Solving the Laplace Equation and Applications in Imaging

Arina Oberoi · 2025

Read Solving the Laplace Equation and Applications in Imaging by Arina Oberoi (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-2508-11896.txt.

Introduction

Arina Oberoi surveys ways to solve the Laplace equation and puts them to work on imaging tasks, in a paper from 2025. The equation governs smooth fields like heat and electric potential at rest. Its solutions, called harmonic functions, obey neat rules about averages and extremes. Those rules make them handy for filling and smoothing data.

The paper covers classic hand methods, such as splitting the field into parts and a disk formula. It also covers chance-based methods, where wandering paths stand in for the math. Imaging uses include cutting noise, filling gaps in data, and sharpening detail. The mean value rule ties the theory to each pixel's neighbors. That tie gives a clean frame for building image tools. The survey bridges pure math and daily practice.

Catalog record

CREATORArina Oberoi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-11896

Browse the full library · About the Tower of Babel

Full text

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

Solving the Laplace Equation and Applications in Imaging Arina Oberoi Abstract . This paper examines solutions to the Laplace equation using analytical techniques, including separation of variables and the Poisson integral formula, and probabilistic methods, such as Brownian motion. We address applications to imaging, including noise reduction, data interpolation, and resolution enhancement, as well as discuss theoretical connections between the Laplace equation, specifically concepts such as harmonic functions, the mean value property, and the maximum principle, to provide a mathematical framework for practical implementation in imaging workflows. 1. Solving the Laplace Equation in a Disc The Laplace equation in Cartesian coordinates is \nabla^{2}u=\frac{\partial^{2}u}{\partial x^{2}}+\frac{\partial^{2}u}{\partial y^{2}}=0 where x=r\cos(\theta) and y=r\sin(\theta) . To solve the Laplace equation \nabla^{2}u=0 in a disc, we convert the equation to polar coordinates (r,\theta) \frac{1}{r}\frac{\partial}{\partial r}\left(r\frac{\partial u}{\partial r}\right)+\frac{1}{r^{2}}\frac{\partial^{2}u}{\partial\theta^{2}}=0 Assume the boundary condition u(R,\theta)=f(\theta) on the disc of radius R and apply separation of variables by writing u(r,\theta)=R(r)\Theta(\theta) , which splits the equation into radial and angular parts. Substituting u(r,\theta)=R(r)\Theta(\theta) into the Laplace equation, we have \Theta(\theta)\left[\frac{1}{r}\frac{d}{dr}\left(r\frac{dR}{dr}\right)\right]+R(r)\frac{1}{r^{2}}\left[\frac{d^{2}\Theta}{d\theta^{2}}\right]=0 \frac{r^{2}}{R(r)}\left[\frac{1}{r}\frac{d}{dr}\left(r\frac{dR}{dr}\right)\right]+\frac{1}{\Theta(\theta)}\left[\frac{d^{2}\Theta}{d\theta^{2}}\right]=0 \frac{r}{R(r)}\frac{d}{dr}\left(r\frac{dR}{dr}\right)+\frac{1}{\Theta(\theta)}\frac{d^{2}\Theta}{d\theta^{2}}=0 Since the first term depends only on r and the second term depends only on \theta , and r and \theta are independent variables, each term must be equal to a constant that sums to zero. Therefore, we set \frac{r}{R(r)}\frac{d}{dr}\left(r\frac{dR}{dr}\right)=-\lambda,\quad\frac{1}{\Theta(\theta)}\frac{d^{2}\Theta}{d\theta^{2}}=\lambda where \lambda is the separation constant. To determine the sign and possible values of \lambda , we consider the angular equation \frac{d^{2}\Theta}{d\theta^{2}}+\lambda\Theta=0 \int_{0}^{2\pi}\Theta\frac{d^{2}\Theta}{d\theta^{2}}\,d\theta+\lambda\int_{0}^{2\pi}\Theta^{2}\,d\theta=0 \int_{0}^{2\pi}\Theta\frac{d^{2}\Theta}{d\theta^{2}}\,d\theta=\left[\Theta\frac{d\Theta}{d\theta}\right]_{0}^{2\pi}-\int_{0}^{2\pi}\left(\frac{d\Theta}{d\theta}\right)^{2}\,d\theta For periodic boundary conditions, \Theta(0)=\Theta(2\pi) and \frac{d\Theta}{d\theta}(0)=\frac{d\Theta}{d\theta}(2\pi) , so the boundary terms disappear \left[\Theta\frac{d\Theta}{d\theta}\right]_{0}^{2\pi}=0 -\int_{0}^{2\pi}\left(\frac{d\Theta}{d\theta}\right)^{2}\,d\theta+\lambda\int_{0}^{2\pi}\Theta^{2}\,d\theta=0 \lambda\int_{0}^{2\pi}\Theta^{2}\,d\theta=\int_{0}^{2\pi}\left(\frac{d\Theta}{d\theta}\right)^{2}\,d\theta Since both \Theta^{2} and \left(\frac{d\Theta}{d\theta}\right)^{2} are non-negative, it follows that \lambda\geq 0 Because the periodic boundary conditions require that \Theta(\theta) be sinusoidal functions (e.g., \cos(n\theta) or \sin(n\theta) ) where n is an integer, we have that \lambda is be a non-negative integer. Does a classical C^{2} solution always exist? For a classical solution to exist, the boundary condition f(\theta) must be sufficiently smooth to ensure uniform convergence of its Fourier expansion. Specifically, f(\theta) should be at least twice differentiable. When f is C^{2} , the Fourier expansion converges uniformly, which guarantees the existence of a well-behaved solution inside. This smoothness ensures that the solution satisfies the Laplace equation, which describes harmonic functions. Even if f is not twice differentiable, a solution can still exist, but with lower regularity. Also, the solution inside the disc is regularized by noise because the Laplace operator is closely tied to Brownian motion. As a result, even if the boundary function f has some roughness, the solution u inside the disc can still be differentiated as many times as necessary. However, if f is not smooth, then the solution near the boundary may reflect that, hence limiting the regularity to the level of f . The Role of the Disc The disc geometry makes sure that the boundary is smooth and compact, which allows for the separation of variables and Fourier expansion; this simplifies solving the equation and ensures the uniqueness of the solution! For example, Brownian motion cannot fully explore regions with spikes or discontinuities because it follows random, continuous paths that are unlikely to precisely hit sharp or narrow areas. In smooth domains, Brownian motion naturally diffuses across the entire boundary; meanwhile when the boundary has a spike or discontinuity, the probability of the random path intersecting or reaching these points is significantly lower resulting in less regular solutions to differential equations. 1.0.1. The Angular Equation First, we solve the angular equation \frac{d^{2}\Theta}{d\theta^{2}}+n^{2}\Theta=0 This is a standard linear second-order ordinary differential equation (ODE) with constant coefficients and its general solution is \Theta(\theta)=A_{n}\cos(n\theta)+B_{n}\sin(n\theta) where A_{n} and B_{n} are constants. An important consideration is that \Theta(\theta) must be 2\pi -periodic because \theta represents the angular coordinate in polar coordinates, which is periodic with period 2\pi . That is, \Theta(\theta)=\Theta(\theta+2\pi) . This periodicity condition imposes restrictions on n . \Theta(\theta+2\pi)=A_{n}\cos(n(\theta+2\pi))+B_{n}\sin(n(\theta+2\pi))=\Theta(\theta) \cos(n(\theta+2\pi))=\cos(n\theta+2\pi n)=\cos(n\theta) \sin(n(\theta+2\pi))=\sin(n\theta+2\pi n)=\sin(n\theta) However, these equalities hold if n is an integer, because \cos(2\pi n)=\cos(0)=1 and \sin(2\pi n)=\sin(0)=0 when n is an integer. Therefore, to satisfy the periodicity condition, n must be an integer. This shows that n^{2} is the square of a natural number. 1.0.2. The Radial Equation Next, we solve the radial equation r^{2}\frac{d^{2}R}{dr^{2}}+r\frac{dR}{dr}-n^{2}R=0 This is a Cauchy-Euler equation, which can be solved using the trial solution R(r)=r^{m} , where m is a constant to be determined. Substituting R(r)=r^{m} into the radial equation r^{2}\left(m(m-1)r^{m-2}\right)+r\left(mr^{m-1}\right)-n^{2}r^{m}=0 m(m-1)r^{m}+mr^{m}-n^{2}r^{m}=0 \left[m(m-1)+m-n^{2}\right]r^{m}=0 \left[m^{2}-n^{2}\right]r^{m}=0 For a nontrivial solution, we need m^{2}-n^{2}=0\implies m=\pm n Therefore, the general solution for R(r) is R(r)=C_{n}r^{n}+D_{n}r^{-n} where C_{n} and D_{n} are constants. 1.1. Combining Solutions and Applying Boundary Conditions Combining the solutions for \Theta(\theta) and R(r) , the general solution to the Laplace equation in the disc is u(r,\theta)=\sum_{n=0}^{\infty}\left(C_{n}r^{n}+D_{n}r^{-n}\right)\left(A_{n}\cos(n\theta)+B_{n}\sin(n\theta)\right) Since we are solving inside a disc (i.e., 0\leq r\leq R ), and to ensure that u(r,\theta) remains finite at r=0 , we must remove the r^{-n} terms for n>0 , as they become singular at r=0 . So, the solution simplifies u(r,\theta)=C_{0}+\sum_{n=1}^{\infty}r^{n}\left(A_{n}\cos(n\theta)+B_{n}\sin(n\theta)\right) Now, we apply the boundary condition u(R,\theta)=f(\theta) . Setting r=R , we get u(R,\theta)=C_{0}+\sum_{n=1}^{\infty}R^{n}\left(A_{n}\cos(n\theta)+B_{n}\sin(n\theta)\right)=f(\theta) This is a Fourier series expansion of f(\theta) in terms of sines and cosines. The coefficients C_{0} , A_{n} , and B_{n} can be determined by expanding f(\theta) in a Fourier series f(\theta)=a_{0}+\sum_{n=1}^{\infty}\left(a_{n}\cos(n\theta)+b_{n}\sin(n\theta)\right) By matching coefficients, we have C_{0}=a_{0},\quad A_{n}=\frac{a_{n}}{R^{n}},\quad B_{n}=\frac{b_{n}}{R^{n}} 1.2. Uniqueness of Solution An important question is whether the solution to the Laplace equation with given boundary conditions is unique. 1.2.1. Proof of the Maximum Principle Let w(r,\theta) be a harmonic function defined on a closed domain \overline{D} (including its boundary \partial D ). If w attains its maximum or minimum value at an interior point of D , then w is constant throughout D . Case 1: Maximum in the Interior Suppose w is harmonic in D (i.e., \nabla^{2}w=0 ) and reaches a local maximum at some interior point (r_{0},\theta_{0}) . Consider the function v(r,\theta)=w(r,\theta)-\epsilon r^{2} for a small \epsilon<0 . Assume v(r,\theta) attains a local maximum at the same point (r_{0},\theta_{0}) . At this maximum point, the first-order partial derivatives of v disappear \frac{\partial v}{\partial r}(r_{0},\theta_{0})=\frac{\partial w}{\partial r}(r_{0},\theta_{0})-2\epsilon r_{0}=0 \frac{\partial v}{\partial\theta}(r_{0},\theta_{0})=\frac{\partial w}{\partial\theta}(r_{0},\theta_{0})=0 The second-order partial derivatives satisfy \frac{\partial^{2}v}{\partial r^{2}}(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial r^{2}}(r_{0},\theta_{0})-2\epsilon \frac{\partial^{2}v}{\partial\theta^{2}}(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial\theta^{2}}(r_{0},\theta_{0}) The Laplacian of v at (r_{0},\theta_{0}) then is \nabla^{2}v(r_{0},\theta_{0})=\frac{\partial^{2}v}{\partial r^{2}}+\frac{1}{r_{0}}\frac{\partial v}{\partial r}+\frac{1}{r_{0}^{2}}\frac{\partial^{2}v}{\partial\theta^{2}} Using the conditions at the maximum point \nabla^{2}v(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial r^{2}}-2\epsilon+\frac{1}{r_{0}^{2}}\frac{\partial^{2}w}{\partial\theta^{2}} Since \nabla^{2}v=4\epsilon and \epsilon<0 , we have \frac{\partial^{2}w}{\partial r^{2}}(r_{0},\theta_{0})-2\epsilon+\frac{1}{r_{0}^{2}}\frac{\partial^{2}w}{\partial\theta^{2}}=4\epsilon It follows that \frac{\partial^{2}w}{\partial r^{2}}(r_{0},\theta_{0})+\frac{1}{r_{0}^{2}}\frac{\partial^{2}w}{\partial\theta^{2}}=2\epsilon However, because w is harmonic ( \nabla^{2}w=0 ), the left-hand side must equal zero. This shows that w cannot attain a maximum at an interior point unless w is constant. Case 2: Minimum in the Interior Assume w(r,\theta) attains a local minimum at an interior point (r_{0},\theta_{0}) . Consider the function v(r,\theta)=w(r,\theta)+\epsilon r^{2} , where \epsilon<0 . By similar reasoning, v is harmonic, but: \nabla^{2}v=-4\epsilon<0 At the minimum point (r_{0},\theta_{0}) , the first-order partial derivatives again disappear \frac{\partial v}{\partial r}(r_{0},\theta_{0})=\frac{\partial w}{\partial r}(r_{0},\theta_{0})+2\epsilon r_{0}=0 \frac{\partial v}{\partial\theta}(r_{0},\theta_{0})=\frac{\partial w}{\partial\theta}(r_{0},\theta_{0})=0 The second-order partial derivatives satisfy \frac{\partial^{2}v}{\partial r^{2}}(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial r^{2}}(r_{0},\theta_{0})+2\epsilon \frac{\partial^{2}v}{\partial\theta^{2}}(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial\theta^{2}}(r_{0},\theta_{0}) The Laplacian of v at (r_{0},\theta_{0}) is \nabla^{2}v(r_{0},\theta_{0})=\frac{\partial^{2}w}{\partial r^{2}}+2\epsilon+\frac{1}{r_{0}^{2}}\frac{\partial^{2}w}{\partial\theta^{2}} Since \nabla^{2}v=-4\epsilon and \epsilon<0 , we have \frac{\partial^{2}w}{\partial r^{2}}(r_{0},\theta_{0})+\frac{1}{r_{0}^{2}}\frac{\partial^{2}w}{\partial\theta^{2}}=2\epsilon This is a contradiction, as w being harmonic requires \nabla^{2}w=0 . So, w cannot attain a minimum at an interior point unless w is constant. Therefore, w attains its maximum or minimum at an interior

Read the complete text