← Proof Notes
Counterexample / Generative models

An omitted off-support condition in the EBGAN equilibrium characterization

Energy-based Generative Adversarial Network ↗

1,152 citations ↗Semantic Scholar · 2026-09-09

2017 · arXiv:1609.03126v4, 6 March 2017; ICLR 2017 · Reviewed 09 September 2026

An explicit example contradicts the selected statement as written.

Paper context

Overview

EBGAN trains a discriminator that assigns an energy score to each sample, with lower energy favored for real data and a margin-based penalty for generated samples. The generator learns to produce samples with low discriminator energy. The paper uses an autoencoder's reconstruction error as one way to define that energy.

Role of the theoretical result

The equilibrium analysis explains what the two-player objective should recover when both networks have unrestricted capacity. Theorem 2 characterizes pairs of generators and energy functions that are claimed to be equilibria, extending the distribution-matching conclusion of Theorem 1.

Original paper ↗

01 / Summary

Summary of the result

The off-support extension of Theorem 2 permits discriminator energies below the constant energy attained on the data support. Such a discriminator is a best response to a matching generator, but the generator can reduce its objective by moving to an accessible lower-energy region. The extension therefore fails as a sufficient Nash-equilibrium condition. The necessary distribution-matching result in Theorem 1 is unaffected.

02 / Statement

Statement under review

Theorem 2 and footnote 1, PDF p. 3; Appendix A, proof of Theorem 2, PDF p. 11. · paraphrased

Theorem 2 characterizes equilibria by a matching generator and a constant discriminator energy γ on the data distribution. Footnote 1 extends the statement to regions where the data density is zero by allowing arbitrary discriminator values between zero and the margin there.

pG=pD,D(x)=γ on supp⁡(pD),0≤γ≤m,D(x)∈[0,m] off support.p_G=p_D,\qquad D(x)=\gamma\text{ on }\operatorname{supp}(p_D),\qquad 0\le\gamma\le m,\qquad D(x)\in[0,m]\text{ off support}.
Relevant assumptions ↘
  • The analysis is the paper's nonparametric game with unrestricted generator and discriminator capacity.
  • Both players minimize their respective objectives; the generator minimizes its expected discriminator energy.
  • The discriminator energy is nonnegative and the hinge margin m is positive.
  • The selected extension concerns data distributions with a region of zero density.

03 / Derivation

Counterexample and derivation

5 steps · complete derivation
  1. 01

    Specify the losses

    For the matched generator, the discriminator sees the same distribution in its two terms. The generator minimizes energy under its own distribution.

    V(G,D)=EpDD(X)+EpG[m−D(X)]+,U(G,D)=EpGD(X).V(G,D)=\mathbb E_{p_D}D(X)+\mathbb E_{p_G}[m-D(X)]_+,\qquad U(G,D)=\mathbb E_{p_G}D(X).
  2. 02

    Construct a pair allowed by the footnote

    Set m=1 and pD=pG=Uniform[0,1]. Choose a continuous nonnegative energy equal to 1/2 on [0,1], decreasing to zero on [1,2], and zero from 2 onward.

    D(x)=12min⁡{1,max⁡{0,2−x}},γ=12.D(x)=\tfrac12\min\{1,\max\{0,2-x\}\},\qquad \gamma=\tfrac12.
  3. 03

    Verify the discriminator is a best response

    On the common data/generated support, D+[1−D]+=1. This is the minimum of the pointwise sum, so no discriminator deviation lowers V.

    V(G,D)=1,d+[1−d]+≥1(d≥0).V(G,D)=1,\qquad d+[1-d]_+\ge1\quad(d\ge0).
  4. 04

    Exhibit the generator's profitable deviation

    Keep D fixed and change the generator distribution to Uniform[2,3]. Its energy is now zero instead of one half.

    U(G,D)=12,U(G′,D)=0,pG′=Uniform⁡[2,3].U(G,D)=\tfrac12,\qquad U(G',D)=0,\qquad p_{G'}=\operatorname{Uniform}[2,3].
  5. 05

    Apply the equilibrium definition

    The required inequality U(G,D)≤U(G′,D) is false. The claimed sufficient characterization, including its off-support footnote, therefore admits a pair that is not an equilibrium.

    12≰0.\tfrac12\not\le0.

Counterexample

Let data and generator be uniform on [0,1], with margin one and discriminator energy 1/2 on that interval. If the energy is zero on [2,3], the generator can reduce its objective from 1/2 to zero by changing support.

pG=pD=Unif⁡[0,1],D∣[0,1]=12,D∣[2,3]=0:U(G′,D)=0<12=U(G,D).p_G=p_D=\operatorname{Unif}[0,1],\quad D|_{[0,1]}=\tfrac12,\quad D|_{[2,3]}=0:\quad U(G',D)=0<\tfrac12=U(G,D).

04 / Implications

Implications and proposed correction

Theoretical implications

Affected result

This refutes the sufficiency of the off-support extension. It does not refute existence of equilibria, Theorem 1's necessary distribution-matching conclusion, or Theorem 2 under its full-support assumption.

Empirical scope

Relation to reported experiments

This is a counterexample to the abstract game characterization. It is not a training experiment or a claim that a particular finite autoencoder architecture exhibits the deviation.

Proposed correction

Sufficient conditions and revised bound

For a matched generator and constant energy γ on its support, require D(x)≥γ almost everywhere in every off-support region reachable by an allowed generator. Together with 0≤γ≤m, this supplies the missing best-response condition. A global constant discriminator is one sufficient special case.

D(x)≥γon generator-reachable off-support regions.D(x)\ge\gamma\quad\text{on generator-reachable off-support regions}.

Implementation implications

Matching generated and data distributions is not, by itself, an equilibrium certificate for a fixed learned energy function. A certificate must also rule out lower-energy reachable regions. This correction does not require replacing the paper's loss functions.

Limits of this review

  • The selected claim is the footnote read as part of Theorem 2's sufficient characterization.
  • The example uses intervals of positive measure, so it does not depend on changing a density at a single point.
  • The proposed energy can be continuous; discontinuity is not needed.

05 / References

Sources and correction history

  1. 01
    EBGAN, exact version ↗

    Section 2.2, Theorem 2 and footnote 1, p. 3; Appendix A, p. 11

  2. 02
    Version history ↗

    Latest listed revision v4, 6 March 2017

  3. 03
    Calibrating Energy-based Generative Adversarial Networks ↗

    Appendix A.2, Eqs. (19)–(23) and Proposition A.1, pp. 11–12

  4. 04
    ICLR discussion page ↗

    Discussion access was incomplete in this audit

Download the arithmetic witnesses · Python, no dependencies ↓
Correction search · 09 September 2026

2026-09-09 The original equilibrium definition, theorem, footnote, and appendix proof were reread and the theorem page visually checked. Current arXiv history and targeted searches for support-related corrections were checked. Calibrating EBGAN Appendix A.2 was reread through Proposition A.1. "Energy-based Generative Adversarial Network" "off-support" "Energy-based Generative Adversarial Network" "constant" "support" equilibrium correction "Energy-based" "GAN" "off-support" equilibrium No source supplying the missing fixed-discriminator off-support lower bound was found in the bounded search. Calibrating EBGAN explicitly analyzes discriminator best responses and then profiles the discriminator as a function of the generator distribution. Its Eq. (21) also includes a data-expectation term. That optimization is not the fixed-discriminator generator deviation required by the original Nash definition, so its inspected result does not repair this sufficient characterization. The original OpenReview discussion and notes endpoints were not fully accessible. The search is bounded and does not establish that the observation is previously unknown.

A bounded search is not evidence of priority or proof that no correction exists.

Entirely AI-generated analysis, including cross-checks by separate AI agents; no independent human verification. Authors have not been contacted. Review standard.

Suggest a correction with a source ↗
Next analysisAn omitted skew-part scale in the consensus damping bound