Precise solutions of the one-dimensional Monge–Kantorovich problem

A. Yu. Plakhov

Sbornik. Mathematics, Tome 195 (2004) no. 9, pp. 1291-1307

Résumé

The Monge–Kantorovich problem on finding a measure realizing the transportation of mass from $\mathbb R$ to $\mathbb R$ at minimum cost is considered. The initial and resulting distributions of mass are assumed to be the same and the cost of the transportation of the unit mass from a point $x$ to $y$ is expressed by an odd function $f(x+y)$ that is strictly concave on $\mathbb R_+$. It is shown that under certain assumptions about the distribution of the mass the optimal measure belongs to a certain family of measures depending on countably many parameters. This family is explicitly described: it depends only on the distribution of the mass, but not on $f$. Under an additional constraint on the distribution of the mass the number of the parameters is finite and the problem reduces to the minimization of a function of several variables. Examples of various distributions of the mass are considered.

Bibliographie

[1] Rachev S. T., Rüschendorf L., Mass transportation problems. V. 1: Theory, Springer, New York, 1998 | MR | Zbl

[2] McCann R. J., “Exact solutions to the transportation problem on the line”, Proc. Roy. Soc. Lond. A. Math. Phys. Eng. Sci., 455 (1999), 1341–1380 | DOI | MR | Zbl

[3] Uckelmann L., “Optimal couplings between one-dimensional distributions”, Distributions with given marginals and moment problems, eds. V. Benes, J. Stepan, Kluwer, Dordrecht, 1997, 275–281 | MR | Zbl

[4] Levin V. L., “Reshenie zadach Monzha i Monzha–Kantorovicha. Teoriya i primery”, Dokl. RAN, 388:1 (2003), 7–10 | MR | Zbl