pith. sign in

arxiv: 2604.14924 · v1 · submitted 2026-04-16 · 🧮 math.OC

Dynamic Lagrange Multipliers in a Non-concave Utility Framework

Yang Liu , Alexander Schied , Zhenyu Shen This is my paper

Pith reviewed 2026-05-10 10:52 UTC · model grok-4.3

classification 🧮 math.OC
keywords non-concave utilitydynamic Lagrange multipliersmartingale dualitydynamic programmingportfolio selectionvalue functionHJB equationshadow price
0
0 comments X

The pith

Dynamic Lagrange multipliers equate the martingale duality multiplier to the wealth derivative of the value function for non-concave utilities.

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

The paper shows that in continuous-time portfolio selection with non-concave utility functions, a newly defined dynamic Lagrange multiplier links the martingale duality method to dynamic programming. It proves that the multiplier function equals the conjugate dual point tied to the value function and that this point is precisely the partial derivative of the value function with respect to wealth. The multiplier process also displays homogeneity expressed through the optimal wealth and pricing kernel, which supplies an economic reading as a dynamic shadow price. If this holds, the two approaches become consistent even when the Hamilton-Jacobi-Bellman equation admits singular solutions, and standard optimal portfolio results can be recovered in complete markets.

Core claim

In a non-concave utility framework for continuous-time portfolio selection, the Lagrangian multiplier function in the martingale duality approach equals the conjugate dual point related to the value function in dynamic programming, and this point is exactly its partial derivative with respect to wealth. The dynamic multiplier process exhibits homogeneity via the optimal wealth and pricing kernel processes, offering intuitive economic interpretations as a dynamic shadow price of the envelope theorem. Classical optimal results are recovered and numerically validated by non-concave utility examples.

What carries the argument

Dynamic Lagrange multipliers, defined to bridge the martingale duality and dynamic programming frameworks by establishing the equality between the duality multiplier and the value function's partial derivative with respect to wealth.

If this is right

  • The equality between the multiplier and the value-function derivative continues to hold when the HJB equation has singular solutions.
  • The multiplier process is homogeneous in the optimal wealth and pricing kernel processes.
  • The multiplier admits an interpretation as a dynamic shadow price through the envelope theorem.
  • Standard optimal portfolio results from concave cases extend directly to the non-concave setting.

Where Pith is reading between the lines

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

  • The homogeneity property could simplify numerical solution of the optimization problem by reducing the state space.
  • The bridging technique may extend to other stochastic control problems where duality and dynamic programming are both applicable.
  • Numerical validation in higher-dimensional markets would test whether the equality remains computationally tractable.

Load-bearing premise

Non-concave utilities still permit the martingale duality approach to apply in complete markets without creating inconsistencies with dynamic programming, even when the Hamilton-Jacobi-Bellman equation has singular solutions.

What would settle it

In a concrete non-concave utility example, compute the Lagrange multiplier function from the martingale duality method and the partial derivative of the value function from dynamic programming, then check whether the two quantities coincide for all times and states.

Figures

Figures reproduced from arXiv: 2604.14924 by Alexander Schied, Yang Liu, Zhenyu Shen.

Figure 6.1
Figure 6.1. Figure 6.1: The value function u(t, x) given by (6.2), where we let the market contain only one risky asset. The coefficients are defined as follows: T = 10, r = 0.05, µ = 0.086, σ = 0.3 [PITH_FULL_IMAGE:figures/full_fig_p022_6_1.png] view at source ↗
Figure 6.2
Figure 6.2. Figure 6.2: The dynamic Lagrange multiplier λ(t, x), which is obtained by Legendre-Fenchel transform of u(t, x) in (6.2). The coefficients are defined as follows: T = 10, r = 0.05, µ = 0.086, σ = 0.3. the terminal time T, the Lagrange multiplier λ(t, x) is convex in x. This convexity fails to hold as t approaches T, where λ(t, x) approaches Λ(x) in (6.1) simultaneously. It is because the non-concavity of the utility… view at source ↗
read the original abstract

In continuous-time portfolio selection for non-concave utility functions, the martingale duality approach is widely adopted in complete markets, while the dynamic programming approach may sometimes lead to singular solutions of the Hamilton-Jacobi-Bellman (HJB) equation. We propose "dynamic Lagrange multipliers" in a non-concave utility framework, bridging two approaches and demonstrating that the Lagrangian multiplier function (in the martingale duality approach) equals the conjugate dual point related to the value function (in dynamic programming), which is exactly its partial derivative with respect to wealth. Moreover, the dynamic multiplier process exhibits homogeneity via the optimal wealth and pricing kernel processes, offering intuitive economic interpretations as a dynamic shadow price of the envelope theorem. Finally, classical optimal results are recovered and numerically validated by non-concave utility examples.

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 manuscript proposes 'dynamic Lagrange multipliers' for continuous-time portfolio selection with non-concave utility functions. It claims to bridge the martingale duality and dynamic programming approaches by showing that the Lagrangian multiplier function equals the conjugate dual point of the value function, which is precisely its partial derivative with respect to wealth. The work further establishes homogeneity of the dynamic multiplier process via optimal wealth and pricing kernel, interprets it as a dynamic shadow price, recovers classical results, and provides numerical validation on non-concave examples.

Significance. If the claimed equality holds rigorously in the non-concave setting, the result would usefully unify two standard methodologies for complete-market problems where dynamic programming yields only singular (viscosity) HJB solutions. The homogeneity property and economic interpretation as an envelope-theorem shadow price are conceptually attractive, and the numerical recovery of classical optima supplies concrete support. These elements address a genuine technical gap without introducing free parameters or ad-hoc constructions.

major comments (2)
  1. [§4, Theorem 4.1] §4, Theorem 4.1 (central equality): the identification of the dynamic Lagrange multiplier with V_x via the conjugate dual point invokes the envelope theorem and first-order conditions. For non-concave utilities the value function is typically only Lipschitz and the HJB equation admits viscosity solutions; the proof must therefore supply a weak-sense justification (e.g., via subdifferential or viscosity-test-function arguments) rather than classical differentiability. Without this step the equality does not automatically extend to the singular case the paper targets.
  2. [§5, Proposition 5.2] §5, Proposition 5.2 (homogeneity): the claimed scaling property of the multiplier process is derived under the optimal wealth and pricing-kernel dynamics. It is unclear whether the same relation continues to hold pathwise when the value function is merely a viscosity solution; an explicit verification or counter-example in the non-smooth regime would strengthen the claim.
minor comments (2)
  1. [§2–3] Notation for the dynamic multiplier process (e.g., λ_t versus Λ_t) is introduced inconsistently across Sections 2 and 3; a single global definition would improve readability.
  2. [§6] The numerical examples in §6 would benefit from an explicit statement of the discretization scheme and convergence diagnostics, even if the focus is illustrative.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and constructive feedback on our manuscript. The two major comments highlight important points regarding rigor in the non-concave, viscosity-solution setting. We address each below and indicate the revisions we will make to strengthen the arguments.

read point-by-point responses

  1. Referee: [§4, Theorem 4.1] §4, Theorem 4.1 (central equality): the identification of the dynamic Lagrange multiplier with V_x via the conjugate dual point invokes the envelope theorem and first-order conditions. For non-concave utilities the value function is typically only Lipschitz and the HJB equation admits viscosity solutions; the proof must therefore supply a weak-sense justification (e.g., via subdifferential or viscosity-test-function arguments) rather than classical differentiability. Without this step the equality does not automatically extend to the singular case the paper targets.

    Authors: We agree that the non-concave setting requires a weak-sense justification. Theorem 4.1 is proved by constructing the dynamic Lagrange multiplier directly from the martingale duality approach (which is valid for merely continuous value functions) and showing it equals the argmax of the conjugate utility. This point lies in the subdifferential of the value function by convex duality. To address the viscosity case explicitly, we will add a new lemma in the revised manuscript that verifies the equality using viscosity test functions: if a smooth test function touches the value function from above or below at a point, the first-order condition for the conjugate holds at that point, consistent with the HJB viscosity solution property. This supplies the requested weak justification without assuming classical differentiability. revision: yes

  2. Referee: [§5, Proposition 5.2] §5, Proposition 5.2 (homogeneity): the claimed scaling property of the multiplier process is derived under the optimal wealth and pricing-kernel dynamics. It is unclear whether the same relation continues to hold pathwise when the value function is merely a viscosity solution; an explicit verification or counter-example in the non-smooth regime would strengthen the claim.

    Authors: The homogeneity in Proposition 5.2 is obtained by direct application of Itô's formula to the product of optimal wealth and the inverse pricing kernel; these processes are semimartingales whose dynamics follow from the budget constraint and market completeness, independent of the regularity of the value function. The scaling therefore holds pathwise by construction. We will revise the proof to include an explicit remark stating that the derivation uses only the semimartingale property and the definition of optimality via duality, not the HJB equation or differentiability of V. No counter-example is required because the argument is general; we will also add a short numerical check in the non-smooth example section to illustrate the property numerically. revision: yes

Circularity Check

0 steps flagged

No significant circularity: equality claim is a derived bridge between independent approaches

full rationale

The paper's central result equates the martingale-duality Lagrange multiplier to the conjugate dual point of the value function (and thus its wealth derivative) in a non-concave setting. This is presented as a demonstration that bridges two standard methods rather than a definitional identity or a prediction obtained by fitting parameters to the target quantity. No self-citation is invoked as load-bearing justification for the equality, no ansatz is smuggled via prior work, and the derivation does not reduce the claimed result to its own inputs by construction. The abstract and description indicate a theorem-style identification that must be proved under the paper's stated assumptions, leaving the chain self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The central claim rests on the existence of the stated equivalence in the non-concave setting; the key new element is the dynamic multiplier itself. No explicit free parameters are mentioned. The framework inherits standard assumptions from duality theory in complete markets.

axioms (2)
  • domain assumption Complete markets in continuous time
    Required for the martingale duality approach to be well-defined, as stated in the abstract.
  • domain assumption Existence of optimal wealth and pricing kernel processes
    Implicit for both duality and dynamic programming methods to apply.
invented entities (1)
  • dynamic Lagrange multipliers no independent evidence
    purpose: Bridging martingale duality and dynamic programming while providing economic interpretation as shadow price
    Newly proposed object whose properties are demonstrated in the paper.

pith-pipeline@v0.9.0 · 5427 in / 1352 out tokens · 55466 ms · 2026-05-10T10:52:53.886669+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

63 extracted references · 63 canonical work pages

  1. [1]

    Bergk, K., Brandtner, M., K\" u rsten, W. (2021). Portfolio selection with tail nonlinearly transformed risk measures-a comparison with mean-CVaR analysis. Quantitative Finance, 21, 1011-1025

    work page 2021

  2. [2]

    B., Kouwenberg, R., Post, G

    Berkelaar, A. B., Kouwenberg, R., Post, G. T. (2004). Optimal portfolio choice under loss aversion. The Review of Economics and Statistics, 86, 973-987

    work page 2004

  3. [3]

    Bernard, C., Vanduffel, S., Ye, J. (2019). Optimal strategies under Omega ratio. European Journal of Operational Research, 275(2), 755-767

    work page 2019

  4. [4]

    Bichuch, M., Sturm, S. (2014). Portfolio optimization under convex incentive schemes. Finance and Stochastics, 18, 873-915

    work page 2014

  5. [5]

    Brighi, B., Chipot, M. (1994). Approximated convex envelope of a function. SIAM Journal on Numerical Analysis, 31, 128-148

    work page 1994

  6. [6]

    G., Zin, S

    Epstein, L. G., Zin, S. E. (1990). `First-order' risk aversion and the equity premium puzzle. Journal of Monetary Economy, 26, 387-407

    work page 1990

  7. [7]

    Carpenter, J. N. (2000). Does option compensation increase managerial risk appetite? Journal of Finance, 55, 2311-2331

    work page 2000

  8. [8]

    Chen, A., Hieber, P., Nguyen, T. (2019). Constrained non-concave utility maximization: An application to life insurance contracts with guarantees. European Journal of Operational Research, 273, 1119-1135

    work page 2019

  9. [9]

    Chen, A., & Hieber. P. (2016). Optimal asset allocation in life insurance: The impact of regulation. Astin Bulletin, 46, 605-626

    work page 2016

  10. [10]

    Chen, A., Hieber, P., Nguyen, T. (2019). Constrained non-concave utility maximization: an application to life insurance contracts with guarantees. European Journal of Operational Research, 273, 1119-1135

    work page 2019

  11. [11]

    Chen, A., Pelsser, A., Vellekoop, M. (2011). Modeling non-monotone risk aversion using SAHARA utility functions. Journal of Economic Theory, 146, 2075-2092

    work page 2011

  12. [12]

    C., Huang, C

    Cox, J. C., Huang, C. (1989). Optimal consumption and portfolio policies when asset prices follow a diffusion process. Journal of Economic Theory, 49, 33-83

    work page 1989

  13. [13]

    Dai, M., Kou, S., Qian, S., Wan, X. (2022). Nonconcave utility maximization with portfolio bounds. Management Science, 68, 8368-8385

    work page 2022

  14. [14]

    Dong, Y., Zheng, H. (2020). Optimal investment with S-shaped utility and trading and Value at Risk constraints: An application to defined contribution pension plan. European Journal of Operational Research, 281, 341-356

    work page 2020

  15. [15]

    J., Kolk, J

    Duistermaat, J. J., Kolk, J. A. C. (2010). Distributions: Theory and Application. Birkhäuser Boston, Massachusetts

    work page 2010

  16. [16]

    Evans, L. C. (2010). Partial Differential Equations. American Mathematical Society, Providence, Rhode Island

    work page 2010

  17. [17]

    Friedman, A. (2008). Partial Differential Equations of Parabolic Type. Dover Publications, New York

    work page 2008

  18. [18]

    Guan, G., Liang, Z., Xia, Y. (2023). Optimal management of DC pension fund under the relative performance ratio and VaR constraint. European Journal of Operational Research, 305, 868-886

    work page 2023

  19. [19]

    M., Pliska, S

    Harrison, J. M., Pliska, S. R. (1981). A stochastic calculus model of continuous trading: Complete markets, Stochastic Processes and their Applications, 15, 313-316

    work page 1981

  20. [20]

    He, L., Liang, Z., Liu, Y., Ma, M. (2019). Optimal control of DC pension plan manager under two incentive schemes. North American Actuarial Journal, 23, 120-141

    work page 2019

  21. [21]

    He, L., Liang, Z., Liu, Y., & Ma, M. (2020). Weighted utility optimization of the participating endowment contract. Scandinavian Actuarial Journal, 2020, 577-693

    work page 2020

  22. [22]

    D., Kou, S

    He, X. D., Kou, S. (2018). Profit sharing in hedge funds. Mathematical Finance, 28, 50-81

    work page 2018

  23. [23]

    D., Zhou, X

    He, X. D., Zhou, X. Y. (2016). Hope, fear, and aspirations. Mathematical Finance, 26, 3-50

    work page 2016

  24. [24]

    Hiriart-Urruty, J., Lemaréchal, C. (2001). Fundamentals of Convex Analysis. Springer Berlin, Heidelberg

    work page 2001

  25. [25]

    E., Jackwerth, J

    Hodder, J. E., Jackwerth, J. C. (2007). Incentive contracts and hedge fund management. Journal of Financial and Quantitative Analysis, 42, 811-826

    work page 2007

  26. [26]

    Q., Zhou X

    Jin, H., Xu, Z. Q., Zhou X. Y. (2007). A convex stochastic optimization problem arising from portfolio selection, Mathematical Finance, 18, 171-183

    work page 2007

  27. [27]

    Hörmander, L. (2003). Distribution Theory and Fourier Analysis. Springer Berlin, Heidelberg

    work page 2003

  28. [28]

    (1979) Prospect theory: An analysis of decision under risk

    Kahneman, D., Tversky, T. (1979) Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291

    work page 1979

  29. [29]

    E.(1991)

    Karatzas, I.,& Shreve, S. E.(1991). Brownian Motion and Stochastic Calculus. Springer, New York

    work page 1991

  30. [30]

    Karatzas, I., Shreve, S. E. (1998). Methods of Mathematical Finance. Springer, New York

    work page 1998

  31. [31]

    P., Shreve, S

    Karatzas, I., Lehoczky, J. P., Shreve, S. E. (1987). Optimal portfolio and consumption decisions for a ``small investor" on a finite horizon. SIAM Journal on Control and Optimization, 25, 1557-1586

    work page 1987

  32. [32]

    Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under risk, Econometrica, 47, 263-291

    work page 1979

  33. [33]

    P., Shreve, S

    Karatzas, I., Lehoczky, J. P., Shreve, S. E., Xu, G. L. (1991). Martingale and duality methods for utility maximization in an incomplete market. SIAM Journal on Control and Optimization, 29, 702-730

    work page 1991

  34. [34]

    Kouwenberg, R., Ziemba, W. T. (2007). Incentives and risk taking in hedge funds. Journal of Banking and Finance, 31, 3291-3310

    work page 2007

  35. [35]

    Kramkov, D., Schachermayer, W. (1999). The asymptotic elasticity of utility functions and optimal investment in incomplete markets, Annals of Applied Probability, 9, 904-950

    work page 1999

  36. [36]

    Li, Y., Yeh, C. (2010). Some characterizations of convex functions. Computers and Mathematics with Applications, 59, 327-337

    work page 2010

  37. [37]

    Liang, Z., Liu, Y. (2020). A classification approach to general S-shaped utility optimization with principals' constraints. SIAM Journal on Control and Optimization, 58, 3734-3762

    work page 2020

  38. [38]

    Liang, Z., Liu, Y. (2024). An asymptotic approach to centrally planned portfolio selection. Advances in Applied Probability, 56, 757-784

    work page 2024

  39. [39]

    Liang, Z., Liu, Y., Ma, M., Vinoth, R. P. (2024). A unified formula of the optimal portfolio for piecewise hyperbolic absolute risk aversion utilities. Quantitative Finance, 24, 281-303

    work page 2024

  40. [40]

    Liang, Z., Liu, Y., Zhang, L. (2025). A framework of state-dependent utility optimisation with general benchmarks. Finance and Stochastics, 29, 469-518

    work page 2025

  41. [41]

    Lin, H., Saunders, D., Weng, C. (2017). Optimal investment strategies for participating contracts. Insurance: Mathematics and Economics, 73, 137-155

    work page 2017

  42. [42]

    Lin, H., Saunders, D., Weng, C. (2019). Portfolio optimization with performance ratios. International Journal of Theoretical and Applied Finance, 22(05), 1950022

    work page 2019

  43. [43]

    Machina, M. J. (1982). Expected utility analysis without the independence axiom. Econometrica, 50, 277-323

    work page 1982

  44. [44]

    Machina, M. J. (1989). Comparative statics and non-expected utility preferences. Journal of Economic Theory, 47, 393-405

    work page 1989

  45. [45]

    Merton, R. C. (1969). Lifetime portfolio selection under uncertainty: The continuous-time case. Review of Economics and Statistics, 51, 247-257

    work page 1969

  46. [46]

    Markowitz, H. (1952). The Utility of Wealth, Journal of Political Economy 60 , 151–158

    work page 1952

  47. [47]

    Merton, R. C. (1971). Optimum consumption and portfolio rules in a continuous-time model. Journal of Economic Theory, 3, 373-413

    work page 1971

  48. [48]

    Nguyen, T., Stadje, M. (2020). Nonconcave optimal investment with value-at-risk constraint: An application to life insurance contracts. SIAM Journal on Control and Optimization, 58, 895-936

    work page 2020

  49. [49]

    Pliska, S. R. (1986). A stochastic calculus model of continuous trading: Optimal portfolios. Mathematics of Operations Research, 11, 371-382

    work page 1986

  50. [50]

    Pratt, J. (1964). Risk aversion in the small and in the large. Econometrica, 32, 122-136

    work page 1964

  51. [51]

    Qian, S., Yang, C. (2026). Non-concave utility maximization with transaction costs. SIAM Journal on Financial Mathematics, forthcoming. arxiv: 2307.02178

    work page arXiv 2026

  52. [52]

    Qin, C., Yang, C., Zheng, H. (2026). Periodic evaluation with non-concave utility. SSRN: 5305617

    work page 2026

  53. [53]

    Quiggin, J. (1982). A Theory of Anticipated Utility. Journal of Economic Behavior and Organization, 3, 323-343

    work page 1982

  54. [54]

    Quiggin, J. (1993). Generalized Expected Utility Theory: The Rank-Dependent Model. Boston: Kluwer

    work page 1993

  55. [55]

    Reichlin, C. (2013). Utility maximization with a given pricing measure when the utility is not necessarily concave. Mathematics and Financial Economics, 7, 531-556

    work page 2013

  56. [56]

    Rockafellar, R. T. (1970). Convex Analysis. Princeton University Press, Princeton

    work page 1970

  57. [57]

    Shreve, S. E. (2004). Stochastic Calculus for Finance I. Springer, New York

    work page 2004

  58. [58]

    M., Shakarchi, R

    Stein, E. M., Shakarchi, R. (2010). Functional Analysis: Introduction to Further Topics in Analysis. Princeton University Press, Princeton

    work page 2010

  59. [59]

    Segal, U., Spivak, A. (1990). First-order versus second-order risk aversion. Journal of Economic Theory, 51, 111-125

    work page 1990

  60. [60]

    Segal, U., Spivak, A. (1997). First-order risk aversion and non-differentiability. Economic Theory, 9, 179-183

    work page 1997

  61. [61]

    Wang, R., Xu, Z., Zhou, X. (2019). Dual utilities on risk aggregation under dependence uncertainty. Finance and Stochastics, 23, 1025-1048

    work page 2019

  62. [62]

    Yong, J., Zhou, X. Y. (1999). Stochastic Controls: Hamiltonian Systems and HJB Equations. Springer. New York

    work page 1999

  63. [63]

    Westray, N., Zheng, H. (2009). Constrained nonsmooth utility maximization without quadratic inf convolution. Stochastic Processes and their Applications, Vol. 119, p1561-1579

    work page 2009