1
$\begingroup$

What I have in mind is that on $(0,T)\times\Omega$, we have a parabolic pde operator $L$, we have unique solution to

$Lu = f$ when f is Hoelder for some coefficient strictly between 0 and 1.

This is a theorem from Friedman's book on pdes. While i can plough through the proofs which were full of estimates, i don't get the intuitions why they work or why wouldn't they work (Or would they?) for an operator such as

$\bigtriangleup u - \partial_{t_1} - \partial_{t_2} = f$

Where f is a suitable function and the domain is something like $(0,T)\times(0,S)\times\Omega$?

The only answer i found regarding intuition was from User's guide to viscosity solution by crandall-ishii-lions page 52. It says

$u_t+ F(x,u,Du,D^2u)=0$ more or less correspond to $\lambda t+ F = 0$ for large $\lambda$.

I don't quite understand what this means? Does it work if we do $u_t+u_s+F = 0$? What is so nice about having a time evolution that does not (or maybe it does, but i am ignorant) work in higher dimensions?

$\endgroup$
2
  • $\begingroup$ The equation you wrote, $$ \nabla u + \partial_{t_1} u +\partial_{t_2} u =f $$ is not parabolic, but hyperbolic, in $\tau$ as in Bazin's answer above. So it is simply a linear transport equation, with a parameter $\sigma$ (using again Bazin's notations). Or it is just a typo, and you meant $\Delta u + \partial_{t_1} u +\partial_{t_2} u =f$ ? $\endgroup$ – username Jan 13 '14 at 9:20
  • $\begingroup$ @AthanagorWurlitzer i mean laplacian... Oh dear $\endgroup$ – lost1 Jan 13 '14 at 9:24
2
$\begingroup$

Change the variables: define $$ u(t,s,x)=v(\underbrace{\frac{t+s}{2}}_{\tau},\underbrace{{t-s}}_{\sigma},x). $$ You get $ \partial_t u+\partial_s u=\partial_{\tau} v $ and the equation becomes $$ \partial_{\tau} v+F(x,v,D_x v,D_x^2 v)=0,$$ where the function $v$ depends on the real parameter $\sigma$.

$\endgroup$
3
  • $\begingroup$ Okay, thanks. Would you be able to comment on the intuition question? $\endgroup$ – lost1 Jan 12 '14 at 22:16
  • $\begingroup$ Sorry i made a gross typo on the question. I wrote grad instead of the laplacian $\endgroup$ – lost1 Jan 13 '14 at 9:25
  • $\begingroup$ Geometrically speaking your parabolic equation is $X+Q$ where $X$ is a real vector field and $Q$ a positive elliptic operator of order 2 (e.g. $-\Delta$). You may note the following equality of vector fields: $$\partial_t+\partial_s=\partial_\tau.$$ $\endgroup$ – Bazin Jan 13 '14 at 16:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.