Pith. sign in

REVIEW 3 major objections 6 minor 39 references

Optimal control of symmetry-breaking dynamics near criticality

T0 review · 3 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read This paper claims that near a pitchfork the optimality conditions reduce to one of three universal control normal forms selected by the cost scaling, with closed-form feedback laws in the intermediate and weak regimes.

desk verdict A genuinely new asymptotic reduction of PMP near a pitchfork, with a real but openly acknowledged sufficiency gap in the weak-control regime; worth refereeing. read the letter →

arxiv 2607.16188 v1 pith:UBOHKJJB submitted 2026-07-17 math.OC

Pearson W. Miller This is my paper · ORCID
classification math.OC MSC 49K1534C2393C1534E1037G10
keywords optimalcontrolpitchforkbifurcationcentermanifoldreductionPontryaginmaximumprincipleasymptoticexpansionnormalformssymmetrybreakingcell-fateselection
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

The paper tries to establish that optimal control inherits the universality of the pitchfork normal form: for any n-dimensional controlled system near a pitchfork with quadratic tracking cost, the leading-order optimal cue is described by one of three universal reduced models, and the choice among them is fixed by a single piece of data—the scaling of the tracking penalty relative to the distance from criticality. When the control is strong, the leading problem is standard linear-quadratic regulation; at intermediate strength it is a scalar Riccati equation on the center manifold with a closed-form tanh feedback; in the weak limit it is fully nonlinear optimal tracking on the center manifold, with feedback given by Hamiltonian energy level sets. The weak-regime reduced system is then solved in the sense that its bifurcation structure is classified and its long-time trajectories are constructed analytically as interacting heteroclinic layers with closed-form positions and cost. A sympathetic reader would care because it converts a hard infinite-dimensional optimal-control problem into a catalog of normal forms, which is exactly the kind of reduction that makes predictions about biological symmetry-breaking decisions testable.

What carries the argument

The load-bearing machinery is an asymptotic expansion of the first-order optimality conditions (the Pontryagin maximum principle) in half-integer powers of the bifurcation parameter, applied simultaneously to the state, control, and costate with tracking penalties scaled to match the control scale. The central object that emerges is the reduced two-dimensional Hamiltonian system for the critical-mode amplitude A and costate amplitude Q, obtained by projecting the n-dimensional optimality system onto the critical eigenmode via Fredholm solvability; system-specific data enter only through the center-manifold coefficients mu, nu and the critical-mode control coupling beta_c. In the weak regime

What would settle it

For the bistable switch of the paper in the weak regime, evaluate the second variation (Eq. A.9) along the Hamiltonian-level-set extremal for parameters where the extra term 6 nu A_bar Q_bar a_tilde_c^2 is negative and large; if the quadratic form admits a negative direction, the 'optimal' law is not a minimizer. Alternatively, run the full numerical solver at T = 1000 over an epsilon range down to 1e-3 and test whether the L-infinity gap between the asymptotic and numerical controls fails to shrink as O(epsilon), which would indicate the leading-order reduction misses a relevant balance.

Watch

Extended reading notes

Core claim

The central claim: near a pitchfork, leading-order optimality for an n-dimensional control system collapses, under joint scaling of state, control, costate, and penalties, into a universal family. In the weak regime this is exact optimal tracking on the center manifold—A' = mu A + nu A^3 - beta_c Q, Q' = zeta_tilde(A* - A) - 3 nu A^2 Q - mu Q—with conserved Hamiltonian and a two-branch level-set feedback. At intermediate strength the same reduction gives a driftless scalar Riccati law with tanh profile; at strong control it recovers LQR. The weak-regime bifurcation diagram is classified, and long-horizon extremals are heteroclinic-layer sequences with closed-form positions and cost.

Load-bearing premise

The reduction rests on the assumption that a true minimizer exists and keeps the trajectory inside the local center-manifold neighborhood, and—in the weak regime—on the unproved positive definiteness of the second variation; if either fails, the derived feedback laws are candidates, not optimal controls.

Editorial extensions

If this is right

  • For any system satisfying the assumptions, the optimal cue near criticality is computable from the normal-form coefficients (mu, nu, beta_c) and the scaling class of the cost, bypassing repeated solves of the full Hamiltonian boundary-value problem.
  • In the intermediate regime the optimal feedback is a tanh-Riccati law that is nearly constant except in a terminal boundary layer, with characteristic rate sqrt(zeta_tilde beta_c) on the slow timescale.
  • In the weak regime, optimal trajectories to a constant target are integrable: the feedback is a two-branch level-set formula, and in the long-horizon limit the trajectory is a sequence of heteroclinic layers whose positions are determined by a backward recursion and whose cost has closed form.
  • The weak- and intermediate-regime flows are topologically identical above the high-kappa saddle-node boundary, re-deriving the regime distinction from the reduced phase portrait rather than from a priori scaling.
  • When the tracking penalty vanishes, the reduced optimality system coincides with the Freidlin–Wentzell Hamiltonian for noise-induced transitions on the center manifold, with beta_c playing the role of noise intensity; tracking tilts that Hamiltonian and thereby selects transition timing.

Reading between the lines

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

  • A natural extension the author leaves implicit: the same distinguished-scaling analysis should carry over to other elementary bifurcations (saddle-node, transcritical) after adapting the symmetry conditions, so the 'three universality classes' may be a general architecture of near-critical optimal control rather than a pitchfork-specific result.
  • If the reduction is quantitatively accurate, it yields a practical test for optimality in cell-fate experiments: fit (mu, nu, beta_c) from perturbation data, then ask whether observed external-cue dynamics lie on the predicted level-set feedback; residuals quantify suboptimality in a parameter-free way.
  • The unresolved second-variation issue in the weak regime invites a numerical check the paper does not perform: computing conjugate points along the level-set extremal for the bistable switch would turn the stationary laws into certified minimizers or expose a saddle, settling the sufficiency question.
  • Because the layer positions are fixed by a stationarity of the restricted action, the matched-asymptotic construction is equivalent to a variational principle for the trajectory itself; this suggests that the O(kappa) tilt selection could be recast as a discrete optimization over layer sequences, potentially making the multi-turnpike selection rigorous.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper develops an asymptotic reduction of Pontryagin optimality conditions for finite-horizon quadratic-cost control of systems near a pitchfork bifurcation. Three distinguished scalings of the tracking penalties with the bifurcation parameter are identified, yielding: a strong-control regime in which the leading problem is full-system LQR; an intermediate regime in which a scalar Riccati equation provides a closed-form feedback on the center manifold; and a weak-control regime in which the reduced problem is a nonlinear Hamiltonian system on the center manifold, with a feedback family parameterized by a conserved energy. The weak-regime reduced system is analyzed further: its phase-portrait bifurcations are classified in a canonical (κ, X_c) plane, and matched-asymptotic boundary-layer solutions are constructed for long horizons, including force-balance equations for front positions and a closed-form cost. The three asymptotic laws are compared with numerical solutions of a two-species bistable switch model. The paper is transparent about its limitations, explicitly stating in the Conclusion that the weak-regime laws are stationary solutions of the first-order conditions rather than certified minimizers, and that the numerical validation uses a single two-dimensional model.

Significance. If the reduction is valid, the three-regime taxonomy is a substantial contribution: it gives a practical way to obtain near-closed-form control laws for high-dimensional systems near a pitchfork, using only the normal-form coefficients (μ, ν, β_c) and the scaling class of the cost. The intermediate- and weak-regime laws are genuinely new (the strong regime reduces to known LQR). The paper also provides a canonical transformation of the reduced Hamiltonian, a bifurcation diagram, and explicit boundary-layer formulas for long horizons, including the important observation that the layer-position solvability condition is the stationarity condition of the reduced action. The connection to Freidlin–Wentzell theory in the zero-tracking limit is insightful. The manuscript includes detailed numerical methodology, validation across three regimes, and an unusually candid discussion of the second-variation obstruction in the weak regime. The main weakness is that the paper's central 'optimal' claims in the weak regime are not supported by a sufficiency argument, a point the paper itself concedes.

major comments (3)
  1. [Sec. 3.3, Eq. (70); Appendix A.4, Eq. (132); Conclusion] The weak-regime feedback family (70) and the reduced problem (66)–(68) are presented as 'optimal', but the second variation of the reduced cost along an extremal contains the uncontrolled term 6νĀQ̄ ã_c^2 (Eq. 132). The paper's own Conclusion states that the weak-regime laws are 'stationary solutions of the first-order conditions rather than certified minimizers'. Since the Abstract and Section 1 claim an 'optimal control law' and 'optimal tracking', the central claim is not established for the weak regime. This affects not only Eqs. (66)–(70) but also the bifurcation and boundary-layer analyses of Section 4, which describe extremals. The authors should either provide a sufficiency check (e.g., conjugate-point analysis along the extremal) or explicitly and consistently reframe the weak-regime results as necessary-condition/candidate extremal laws throughout, including the Abstract, Secti
  2. [Sec. 3.3 and Sec. 4.2] The asymptotic expansion in ε^{1/2} is formal, and no proof is given of its uniformity over the horizon when T~ε^{-1} (equivalently τ∈[0,T_τ] with T_τ=εT large). The derivation of the weak-regime amplitude equations (66)–(67) assumes τ=O(1), while the boundary-layer construction of Sec. 4.2 uses S≈T_τ large and produces formulas such as (106)–(107) and the cost (119) that rely on the expansion being valid over the full long horizon. The numerical validation uses fixed physical T=1000 for all ε, so T_τ ranges from 1 (ε=10^{-3}) to 100 (ε=10^{-1}); for the larger ε values the reduced-time horizon is not O(1), and the expansion's remainder is uncontrolled. The claim that the reduced normal form captures the optimal solution over the full horizon therefore needs an explicit uniformity estimate or a restriction of the claims to τ=O(1).
  3. [Sec. 3.3, Eq. (70); Figs. 6–7] The weak-regime validation is weakened because the conserved quantity E in the feedback family (70) is fitted to the numerical solutions being compared. The text states: 'Fitting E to the numerical solutions of the optimal control problem, as shown in Figs. 6 and 7, we find that this family of feedback laws captures the main features...' This makes the comparison a test of the expressive power of the family, not a predictive test of the asymptotic law. A predictive validation would require determining E from the boundary conditions (e.g., by shooting) before comparison. The error-scaling claims for the weak regime should be qualified accordingly.
minor comments (6)
  1. [Sec. 1] Typo: 'asympototic reduction' should be 'asymptotic reduction'.
  2. [Remark 4.1] Typo: 'it's reflection' should be 'its reflection'.
  3. [Sec. 3.3, Eq. (70)] The notation is confusing: E is used both as the energy level of H_reduced and as a function E(A_0,T). Clarify the branch selection and the role of E(A_0,T).
  4. [Sec. 4.1, Eqs. (73)–(76)] The rescaling uses H both for the original and rescaled Hamiltonian; define the rescaled quantity with a different symbol (e.g., H_canon) to avoid ambiguity.
  5. [Table 2] Some entries are ambiguous, e.g., '-3/2 X_c + 1/2'; use parentheses or spacing so the reader can distinguish (-3/2)X_c + 1/2 from -(3/2 X_c + 1/2).
  6. [Appendix A, Eq. (120)] Stray character 'x‘' appears in the display of δ²J; remove it.

Circularity Check

1 steps flagged · score 4.0 of 10

The asymptotic derivation is self-contained; the only substantive circular loop is in the weak-regime validation, where the conserved level E is fitted to the numerical solutions before the comparison is made.

  1. fitted input called prediction [Sec. 3.3, Eq. (70) and the paragraph before Figs. 6-7]
    "Fitting E to the numerical solutions of the optimal control problem, as shown in Figs. 6 and 7, we find that this family of feedback laws captures the main features of the optimal control solution, with the particular branch and value of E determined by the initial conditions and terminal time."

    The weak-regime feedback family (70) is parameterized by the conserved level E, and the paper states that it does not have a closed-form expression for E. The validation in Figs. 6-7 then takes the numerical solution of the full problem, fits E to that same solution, and compares the feedback family back to it. Thus one scalar degree of freedom per trajectory is removed by construction. The agreement does not constitute an independent prediction of the law's parameters; it tests the shape and branch family of (70) after E has been chosen using the very data being validated. This weakens the validation claim without making the derivation of (66)-(67) itself circular, since those equations follow from PMP solvability rather than from a fit.

full rationale

The central derivation is not circular. The paper starts from the Pontryagin maximum principle (Eq. 11), imposes the pitchfork scalings, and obtains the reduced amplitude equations (66)-(67) and reduced Hamiltonian (69) by Fredholm solvability, not by assuming the target result. The strong- and intermediate-regime feedback laws are compared with the full numerical solution without fitted parameters, so those comparisons are independent (modulo the usual caveat that the solver is warm-started by the asymptotic law). The genuine circular loop is confined to the weak-regime validation: the feedback law (70) contains E, for which the paper explicitly has no closed-form expression, and Figs. 6-7 are produced after fitting E to the numerical solutions. That makes the weak-regime agreement partially constructed rather than predicted, but it does not reduce the main asymptotic derivation to its inputs. The paper also self-reports in Appendix A.4 that the weak-regime second variation is not positive definite and that the weak-regime laws are stationary points rather than certified minimizers; this is a correctness limitation, not a circularity. There are no load-bearing self-citations or imported uniqueness theorems. Overall score 4 reflects the partial validation loop in the weak regime, while the core derivation retains independent content.

Assumptions & free parameters 1 free parameters · 8 assumptions · 0 invented entities

The central derivation rests on standard PMP/center-manifold machinery plus several domain restrictions (single critical mode, direct controllability, local targets) and one explicit unproved existence assumption. The weak-regime 'closed-form' feedback contains a fitted conserved quantity E, the main free parameter. No invented entities are introduced.

free parameters (1)
  • E (reduced Hamiltonian energy) = fitted numerically; values not reported
    Eq. (70) parameterizes the weak-regime feedback by E(A0,T); the paper does not provide a closed-form or numerical method for E beyond shooting, and states it fits E to the numerical solutions for the validation figures.
assumptions (8)
  • domain assumption Pitchfork bifurcation assumptions (Assumptions 1-3): single zero eigenvalue, stable others, nonzero crossing speed, vanishing quadratic/parametric terms, mu, nu nonzero with mu*nu < 0.
    Defines the class of systems; standard for center-manifold reduction but restricts to symmetric pitchfork degeneracies.
  • domain assumption Critical-mode controllability (Assumption 4): w_c^i r_i_alpha != 0.
    Excludes systems where control reaches the critical mode only through stable modes; the paper notes these fall under control-bifurcation theory and are not addressed.
  • domain assumption Locality of states and targets (Assumption 5): |c(0)-c_ss|, |c*-c_ss| = O(epsilon^{1/2}).
    Keeps trajectories in the center-manifold neighborhood; if targets are far away, global nonlinear effects invalidate the reduction.
  • ad hoc to paper Existence of a minimizer (Assumption 6).
    The paper explicitly does not address existence, but the label 'optimal' depends on it.
  • ad hoc to paper Uniform validity of the formal epsilon^{1/2} expansion over the full horizon, including T ~ epsilon^{-1} in the weak regime.
    No error bounds are given; the matched-asymptotic construction assumes the expansion and boundary-layer decomposition remain ordered over the long horizon.
  • standard math Center manifold theorem and Fredholm solvability for projection of the expanded optimality conditions.
    Used in Secs. 3.2-3.3 to project the state/costate equations onto the critical mode and solve for stable-mode amplitudes.
  • standard math Pontryagin maximum principle first-order necessary conditions.
    The reduction starts from Eqs. (11)-(15); sufficiency is not established, and App. A.4 shows the weak-regime second variation is not positive definite.
  • domain assumption Boundary-layer separation assumptions in Sec. 4.2: adjacent saddles well separated, exponential tails, terminal-layer contribution exponentially small in the backward sweep.
    Needed for the force-balance recursion; validated only for one parameter set (Fig. 9, S ~ 59.22).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Optimal control of symmetry-breaking dynamics near criticality." pith.science (2026). https://pith.science/paper/UBOHKJJB

@misc{pith2026260716188,
  author       = {Pith},
  title        = {Pith review of: Optimal control of symmetry-breaking dynamics near criticality},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UBOHKJJB}},
  note         = {Machine review of arXiv:2607.16188}
}
read the original abstract

We study the problem of optimal control for dynamical systems near a pitchfork bifurcation, motivated by the role of external cues in guiding symmetry-breaking transitions in cell-fate selection and other natural processes. Using an asymptotic expansion of the optimality conditions obtained from the Pontryagin maximum principle, the leading-order optimal control law for a general n-dimensional system is examined across three dynamical regimes distinguished by scaling of control strength with respect to the distance from criticality. While in the strong control limit the results reduce to known approximations from linear-quadratic control, we derive generalized amplitude equations for the co-evolution of state and costate variables describing the optimized trajectory in the weak and intermediate control regimes. These control normal forms are validated against numerical solutions of the full optimal control problem for a canonical model of a bistable biochemical switch. The bifurcation structure of the optimal control problem is analyzed in the weak control regime. Finally, we demonstrate the construction of asymptotic solutions in the long time limit in this regime using boundary-layer methods.

Figures

Figures reproduced from arXiv: 2607.16188 by the authors.

Figure 1
Figure 1. Standard behavior of the pitchfork bifurcation. (left) A supercritical pitchfork, for which a [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Strong control regime validation (Unstable [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Strong control regime validation (Stable [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Intermediate control regime validation (Unstable [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Intermediate control regime validation (Stable [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Weak control regime validation (Unstable [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: Weak control regime validation (Stable→Stable transition). Layout as in [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]
Figure 8
Figure 8. Figure 8: Bifurcation diagram of the canonical Hamiltonian flow in the ( [PITH_FULL_IMAGE:figures/full_fig_p021_8.png]
Figure 9
Figure 9. Figure 9: Boundary-layer validation: comparison of the projected amplitude trajectory from the [PITH_FULL_IMAGE:figures/full_fig_p027_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 1 linked inside Pith

  1. [1]

    Local feedback stabilization and bifurcation control, i

    Eyad H Abed and Jyun-Horng Fu. Local feedback stabilization and bifurcation control, i. hopf bifurcation.Systems & Control Letters, 7(1):11–17, 1986

    1986

  2. [2]

    Local feedback stabilization and bifurcation control, ii

    Eyad H Abed and Jyun-Horng Fu. Local feedback stabilization and bifurcation control, ii. stationary bifurcation.Systems & Control Letters, 8(5):467–473, 1987

    1987

  3. [3]

    Birkh¨ auser, 2012

    Hisham Abou-Kandil, Gerhard Freiling, Vlad Ionescu, and Gerhard Jank.Matrix Riccati equations in control and systems theory. Birkh¨ auser, 2012

    2012

  4. [4]

    On the optimal stabilization of nonlinear systems.Journal of Applied Mathe- matics and Mechanics, 25(5):1254–1266, 1961

    EG Al’Brekht. On the optimal stabilization of nonlinear systems.Journal of Applied Mathe- matics and Mechanics, 25(5):1254–1266, 1961

    1961

  5. [5]

    Optimal control in soft and active matter.Annual Review of Condensed Matter Physics, 17, 2026

    Jos´ e Alvarado, Erin G Teich, David A Sivak, and John Bechhoefer. Optimal control in soft and active matter.Annual Review of Condensed Matter Physics, 17, 2026

    2026

  6. [6]

    Information bottleneck in molecular sensing.PRX Life, 1(2):023005, 2023

    Marianne Bauer and William Bialek. Information bottleneck in molecular sensing.PRX Life, 1(2):023005, 2023

    2023

  7. [7]

    Wiley, 1988

    Alain Bensoussan.Perturbation Methods in Optimal Control. Wiley, 1988

    1988

  8. [8]

    Jack Carr and Robert L. Pego. Metastable patterns in solutions ofu t =ε 2uxx −f(u).Com- munications on Pure and Applied Mathematics, 42:523–576, 1989

    1989

Show all 39 references
  1. [9]

    Additive noise destroys a pitchfork bifurcation.Journal of Dynamics and Differential Equations, 10(2):259–274, 1998

    Hans Crauel and Franco Flandoli. Additive noise destroys a pitchfork bifurcation.Journal of Dynamics and Differential Equations, 10(2):259–274, 1998. 34 P. W. Miller Optimal control of symmetry-breaking dynamics near criticality

    1998

  2. [10]

    Bistability, bifurcations, and waddington’s epigenetic landscape.Current Biology, 22(11):R458–R466, 2012

    James E Ferrell. Bistability, bifurcations, and waddington’s epigenetic landscape.Current Biology, 22(11):R458–R466, 2012

    2012

  3. [11]

    On small random perturbations of dynamical systems.Russian Mathematical Surveys, 25(1):1–55, 1970

    Mark I Freidlin. On small random perturbations of dynamical systems.Russian Mathematical Surveys, 25(1):1–55, 1970

    1970

  4. [12]

    Giorgio Fusco and Jack K. Hale. Slow-motion manifolds, dormant instability, and singular perturbations.Journal of Dynamics and Differential Equations, 1:75–94, 1989

    1989

  5. [13]

    Non-equilibrium transitions in multiscale systems with a bifurcating slow manifold.Journal of Statistical Mechanics: Theory and Experiment, 2017(9):093208, 2017

    Tobias Grafke and Eric Vanden-Eijnden. Non-equilibrium transitions in multiscale systems with a bifurcating slow manifold.Journal of Statistical Mechanics: Theory and Experiment, 2017(9):093208, 2017

    2017

  6. [14]

    Probing the limits to positional information.Cell, 130(1):153–164, 2007

    Thomas Gregor, David W Tank, Eric F Wieschaus, and William Bialek. Probing the limits to positional information.Cell, 130(1):153–164, 2007

    2007

  7. [15]

    Springer-Verlag, New York, 1983

    John Guckenheimer and Philip Holmes.Nonlinear Oscillations, Dynamical Systems, and Bi- furcations of Vector Fields, volume 42 ofApplied Mathematical Sciences. Springer-Verlag, New York, 1983

    1983

  8. [16]

    Boumediene Hamzi, Wei Kang, and Arthur J. Krener. The controlled center dynamics.Mul- tiscale Modeling & Simulation, 3(4):838–852, 2005

    2005

  9. [17]

    Bifurcation and normal form of nonlinear control systems, part I.SIAM Journal on Control and Optimization, 36(1):193–212, 1998

    Wei Kang. Bifurcation and normal form of nonlinear control systems, part I.SIAM Journal on Control and Optimization, 36(1):193–212, 1998

    1998

  10. [18]

    Courier Corporation, 2004

    Donald E Kirk.Optimal control theory: an introduction. Courier Corporation, 2004

    2004

  11. [19]

    Khalil, and John O’Reilly.Singular Perturbation Methods in Control: Analysis and Design

    Petar Kokotovi´ c, Hassan K. Khalil, and John O’Reilly.Singular Perturbation Methods in Control: Analysis and Design. Classics in Applied Mathematics. SIAM, 1999

    1999

  12. [20]

    Al’brekht’s method for time varying, finite horizon optimal control problems

    Arthur J Krener. Al’brekht’s method for time varying, finite horizon optimal control problems. IF AC-PapersOnLine, 59(19):632–637, 2025

    2025

  13. [21]

    Krener, Wei Kang, and Dong Eui Chang

    Arthur J. Krener, Wei Kang, and Dong Eui Chang. Control bifurcations.IEEE Transactions on Automatic Control, 49(8):1231–1246, 2004

    2004

  14. [22]

    Nonlinear aeroelastic analysis of airfoils: bifurcation and chaos.Progress in Aerospace Sciences, 35(3):205–334, 1999

    BHK Lee, SJ Price, and YS Wong. Nonlinear aeroelastic analysis of airfoils: bifurcation and chaos.Progress in Aerospace Sciences, 35(3):205–334, 1999

    1999

  15. [23]

    Dahlard L. Lukes. Optimal regulation of nonlinear dynamical systems.SIAM Journal on Control, 7(1):75–100, 1969

    1969

  16. [24]

    Are biological systems poised at criticality?Journal of Statistical Physics, 144(2):268–302, 2011

    Thierry Mora and William Bialek. Are biological systems poised at criticality?Journal of Statistical Physics, 144(2):268–302, 2011

    2011

  17. [25]

    Carmeliza Navasca and Arthur J. Krener. Patchy solutions of Hamilton–Jacobi–Bellman par- tial differential equations. InModeling, Estimation and Control, volume 364 ofLecture Notes in Control and Information Sciences, pages 251–270. Springer, 2007

    2007

  18. [26]

    Optimal control of gene regulatory networks for morphogen-driven tissue patterning.Cell Systems, 14(11):940–952, 2023

    Alberto Pezzotta and James Briscoe. Optimal control of gene regulatory networks for morphogen-driven tissue patterning.Cell Systems, 14(11):940–952, 2023

    2023

  19. [27]

    Routledge, 2018

    Lev Semenovich Pontryagin.Mathematical theory of optimal processes. Routledge, 2018. 35 P. W. Miller Optimal control of symmetry-breaking dynamics near criticality

    2018

  20. [28]

    Long time versus steady state optimal control.SIAM Journal on Control and Optimization, 51(6):4242–4273, 2013

    Alessio Porretta and Enrique Zuazua. Long time versus steady state optimal control.SIAM Journal on Control and Optimization, 51(6):4242–4273, 2013

    2013

  21. [29]

    Remarks on long time versus steady state optimal control

    Alessio Porretta and Enrique Zuazua. Remarks on long time versus steady state optimal control. InMathematical Paradigms of Climate Science, volume 15 ofSpringer INdAM Series, pages 67–89. Springer, 2016

    2016

  22. [30]

    Ge- ometry of gene regulatory dynamics.Proceedings of the National Academy of Sciences, 118(38):e2109729118, 2021

    David A Rand, Archishman Raju, Meritxell S´ aez, Francis Corson, and Eric D Siggia. Ge- ometry of gene regulatory dynamics.Proceedings of the National Academy of Sciences, 118(38):e2109729118, 2021

    2021

  23. [31]

    Front interaction and nonhomoge- neous equilibria for tristable reaction-diffusion equations.SIAM Journal on Applied Mathe- matics, 53(6):1669–1685, 1993

    Jacob Rubinstein, Peter Sternberg, and Joseph B Keller. Front interaction and nonhomoge- neous equilibria for tristable reaction-diffusion equations.SIAM Journal on Applied Mathe- matics, 53(6):1669–1685, 1993

    1993

  24. [32]

    Statistically derived geometrical landscapes capture principles of decision- making dynamics during cell fate transitions.Cell Systems, 13(1):12–28, 2022

    Meritxell S´ aez, Robert Blassberg, Elena Camacho-Aguilar, Eric D Siggia, David A Rand, and James Briscoe. Statistically derived geometrical landscapes capture principles of decision- making dynamics during cell fate transitions.Cell Systems, 13(1):12–28, 2022

    2022

  25. [33]

    The turnpike property in nonlinear optimal control—a geometric approach.Automatica, 134:109939, 2021

    Noboru Sakamoto, Dario Pighin, and Enrique Zuazua. The turnpike property in nonlinear optimal control—a geometric approach.Automatica, 134:109939, 2021

    2021

  26. [34]

    van der Schaft

    Noboru Sakamoto and Arjan J. van der Schaft. Analytical approximation methods for the sta- bilizing solution of the Hamilton–Jacobi equation.IEEE Transactions on Automatic Control, 53(10):2335–2350, 2008

    2008

  27. [35]

    Optimal control of short-time attractors in active nematics.Physical Review Letters, 132(21):218302, 2024

    Carlo Sinigaglia, Francesco Braghin, and Mattia Serra. Optimal control of short-time attractors in active nematics.Physical Review Letters, 132(21):218302, 2024

    2024

  28. [36]

    The turnpike property in finite-dimensional nonlinear optimal control.Journal of Differential Equations, 258(1):81–114, 2015

    Emmanuel Tr´ elat and Enrique Zuazua. The turnpike property in finite-dimensional nonlinear optimal control.Journal of Differential Equations, 258(1):81–114, 2015

    2015

  29. [37]

    Turnpike in optimal control and beyond: a survey

    Emmanuel Tr´ elat and Enrique Zuazua. Turnpike in optimal control and beyond: a survey. arXiv preprint arXiv:2503.20342, 2025

    2025 arXiv

  30. [38]

    Waddington.The Strategy of the Genes

    Conrad H. Waddington.The Strategy of the Genes. Allen & Unwin, London, 1957

    1957

  31. [39]

    Quantifying the Waddington landscape and biological paths for development and differentiation.Proceedings of the National Academy of Sciences, 108(20):8257–8262, 2011

    Jin Wang, Kun Zhang, Li Xu, and Erkang Wang. Quantifying the Waddington landscape and biological paths for development and differentiation.Proceedings of the National Academy of Sciences, 108(20):8257–8262, 2011. Acknowledgments We thank Vishal Patil, Natalia Komarova, and Mar...

    2011

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.