Conditioning of integrable determinantal point processes
(2023)
Files
ThesisGabrielGlesnerfinal.pdf
Open Access - Adobe PDF
- 2.24 MB
Details
- Authors
- Supervisors
- Claeys, Tom
- Abstract
- In this thesis, we emphasise the role of a particular 'integrable' structure in the study of determinantal point processes, namely kernels of integrable form. We introduce an observation procedure of a point process involving the notions of marking and thinning, and then condition that point process on the resulting observation. This is shown to preserve important subclasses of point processes: -determinantal point processes, i.e. point processes whose correlation functions can be written as determinants of a so-called correlation kernel: the transformation makes use of Fredholm determinants and minors; -determinantal point processes induced by kernels of projections; -determinantal point processes with kernels of integrable form: the transformation can be characterised by the solution to a Riemann-Hilbert problem (RHP). We then study Jánossy densities associated to observations of the Airy point process, which are, informally speaking, likelihoods of observations. We prove that we can apply the conditioning transformation to this process and that the kernel of the conditioned point process can be characterized by an RHP. Moreover: -that RHP’s solution satisfies linear differential equations and undergoes isomonodromic deformations; -from the compatibility conditions of these differential equations we obtain an isospectral deformation of the Stark operator whose potential is expressible in terms of the Jánossy density and satisfies the cylindrical Korteweg-de Vries equation; -we prove a trace formula implying that the Stark operator's eigenfunctions are solutions to an infinitely coupled version of the Painlevé II differential equation.
- Affiliations
Citations
Glesner, G. (2023). Conditioning of integrable determinantal point processes. https://hdl.handle.net/2078.5/102290