Symmetry breaking for local minimizers of a free discontinuity problem

Massimo Gobbino; Nicola Picenni

REVIEW 1 major objections 4 minor 26 references

Symmetry breaking for local minimizers of a free discontinuity problem

T0 review · 1 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read In two dimensions, a free-discontinuity energy with linear forcing admits entire local minimizers—explicit bi-staircases—whose jump set is not a family of parallel hyperplanes, for $\theta=0$.

desk verdict A genuinely new symmetry-breaking construction for theta=0 with a real but likely fixable gap in the extension from piecewise-constant to general competitors. read the letter →

arxiv 2506.05270 v2 pith:7ECBY6N6 submitted 2025-06-05 math.AP math.OC

Massimo Gobbino , Nicola Picenni This is my paper · ORCID

classification math.APmath.OC MSC 49Q2049K0549K10

keywords symmetrybreakingentirelocalminimizersfreediscontinuityproblemcalibrationmethodPerona-Malikfunctionalstaircasingpurejumpfunctionsbi-staircase

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

This paper studies a variational problem in which functions pay a cost for their jump set and a quadratic penalty for deviating from a fixed linear function, the forcing term. In one dimension, the entire local minimizers are completely classified: they are staircases whose step length and height are fixed by the parameters. The main result is that in two and higher dimensions this one-dimensional picture is incomplete. For jump exponent $\theta=0$, there exist entire local minimizers—explicitly constructed bi-staircases—whose level sets are not stripes orthogonal to the forcing gradient, so the symmetry of the problem is broken. The proof works through a calibration method that manipulates auxiliary functions directly rather than their derivatives, which is why the construction requires less regularity than earlier calibration arguments.

What carries the argument

The calibration method for free-discontinuity problems is the central tool. One chooses a vector field, or a family of differential forms $\omega_z=A_\theta(x,z)\,dx+F_\theta(x,z)\,dy$, such that the induced functional $G(\Omega,v)$ depends only on the values of $v$ near $\partial\Omega$, satisfies $G(\Omega,v)\le JF(\Omega,v)$ for every competitor $v$, and satisfies equality $G(\Omega,u)=JF(\Omega,u)$ for the candidate $u$. In one dimension the calibration reduces to a scalar function $F_\theta$ built from a truncated cubic, and no continuity of $F_\theta$ is needed. In two dimensions the calibration uses the pair $(A_\theta,F_\theta)$ and reduces to the circle-type inequality $(A_\theta(x,z_2)-A_\theta(x,z_1))^2+(F_\theta(x,z_2)-F_\theta(x,z_1))^2\le \alpha_\theta^2(z_2-z_1)^{2\theta}$, with equality on the jump set of the bi-staircase. A slicing argument extends minimizers from $\mathbb{R}^{d_1}$ to $\mathbb{R}^{d_1+d_2}$ by ignoring extra variables, which both proves that staircases remain minimizers in all dimensions and lifts the bi-staircase from $\mathbb{R}^2$ to higher dimensions.

What would settle it

Compute the explicit calibration data for $\theta=0$ on a fine grid: any pair $(x,z_1,z_2)\in[0,1]\times\mathbb{R}^2$ with $(A_0(x,z_2)-A_0(x,z_1))^2+(F_0(x,z_2)-F_0(x,z_1))^2 > 16$ would violate inequality (5.9) and invalidate Theorem 2.5(2). Equivalently, a numerical descent that lowers the energy of the bi-staircase on a large rectangle below the value predicted by the equality cases would refute the claim.

Watch

Extended reading notes

Core claim

Theorem 2.5(2) states that when $\theta=0$, the canonical $(H,V)$-staircase in the direction $\xi/M$ and its oblique translations are not the only entire local minimizers of the functional (1.4). The paper constructs a canonical bi-staircase in $\mathbb{R}^2$: above a 2-periodic interface curve $y=f_0(x)$ it agrees with the staircase $S(x)$, and below it agrees with the shifted staircase $S(x-1)+1$. Its jump set consists of the graph of $f_0$ together with vertical half-lines at integer positions, so it is not a union of parallel lines. Using a calibration built from the one-dimensional calibration function $F_0$ and an auxiliary piecewise-affine function $A_0$, the paper proves that this bi-staircase is an entire local minimizer for the rescaled parameters $\alpha=4$, $\beta=3$, $\xi=(1,0)$, and then slicing extends the example to every dimension $d\ge 3$. For $\theta>0$, the same candidate is conjectured to minimize, but the proof is given only for $\theta=0$.

Load-bearing premise

The argument that every pure-jump competitor can be approximated, with converging energy, by piecewise-constant functions that also match the candidate near the boundary is asserted in Section 5 as an adaptation of results in [4] and [5], but the adaptation is not proved in detail.

Editorial extensions

If this is right

In one dimension the classification is complete: for every $\theta\in[0,1)$, $\alpha,\beta>0$, and $M\neq 0$, the entire local minimizers of (1.3) are exactly the oblique translations of the $(H,V)$-staircase with $H$ and $V$ given by (2.2).
In every dimension $d\ge 2$, the classical staircases remain entire local minimizers, because any minimizer that ignores extra variables extends by slicing.
For $\theta=0$, non-staircase entire local minimizers exist in every dimension $d\ge 2$: the bi-staircase in $\mathbb{R}^2$ and its slicing extensions to $\mathbb{R}^d$.
The calibration constructed here does not require differentiability of the auxiliary functions, so the three-condition scheme (boundary independence, lower bound, equality on the candidate) can certify minimality in settings where classical calibrations would demand too much regularity.
Because these functionals arise as $\Gamma$-limits of blow-ups of Perona-Malik regularizations, the exotic minimizers are new candidates for the asymptotic staircasing patterns in two dimensions; whether they actually appear there is left open in the paper.

Reading between the lines

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

The $\theta=0$ construction is probably not an isolated phenomenon: the paper's own numerical experiments suggest that the same bi-staircase may minimize for $\theta\in(0,1)$, and the only missing ingredient is an $A_\theta$ satisfying the analogous inequality (5.9) with exponent $2\theta$.
If exotic minimizers do appear as blow-up limits in the Perona-Malik models, the effective description of two-dimensional staircasing would have to be enlarged from one-dimensional profiles to patterns with triple junctions, with the interface curve $f_\theta$ carrying the microstructural information.
The low-regularity calibration suggests that the same method could be applied to energies with anisotropic or nonlocal jump costs, where smooth null Lagrangians may not exist; a natural test is whether an analogous bi-staircase minimizes when the jump cost is a general function of jump height.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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

Colleague,

You can skip the one-dimensional half of this paper and read the construction: for theta=0, they build an explicit entire local minimizer in the plane that is not a straight staircase. That is a genuine counterexample to the natural "stripes" intuition, and the calibration machinery adapted from Alberti-Bouchitté-Dal Maso fits the problem well. The bi-staircase and the piecewise-affine psi that calibrates it are new, and the reduced-regularity variant of calibration (working with functions rather than vector fields) is a useful trick for this class of problems.

The soft spot is exactly where the reader's report puts it. Proposition 5.3 proves minimality among piecewise-constant competitors with regular level sets. For general pure-jump competitors the authors rely on a density statement in the final paragraph of Section 5: any competitor equal to the candidate near the boundary can be approximated in L2 and in jump energy by piecewise-constant functions with the same boundary values. They cite [4,5] and say the constructions can be adapted. That is asserted, not proved. The cited results are density theorems without prescribed boundary values, and the adaptation is not a formality—triple junctions and the quadratic fidelity term both deserve careful checking. If the density statement fails, the theorem only establishes minimality in a restricted class, not the stated result. I consider this a genuine gap, but a fixable one; this is exactly what a referee should ask for.

The one-dimensional uniqueness theorem (Theorem 2.2) is also sketched rather than proved, but it points to prior work for the details, so I would treat it as background. The self-citations to [19,24] are appropriate; the one-dimensional characterization is genuinely prior work.

On balance, the paper deserves a serious referee. The construction is explicit, the calibration inequalities are verified in detail on piecewise-constant competitors, and the symmetry-breaking conclusion is worth publishing if the density step can be supplied. I would recommend conditional acceptance: ask for a complete proof or a precise reference for the boundary-preserving density approximation, and for a clear statement of what is proved without it.

Referee Report

1 major / 4 minor

Summary. The paper studies the functional (1.4) on pure-jump functions, combining a jump penalization with a quadratic fidelity to a linear forcing term. It proves a complete characterization of entire local minimizers in one dimension (Theorem 2.2) and, in the main theorem (Theorem 2.5), shows that for θ=0 in two dimensions there exist entire local minimizers that are not oblique translations of the standard staircase in the forcing direction. The proof constructs a 'bi-staircase' whose jump set combines the graph of a periodic curve with vertical half-lines, and verifies minimality through an explicit calibration argument. The slicing method then extends the two-dimensional example to higher dimensions.

Significance. If fully established, the main theorem is a valuable and surprising result: it demonstrates symmetry breaking for a free-discontinuity problem with a linear fidelity and provides an explicit family of exotic entire minimizers. The calibration method is adapted to a low-regularity setting in an interesting way, and the verification for piecewise-constant competitors (Proposition 5.3 together with Lemma 5.5) is explicit and internally consistent. The main caveat is that the final density step from piecewise-constant to all pure-jump competitors is asserted rather than proved, and that step is load-bearing for Theorem 2.5(2).

major comments (1)

[Section 5, proof of Proposition 5.3, final paragraph ('Inequality for general competitors')] The proof establishes the inequality JF(Ω,v) ≥ JF(Ω,Ŝθ) only for v∈PC(Ω). The extension to arbitrary v∈PJ(Ω) is contained in the assertion that v can be approximated by v_n∈PC(Ω) with v_n=Ŝθ near ∂Ω, v_n→v in L2, and convergence of the jump energy, citing [4,5] and stating that the constructions 'can be adapted.' This is the load-bearing step for Theorem 2.5(2), and no proof is given. The cited density results are formulated without prescribed boundary values, and forcing the approximants to equal the fixed bi-staircase in a neighborhood of ∂Ω is a nontrivial constraint, especially at the triple junctions where the graph of fθ meets the vertical half-lines. In addition, the fidelity term is quadratic, so the required L2 convergence is stronger than the L1-type approximation commonly provided by BV/SBV density results. The paper should either prove this boundary-preserving density lemma or state it as a separate theorem with a complete proof.

minor comments (4)

[Equation (5.3)] The integrand uses x both as the integration variable and as the upper limit; please replace the dummy variable, e.g. fθ(x) := ∫_0^{|x|} gθ(t) / sqrt(αθ² - gθ(t)²) dt.
[Abstract and Section 2] The abstract says the functional is defined on 'piecewise constant functions,' but the actual space PJloc allows countably many jumps; please align the terminology with the pure-jump setting used in the paper.
[Section 5, end of proof of Proposition 5.3] The statement that 'regular open sets where Ŝθ is piecewise constant exhaust the whole plane' is used to complete the proof, but no proof or precise formulation is given. Please clarify that for every bounded open set and every compact subset there is a regular open set Ω' with K⊂Ω'⊂⊂Ω and Ŝθ∈PC(Ω').
[Step 4 of the proof of Theorem 2.2] The uniqueness part of Theorem 2.2 is presented only as a sketch with references to [19,24]. Since Theorem 2.2 is stated as a full characterization, please state explicitly which arguments are imported from those references and how they adapt to all θ∈[0,1).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the exotic minimizer is verified by an explicit calibration, not derived from the target conclusion.

full rationale

The central claim (Theorem 2.5(2)) is an existence statement for the canonical bi-staircase. The candidate is motivated heuristically in Remark 5.2 from the Euler-Lagrange equation for the separating curve, but minimality is then proved independently by Proposition 5.3: one constructs a piecewise affine function A0 in Step 5 from Lemma 5.5 and verifies conditions (5.6)-(5.9). The equalities used on the candidate's jump set are consistency relations: (5.5) identifies Fθ(x,1)-Fθ(x,0) with gθ, and (5.7) is a matching condition that the calibration vector has length αθ. This is the standard matching step of a calibration argument, not a derivation of the theorem from its own conclusion. The one-dimensional classification in Theorem 2.2 borrows its uniqueness direction from the authors' earlier papers [19,24]; however, the burden of [19,24] is only the converse of the one-dimensional statement, and the exotic minimizer claim in Theorem 2.5(2) is unaffected because the minimality proof for the bi-staircase never invokes that classification. The slicing extension cites [17] (Federer) for the standard section-area formula, an external tool. The only delicate step is the approximation of general pure-jump competitors by piecewise-constant competitors with prescribed boundary data, attributed to [4,5] with the assertion that 'the constructions can be adapted to produce a sequence that satisfies this additional condition'. This adaptation is not proved in detail and is a genuine correctness risk, but it is not circular: the density of polyhedral partitions is an external result, and the adaptation is a technical approximation claim rather than a restatement of the target inequality. No fitted parameter is relabeled as a prediction, no uniqueness theorem is imported from the authors' prior work to forbid alternatives, and no known result is renamed as a new one. Therefore the derivation chain is substantially self-contained.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters are fitted to data; the only parameters (theta, alpha, beta, xi) are part of the problem statement. The central existence proof for theta=0 relies on explicit calibration functions F and A, whose defining formulas are given. The main external input is the density and approximation result for piecewise constant functions with boundary preservation, which is stated rather than fully proved. The one-dimensional characterization also uses prior work [19,24].

assumptions (3)

standard math SBV/GSBV structural facts, including rectifiability of jump sets and the slicing identity [17, Theorem 3.2.22], hold for pure jump functions.
Used throughout Section 2 and Proposition 4.1 to justify the functional and the slicing extension.
domain assumption Every pure-jump competitor can be approximated by piecewise constant functions that match the candidate near the boundary and have convergent energy.
Stated and justified by appeal to [4,5] plus an assertion that the constructions adapt; central for lifting minimality from piecewise-constant to general competitors in Section 5.
domain assumption The one-dimensional uniqueness part of Theorem 2.2 is inherited from [19,24] via a sketch.
Only needed for the characterization in one dimension; the existence of staircases as minimizers is proved by calibration in Section 3.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Symmetry breaking for local minimizers of a free discontinuity problem." pith.science (2026). https://pith.science/paper/7ECBY6N6

@misc{pith2026250605270,
  author       = {Pith},
  title        = {Pith review of: Symmetry breaking for local minimizers of a free discontinuity problem},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7ECBY6N6}},
  note         = {Machine review of arXiv:2506.05270}
}

read the original abstract

We study a functional defined on the class of piecewise constant functions, combining a jump penalization, which discourages discontinuities, with a fidelity term that penalizes deviations from a given linear function, called the forcing term. In one dimension, it is not difficult to see that local minimizers form staircases that approximate the forcing term. Here we show that in two dimensions symmetry breaking occurs, leading to the emergence of exotic minimizers whose level sets are not simple stripes with boundaries orthogonal to the gradient of the forcing term. The proof relies on a suitable adaptation of the calibration method for free discontinuity problems; as a side benefit, our version requires less regularity than the classical one.

Figures

Figures reproduced from arXiv: 2506.05270 by the authors.

**Figure 2.** Some level sets of the canonical bi-staircase. The separ [PITH_FULL_IMAGE:figures/full_fig_p020_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 26 canonical work pages

[1]

Alberti , G

G. Alberti , G. Bouchitt ´ e, G. Dal Maso . The calibration method for the Mumford-Shah functional and free-discontinuity problems. Calc. Var. Partial Dif- ferential Equations 16 (2003), no. 3, 299–333

work page 2003
[2]

Ambrosio , N

L. Ambrosio , N. Fusco , D. Pallara. Functions of bounded variation and free discontinuity problems. Oxford Mathematical Monographs. The Clarendon Press, Oxford University Press, New York, 2000

work page 2000
[3]

W. Ao, A. DelaTorre , M. d. M. Gonz ´alez. Symmetry and symmetry break- ing for the fractional Caﬀarelli-Kohn-Nirenberg inequality. J. Funct. Anal. 282 (2022), no. 11, Paper No. 109438, 58

work page 2022
[4]

Bellettini , A

G. Bellettini , A. Chambolle , M. Goldman . The Γ-limit for singularly per- turbed functionals of Perona-Malik type in arbitrary dimension. Math. Models Methods Appl. Sci. 24 (2014), no. 6, 1091–1113

work page 2014
[5]

Braides , S

A. Braides , S. Conti , A. Garroni . Density of polyhedral partitions. Calc. Var. Partial Diﬀerential Equations 56 (2017), no. 2, Paper No. 28, 10

work page 2017
[6]

Brock , V

F. Brock , V. Ferone , B. Kawohl . A symmetry problem in the calculus of variations. Calc. Var. Partial Diﬀerential Equations 4 (1996), no. 6, 593–599

work page 1996
[7]

Caffarelli, R

L. Caffarelli, R. Kohn , L. Nirenberg . First order interpolation inequalities with weights. Compositio Math. 53 (1984), no. 3, 259–275

work page 1984
[8]

Catrina , Z.-Q

F. Catrina , Z.-Q. W ang. On the Caﬀarelli-Kohn-Nirenberg inequalities: sharp constants, existence (and nonexistence), and symmetry of ext remal functions. Comm. Pure Appl. Math. 54 (2001), no. 2, 229–258. 28

work page 2001

Show all 26 references

[9]

Dacorogna, W

B. Dacorogna, W. Gangbo , N. Sub´ıa. Sur une g´ en´ eralisation de l’in´ egalit´ e de Wirtinger. Ann. Inst. H. Poincar´ e C Anal. Non Lin´ eaire9 (1992), no. 1, 29–50

1992
[10]

Daneri , A

S. Daneri , A. Kerschbaum , E. Runa . One-dimensionality of the minimizers for a diﬀuse interface generalized antiferromagnetic model in gene ral dimension. J. Funct. Anal. 283 (2022), no. 12, Paper No. 109715, 53

2022
[11]

Daneri , E

S. Daneri , E. Runa . Exact periodic stripes for minimizers of a local/nonlocal interaction functional in general dimension. Arch. Ration. Mech. Anal. 231 (2019), no. 1, 519–589

2019
[12]

Daneri, E

S. Daneri, E. Runa. Pattern formation for a local/nonlocal interaction functional arising in colloidal systems. SIAM J. Math. Anal. 52 (2020), no. 3, 2531–2560

2020
[13]

Daneri , E

S. Daneri , E. Runa . One-dimensionality of the minimizers in the large volume limit for a diﬀuse interface attractive/repulsive model in general dim ension. Calc. Var. Partial Diﬀerential Equations 61 (2022), no. 1, Paper No. 12, 31

2022
[14]

De Nitti , F

N. De Nitti , F. Glaudo , T. K ¨onig. Non-degeneracy, stability and symmetry for the fractional Caﬀarelli-Kohn-Nirenberg inequality. ArXiv:2403 .02303 (2024)

2024
[15]

Dolbeault , M

J. Dolbeault , M. J. Esteban , M. Loss . Rigidity versus symmetry breaking via nonlinear ﬂows on cylinders and Euclidean spaces. Invent. Math. 206 (2016), no. 2, 397–440

2016
[16]

M. J. Esteban . Nonsymmetric ground states of symmetric variational problems. Comm. Pure Appl. Math. 44 (1991), no. 2, 259–274

1991
[17]

H. Federer . Geometric measure theory , Die Grundlehren der mathematischen Wissenschaften, volume Band 153. Springer-Verlag New York, Inc., New York, 1969

1969
[18]

Ghisi , M

M. Ghisi , M. Gobbino , G. Rovellini . Symmetry-breaking in a generalized Wirtinger inequality. ESAIM Control Optim. Calc. Var. 24 (2018), no. 4, 1381– 1394

2018
[19]

Gobbino , N

M. Gobbino , N. Picenni . A quantitative variational analysis of the staircas- ing phenomenon for a second order regularization of the Perona-M alik functional. Trans. Amer. Math. Soc. 376 (2023), no. 8, 5307–5375

2023
[20]

Gobbino , N

M. Gobbino , N. Picenni . Multi-scale analysis of minimizers for a second or- der regularization of the Perona–Malik functional. Calc. Var. Partial Diﬀerential Equations 64 (2025), no. 5, Paper No. 156

2025
[21]

Goldman , E

M. Goldman , E. Runa . On the optimality of stripes in a variational model with non-local interactions. Calc. Var. Partial Diﬀerential Equations 58 (2019), no. 3, Paper No. 103, 26

2019
[22]

B. Kawohl. Symmetry or not? Math. Intelligencer 20 (1998), no. 2, 16–22. 29

1998
[23]

N. Picenni . The Perona-Malik problem: singular perturbation and semi- discrete approximation . Ph.D. thesis, Scuola Normale Superiore, 2023. https://dx.doi.org/10.25429/picenni-nicola phd2023-11-23

2023 doi
[24]

N. Picenni . Staircasing eﬀect for minimizers of the one-dimensional discrete Perona–Malik functional. ESAIM Control Optim. Calc. Var. 30 (2024), Paper No. 44, 45

2024
[25]

Pluda , M

A. Pluda , M. Pozzetta. Minimizing properties of networks via global and local calibrations. Bull. Lond. Math. Soc. 55 (2023), no. 6, 3029–3052

2023
[26]

M. Solci . Local interpolation techniques for higher-order singular pertur bations of non-convex functionals: free-discontinuity problems, 2024. A rXiv:2402.10656. 30

2024 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Alberti , G

G. Alberti , G. Bouchitt ´ e, G. Dal Maso . The calibration method for the Mumford-Shah functional and free-discontinuity problems. Calc. Var. Partial Dif- ferential Equations 16 (2003), no. 3, 299–333

work page 2003

[2] [2]

Ambrosio , N

L. Ambrosio , N. Fusco , D. Pallara. Functions of bounded variation and free discontinuity problems. Oxford Mathematical Monographs. The Clarendon Press, Oxford University Press, New York, 2000

work page 2000

[3] [3]

W. Ao, A. DelaTorre , M. d. M. Gonz ´alez. Symmetry and symmetry break- ing for the fractional Caﬀarelli-Kohn-Nirenberg inequality. J. Funct. Anal. 282 (2022), no. 11, Paper No. 109438, 58

work page 2022

[4] [4]

Bellettini , A

G. Bellettini , A. Chambolle , M. Goldman . The Γ-limit for singularly per- turbed functionals of Perona-Malik type in arbitrary dimension. Math. Models Methods Appl. Sci. 24 (2014), no. 6, 1091–1113

work page 2014

[5] [5]

Braides , S

A. Braides , S. Conti , A. Garroni . Density of polyhedral partitions. Calc. Var. Partial Diﬀerential Equations 56 (2017), no. 2, Paper No. 28, 10

work page 2017

[6] [6]

Brock , V

F. Brock , V. Ferone , B. Kawohl . A symmetry problem in the calculus of variations. Calc. Var. Partial Diﬀerential Equations 4 (1996), no. 6, 593–599

work page 1996

[7] [7]

Caffarelli, R

L. Caffarelli, R. Kohn , L. Nirenberg . First order interpolation inequalities with weights. Compositio Math. 53 (1984), no. 3, 259–275

work page 1984

[8] [8]

Catrina , Z.-Q

F. Catrina , Z.-Q. W ang. On the Caﬀarelli-Kohn-Nirenberg inequalities: sharp constants, existence (and nonexistence), and symmetry of ext remal functions. Comm. Pure Appl. Math. 54 (2001), no. 2, 229–258. 28

work page 2001