Solving Mass Transportation Problems Using Liouville Equations

Transporting objects and moving from one place to another occurs in everyday life. We can observe it at all scales: molecules are transported inside cells; rockets transport satellites into Earth’s orbit. This is an age-old activity that has developed its rules and techniques for effectiveness and convenience. Thus, by looking carefully at any transport phenomena, we recognize a methodology. This is true for movement of human beings through different places to perform different tasks, transfer of data, which is the core concept of internet, televisions, radios and radar systems, migration of birds and fishes, the flow of blood in tissues, etc.; all these represent some form of transportation. One could argue that transportation is an expression of life.

Transportation is a complex and precise job and thus it becomes essential to study various methods and strategies of time-effective transportation with minimal effort. This need leads to the concept of optimal transportation that is the focus of the mathematical theory of transport. The main components in optimal transportation include the objects to be transported, which are represented by sets, the dynamics of the process, which is usually given by a time-evolution differential equation and includes a control function whose job is to drive the process to its optimal configuration and a cost functional, which is to be optimized. Such problems were first studied mathematically by A. N. Tolstoi in 1920.

Subsequently, major advances in this field were made by Leonid Kantorovich. One early special class of transport problems is the transport of mass that represents the displacement of a collection of similar objects from an original location to a desired one: the famous Monge-Kantorovich problem. For this purpose, the dynamics of a single particle is extended to the dynamics of multi-particle systems and this is usually done through Liouville equations, which are a set of hyperbolic partial differential equations (PDE). However, although the Liouville equation in space coordinates and its counterpart in phase-space are central in classical continuous mechanics, we notice that optimal transport problems governed by these equations have not been a research focus. Investigation of robust control mechanisms using the Liouville framework appears natural for mass transportation problems.

In a recent work, Nikolay Pogodaev proposed a challenging Liouville optimal control problem where the controller has the purpose to transport an initial probability measure to maximize the measure of a target set at a given final time. This setting accommodates multi-agent control problems as well as the problem of the control of a beam of charged particles, and it can be considered an approximation to the classical mass transportation problem.

Pursuing this challenging work, Alfio Borzì and Souvik Roy from the University of Würzburg developed a new computational framework for solving the Liouville control problems proposed by Pogodaev. Two major breakthroughs of this new research effort are the analysis of high-order conservative and positive preserving approximation schemes for continuity-type equations and the development of a new methodology for the fast solution of Liouville transport problems that are formulated in the framework of the Pontryagin’s maximum principle (PMP).

The major focus of this work was to build a novel iterative scheme for implementing the PMP for an optimal mass transportation problem governed by Liouville PDE. A major advantage of the maximum principle framework for optimal control problems is the non-differentiability with respect to the control function, which helps incorporate a wide class of controllers to be used for real-life mass transportation. Moreover, the concept of maximum principle for PDE control problems is not well-studied in literature. Thus, the scope of this research involves optimization strategies with PMP for various PDE control problems.

With this generic idea at the back of mind, a time-split nonlinear collective update scheme is proposed. This is an iterative scheme which solves the PMP optimality condition pointwise in time which includes a third order conservative numerical scheme for solving the Liouville equation. In fact, the iterative scheme does not change with the choice of discretization schemes for PDE’s, thus enabling it to solve any PDE optimal control problem. Various numerical experiments presented in the paper demonstrated the efficiency and robustness of the proposed optimization framework for mass transportation problems.

These findings are described in the article entitled Numerical Investigation of a Class of Liouville Control Problems, published in the Journal of Scientific Computing. This work was led by Souvik Roy from Universität Würzburg.

About The Author

Souvik Roy

Souvik Roy is an assistant professor at the University of Texas-Arlington in the Department of Mathematics. . His primary area of interest is in the field of inverse problems in fluid flows and medical imaging. For inverse problems in fluid flows, he is interested in determining the flow properties like velocity, vorticity, density, given image data which is advected by the fluid flow field. He studies both the theoretical and computational aspects of such models.

Speak Your Mind!


A Warmer And Wetter Early Mars

Four billion years ago, an interstellar traveler approaching the now-red planet with his spaceship would probably witness beautiful blue skies with fluffy clouds delivering rain and snow, ice-capped mountain ridges, rivers flowing through incised valleys, lakes, and even an ocean covering most of one hemisphere. A pretty much favorable condition for the birth of life […]

Positive Feedback Loop Examples

A positive feedback loop is a system where one variable increases the quality of another variable which in turn increases the quantity/occurrence of the first variable. Positive feedback loops are processes which occur within feedback loops in general, and their conceptual opposite is a negative feedback loop. The mathematical definition of a positive feedback loop […]

Are Our Houses Demanding More From Us?

Winston Churchill once said, “We shape our buildings, and afterward, our buildings shape us.” Buildings grow, adapt, and evolve just like the users that occupy them — in a mutually dependent, co-evolving, and reiterative process. If our houses shape us just as we shape them, what does this say about the escalating energy demands in […]

Decoding Ancient Deep-Earth Geodynamics Using High-Mg Basalts From Central South China

Published by Xuan-Ce Wang The School of Earth Science and Resources, Chang’an University, and Department of Applied Geology, Curtin University These findings are described in the article entitled The 825 Ma Yiyang high–MgO basalts of central South China: Insights from Os–Hf–Nd data, recently published in the journal Chemical Geology (Chemical Geology 502 (2018) 107-121). This work […]

An Improved Method To Remove Debris From Cyst Nematode Egg Suspensions And Computer-Aided Technologies For Egg Counting

Plant-parasitic nematodes infect the roots of plants, causing billions of dollars of crop loss worldwide. One such example is the soybean cyst nematode (SCN), which is the most damaging soybean pathogen in North America. Knowing the population density (number) of nematode eggs in the soil is crucial for the detection, monitoring, and effective management of […]

Using Remote Sensing To Measure Crop Growth And Management

In most East and Southeast Asia countries, mosaic crop production systems usually consist of relatively small land holdings ( < 2 ha), which are managed by individual farmers/families (Fig. 1). Diverse field water and nutrient management practices exist among spatially adjacent fields in main crop plains, due to differences in individual farmers’ preference. Seasonal strengths […]

What Is Black Tourmaline?

Black tourmaline is a particular variant of tourmaline, a crystalline borosilicate mineral. It is thought that the particular hue of black tourmaline is due to traces of titanium embedded in the minerals molecular structure. Black tourmaline, sometimes referred to as schorl,  is one of the most popular semi-precious gemstones in the world and it is commonly […]