Fact-checked by Grok 2 weeks ago

Pontryagin's maximum principle

Pontryagin's maximum principle is a fundamental theorem in optimal control theory that provides necessary conditions for the existence of an optimal control in problems involving dynamical systems governed by ordinary differential equations.^[1] Developed by Lev Semenovich Pontryagin and his collaborators at the Steklov Mathematical Institute in the mid-1950s, the principle addresses the optimization of a performance index subject to state constraints and bounded controls, extending classical calculus of variations to modern control scenarios.^[2] The principle was first systematically presented in the 1962 English edition of The Mathematical Theory of Optimal Processes by L. S. Pontryagin, V. G. Boltyanskii, R. V. Gamkrelidze, and E. F. Mishchenko, following its initial Russian publication in 1961.^[3] At its core, the maximum principle requires the introduction of an adjoint vector \psi(t) that satisfies a system of differential equations derived from the system's dynamics, ensuring the overall solution remains nontrivial (i.e., not identically zero).^[4] The key optimality condition mandates that the optimal control u^*(t) maximizes the Hamiltonian function H(\psi(t), x(t), u) = \psi(t) \cdot f(x(t), u) almost everywhere along the optimal trajectory x(t), where f describes the system's vector field.^[4] Additional requirements include transversality conditions at the boundaries, such as \psi(t_1) = 0 for free terminal states, and the normalization H(\psi(t_1), x(t_1), u(t_1)) = 0.^[4] This formulation unifies various optimal control problems, including time-optimal control, fuel-optimal trajectories, and resource allocation, by reducing them to boundary value problems solvable via numerical methods or analytical techniques.^[1] Since its inception, the principle has been generalized to handle constraints like state inequalities, stochastic systems, and infinite-dimensional settings, influencing fields from aerospace engineering (e.g., spacecraft trajectory optimization) to economics (e.g., growth models) and biology (e.g., epidemic control).^[5] Its equivalence to dynamic programming approaches, such as Bellman's principle of optimality, underscores its versatility, though it excels in problems with non-smooth or discontinuous controls.^[6]

Introduction

Overview and significance

Pontryagin's maximum principle (PMP) serves as a necessary condition for optimality in continuous-time optimal control problems, where the goal is to minimize or maximize a cost functional subject to dynamic constraints. It identifies candidate optimal solutions by requiring that the control input maximizes (or minimizes, depending on the formulation) a constructed Hamiltonian function pointwise along the optimal trajectory.^[1]^[7] The principle emerged from research conducted by Lev Pontryagin and his collaborators at the Steklov Mathematical Institute in Moscow during the 1950s, a period marked by Cold War tensions and intense competition in aerospace technology between the Soviet Union and the United States. This work was driven by practical needs in trajectory optimization for rocketry and missile systems, enabling solutions to complex guidance problems that were previously intractable.^[8]^[9] PMP holds broad significance in diverse disciplines, including robotics for efficient path planning in manipulators and mobile agents, economics for dynamic resource allocation and growth models, and physics for steering quantum systems toward desired states. It builds on the Euler-Lagrange equations of the calculus of variations by incorporating explicit control variables and constraints, thus addressing a wider class of optimization challenges in controlled dynamical systems.^[10]^[11]^[7]^[12] A crucial aspect is that PMP yields local, pointwise conditions on controls rather than ensuring global optimality, often requiring supplementary analysis to confirm true minima among candidates.^[1]

Historical background

Pontryagin's maximum principle emerged in the mid-1950s through the collaborative efforts of the blind Russian mathematician Lev Semenovich Pontryagin and his students Vladimir Grigorievich Boltyanskii, Revaz Valerianovich Gamkrelidze, and Evgenii Fedorovich Mishchenko at the Steklov Mathematical Institute in Moscow.^[13]^[9] The development was spurred by practical demands from the Soviet rocketry and space program, particularly the need to optimize trajectories for minimum-time problems in aerospace engineering during the Cold War era.^[9]^[14] The principle's initial formulation appeared in a seminal 1956 paper titled "On the Theory of Optimal Processes," published in Doklady Akademii Nauk SSSR by Pontryagin, Boltyanskii, and Gamkrelidze, marking the first public announcement of the maximum principle as a necessary condition for optimal control.^[14] This work laid the groundwork for a comprehensive theory, which was formalized in the 1961 Russian monograph Matematicheskaya teoriya optimal'nykh protsessov (translated into English in 1962 as The Mathematical Theory of Optimal Processes by Interscience Publishers), co-authored by the full group including Mishchenko.^[13]^[15] For these contributions, Pontryagin and his collaborators received the prestigious Lenin Prize in 1962, recognizing their foundational impact on optimal control theory.^[13]^[9] Early presentations of the principle featured heuristic arguments rather than fully rigorous proofs, which were subsequently refined by the students: Gamkrelidze established it for linear systems in 1957, while Boltyanskii provided a general proof in 1958, confirming its status as a necessary optimality condition.^[13]^[9] This Soviet approach contrasted with contemporaneous Western developments, such as Richard Bellman's dynamic programming method introduced in the 1950s, which emphasized recursive value functions rather than the Hamiltonian-based maximization central to Pontryagin's framework.^[9] The principle's global influence grew in the 1960s through English and German translations of the 1962 book, bridging classical calculus of variations with emerging modern control theory and facilitating its adoption in Western academic and engineering communities.^[9]^[14]

Mathematical Foundations

Problem formulation

Pontryagin's maximum principle provides necessary conditions for optimality in a broad class of optimal control problems, where the objective is to select a control function that minimizes a cost functional while satisfying given dynamic constraints. The standard problem is formulated as follows: find an admissible control u(\cdot) that minimizes the performance index

J(x, u) = \int_{t_0}^{T} \phi(x(t), u(t), t) \, dt + \psi(x(T)),

where the state trajectory x(\cdot) evolves according to the ordinary differential equation

\dot{x}(t) = f(x(t), u(t), t), \quad x(t_0) = x_0,

over the time interval [t_0, T], and the control satisfies u(t) \in U for all t \in [t_0, T]. Here, x(t) \in \mathbb{R}^n represents the state vector, u(t) \in \mathbb{R}^m the control vector, x_0 \in \mathbb{R}^n the fixed initial state, \phi: \mathbb{R}^n \times \mathbb{R}^m \times \mathbb{R} \to \mathbb{R} the running cost, \psi: \mathbb{R}^n \to \mathbb{R} the terminal cost, and f: \mathbb{R}^n \times \mathbb{R}^m \times \mathbb{R} \to \mathbb{R}^n the dynamics.^[16]^[17] The terminal time T can be either fixed or free, depending on the specific application; when free, additional conditions such as transversality must hold at T. The control set U \subset \mathbb{R}^m is typically required to be closed and compact to ensure the existence of optimal controls and the applicability of the principle. Boundary conditions include the fixed initial state x(t_0) = x_0, while the terminal state x(T) may be fixed to a target, free, or constrained to a manifold, influencing the form of the necessary conditions.^[16]^[18] Common assumptions on the problem functions ensure well-posedness and smoothness for deriving the principle: f and \phi are continuously differentiable in all arguments, with f Lipschitz continuous in x to guarantee unique solutions to the state equation for given controls. In many cases, explicit time dependence is suppressed for simplicity, yielding time-invariant forms f(x, u) and \phi(x, u), though the general time-varying case is fully addressed by the principle.^[16]^[17] This setup extends the classical calculus of variations, which optimizes integral functionals without explicit control constraints or bounded inputs, by allowing piecewise continuous controls in compact sets U and accommodating non-smooth or discontinuous objectives through appropriate regularity conditions.^[16]

Notation and key assumptions

In the standard formulation of Pontryagin's maximum principle (PMP) for optimal control problems, the state trajectory is denoted by x(\cdot): [t_0, T] \to \mathbb{R}^n, representing the evolution of the system over the fixed time interval [t_0, T]. The control input is u(\cdot): [t_0, T] \to U, where U \subset \mathbb{R}^m is a compact set ensuring bounded admissible controls. The system dynamics are governed by the ordinary differential equation \dot{x}(t) = f(t, x(t), u(t)), with f: [t_0, T] \times \mathbb{R}^n \times U \to \mathbb{R}^n. The objective is typically to minimize a cost functional involving a running cost \phi(t, x(t), u(t)) that is locally Lipschitz continuous, along with possible terminal costs.^[17] The Hamiltonian function, central to the PMP, is defined as H(t, x, \psi, u) = \psi \cdot f(t, x, u) - \phi(t, x, u), where \psi \in \mathbb{R}^n is the adjoint (or costate) variable. This construction incorporates the sign convention for minimization problems, where the running cost \phi is subtracted to reflect the optimization direction; note that some formulations for maximization of payoffs adjust the sign accordingly. The adjoint \psi(\cdot) satisfies a backward differential equation derived from the principle.^[19]^[17] Key assumptions ensure the existence and regularity of solutions. The vector field f is assumed continuous in (t, u) and locally Lipschitz continuous in x, uniformly with respect to t and u in compact sets, guaranteeing uniqueness of solutions to the state equation via Filippov theory for measurable controls. The running cost \phi is locally Lipschitz in x and continuous in (t, u). The control set U is compact, often convex, to allow maximization over u \in U. These conditions hold in the continuous-time setting, distinguishing it from discrete-time variants where summation replaces integration and Carathéodory measurability is not required.^[17]^[12] An extremal is characterized by the tuple (x(\cdot), u(\cdot), \psi(\cdot), \lambda_0), where \lambda_0 \geq 0 is a scalar multiplier (often called the normality constant). The nontriviality condition requires that not both \lambda_0 and \psi(\cdot) vanish identically, i.e., (\lambda_0, \psi) \neq (0, 0). For normalization, one typically scales so that \lambda_0 + \|\psi(\cdot)\|_{L^\infty} = 1. In normal cases, abnormal multipliers are excluded by setting \lambda_0 = 1, simplifying the analysis; abnormal extremals (\lambda_0 = 0) arise in problems with nonstandard constraints but are often ruled out under additional regularity assumptions.^[19]^[12]

Core Statement

Hamiltonian construction

In optimal control theory, the Hamiltonian function serves as the central construct in Pontryagin's maximum principle, embedding both the system dynamics and the objective functional into a single expression to facilitate the derivation of necessary optimality conditions.^[20] This formulation, introduced in the seminal work on optimal processes, allows for the analysis of control problems by treating the state and control variables in a unified framework. For a standard minimization problem, where the goal is to minimize the integral of a running cost L(x, u, t) subject to dynamics \dot{x} = f(x, u, t), the Hamiltonian is defined as

H(x, \psi, u, t) = \langle \psi, f(x, u, t) \rangle - L(x, u, t),

with \langle \cdot, \cdot \rangle denoting the inner product, x \in \mathbb{R}^n the state vector, \psi \in \mathbb{R}^n the costate vector, and u \in U the control from a admissible set U.^[20] The term \langle \psi, f(x, u, t) \rangle captures the evolution of the state influenced by the costate, while -L(x, u, t) incorporates the instantaneous cost. For maximization problems, the sign convention flips, yielding H(x, \psi, u, t) = L(x, u, t) + \langle \psi, f(x, u, t) \rangle, adjusting the principle accordingly to seek minimization of the negative Hamiltonian.^[20] The Hamiltonian admits an interpretation analogous to the total energy in classical mechanics, where \langle \psi, f \rangle represents a momentum-like contribution driving the system forward, balanced against the dissipative cost term L, thus quantifying the trade-off between dynamic progress and objective penalty at each instant.^[21] A key property is that, for an optimal control u^*, the Hamiltonian achieves its maximum value over u \in U pointwise at every time t along the optimal trajectory, i.e., H(x(t), \psi(t), u^*(t), t) = \max_{u \in U} H(x(t), \psi(t), u, t).^[20] This maximization condition highlights the principle's variational nature; however, if the Hamiltonian is flat (independent of u) over a subinterval of U, optimal controls may not be unique, allowing for a continuum of solutions.^[19] The costate \psi functions as the conjugate variable to the state x, establishing a duality that mirrors the symplectic structure of phase space in Hamiltonian mechanics, where the pair (x, \psi) evolves on a contact manifold akin to the cotangent bundle T^* \mathbb{R}^n.^[22] This geometric perspective underscores the principle's roots in variational calculus, with the costate encoding sensitivity to state perturbations much like momentum in physical systems.^[22] A illustrative case arises in linear systems, where the dynamics take the form \dot{x} = Ax + [Bu](/page/BU) and the Hamiltonian becomes affine in u, expressed as H(x, \psi, u, t) = \psi^T (Ax + Bu) - L(x, u, t).^[23] The affinity implies that the maximizer u^* occurs at the boundaries of U, typically yielding bang-bang controls that switch abruptly between extremal values, such as \pm 1 for bounded scalar controls, to extremize the linear term \psi^T [Bu](/page/BU).^[23]

Adjoint system and necessary conditions

In Pontryagin's maximum principle, the adjoint system consists of differential equations governing the costate variables \psi(t) \in \mathbb{R}^n, which are introduced to incorporate the optimality conditions alongside the state dynamics \dot{x}(t) = f(x(t), u(t), t). The adjoint equation is given by

\dot{\psi}(t) = -\frac{\partial H}{\partial x}(x(t), \psi(t), u(t), t),

where H(x, \psi, u, t) = \langle \psi, f(x, u, t) \rangle - L(x, u, t) is the Hamiltonian function.^[17] This equation holds along the optimal trajectory (x^*(t), u^*(t)) for t \in [t_0, T]. For problems with a terminal cost \phi(x(T)) and free final state (subject to possible endpoint constraints), the terminal condition is

\psi(T) = \frac{\partial \phi}{\partial x}(x(T)),

ensuring transversality at the endpoint.^[17] Smoothness assumptions on f and the terminal cost, such as Lipschitz continuity and differentiability, guarantee the existence of such \psi.^[24] The full necessary conditions for optimality require the existence of a nontrivial multiplier pair (\psi(t), \lambda_0), where \lambda_0 \geq 0 is a scalar multiplier for the objective functional, such that the state and adjoint equations are satisfied (in the normal case \lambda_0 = 1):

\dot{x}(t) = \frac{\partial H}{\partial \psi}(x(t), \psi(t), u(t), t), \quad \dot{\psi}(t) = -\frac{\partial H}{\partial x}(x(t), \psi(t), u(t), t).

^[23] Additionally, if the problem has no explicit time dependence in the dynamics or cost and the final time T is free, the condition

H(x(t), \psi(t), u(t), t) = 0

holds along the entire optimal trajectory in the normal case.^[24] The nontriviality condition stipulates that (\psi(t), \lambda_0) \not\equiv (0, 0) for all t \in [t_0, T], preventing the degenerate case where the multipliers vanish identically.^[17] For problems with free initial time t_0, a transversality condition requires H(t_0) = 0 at the optimal starting point.^[23] In abnormal cases, where \lambda_0 = 0, the conditions reduce to constraints driven solely by the endpoint requirements without influence from the running or terminal costs, often arising in problems where the objective functional does not affect the optimality structure.^[24] These cases highlight the need for careful normalization, typically setting \lambda_0 = 1 in normal scenarios to avoid scaling ambiguities.^[17]

Maximization condition

The maximization condition forms the cornerstone of Pontryagin's maximum principle, dictating that an optimal control must pointwise maximize the Hamiltonian function along the optimal trajectory. Specifically, for an optimal state trajectory x^*(t) and control u^*(t), there exists a nontrivial adjoint trajectory \psi(t) such that the Hamiltonian H(x^*(t), \psi(t), u, t) \leq H(x^*(t), \psi(t), u^*(t), t) holds for all admissible controls u \in U and almost everywhere t \in [t_0, T].^[25] This condition ensures that at each instant, the control selection aligns the system's evolution to minimize the overall cost functional.^[19] The implications of this condition profoundly influence the structure of optimal controls. When the Hamiltonian is linear in the control variable, the maximization occurs at extreme points of the convex control set U, often resulting in bang-bang controls that switch abruptly between boundary values based on the sign of the switching function \partial H / \partial u.^[26] If the switching function vanishes over a finite interval, singular controls may arise, requiring higher-order analysis for optimality; otherwise, chattering or mixed controls can emerge in cases of non-uniqueness.^[1] These structures highlight how the principle reduces complex infinite-dimensional problems to finite sets of candidate trajectories.^[23] In the normal case, where the multiplier \lambda_0 = 1 (normalizing the costate to avoid triviality), the inequality is typically strict unless the control is indifferent, meaning multiple u achieve the maximum without affecting the cost.^[19] Abnormal cases (\lambda_0 = 0) relax this, potentially allowing non-strict maximization independent of the objective, though such scenarios are rare and often excluded by problem assumptions like controllability.^[27] Geometrically, the maximization condition interprets the adjoint \psi(t) as a momentum covector that, when maximized, aligns the direction of the dynamics f(x, u, t) to steer the state trajectory toward the desired terminal conditions while minimizing the integrated cost, akin to Hamiltonian mechanics on the cotangent bundle.^[17] This alignment ensures the optimal path follows extremals in the phase space, balancing forward state evolution with backward adjoint propagation.^[25] For problems with state equality constraints, such as g(x(t), t) = 0, the formulation incorporates additional multipliers \eta(t) into the Hamiltonian and adjoint equations, modifying the maximization to H(x^*(t), \psi(t) + \eta(t) \nabla g, u, t) \leq H(x^*(t), \psi(t) + \eta(t) \nabla g, u^*(t), t) while ensuring \eta(t) \geq 0 and complementary slackness.^[28] This extension preserves the core principle but accounts for the constrained manifold, with jumps in the adjoint possible at entry/exit points.^[29]

Derivation and Proof

Derivation of adjoint equations

The derivation of the adjoint equations in Pontryagin's maximum principle proceeds via a variational approach to the optimal control problem, where the goal is to minimize the cost functional J = \int_0^T L(x(t), u(t), t) \, dt + g(x(T)) subject to the dynamics \dot{x}(t) = f(x(t), u(t), t) and initial condition x(0) = x_0. Consider an optimal trajectory (x^*, u^*) and introduce small perturbations \delta x(t) and \delta u(t) around it. The first-order variation in the cost is then

\delta J = \int_0^T \left( \frac{\partial L}{\partial x} \delta x + \frac{\partial L}{\partial u} \delta u \right) dt + \frac{\partial g}{\partial x}(x^*(T)) \cdot \delta x(T),

with \delta x(0) = 0. These perturbations satisfy the linearized state equation

\delta \dot{x} = \frac{\partial f}{\partial x} \delta x + \frac{\partial f}{\partial u} \delta u,

evaluated along the optimal trajectory. To enforce this constraint while ensuring \delta J = 0 for optimality, introduce the adjoint variable \psi(t), a row vector, such that the integral constraint

\int_0^T \psi \cdot \left( \delta \dot{x} - \frac{\partial f}{\partial x} \delta x - \frac{\partial f}{\partial u} \delta u \right) dt = 0

holds for admissible variations. Apply integration by parts to the \delta \dot{x} term:

\int_0^T \psi \cdot \delta \dot{x} \, dt = \left[ \psi \cdot \delta x \right]_0^T - \int_0^T \dot{\psi} \cdot \delta x \, dt.

With \delta x(0) = 0, the lower boundary term vanishes, yielding

\psi(T) \cdot \delta x(T) - \int_0^T \dot{\psi} \cdot \delta x \, dt - \int_0^T \psi \cdot \frac{\partial f}{\partial x} \delta x \, dt - \int_0^T \psi \cdot \frac{\partial f}{\partial u} \delta u \, dt = 0.

Substitute into \delta J = 0 and collect coefficients of \delta x and \delta u. The terms involving \delta u are eliminated using the maximization condition on the Hamiltonian H(x, u, \psi, t) = \psi \cdot f(x, u, t) - L(x, u, t), leading to the adjoint equation for the costate dynamics:

\dot{\psi} = -\frac{\partial H}{\partial x} = -\psi \frac{\partial f}{\partial x} + \frac{\partial L}{\partial x}.

The terminal boundary condition follows from the coefficient of \delta x(T): for fixed final time T, \psi(T) = \frac{\partial g}{\partial x}(x^*(T)). If the final time T is free, the variation \delta T introduces an additional transversality condition H(x^*(T), u^*(T), \psi(T), T) + \frac{\partial g}{\partial t}(x^*(T)) = 0. In the full necessary conditions, a nonnegative multiplier \lambda_0 \geq 0 scales the cost variation as \lambda_0 \delta J, with the nontriviality condition (\lambda_0, \psi) \neq (0, 0) to avoid the trivial solution. Typically, \lambda_0 = 1 if the cost is nondegenerate (normal case); otherwise, \lambda_0 = 0 corresponds to an abnormal case where the adjoint scales the dynamics alone. This derivation assumes that f and L are continuously differentiable with respect to x and u, ensuring the partial derivatives exist and the linearized equations are well-defined. For problems with additional equality constraints, Lagrange multipliers are incorporated into an augmented Hamiltonian to handle them variationally.

Proof of the maximum principle

The proof of Pontryagin's maximum principle relies on variational techniques to establish the necessity of the maximization condition, assuming the existence of an optimal trajectory (x^*, u^*) and corresponding adjoint state \psi^* satisfying the adjoint equations derived previously. The core idea is to perturb the optimal control u^* in a controlled manner and analyze the first-order variation in the cost functional J, showing that optimality implies the Hamiltonian H(x^*, \psi^*, u^*, t) is maximized pointwise over admissible controls.

Needle Variation Method

The needle variation method, introduced in the original formulation, applies to problems where the optimal control u^* is piecewise continuous or can be approximated as such. Consider a point t_0 \in [t_0, T] where u^*(t_0) = \bar{u}, and select an alternative control \tilde{u} \in U with \tilde{u} \neq \bar{u}. Construct a perturbed control u^\epsilon by inserting a "needle" of alternative control \tilde{u} over a small interval [t_0, t_0 + \epsilon h], where h > 0 is fixed and \epsilon > 0 is small, while keeping u^\epsilon(t) = u^*(t) elsewhere; the interval length ensures the perturbation remains admissible.^[30] The corresponding state perturbation \delta x = x^\epsilon - x^* satisfies, to first order,

\dot{\delta x}(t) = \frac{\partial f}{\partial x}(x^*(t), u^*(t), t) \delta x(t) + \epsilon h \left[ f(x^*(t_0), \tilde{u}, t_0) - f(x^*(t_0), \bar{u}, t_0) \right] \delta_{t_0}(t),

where \delta_{t_0} is a Dirac measure at t_0, but for finite \epsilon, it is a rectangular pulse. The variation in the cost \delta J = J(u^\epsilon) - J(u^*) is then

\delta J = \left( \psi^*(t_0) \cdot \left[ f(x^*(t_0), \bar{u}, t_0) - f(x^*(t_0), \tilde{u}, t_0) \right] + L(x^*(t_0), \bar{u}, t_0) - L(x^*(t_0), \tilde{u}, t_0) \right) (\epsilon h) + o(\epsilon),

using the adjoint property \dot{\psi}^* = -\frac{\partial H}{\partial x}(x^*, \psi^*, u^*, t). Since u^* is optimal, \delta J \geq 0 for small \epsilon > 0, implying

\psi^*(t_0) \cdot f(x^*(t_0), \tilde{u}, t_0) - L(x^*(t_0), \tilde{u}, t_0) \leq \psi^*(t_0) \cdot f(x^*(t_0), \bar{u}, t_0) - L(x^*(t_0), \bar{u}, t_0),

or equivalently, H(x^*, \psi^*, \tilde{u}, t_0) \leq H(x^*, \psi^*, u^*(t_0), t_0). Extending this to a dense set of points t_0 and using continuity of H, the pointwise maximization holds almost everywhere.^[30]

Spike Construction

For general measurable controls u^* \in L^1([t_0, T]; U), where U may be nonconvex, the needle method is generalized via spike perturbations to handle essential suprema. A spike variation at t_0 replaces u^*(t) with \tilde{u} over a shrinking interval [t_0, t_0 + \epsilon], but weighted to approximate a Dirac delta: specifically, u^\epsilon(t) = (1 - \frac{\theta(t - t_0)}{\epsilon}) u^*(t) + \frac{\theta(t - t_0)}{\epsilon} \tilde{u} for t \in [t_0, t_0 + \epsilon], where \theta is the unit step function, ensuring the average control approximates the unperturbed one except at t_0. The state variation \delta x^\epsilon leads to

\delta J = \int_{t_0}^{t_0 + \epsilon} \left[ H(x^*(t), \psi^*(t), u^*(t), t) - H(x^*(t), \psi^*(t), \tilde{u}, t) \right] dt + o(\epsilon).

As \epsilon \to 0, by the fundamental lemma of calculus of variations and measurability of u^*, this yields \delta J / \epsilon \to H(x^*(t_0), \psi^*(t_0), u^*(t_0), t_0) - H(x^*(t_0), \psi^*(t_0), \tilde{u}, t_0). Optimality requires \delta J \geq 0 for all small \epsilon > 0 and all \tilde{u} \in U, implying the pointwise maximization H(x^*, \psi^*, u, t_0) \leq H(x^*, \psi^*, u^*(t_0), t_0) almost everywhere. Finite combinations of spikes approximate arbitrary measurable perturbations, ensuring the condition holds in the essential supremum sense.

Handling Free Time

In problems with free terminal time T, the proof incorporates a time rescaling variation alongside control perturbations. Parameterize the trajectory by a new time \tau = t / T, transforming the problem to fixed interval [0, 1] with adjusted dynamics \dot{x} = T f(x, u, t(\tau)). The augmented Hamiltonian includes a term for time scaling, and variations in T yield the transversality condition \partial H / \partial T = 0 at optimum, implying H(x^*, \psi^*, u^*, t) = 0 along the trajectory if the integrand L and dynamics f lack explicit t-dependence. For explicit t-dependence, H is constant but nonzero, derived by differentiating the cost under rescaled perturbations and using the adjoint to absorb time effects.^[23]

Completeness

The proof assumes an optimal solution exists and derives the necessity of the conditions; it establishes a local necessary condition for weak minima, but sufficiency requires additional convexity assumptions on H in u (e.g., linear-quadratic problems). Nondegeneracy (e.g., \psi^* \not\equiv 0) follows from the maximum condition and boundedness of U. For relaxed controls, where u is a probability measure on U to handle chattering limits, convex analysis embeds the problem in the space of Young measures, with the maximum principle stated via convex hull maximization: \bar{H}(x^*, \psi^*, \mu, t) = \int_U H(x^*, \psi^*, v, t) d\mu(v) \leq H(x^*, \psi^*, u^*, t) for measures \mu. This uses lower semicontinuity of the relaxed functional and Krein-Milman theorem for extreme points. For discontinuous trajectories, Filippov's lemma constructs absolutely continuous solutions to \dot{x} \in F(x, u, t) via convex valued selections, adapting the spike method to differential inclusions where the Hamiltonian maximizes over the Filippov set \text{co} F(x^*, u^*, t).

Applications and Examples

Classical optimal control problems

Classical optimal control problems provide concrete illustrations of Pontryagin's maximum principle (PMP) in engineering applications, particularly in aerospace and mechanical systems where the goal is to optimize trajectories under physical constraints. These examples demonstrate how PMP yields necessary conditions for optimality, often resulting in explicit control laws or switching structures that can be verified through simulation. The standard solution approach involves forming the Hamiltonian from the system dynamics and cost functional, solving the adjoint equations backward from the terminal conditions, and determining the control by maximizing the Hamiltonian, typically along loci where the switching function \partial H / \partial u = 0.^[23] A seminal example is Goddard's rocket problem, which seeks to maximize final mass m(t_f) (equivalently minimizing fuel consumption) for achieving a specified ascent altitude in a vertical launch under gravity and atmospheric drag. The state variables are altitude h \geq 0, velocity v \geq 0, and mass m > 0, with initial conditions v(0) = 0, h(0) = 0, m(0) = m_0, and terminal constraints h(t_f) = h_f, v(t_f) = 0, m(t_f) \geq m_{\min}, at free final time t_f > 0. The dynamics are governed by

\dot{h} = v, \quad \dot{v} = -g + \frac{T}{m} u - \frac{k}{2m} v^2, \quad \dot{m} = -\alpha u,

where g > 0 is gravitational acceleration, T > 0 is maximum thrust, k > 0 is the drag coefficient, \alpha > 0 is the mass flow rate per unit thrust, and the control u \in [0, 1] represents the normalized thrust level.^[31] Applying PMP, the Hamiltonian is

H = \psi_h v + \psi_v \left( -g + \frac{T}{m} u - \frac{k}{2m} v^2 \right) + \psi_m (-\alpha u),

with adjoint variables \psi_h, \psi_v, \psi_m satisfying

\dot{\psi}_h = 0, \quad \dot{\psi}_v = \frac{k v}{m} \psi_v - \psi_h, \quad \dot{\psi}_m = \psi_v \left( \frac{T u}{m^2} - \frac{k v^2}{2 m^2} \right),

and transversality conditions \psi_m(t_f) = 1 (free m(t_f)), \psi_h(t_f) and \psi_v(t_f) free, with H(t_f) = 0 for free time. The switching function is \partial H / \partial u = (T/m) \psi_v - \alpha \psi_m, yielding bang-off-bang control: full thrust u=1 when positive, coast u=0 when negative, and potential singular arcs where zero over an interval. Optimal trajectories typically feature an initial full-thrust phase to overcome drag, a coasting phase, and a final thrust adjustment, confirmed by numerical simulation showing minimal fuel trajectories satisfying constraints.^[31] Another classic case is time-optimal control to the origin for the double integrator, modeling a point mass under bounded acceleration, such as spacecraft rendezvous or robotic positioning. The states are position x_1 and velocity x_2, with dynamics

\dot{x}_1 = x_2, \quad \dot{x}_2 = u, \quad |u| \leq 1,

initial state (x_1(0), x_2(0)) = (X_1, X_2), and terminal goal (x_1(t_f), x_2(t_f)) = (0, 0) at minimal free time t_f > 0. The cost is simply J = t_f.^[23] The Hamiltonian is H = -1 + p_1 x_2 + p_2 u, with adjoints \dot{p}_1 = 0, \dot{p}_2 = -p_1, and terminal p_1(t_f) = \nu_1, p_2(t_f) = \nu_2 (Lagrange multipliers \nu). Maximizing H gives bang-bang control u = \operatorname{sign}(p_2), with at most one switch since \ddot{p}_2 = 0. The switching curve in state space is x_2 + \sqrt{2 x_1} = 0 (for x_1 > 0), derived by integrating backward from the origin, leading to unique trajectories with one switch for most initial conditions. Simulations verify optimality by showing no shorter paths exist under the bound. As noted briefly, this exemplifies bang-bang controls from the maximization condition.^[23] For quadratic performance criteria, PMP recovers the linear quadratic regulator (LQR) for linear systems \dot{x} = A x + B u, x(0) = x_0, minimizing J = \frac{1}{2} \int_0^T (x^T Q_x x + u^T Q_u u) \, dt + \frac{1}{2} x(T)^T P_1 x(T), with Q_x \geq 0, Q_u > 0, P_1 \geq 0. The Hamiltonian H = \frac{1}{2} (x^T Q_x x + u^T Q_u u) + \lambda^T (A x + B u) yields \dot{\lambda} = -Q_x x - A^T \lambda, \lambda(T) = P_1 x(T), and unconstrained minimizer u = -Q_u^{-1} B^T \lambda. Assuming \lambda(t) = P(t) x(t), the adjoint substitutes into a differential Riccati equation -\dot{P} = P A + A^T P - P B Q_u^{-1} B^T P + Q_x, P(T) = P_1, providing the feedback law u = -Q_u^{-1} B^T P x without solving the full two-point boundary value problem. For infinite horizon, it reduces to the algebraic Riccati equation. This derivation highlights PMP's role in synthesizing stabilizing controllers, verified by Lyapunov analysis showing cost reduction along closed-loop trajectories.^[32]

Economic interpretations

In economic optimization problems, Pontryagin's maximum principle (PMP) provides a framework for deriving necessary conditions for intertemporal efficiency, where the adjoint variables are interpreted as shadow prices representing the marginal value of state variables such as capital or resource stocks.^[33] The Hamiltonian maximization condition then corresponds to selecting controls—like consumption or extraction rates—that balance immediate payoffs against the future value preserved through these shadow prices.^[33] This interpretation aligns with economic notions of opportunity cost and dynamic efficiency, distinguishing it from engineering applications by emphasizing resource allocation under scarcity and discounting.^[11] A canonical application is the Ramsey model of optimal economic growth, which seeks to maximize the discounted integral of utility from consumption, \int_0^\infty u(c_t) e^{-(\rho - n)t} \, dt, subject to the per capita capital accumulation constraint \dot{k}_t = f(k_t) - c_t - (n + \delta)k_t with c_t \geq 0, where k_t is capital per worker, f(k_t) is production, \rho is the discount rate, n is population growth, and \delta is depreciation.^[34] The current-value Hamiltonian is H = u(c) + \psi [f(k) - c - (n + \delta)k], where \psi(t) is the shadow price of capital.^[34] The adjoint equation \dot{\psi} = -\partial H / \partial k = \psi [ \rho + n + \delta - f'(k) ] combined with the maximization condition \partial H / \partial c = u'(c) - \psi = 0 yields the Euler equation f'(k) = \rho + \delta + \frac{\dot{\psi}}{\psi}, or equivalently, the consumption growth rule \frac{\dot{c}}{c} = \frac{1}{\theta(c)} (f'(k) - \rho - \delta), where \theta(c) = -c u''(c)/u'(c) is the intertemporal elasticity of substitution.^[34] Here, \psi(t) captures the marginal utility of additional capital at time t, guiding the trade-off between current consumption and future growth.^[33] In resource extraction, PMP derives the Hotelling rule for non-renewable resources, maximizing discounted profits \int_0^\infty e^{-\rho t} [p u(t) - c(u(t), x(t))] \, dt subject to stock depletion \dot{x}(t) = -u(t), where x(t) is the resource stock, u(t) is extraction, p is price, and c(u, x) = \gamma u^2 / x is extraction cost.^[35] The Hamiltonian is H = e^{-\rho t} [p u - \gamma u^2 / x] + \lambda (-u), with shadow price \lambda(t).^[35] Maximization gives u(t) = [p x(t) - \lambda(t) x(t)] / (2 \gamma), and the adjoint \dot{\lambda} = \rho \lambda - \partial H / \partial x simplifies under optimal conditions to \dot{\lambda} = \rho \lambda, implying \lambda(t) = \lambda(0) e^{\rho t}.^[35] This Hotelling rule states that the shadow price (resource rent) grows at the discount rate \rho, reflecting the increasing scarcity value; for exhaustible resources, extraction often follows bang-bang patterns (full or zero) until depletion, contrasting with smoother paths for renewables.^[35] The shadow price \lambda(t) thus measures the present value of foregone future rents from current extraction.^[35] These applications highlight key differences from general control problems: states represent economic stocks like capital or resources, controls denote policy choices such as investment or extraction rates, and problems typically span infinite horizons with exponential discounting to model perpetual scarcity.^[33] Maximizing the Hamiltonian equates to profit or utility optimization by equating marginal current benefits to shadow-adjusted future costs.^[33] In practice, PMP conditions inform dynamic programming approximations for empirical policy analysis, such as evaluating fiscal rules or environmental regulations in growth models.^[36]

Extensions

Discrete-time formulations

In discrete-time optimal control problems, the goal is to minimize a total cost functional of the form \sum_{k=0}^{N-1} \phi(x_k, u_k, k) + \psi(x_N), where x_k \in \mathbb{R}^n denotes the state at time step k, u_k \in U \subseteq \mathbb{R}^m is the control input constrained to a compact set U, and the state evolves according to the difference equation x_{k+1} = f(x_k, u_k, k) with initial condition x_0 fixed and terminal state x_N free or constrained, assuming f and \phi are continuously differentiable and U is convex.^[37]^[28] This setup arises naturally in sampled-data systems where the time step h=1 without loss of generality, by rescaling time. The discrete-time analogue of Pontryagin's maximum principle introduces a Hamiltonian function at each step k defined as H_k(x, \psi_{k+1}, u) = \psi_{k+1}^\top f(x, u, k) - \phi(x, u, k), where \psi_{k+1} \in \mathbb{R}^n is the adjoint variable. For an optimal trajectory (x_k^*, u_k^*), the necessary condition requires that the optimal control maximizes the Hamiltonian pointwise: H_k(x_k^*, \psi_{k+1}^*, u) \leq H_k(x_k^*, \psi_{k+1}^*, u_k^*) for all u \in U and k = 0, \dots, N-1, with the adjoint initialized at the terminal time by the transversality condition. For a free terminal state, \psi_N^* = \nabla_{x_N} \psi(x_N^*); for a constrained terminal state x_N \in X_N, \psi_N^* \in \nabla_{x_N} \psi(x_N^*) + N_{X_N}(x_N^*), where N_{X_N} denotes the normal cone to X_N at x_N^*.^[37]^[28] If U is open or the control is unconstrained, this reduces to the first-order stationarity \nabla_u H_k(x_k^*, \psi_{k+1}^*, u_k^*) = 0.^[38] The adjoint variables satisfy a backward recursion \psi_k^* = \frac{\partial H_k}{\partial x}(x_k^*, \psi_{k+1}^*, u_k^*) = \nabla_x f(x_k^*, u_k^*, k)^\top \psi_{k+1}^* - \nabla_x \phi(x_k^*, u_k^*, k), propagating from the terminal condition to \psi_0^*.^[37]^[28] Together with the forward state dynamics and maximization condition, these form a two-point boundary-value problem solvable via shooting methods or numerical continuation.^[38] As the time step h \to 0 in an Euler discretization of a continuous-time problem, the discrete-time conditions converge to the continuous Pontryagin maximum principle, with error bounds of order O(h) under Lipschitz assumptions on the dynamics and cost, often improved to higher order using symplectic discretizations.^[39] This convergence justifies using discrete formulations for numerical approximation of continuous systems, particularly in indirect methods. Applications of the discrete-time principle include digital control systems, where it provides necessary conditions for sampled-data feedback design, and dynamic games, such as nonzero-sum differential games discretized over finite horizons to derive Nash equilibria via coupled Hamiltonian maximizations.^[40] For mixed-integer problems, where controls include discrete decisions, the principle applies via convex relaxation of the control set U, embedding integer constraints in a continuous optimization framework before branch-and-bound refinement.^[41]^[42]

Stochastic variants

Stochastic variants of Pontryagin's maximum principle extend the deterministic framework to controlled systems governed by stochastic differential equations (SDEs), addressing uncertainty through Brownian motion or other noise sources. Consider an optimal control problem minimizing the expected cost \mathbb{E}\left[\int_0^T \phi(t, X_t, u_t) \, dt + \psi(X_T)\right], where the state process X satisfies the forward SDE dX_t = f(t, X_t, u_t) \, dt + \sigma(t, X_t, u_t) \, dW_t, with W a standard Brownian motion, u an adapted control process taking values in a measurable set U, and \phi, \psi the running and terminal cost functions, respectively. This setup assumes standard Itô calculus conditions, such as Lipschitz continuity and linear growth on f and \sigma to ensure existence and uniqueness of solutions.^[43] The stochastic Hamiltonian is defined as H(t, x, u, p, [q](/page/Q)) = p \cdot f(t, x, u) + [q](/page/Q) \cdot \sigma(t, x, u) - \phi(t, x, u), where p and q are the adjoint processes. Unlike the deterministic case, the maximum condition requires that the optimal control u_t^* maximizes the conditional expectation \mathbb{E}[H(t, X_t, u, p_t, q_t) \mid \mathcal{F}_t] over u \in U almost surely, reflecting the role of filtration \mathcal{F}_t generated by the noise. This formulation accounts for the diffusion term via the adjoint q, which captures the sensitivity to stochastic perturbations.^[43]^[44] The adjoint processes satisfy a backward stochastic differential equation (BSDE): -dp_t = \frac{\partial H}{\partial x}(t, X_t, u_t^*, p_t, q_t) \, dt + q_t \, dW_t, with terminal condition p_T = \frac{\partial \psi}{\partial x}(X_T). This BSDE structure, introduced in the context of stochastic control, ensures the adjoint evolves backward from the terminal time, incorporating both drift from the Hamiltonian and martingale terms from the diffusion. Seminal results establishing necessary conditions for optimality, including this BSDE adjoint and the conditional maximization, were developed by Bismut in the late 1970s for convex problems and generalized by Peng in 1990 to non-convex control domains using spike variations and duality arguments.^[44]^[43] Verification theorems for sufficient conditions often rely on the Bismut-Peng decomposition, which expresses the cost functional via martingale representations, or Girsanov's change of measure to transform the problem into a deterministic-like verification under an equivalent martingale measure. These methods confirm that pairs (u^*, p^*, q^*) satisfying the stochastic Hamiltonian system yield global optima under convexity of the Hamiltonian in u. Necessary conditions are derived probabilistically via needle variations, adapting the deterministic spike method to account for the law of iterated logarithm or local time properties of Brownian paths.^[45]^[43] Applications include portfolio optimization, where the principle determines optimal investment strategies in risky assets to minimize expected shortfall or maximize utility under stochastic returns, as in extensions of the Merton problem. In risk management, it aids in controlling stochastic processes for Value-at-Risk minimization. The principle also reveals a duality with the Hamilton-Jacobi-Bellman (HJB) equation in stochastic control, where PMP provides pointwise necessary conditions complementing the viscosity solution framework of HJB for global verification.^[46]^[47] Challenges arise from the inherent non-uniqueness of optimal controls due to noise, as multiple u may achieve the same conditional maximum if the Hamiltonian is flat in certain directions; this contrasts with deterministic uniqueness under strict convexity. The theory requires stringent Itô integrability assumptions, and extensions to jumps or partial observations demand additional filtration enlargements.^[43] Recent developments (2020–2025) include probabilistic extensions of the principle for model-based reinforcement learning by minimizing the mean Hamiltonian under epistemic uncertainty, applications of discrete-time variants to training convolutional neural networks via forward-backward sweeps, and generalizations to interval-valued controls and Volterra integral equations.^[48]^[49]^[50]

References

[1]
Pontryagins Maximum Principle - an overview | ScienceDirect Topics
The Pontryagin Maximum Principle (PMP) is defined as a method in optimal control theory that provides necessary conditions for optimality in control problems, ...
[2]
History of the Discovery of the Pontryagin Maximum Principle
Jun 6, 2019 · We briefly describe the history of the discovery of the Pontryagin maximum principle in optimal control theory in the mid-1950s.
[3]
Mathematical Theory of Optimal Processes - 1st Edition - L.S. Pontryag
The fourth and final volume in this comprehensive set presents the maximum principle as a wide ranging solution to nonclassical, variational problems.
[4]
Pontryagin Maximum Principle -- from Wolfram MathWorld
The function H(psi(t),x(t),u) attains its maximum at the point u=u(t) almost everywhere in the interval t_0<=t<=t_1.
[5]
The Pontryagin Maximum Principle and Transversality Conditions ...
This paper gives a survey of the various forms of Pontryagin's maximum principle for optimal control problems with state variable inequality constraints. The ...
[6]
Pontriagin's maximum principle and the principle of optimality
This paper demonstrates that Pontriagin's Maximum Principle may be derived from the principle of Optimality. It considers a control system described by ẋ ...
[7]
Introduction to the Pontryagin Maximum Principle for Quantum ...
Sep 14, 2021 · This tutorial describes modern aspects of optimal control theory, with a particular focus on the Pontryagin maximum principle.
[8]
The Cold War and the Maximum Principle of Optimal Control
Russian mathematicians around Pontryagin in the Steklov Institute got, with the maximum princi-ple, new results beyond former frontiers while the Americans ...Missing: Soviet | Show results with:Soviet
[9]
[PDF] The Maximum Principle of optimal control - Biblioteka Nauki
Abstract: On the occasion of (more than) 50 years of optimal control theory, this paper deals with the development of the maxi- mum principle of optimal ...
[10]
[PDF] Extremal Trajectories for Bounded Velocity Differential Drive Robots
This paper applies Pontryagin's Maximum Principle to the time optimal control of differential drive mobile robots with velocity bounds.
[11]
[PDF] Applications of Pontryagin's Maximum Principle to Economics
Mathematical emphasis on problems where application of the Maximum Principle requires the study of dynamical systems with nonunique flows. Lecture II: The two- ...
[12]
[PDF] The Maximum Principle of Pontryagin in control and in optimal control
May 23, 2006 · The way to think of the Euler–Lagrange equations is as analogous to the condition that the derivative of a function at a minimum must be ...
[13]
Steklov Mathematical Institute
And already in 1961, Pontryagin, Boltyanskii, Gamkrelidze, and Mishchenko wrote the monograph Mathematical Theory of Optimal Processes, for which the entire ...
[14]
Discovery of the Maximum Principle - Journal of Dynamical and Control Systems
### Summary of Pontryagin's Maximum Principle Discovery and Development
[15]
The Mathematical Theory of Optimal Processes - Google Books
The Mathematical Theory of Optimal Processes. Front Cover. Lev Semenovich ... Interscience Publishers, 1962. Original from, the University of Michigan.Missing: publication | Show results with:publication
[16]
[PDF] cvoc.pdf - Daniel Liberzon
Aug 9, 2011 · 3.3 Optimal control problem formulation and assumptions . ... Note that this is not an optimal control problem because we are not asking to ...
[17]
[PDF] An Introduction to Mathematical Optimal Control Theory Spring ...
This chapter explains the basic theory, works out some examples, and discusses connections with the Pontryagin Maximum. Principle. • Chapter 6: Game theory. We ...Missing: non- | Show results with:non-
[18]
[PDF] Elements of Optimal Control Theory Pontryagin's Principle - UCCS
It can be formulated as a Pontryagin Maximum Principle problem as follows. Write the second order ODE for x = x(t) as a system of two equations for two state ...
[19]
[PDF] The Pontryagin Maximum Principle M403 Lecture Notes by Philip D ...
In all cases, the adjoint equation (a), the state equation (b), and the maximum condition (c) must hold. The difference usually centres on the form of the ...Missing: assumptions | Show results with:assumptions
[20]
[PDF] 16 Pontryagin's maximum principle
. Define for λ ∈ Rd the Hamiltonian. H(t, x, u, λ) = λT b(t, x, u) − c(t, x, u). Pontryagin's maximum principle states that, if (xt,ut)t妻τ is optimal ...
[21]
Hamiltonian Function - an overview | ScienceDirect Topics
... momentum,” and H is called the Hamiltonian function. For many situations, the Hamiltonian H is the total energy of the system. ... cost of completing the ...
[22]
[PDF] contact geometry of the pontryagin maximum principle
It is well known that a necessary condition for optimality of the Pontryagin maximum principle may be interpreted as a Hamiltonian system, and so its geometric ...Missing: source | Show results with:source<|control11|><|separator|>
[23]
[PDF] Optimal Control Theory - Module 3 - Maximum Principle
This is, therefore, a fixed-time optimal control problem in which the terminal point is free. ... Pontryagin's maximum Principle (PMP) is a necessary ...
[24]
Mathematical Theory of Optimal Processes - Google Books
Mar 6, 1987 · The fourth and final volume in this comprehensive set presents the maximum principle as a wide ranging solution to nonclassical, variational problems.
[25]
[PDF] 13 Pontryagin's Maximum Principle - 13.1 Heuristic derivation
Pontryagin's maximum principle (PMP) states a necessary condition that must hold on an optimal trajectory. It is a calculation for a fixed initial value of the ...
[26]
[PDF] Method for Solving Bang-Bang and Singular Optimal Control ... - arXiv
Jan 27, 2022 · If the Hamiltonian is affine in control, the first derivatives are computed and represent the switching function of the system. If the ...
[27]
[PDF] Optimal Control Lectures 25-27: Maximum Principles - EPFL
Normal case: λ0(t) > 0. ▻ λ0,λ1,...,λnx defined up to a constant only. ▻ Need to fix λ0, e.g., λ0(t)=1,∀t. Abnormal case: λ0(t)=0. ▻ λ0,λ1,...,λnx uniquely ...
[28]
[PDF] Pontryagins Maximum Principle in a Nutshell - syscop
Summary. ▷ Pontryagin's Maximum Principle can be derived by in two ways (HJB/Calc. of Variations). ▷ Controls must be explicitly derived. On singular arcs, need.
[29]
Pontryagin maximum principle for state constrained optimal sampled ...
Roughly speaking, the PMP ensures the existence of a costate vector (also called adjoint vector) satisfying some terminal conditions (called transversality ...
[30]
On the proof of Pontryagin's maximum principle by means of needle ...
Dec 7, 2014 · We propose a proof of the maximum principle for the general Pontryagin type optimal control problem, based on packages of needle variations.Missing: method | Show results with:method
[31]
[PDF] Optimal Control
Solve the optimal control problem min Z. 1. 0 u4dt + x(1). ˙x = −x + u x(0) = 0. 3. Page 4. Goddard's Rocket Problem revisited. How to send a rocket as high up ...
[32]
[PDF] Optimization-Based Control
Use the maximum principle to show that any optimal trajectory consists of segments in which the robot is traveling at maximum velocity in either the forward or ...
[33]
None
### Summary of Economic Interpretations of Pontryagin's Maximum Principle in Optimal Economic Growth
[34]
None
### Summary of Pontryagin's Maximum Principle in the Ramsey Model
[35]
None
### Summary of Hotelling's Rule Derivation Using Pontryagin's Maximum Principle
[36]
Dynamic Programming for Optimal Control Problems in Economics
Aug 29, 2013 · ... Dynamic Programming and the Pontryagin Maximum Principle. The main examples motivating such theory usually came from physics and engineering ...
[37]
[PDF] Pontryagin's maximum principle and indirect methods
May 3, 2023 · Example: Minimum fuel for a control-affine system. The Hamiltonian is ... which for η = 1 is an example of “bang-off-bang” control. 25 ...
[38]
The Discrete-Time Geometric Maximum Principle | SIAM Journal on ...
We establish a variety of results extending the well-known Pontryagin maximum principle of optimal control to discrete-time optimal control problems posed ...
[39]
Convergence rates of symplectic Pontryagin approximations in ...
The maximum principle, apart from its sound mathematical basis, also has the advantage that it leads to a system of ordinary differential equations, whereas ...
[40]
(PDF) The maximum principle for discrete-time control systems and ...
The maximum principle for discrete-time control systems and applications to dynamic games. February 2019; Journal of Mathematical Analysis and Applications ...Missing: digital | Show results with:digital
[41]
Multiphase mixed-integer nonlinear optimal control of hybrid electric ...
Indirect methods stem from the application of Pontryagin's maximum principle, which provides the necessary conditions for optimality. The solution of the ...
[42]
[PDF] POD-Based Mixed-Integer Optimal Control of Convection-Diffusion ...
We use relax- ation techniques for the optimal control problem which have been already used in mixed-integer optimal control problems for ordinary differential ...
[43]
A General Stochastic Maximum Principle for Optimal Control Problems
The maximum principle for nonlinear stochastic optimal control problems in the general case is proved. The control domain need not be convex.
[44]
[PDF] an introductory approach to duality in optimal stochastic control
The maximum principle. Under very stringent convexity assumptions the generalized maximum principle for Ito equations can be written in the following way, as ...
[45]
Linear Quadratic Optimal Stochastic Control with Random Coefficients
By the dynamic programming principle, Doob--Meyer decomposition, and inverse flow technique, the existence and uniqueness of the solution for the BSREJ are ...
[46]
Maximum Principle for a Stochastic Optimal Control Problem and ...
A Pontryagin local maximum principle is obtained by using classical variational methods. We apply the result to make optimal portfolio and consumption decisions ...
[47]
On some recent aspects of stochastic control and their applications
We discuss the two main historical approaches, Bell- man's optimality principle and Pontryagin's maximum principle, and their modern exposition with viscosity ...