Cole–Hopf transformation explained

The Cole–Hopf transformation is a change of variables that allows to transform a special kind of parabolic partial differential equations (PDEs) with a quadratic nonlinearity into a linear heat equation. In particular, it provides an explicit formula for fairly general solutions of the PDE in terms of the initial datum and the heat kernel.

Consider the following PDE:u_ - a\Delta u + b\|\nabla u\|^ = 0, \quad u(0,x) = g(x)where

x\inRn

,

a,b

are constants,

\Delta

is the Laplace operator,

\nabla

is the gradient, and

\|\|

is the Euclidean norm in

Rn

. By assuming that

w=\phi(u)

, where

\phi()

is an unknown smooth function, we may calculate:w_ = \phi'(u)u_, \quad \Delta w = \phi'(u)\Delta u + \phi(u)\|\nabla u\|^Which implies that:\beginw_ = \phi'(u)u_ &= \phi'(u)\left(a\Delta u - b\|\nabla u\|^\right) \\&= a\Delta w - (a\phi + b\phi')\|\nabla u\|^ \\&= a\Delta w\endif we constrain

\phi

to satisfy

a\phi''+b\phi'=0

. Then we may transform the original nonlinear PDE into the canonical heat equation by using the transformation:

This is the Cole-Hopf transformation.[1] With the transformation, the following initial-value problem can now be solved:w_ - a\Delta w = 0, \quad w(0,x) = e^The unique, bounded solution of this system is:w(t,x) = \int_ e^dySince the Cole–Hopf transformation implies that

u=-(a/b)logw

, the solution of the original nonlinear PDE is:u(t,x) = -\log \left[{1\over{(4\pi at)^{n/2}}} \int_{\mathbb{R}^{n}} e^{-\|x-y\|^{2}/4at - bg(y)/a}dy \right]

Applications

References

  1. Book: Evans, Lawrence C. . Lawrence C. Evans

    . Lawrence C. Evans. Partial Differential Equations . American Mathematical Society . 2010 . 2nd . Graduate Studies in Mathematics . 19 . 206–207.

  2. Cole . Julian D. . Julian Cole . 1951 . On a quasi-linear parabolic equation occurring in aerodynamics . Quarterly of Applied Mathematics . en . 9 . 3 . 225–236 . 10.1090/qam/42889 . 0033-569X. free .
  3. Hopf . Eberhard . Eberhard Hopf . 1950 . The partial differential equation ut + uux = μxx . Communications on Pure and Applied Mathematics . en . 3 . 3 . 201–230 . 10.1002/cpa.3160030302.