24:16.069933+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a decomposition satisfying (D), (S), and (P) using a nonmeasurable additive map φ: R^n → R with φ(1)=0, giving G ≠ F and Δ ≠ D(·||π); that would refute the conclusion if (R) is dropped. Alternatively, a proof that any such φ must be unbounded below on the hypersurface {h: ⟨Q,e^h⟩=1} would show (R) is redundant and strengthen the theorem.","supporting_citations":[],"review_version":1}