B03Numerics of Riemann-Hilbert Problems and Operator Determinants

The Search for the Optimal Contour

Riemann-Hilbert Problems (RHP) are another way of expressing equations satisfying a special property and have some advantages over the traditional forms. Take for example an equation describing the motion of a water wave and its current state: Both the traditional form and the RHP form of the equation enables us to calculate the state of the wave at any point in time. But with the RHP form we can accomplish this without knowing or calculating anything about the state of the wave in between.

Scientific Details+

Though Riemann-Hilbert Problems (RHP) offer some advantages over traditional forms of equations they also introduce new problems. In general, solving a RHP with a computer does not yield good results. As computers can only store finitely many digits of a number, storing a number in a computer always causes a small error. These small errors can add up to a substantial error during a calculation and in the end lead to a result which is far away from the correct solution. In the case of RHPs this happens regularly, but fortunately it is possible to avoid this situation.

A RHP depends on a set of lines in the plane, which is also called a contour. This contour can be modified in specific ways without changing the solution of the RHP. One can think of these modifications as moving the lines around in the plane while honouring some restrictions like no line may cross another line during the movement''. Although the solution of a RHP is not changed by these modifications they do have an effect on the calculation error that occurs when the RHP is solved with a computer. Some modifications will increase the error while others will decrease it. So the mission of this project is to develop an algorithm which is capable of determining for a given starting contour a modification of it which minimizes this calculation error. This modified contour can then be used to actually compute the correct solution of the RHP.

Publications+

Papers
Numerical Methods for the Discrete Map $Z^a$

Authors: Bornemann, F. and Its, A. and Olver, S. and Wechslberger, G.
In Collection: Advances in Discrete Differential Geometry, Springer
Date: 2016
Download:

Singular values and evenness symmetry in random matrix theory

Authors: Bornemann, Folkmar and Forrester, Peter J.
Journal: Forum Math. (ahead of print); 19pp
Date: Oct 2015
DOI: 10.1515/forum-2015-0055
Download:

The Singular Values of the GOE

Authors: Bornemann, Folkmar and Croix, Michael La
Journal: Random Matrices: Theory Appl. 04, 1550009 (2015) (32 pages)
Date: Jun 2015
DOI: 10.1142/S2010326315500094
Download:

A note on the expansion of the smallest eigenvalue distribution of the LUE at the hard edge

Author: Bornemann, Folkmar
Note: To appear in Ann. Appl. Probab.
Date: Mar 2015
Download:

An improved Talbot method for numerical Laplace transform inversion

Authors: Dingfelder, Benedict and Weideman, J.A.C.
Journal: Numerical Algorithms, 68(1):167-183
Date: 2015
DOI: 10.1007/s11075-014-9895-z
Download:

Automatic Deformation of Riemann–Hilbert Problems with Applications to the Painlevé II Transcendents

Authors: Wechslberger, Georg and Bornemann, Folkmar
Journal: Constructive Approximation, pages 1-21
Date: Jun 2013
DOI: 10.1007/s00365-013-9199-x
Download:

Joint distribution of the first and second eigenvalues at the soft edge of unitary ensembles

Authors: Witte, N.S. and Bornemann, F. and Forrester, P.J.
Journal: Nonlinearity, Volume 26, Number 6, pp. 1799-1822
Date: Jun 2013
DOI: 10.1088/0951-7715/26/6/1799
Download:

PhD thesis
Automatic Contour Deformation of Riemann-Hilbert Problems

Author: Wechslberger, G.
Date: Jul 2015
Download:

Prof. Dr. Folkmar Bornemann   +

Projects: B03
University: TU München
E-Mail: bornemann[at]tum.de
Website: http://www-m3.ma.tum.de/Allgemeines/FolkmarBornemann

Dominik Volland   +

Projects: B03
University: TU München
E-Mail: volland[at]ma.tum.de

Dr. Georg Wechslberger   +

Projects: B03
University: TU München
E-Mail: wechslbe[at]ma.tum.de