An explicit construction of heat kernels and Green’s functions in measure spaces

Palle Jorgensen and Jay Jorgenson and Lejla Smajlović · 2025

Read An explicit construction of heat kernels and Green’s functions in measure spaces by Palle Jorgensen and Jay Jorgenson and Lejla Smajlović (2025) 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-2512-24348.txt.

Introduction

Palle Jorgensen, Jay Jorgenson, and Lejla Smajlovic construct heat kernels and Green's functions in measure spaces. Heat kernels describe how heat spreads from a point source. Green's functions solve the associated steady problems. The setting is broad: a locally compact Hausdorff space with Borel structure and extra measure data. Function spaces over it include the Lebesgue and Hilbert cases.

The construction is explicit. A Neumann series builds the kernel term by term. The input is a parametrix, a small-time approximation to the kernel. The method adapts classical work of Minakshisundaram and Pleijel. New instances follow from the general machine. Normalized Laplacians on finite and infinite graphs are covered. Hilbert spaces with reproducing kernels are covered as well. Analysts gain a uniform route to kernels on spaces far from smooth manifolds.

Catalog record

CREATORPalle Jorgensen and Jay Jorgenson and Lejla Smajlović
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE105 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-24348

Browse the full library · About the Tower of Babel

Full text

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

An explicit construction of heat kernels and Green’s functions in measure spaces Palle Jorgensen and Jay Jorgenson and Lejla Smajlović Abstract . We explicitly construct a heat kernel as a Neumann series for certain function spaces, such as L 1 L^{1} , L 2 L^{2} , and Hilbert spaces, associated to a locally compact Hausdorff space 𝔛 \mathfrak{X} with Borel σ \sigma -algebra ℬ \mathcal{B} , and endowed with additional measure-theoretic data. Our approach is an adaptation of classical work due to Minakshishundaram and Pleijel, and it requires as input a parametrix or small time approximation to the heat kernel. The methodology developed in this article applies to yield new instances of heat kernel constructions, including normalized Laplacians on finite and infinite graphs as well as Hilbert spaces with reproducing kernels. † † Keywords: heat kernels, Neumann series, generalized time convolution, parametrix † † AMS Classification numbers. 35K08, 58J35, 44A35, 31C20, 46E30, 46E22, 47A10 Dedicated to the memory of Professor Robert Strichartz. 1. Introduction 1.1. Imagining the heat kernel in the physical world From its inception, a certain class of heat kernels K(x,y;t) consists of mathematical objects which are rooted in the physical world. For example, one can imagine introducing a unit burst of heat at a point x in a domain M at time zero, after which one then can measure the proportion of heat which has diffused to the point y after time t . This amount of heat is denoted by K(x,y;t) . Equivalently, one can envision a particle which undertakes a random walk on M beginning at x at time zero, so then K(x,y;t) is interpreted as the probability that the particle is at the point y at time t . Without exaggeration, there are hundreds of mathematics articles which study the heat kernel in various settings for M , such as Euclidean domains, Riemannian manifolds, projective Kähler varieties, symmetric spaces, and so on, and with some articles considering different possibilities for time, including positive real time, positive integer time, or arbitrary time-scales; see [ 8 ] . In our opinion, it is valid to keep the physical or probabilistic language when describing properties of a heat kernel K(x,y;t) in the case when M is modeled locally on a Euclidean domain since this means that in one way or another one can draw M , hence one can imagine diffusion or a random walk. 1.2. A general point of view Beyond a physical or probabilistic setting, there is an extensive body of literature which considers heat kernels more generally, such as when M is replaced by a locally compact Hausdorff space \mathfrak{X} with a Borel \sigma -algebra \mathcal{B} of subsets and possessing a \sigma -finite measure \lambda on (\mathfrak{X},\mathcal{B}) . We refer the reader to the particularly eloquent article [ 13 ] where the author defines and studies a heat kernel in a metric measure space; see, specifically, Definition 2.1 on page 146 of [ 13 ] . With various assumptions, it is shown in [ 13 ] that a heat kernel can give rise to a Markov process for which the heat kernel is the associated transition density; see loc. cit., page 147. In this regard, the heat kernel retains its connection to probability theory. We also note that the heat kernel in these instances has an associated semigroup of operators, and a heat operator is obtained by considering the infinitesimal generator of this semigroup. In essence, the infinitesimal generator is a type of Laplacian operator. Our main goal in the present article is to explicitly construct a heat kernel as an integral kernel in certain measure space settings, similar to those considered in [ 13 ] , with positive real time; see Theorems 6.3 , 6.6 and 6.9 . In effect, we assume that, somehow, we have obtained a parametrix or small time approximation for the heat kernel as an integral kernel. Given this approximation, the heat kernel itself is computed using a convergent Neumann series which is defined using the parametrix. As is commonplace, the parametrix itself has reasonably few requirements, so in many instances our construction yields an easily attainable result. The parametrix approach we present is modeled on the classical method from the Riemannian setting as given the seminal paper [ 33 ] , which itself attributes some of the main ideas to Carleman; see loc. cit., page 243. In the recent articles [ 10 ] and [ 25 ] the authors developed the parametrix approach for constructing heat kernels in the setting of finite and certain infinite graphs, respectively. Again, there is considerable flexibility in choosing a parametrix. For example, it is shown in [ 10 ] and [ 25 ] that different choices of a parametrix lead to different identities, including expressions for the classical I -Bessel function. With that in mind, one can say that the parametrix construction of the heat kernel has been shown to be quite robust. There are numerous studies in the literature which study heat kernels in metric measure spaces; see, for example [ 13 ] and references therein. As far as we can see, the present article is the first which explores an explicit construction of a heat kernel starting with a parametrix in such settings. 1.3. Related studies Our study is motivated with a view toward analysis on infinite graphs in a general framework, namely: (i) Generalized integral kernels and transition kernels, with an emphasis on generalized graph Laplacians, their induced Green’s functions and potential theoretic properties; (ii) the relevant potential theoretic constructions which we shall make use of; and (iii) families of associated stochastic processes, directly related to (i) and (ii). Our discussion below of these themes will be supplemented by background references. While the general literature is vast, we, of course, will focus on the details that are tailored to our needs. These ideas are connected to recent papers by P. Jorgensen (first named author) and his (other) co-authors, especially the following: With S. Bezuglyi for (i) (see [ 6 ] and references therein), with E. Pearse for (ii) (see [ 15 ] and references therein) as well as [ 23 ] , and with F. Tian for (iii) (see [ 19 ] and references therein). While there is a rich variety of frameworks which involve incarnations of Green’s functions and associated diffusions, each case and each application entails unique choices of both function spaces and associated Laplacians. An interesting new development which has served as motivation for our present results is an adaptation of graph Laplacians to a continuous Borel framework. In the various settings one then arrives at a natural choice of a measure \mu and an associated L^{p} space where the Laplacian has dense domain. This is the framework we follow. We are then able to obtain global results where the time variable enters at the outset and, therefore, yields a more direct solution to the particular diffusion equations for the setting at hand. We refer to the following earlier papers which deal with a variety of settings of generalized Laplacians, heat kernels, Green’s functions, boundary spaces, and potential theory: [ 1 , 4 , 7 , 18 , 17 , 20 , 35 ] . This

Read the complete text