Linear quadratic optimal transport and interpolation inequalities

Luca Rizzi, Alec J. A. Schiavoni Piazza · 2026

Read Linear quadratic optimal transport and interpolation inequalities by Luca Rizzi, Alec J. A. Schiavoni Piazza (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-23608.txt.

Introduction

Rizzi and Piazza study the optimal transport problem within linear square optimal control systems. They establish soundness of the Monge problem and analyze regularity of the final optimal transport map, extending results for costs at or above zero. They also study displacement interpolation of measures and derive general interpolation inequalities for entropy functionals. The analysis is motivated by these systems as natural model spaces for comparison theory in sub-Riemannian geometry.

Catalog record

CREATORLuca Rizzi, Alec J. A. Schiavoni Piazza
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE136987
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.23608

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Linear quadratic optimal transport and interpolation inequalities Thanks: This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement GEOSUB, No. 945655). The authors also acknowledge the INdAM support. Luca Rizzi, Alec J. A. Schiavoni Piazza Abstract . This paper investigates the optimal transport problem within the framework of Linear Quadratic optimal control systems. We establish the well-posedness of the Monge problem and analyze the regularity of the resulting optimal transport map, extending the results obtained in [ 28 ] for non-negative costs. Furthermore, we study the displacement interpolation of measures and derive general interpolation inequalities for entropy functionals. Our analysis is motivated by the role of these systems as natural model spaces for comparison theory in sub-Riemannian geometry. 1. Introduction The theory of optimal transport, introduced by Monge in the 18^{th} century [ 35 ] , in recent decades found new links with differential geometry and dynamical systems. One of the most relevant ones is the Lott-Sturm-Villani synthetic formulation of Ricci curvature lower bounds for metric measure spaces [ 31 , 38 , 37 ] , realized via entropy interpolation inequalities along Wasserstein geodesics. Recently, Barilari-Rizzi-Mondino [ 11 , 12 , 13 , 14 ] generalized these to the sub-Riemannian setting, building on previous work for Heisenberg and corank 1 Carnot groups [ 9 , 10 ] . In this context, Linear Quadratic (LQ) optimal control problems [ 4 , 29 ] emerged as the natural “constant curvature” models. The study of optimal transport in this framework was initiated in [ 28 ] , specifically for -Q\geq 0 , corresponding to “negative curvature” and costs bounded from below. The purpose of this work is to provide a comprehensive study of the problem: we establish the well-posedness of the Monge problem for general LQ costs, analyze displacement interpolations, and derive entropy inequalities, aiming to provide a reference for future research. 1.1. LQ problems Let n,m\in\mathbb{N} , m\leq n , and let A,Q\in\mathbb{R}^{n\times n} and B\in\mathbb{R}^{n\times m} be matrices, with Q=Q^{*} . The LQ optimal control problem (A,B,Q,T) consists in minimizing the functional (2) \mathcal{J}(u)=\int_{0}^{T}\left(|u|^{2}-x^{*}Qx\right)dt among all trajectories x:[0,T]\to\mathbb{R}^{n} with fixed endpoints satisfying (3) \dot{x}=Ax+Bu, for some u\in L^{2}([0,T],\mathbb{R}^{m}) . Optimal trajectories are projections on \mathbb{R}^{n} of integral curves of the quadratic Hamiltonian \mathcal{H}:T^{*}\mathbb{R}^{n}(\simeq\mathbb{R}^{2n})\to\mathbb{R} , given by (4) \mathcal{H}(p,x)=\frac{1}{2}\left(p^{*}BB^{*}p+2p^{*}Ax+x^{*}Qx\right). A time t>0 is called a conjugate time if there exists an integral curve of ( 4 ) whose projection x(\cdot) onto \mathbb{R}^{n} satisfies x(0)=x(t)=0 . LQ problems exhibit a dichotomy: they either possess no conjugate times or an infinite, discrete set of them that does not accumulate at t=0 [ 3 ] . Let t^{*}>0 denote the first conjugate time. We observe that if -Q\geq 0 (the case treated in [ 28 ] ), then t^{*}=+\infty and the cost functional ( 2 ) is non-negative. If T<t^{*} , for any pair of endpoints x,y\in\mathbb{R}^{n} , there is a unique optimal trajectory with these endpoints minimizing ( 2 ). Such minimum, called LQ cost function, is an homogeneous polynomial of degree 2 , of the form (5) c(x,y)=\frac{1}{2}\left(x^{*}Cx+2x^{*}Dy+y^{*}Ey\right),\qquad\forall x,y\in\mathbb{R}^{n}, with C,D,E\in\mathbb{R}^{n\times n} and C,E symmetric. The matrices C,D and E are explicitly computable in terms of the Hamiltonian flow of the system (and of the backwards one), and depend smoothly on T\in(0,t^{*}) . Note that, unless -Q\geq 0 , the LQ cost function is not bounded from below. 1.2. Optimal transport Given a cost function c:\mathcal{X}\times\mathcal{Y}\to\mathbb{R} between Polish spaces and probability measures \mu\in P(\mathcal{X}) , \nu\in P(\mathcal{Y}) , the Monge optimal transport problem [ 35 ] corresponds to minimizing the quantity (6) \int_{\mathcal{X}}c(x,T(x))d\mu(x), among all transport maps, i.e. measurable maps T:\mathcal{X}\to\mathcal{Y} , that satisfy T_{\sharp}\mu=\nu . A relaxed version of it, the Kantorovich formulation [ 30 ] , is the problem of minimizing (7) \int_{\mathcal{X}\times\mathcal{Y}}c(x,y)d\alpha(x,y), among all transport plans, i.e. probability measures \alpha\in P(\mathcal{X}\times\mathcal{Y}) , such that (\pi_{\mathcal{X}})_{\sharp}\alpha=\mu and (\pi_{\mathcal{Y}})_{\sharp}\alpha=\nu . A fundational result in the field was the one of Brenier [ 16 ] , for the case of the Euclidean squared-distance cost, proving existence and uniquness of solutions to the Monge problem, for \mu\ll\mathscr{L}^{n} . Since then, many authors contributed to the study of existence and regularity of optimal plans and maps (see [ 39 , 6 ] and references therein). Under suitable assumptions on the measures, existence and uniqueness of optimal transport maps is known for: • compact Riemannian manifolds without boundary and squared-distance cost [ 34 ] , • infinite-dimensional Hilbert spaces and squared-distance cost [ 5 ] , • Heisenberg groups [ 7 ] and a class of sub-Riemannian manifolds [ 25 ] , with cost equal to the Carnot-Carathéodory squared-distance, • cost functions induced by Tonelli Lagrangians on compact manifolds [ 15 ] , and the generalization to non-compact manifolds [ 23 ] , • cost functions induced by sub-Riemannian-type Lagrangians bounded from below [ 2 ] , • metric measure spaces that satisfy suitable MCP-like and non-branching assumptions, with cost equal to a convex non-decreasing function of the distance [ 19 , 20 ] , • cost functions of LQ optimal control problems with -Q\geq 0 [ 28 ] . We first extend the results in [ 28 ] to any LQ problem, and to a broader class of measures, possibly with non-compact support, resembling the sharp version of Brenier theorem in [ 27 ] . In addition, we characterize optimal maps in terms of the LQ exponential map applied to the gradient of a semi-convex function. The next result corresponds to Theorem 3.12 . Theorem 1.1 (Well-posedness) . Let (A,B,Q,T) be a LQ optimal control problem satisfying the Kalman condition, with T<t^{*} . Let \mu,\nu\in P_{2}(\mathbb{R}^{n}) , and assume that \mu does not give mass to any c-c -hypersurface. Then, there exists a unique optimal transport plan \alpha\in P_{2}(\mathbb{R}^{n}\times\mathbb{R}^{n}) between \mu and \nu for the LQ optimal transport problem, induced by a map T:\mathbb{R}^{n}\to\mathbb{R}^{n} . In addition, any Kantorovich potential \psi associated with \alpha is \mu -almost everywhere differentiable and the optimal transport map T satisfies (8) T(x)=\exp_{x,T}(\nabla\psi(x)),\qquad\qquad\mu\text{-a.e. }x\in\mathbb{R}^{n}, where \exp_{x,\tau}:T_{x}\mathbb{R}^{n}(\cong\mathbb{R}^{n})\to\mathbb{R}^{n} , for \tau\in[0,T] , denotes the LQ exponential map. If \mu\ll\mathscr{L}^{n} , then \psi is \mu -almost everywhere twice differentiable. In particular, T is \mu -almost everywhere differentiable. Remark 1.2 . Due to the quadratic nature of the LQ cost function, it is natural to choose \mu,\nu\in P_{2}(\mathbb{R}^{n}) , the Euclidean 2 -Wasserstein space. Remark 1.3 . Since c-c -hypersurfaces are (n-1) -rectifiable, Theorem 1.1 applies to all \mu\in P_{2}(\mathbb{R}^{n}) that do not give mass to (n-1) -rectifiable sets. Following [ 28 ] , and relying on the regularity theory for the Euclidean Monge-Ampère equation [ 17 ] , we obtain the following result, which corresponds to Theorem 3.14 . Theorem 1.4 (Continuity of the transport map) . Consider two compactly supported measures \mu,\nu\in P_{2}(\mathbb{R}^{n}) , with \mu,\nu\ll\mathscr{L}^{n} . Assume, in addition, that \mu=f\mathscr{L}^{n} , \nu=g\mathscr{L}^{n} with f and g bounded from above and below on \supp(\mu) and \supp(\nu) , respectively, and \supp(\mu) has connected interior, whilst \supp(\nu) is convex. Then, the unique optimal transport map T:\mathbb{R}^{n}\to\mathbb{R}^{n} between \mu and \nu is \beta -Hölder continuous, for some \beta\in(0,1) . We refer to Remark 3.15 for higher-order regularity results. 1.3. Displacement interpolations

Read the complete text