pith. sign in

arxiv: 2602.08906 · v2 · pith:GOMZ4KBFnew · submitted 2026-02-09 · 🧮 math.OC

Switching Point Optimization for Abstract Parabolic Equations

Christoph Buchheim , Christian Meyer , Alimhan Musalatov This is my paper

Pith reviewed 2026-05-22 10:54 UTC · model grok-4.3

classification 🧮 math.OC
keywords switching point optimizationsemilinear parabolic equationsoptimal controlFréchet differentiabilitymaximal parabolic regularityproximal gradient methodconvex hull formulation
0
0 comments X

The pith

The reduced objective in switching point optimization for abstract parabolic equations is continuously differentiable with respect to the switching points.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper proves that the mapping from switching points to the resulting control is continuously Fréchet-differentiable when its range is taken in the dual of Hölder continuous functions of time. Treating the underlying state equation in a weak form that relies on maximal parabolic regularity then transfers this differentiability to the reduced objective function. The resulting smoothness opens the door to gradient-based local minimization algorithms such as the proximal gradient method. The authors also supply an explicit description of the convex hull of all feasible switching functions, which may later support global solution techniques.

Core claim

The switching-point-to-control mapping is continuously Fréchet-differentiable when considered with values in the dual of Hölder continuous functions in time. By treating the state equation in weak form based on the concept of maximal parabolic regularity, one can then show that the reduced objective is continuously differentiable with respect to the switching points.

What carries the argument

The switching-point-to-control mapping, shown to be continuously Fréchet-differentiable into the dual of Hölder continuous functions, together with the weak formulation of the state equation that uses maximal parabolic regularity.

If this is right

  • Gradient-based algorithms such as the proximal gradient method become applicable to the minimization of the reduced objective.
  • Local solutions obtained this way will in general not be globally optimal because the switching-point-to-control map is non-convex.
  • The convex hull of the set of feasible switching functions admits an explicit extended formulation that could be used inside branch-and-bound procedures.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same differentiability technique might carry over to other classes of evolution equations once an analogous weak regularity notion is available.
  • Embedding the convex-hull description into a global solver could turn the present local method into a practical tool for finding globally optimal switching times.
  • Numerical verification of the derivative formula at randomly sampled switching configurations would provide an independent check of the abstract result.

Load-bearing premise

The state equation admits a weak formulation based on maximal parabolic regularity that is sufficient to pass differentiability from the control-to-state map to the reduced objective with respect to the switching times.

What would settle it

A concrete parabolic equation satisfying maximal parabolic regularity for which the reduced objective is nevertheless non-differentiable at some admissible set of switching points would falsify the central claim.

read the original abstract

This work is concerned with a switching point optimization problem governed by a semilinear parabolic equation in abstract function spaces. It is shown that the switching-point-to-control mapping is continuously Fr\'echet-differentiable when considered with values in the dual of H\"older continuous functions in time. By treating the state equation in weak form based on the concept of maximal parabolic regularity, one can then show that the reduced objective is continuously differentiable w.r.t. the switching points, which allows to use gradient-based methods like the proximal gradient method for its minimization. Numerical experiments confirm our theoretical findings, but also illustrate that such a method will in general not be able to solve the problem to global optimality due to the non-convex nature of the switching-point-to-control map. We therefore give a precise characterization of the convex hull of set of feasible switching functions in terms of an extended formulation. The latter might be useful for a branch-and-bound approach for the computation of global minimizers, but this is subject to future research.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper studies switching-point optimization for controls in abstract semilinear parabolic equations. It proves that the map from switching points to the control is continuously Fréchet differentiable when taking values in the dual of Hölder-continuous functions in time. Treating the state equation in weak form via maximal parabolic regularity then yields continuous differentiability of the reduced objective with respect to the switching times, enabling gradient-based methods such as proximal gradient. Numerical examples confirm the theory but illustrate that local methods generally fail to reach global optimality due to non-convexity; an extended formulation characterizing the convex hull of feasible switching functions is derived for possible future branch-and-bound use.

Significance. If the differentiability results hold, the work supplies a functional-analytic foundation for gradient optimization of switching times in infinite-dimensional parabolic control problems, extending standard bang-bang theory to semilinear abstract settings. The use of maximal parabolic regularity in weak form is a technical strength that broadens applicability. The convex-hull characterization and numerical illustrations add practical value, though the non-convexity discussion correctly cautions against expecting global solutions from local methods.

major comments (2)
  1. [§3.4, Theorem 3.7] §3.4, Theorem 3.7 and the subsequent chain-rule argument: the continuous Fréchet differentiability of the reduced objective is obtained by composing the switching-point-to-control map with the control-to-state operator in the weak formulation. The argument invokes maximal parabolic regularity for the linearized equation, but no explicit uniform bound (independent of the switching-time perturbation) is derived for the regularity constant when the semilinear term is present; the Lipschitz constant of the nonlinearity may depend on the state, which itself jumps with the control. Without a localization argument or an a-priori bound on the state in a suitable Hölder space, the estimates may only guarantee directional derivatives rather than continuous Fréchet differentiability.
  2. [§5, Proposition 5.2] §5, Proposition 5.2: the extended formulation is asserted to describe the convex hull of the set of feasible switching functions. The proof sketch relies on convex combinations of Dirac measures at the switching instants, yet it is not shown that every point in the convex hull can be realized by a control whose switching times lie in the original admissible set; an explicit counter-example or a tightness result would be needed to confirm that the formulation is exact rather than a relaxation.
minor comments (2)
  1. [§2 and §3] The notation for the dual of the Hölder space C^α([0,T]) is introduced in §2 but used without the dual pairing symbol in several places in §3; adding the explicit duality bracket would improve readability.
  2. [Figure 3] Figure 3 (proximal-gradient convergence) lacks a comparison with the theoretical linear rate that would follow from the established differentiability; including such a reference line would strengthen the numerical validation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of our manuscript and the constructive comments. We address the two major comments point by point below, providing clarifications and indicating where revisions will strengthen the presentation.

read point-by-point responses

  1. Referee: [§3.4, Theorem 3.7] §3.4, Theorem 3.7 and the subsequent chain-rule argument: the continuous Fréchet differentiability of the reduced objective is obtained by composing the switching-point-to-control map with the control-to-state operator in the weak formulation. The argument invokes maximal parabolic regularity for the linearized equation, but no explicit uniform bound (independent of the switching-time perturbation) is derived for the regularity constant when the semilinear term is present; the Lipschitz constant of the nonlinearity may depend on the state, which itself jumps with the control. Without a localization argument or an a-priori bound on the state in a suitable Hölder space, the estimates may only guarantee directional derivatives rather than continuous Fréchet differentiability.

    Authors: We appreciate this observation on the uniformity of the estimates. In the proof of Theorem 3.7 the continuous dependence of the state on the control (which follows from maximal parabolic regularity in weak form) implies that, for sufficiently small perturbations of the switching times, the states remain in a bounded set of the Hölder space. This bound is independent of the particular perturbation within a neighborhood and allows localization of the Lipschitz constant of the nonlinearity. We will add an explicit remark after the proof that records this a-priori bound and the resulting uniform control on the regularity constants, thereby making the passage from directional to continuous Fréchet differentiability fully transparent. revision: yes

  2. Referee: [§5, Proposition 5.2] §5, Proposition 5.2: the extended formulation is asserted to describe the convex hull of the set of feasible switching functions. The proof sketch relies on convex combinations of Dirac measures at the switching instants, yet it is not shown that every point in the convex hull can be realized by a control whose switching times lie in the original admissible set; an explicit counter-example or a tightness result would be needed to confirm that the formulation is exact rather than a relaxation.

    Authors: We agree that the argument in Proposition 5.2 is presented as a sketch and would benefit from additional detail. The extended formulation is constructed so that its extreme points are precisely the original switching functions with switching times in the admissible set; any convex combination can be realized as a weak*-limit of feasible controls. To confirm exactness we will insert a short tightness argument showing that the convex hull is attained within the original admissible set. If the referee prefers, we can also discuss a simple one-dimensional example illustrating that the relaxation is tight. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation uses standard functional-analytic tools without self-referential reductions

full rationale

The paper establishes continuous Fréchet differentiability of the switching-point-to-control map (valued in the dual of Hölder functions) and transfers it to differentiability of the reduced objective by treating the state equation in weak form via maximal parabolic regularity. These steps invoke established concepts from PDE theory that are independent of the paper's own constructions or fitted quantities. No equation or claim reduces by definition to a prior result of the same authors, no parameter is fitted to data and then relabeled as a prediction, and the numerical experiments serve only as confirmation. The derivation chain therefore remains self-contained against external mathematical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claims rest on the background theory of maximal parabolic regularity and on standard properties of Fréchet differentiability in Banach spaces; no new free parameters or invented entities are introduced in the abstract.

axioms (1)
  • domain assumption The state equation admits a weak formulation based on maximal parabolic regularity.
    Invoked to obtain the reduced objective's differentiability with respect to switching points.

pith-pipeline@v0.9.0 · 5702 in / 1325 out tokens · 39203 ms · 2026-05-22T10:54:27.881066+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages

  1. [1]

    The Switch Point Algorithm

    M. Aghaee and W. W. Hager. “The Switch Point Algorithm”. In:SIAM Journal on Control and Optimization59.4 (Jan. 2021), pp. 2570–2593.doi:10.1137/21m1393315.url:http: //dx.doi.org/10.1137/21M1393315

    work page doi:10.1137/21m1393315.url:http: 2021

  2. [2]

    Alt.Numerische Verfahren der konvexen, nichtglatten Optimierung

    W. Alt.Numerische Verfahren der konvexen, nichtglatten Optimierung. Wiesbaden: Vieweg+Teubner Verlag, 2004.doi:10.1007/978-3-322-80083-1

    work page doi:10.1007/978-3-322-80083-1 2004

  3. [3]

    Linear parabolic problems involving measures

    H. Amann. “Linear parabolic problems involving measures”. In:RACSAM. Rev. R. Acad. Cienc. Exactas F´ ıs. Nat. Ser. A Mat.95.1 (2001)

    work page 2001

  4. [4]

    Nonautonomous parabolic equations involving measures

    H. Amann. “Nonautonomous parabolic equations involving measures”. In:Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI)306 (2003), pp. 16–52, 229.doi: 10.1007/s10958-005-0376-8

    work page doi:10.1007/s10958-005-0376-8 2003

  5. [5]

    Amann.Linear and quasilinear parabolic problems

    H. Amann.Linear and quasilinear parabolic problems. Vol. I. Vol. 89. Monographs in Math- ematics. Abstract linear theory. Birkh¨ auser Boston, Inc., Boston, MA, 1995, pp. xxxvi+335. doi:10.1007/978-3-0348-9221-6

    work page doi:10.1007/978-3-0348-9221-6 1995

  6. [6]

    Beck.First-Order Methods in Optimization

    A. Beck.First-order methods in optimization. Vol. 25. MOS-SIAM Series on Optimization. Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA; Mathematical Optimization Society, Philadelphia, PA, 2017, pp. xii+475.doi:10.1137/1.9781611974997. ch1

    work page doi:10.1137/1.9781611974997 2017

  7. [7]

    A fast iterative shrinkage-thresholding algorithm for linear inverse problems

    A. Beck and M. Teboulle. “A fast iterative shrinkage-thresholding algorithm for linear inverse problems”. In:SIAM J. Imaging Sci.2.1 (2009), pp. 183–202.doi:10 . 1137 / 080716542. REFERENCES 37

    work page 2009

  8. [8]

    Parabolic optimal control problems with com- binatorial switching constraints, Part I: Convex relaxations

    C. Buchheim, A. Gr¨ utering, and C. Meyer. “Parabolic optimal control problems with com- binatorial switching constraints, Part I: Convex relaxations”. In:SIAM J. Optim.34.2 (2024), pp. 1187–1205.doi:10.1137/22M1490260

    work page doi:10.1137/22m1490260 2024

  9. [9]

    Parabolic optimal control problems with com- binatorial switching constraints, Part II: Outer approximation algorithm

    C. Buchheim, A. Gr¨ utering, and C. Meyer. “Parabolic optimal control problems with com- binatorial switching constraints, Part II: Outer approximation algorithm”. In:SIAM J. Optim.34.2 (2024), pp. 1295–1315.doi:10.1137/22M1490284

    work page doi:10.1137/22m1490284 2024

  10. [10]

    Parabolic optimal control problems with combi- natorial switching constraints, part III: branch-and-bound algorithm

    C. Buchheim, A. Gr¨ utering, and C. Meyer. “Parabolic optimal control problems with combi- natorial switching constraints, part III: branch-and-bound algorithm”. In:Comput. Optim. Appl.90.3 (2025), pp. 649–689.doi:10.1007/s10589-025-00654-3

    work page doi:10.1007/s10589-025-00654-3 2025

  11. [11]

    Sparse Switching Times Optimization and a Sweeping Hes- sian Proximal Method

    A. De Marchi and M. Gerdts. “Sparse Switching Times Optimization and a Sweeping Hes- sian Proximal Method”. In:Operations Research Proceedings 2019. Ed. by J. S. Neufeld, U. Buscher, R. Lasch, D. M¨ ost, and J. Sch¨ onberger. Cham: Springer International Publishing, 2020, pp. 89–95

    work page 2019

  12. [12]

    L p regularity for abstract differential equations

    G. Dore. “L p regularity for abstract differential equations”. In:Functional analysis and related topics, 1991 (Kyoto). Vol. 1540. Lecture Notes in Math. Springer, Berlin, 1993, pp. 25–38.doi:10.1007/BFb0085472

    work page doi:10.1007/bfb0085472 1991

  13. [13]

    Transition-time optimization for switched-mode dynamical systems

    M. Egerstedt, Y. Wardi, and H. Axelsson. “Transition-time optimization for switched-mode dynamical systems”. In:IEEE Transactions on Automatic Control51.1 (2006), pp. 110– 115.doi:10.1109/TAC.2005.861711

    work page doi:10.1109/tac.2005.861711 2006

  14. [14]

    Optimal control of switching times in switched dynamical systems

    M. Egerstedt, Y. Wardi, and F. Delmotte. “Optimal control of switching times in switched dynamical systems”. In:42nd IEEE International Conference on Decision and Control (IEEE Cat. No.03CH37475). Vol. 3. 2003, 2138–2143 Vol.3.doi:10 . 1109 / CDC . 2003 . 1272934

    work page 2003

  15. [15]

    Discretized switching time optimiza- tion problems

    K. Flaßkamp, T. D. Murphey, and S. Ober-Bl¨ obaum. “Discretized switching time optimiza- tion problems”. In:2013 European Control Conference (ECC)(2013), pp. 3179–3184.url: https://api.semanticscholar.org/CorpusID:11685156

    work page 2013

  16. [16]

    On Nemytskij operators inL p-spaces of abstract functions

    H. Goldberg, W. Kampowsky, and F. Tr¨ oltzsch. “On Nemytskij operators inL p-spaces of abstract functions”. In:Math. Nachr.155 (1992), pp. 127–140.doi:10 . 1002 / mana . 19921550110

    work page 1992

  17. [17]

    Relaxation methods for mixed-integer optimal control of partial differential equations

    F. M. Hante and S. Sager. “Relaxation methods for mixed-integer optimal control of partial differential equations”. In:Comput. Optim. Appl.55.1 (2013), pp. 197–225.doi:10.1007/ s10589-012-9518-3

    work page 2013

  18. [18]

    Existence of solutions of a thermoviscoplastic model and associated optimal control problems

    R. Herzog, C. Meyer, and A. St¨ otzner. “Existence of solutions of a thermoviscoplastic model and associated optimal control problems”. In:Nonlinear Anal. Real World Appl.35 (2017), pp. 75–101.doi:10.1016/j.nonrwa.2016.10.008

    work page doi:10.1016/j.nonrwa.2016.10.008 2017

  19. [19]

    Optimal control for the thermistor problem

    D. H¨ omberg, C. Meyer, J. Rehberg, and W. Ring. “Optimal control for the thermistor problem”. In:SIAM J. Control Optim.48.5 (2009/10), pp. 3449–3481.doi:10 . 1137 / 080736259

    work page 2009

  20. [20]

    Sequential linear integer programming for integer optimal control with total variation regularization

    S. Leyffer and P. Manns. “Sequential linear integer programming for integer optimal control with total variation regularization”. In:ESAIM Control Optim. Calc. Var.28 (2022), Paper No. 66, 34.doi:10.1051/cocv/2022059

    work page doi:10.1051/cocv/2022059 2022

  21. [21]

    Multidimensional sum-up rounding for elliptic control systems

    P. Manns and C. Kirches. “Multidimensional sum-up rounding for elliptic control systems”. In:SIAM J. Numer. Anal.58.6 (2020), pp. 3427–3447.doi:10.1137/19M1260682

    work page doi:10.1137/19m1260682 2020

  22. [22]

    Integer optimal control problems with total variation reg- ularization: optimality conditions and fast solution of subproblems

    J. Marko and G. Wachsmuth. “Integer optimal control problems with total variation reg- ularization: optimality conditions and fast solution of subproblems”. In:ESAIM Control Optim. Calc. Var.29 (2023), Paper No. 81, 30.doi:10.1051/cocv/2023065

    work page doi:10.1051/cocv/2023065 2023

  23. [23]

    Optimization methods for the ver- ification of second order sufficient conditions for bang–bang controls

    H. Maurer, C. B¨ uskens, J.-H. R. Kim, and C. Y. Kaya. “Optimization methods for the ver- ification of second order sufficient conditions for bang–bang controls”. In:Optimal Control Applications and Methods26.3 (2005), pp. 129–156.doi:https://doi.org/10.1002/oca. 756

    work page doi:10.1002/oca 2005

  24. [24]

    Analysis and Optimal Control of Quasilinear Parabolic Evolution Equa- tions in Divergence Form on Rough Domains

    H. Meinlschmidt. “Analysis and Optimal Control of Quasilinear Parabolic Evolution Equa- tions in Divergence Form on Rough Domains”. PhD thesis. TU Darmstadt, 2017

    work page 2017

  25. [25]

    A method for solving the convex programming problem with convergence rateO(1/k 2)

    Y. E. Nesterov. “A method for solving the convex programming problem with convergence rateO(1/k 2)”. In:Dokl. Akad. Nauk SSSR269.3 (1983), pp. 543–547

    work page 1983

  26. [26]

    Semigroups of linear operators and applications to partial differential equa- tions, volume 44 of Applied Mathematical Sciences

    A. Pazy.Semigroups of linear operators and applications to partial differential equations. Vol. 44. Applied Mathematical Sciences. Springer-Verlag, New York, 1983, pp. viii+279. doi:10.1007/978-1-4612-5561-1. 38 REFERENCES

    work page doi:10.1007/978-1-4612-5561-1 1983

  27. [27]

    Hamiltonian Pontryagin’s principles for control problems governed by semilinear parabolic equations

    J. P. Raymond and H. Zidani. “Hamiltonian Pontryagin’s principles for control problems governed by semilinear parabolic equations”. In:Appl. Math. Optim.39.2 (1999), pp. 143– 177.doi:10.1007/s002459900102

    work page doi:10.1007/s002459900102 1999

  28. [28]

    Optimal switching for hybrid semilinear evolutions

    F. R¨ uffler and F. M. Hante. “Optimal switching for hybrid semilinear evolutions”. In: Nonlinear Anal. Hybrid Syst.22 (2016), pp. 215–227.doi:10.1016/j.nahs.2016.05.001

    work page doi:10.1016/j.nahs.2016.05.001 2016

  29. [29]

    Combinatorial integral approximation

    S. Sager, M. Jung, and C. Kirches. “Combinatorial integral approximation”. In:Math. Methods Oper. Res.73.3 (2011), pp. 363–380.doi:10.1007/s00186-011-0355-4

    work page doi:10.1007/s00186-011-0355-4 2011

  30. [30]

    Salmon.Isotonic Regression

    J. Salmon.Isotonic Regression. Nov. 10, 2024.url:https : / / josephsalmon . eu / blog / isotonic/

    work page 2024

  31. [31]

    Second-Order Switching Time Opti- mization for Switched Dynamical Systems

    B. Stellato, S. Ober-Bl¨ obaum, and P. J. Goulart. “Second-Order Switching Time Opti- mization for Switched Dynamical Systems”. In:IEEE Transactions on Automatic Control 62.10 (2017), pp. 5407–5414.doi:10.1109/TAC.2017.2697681

    work page doi:10.1109/tac.2017.2697681 2017

  32. [32]

    Triebel.Interpolation theory, function spaces, differential operators

    H. Triebel.Interpolation theory, function spaces, differential operators. Vol. 18. North- Holland Mathematical Library. North-Holland Publishing Co., Amsterdam-New York, 1978, p. 528

    work page 1978

  33. [33]

    Wloka.Partial differential equations

    J. Wloka.Partial differential equations. Translated from the German by C. B. Thomas and M. J. Thomas. Cambridge University Press, Cambridge, 1987, pp. xii+518.doi:10.1017/ CBO9781139171755

    work page 1987

  34. [34]

    Optimal control of hybrid switched systems: a brief survey

    F. Zhu and P. J. Antsaklis. “Optimal control of hybrid switched systems: a brief survey”. In:Discrete Event Dyn. Syst.25.3 (2015), pp. 345–364.doi:10.1007/s10626-014-0187-5. Technische Universit ¨at Dortmund, F akult ¨at f ¨ur Mathematik, Lehrstuhl LSX, Vo- gelpothsweg 87, 44227 Dortmund, Germany Email address:alimhan.musalatov@tu-dortmund.de Technische U...

    work page doi:10.1007/s10626-014-0187-5 2015