       • To: mathgroup at smc.vnet.net
• Subject: [mg129440] bad PDE solution...
• From: Fred Bartoli <""@news.free.fr>
• Date: Mon, 14 Jan 2013 00:03:10 -0500 (EST)
• Delivered-to: l-mathgroup@mail-archive0.wolfram.com
• Delivered-to: l-mathgroup@wolfram.com
• Delivered-to: mathgroup-newout@smc.vnet.net
• Delivered-to: mathgroup-newsend@smc.vnet.net
• Reply-to: myname_with_a_dot_inbetween at free.fr

```Hello,

As a first simple test case of a more complicated pb I want to solve the
plane diffusion equation.
It seems NDSolve gives a bad answer to the simple step case but I can't
find what's happening...

(* Set up the equation: *)

eq1=D[h[t,x],x,x]- k1 D[h[t,x],t]==0

const=k1-> 10000

(*"solve" it...*)

sol = NDSolve[
{D[h[t, x], {x, 2}] - k1 D[h[t, x], t] == 0, h[0, x] == 0,
h[t, 1] == 0, h[t, 0] == 1 - Exp[-10^6*t]} /. const,
h, {t, 0, 10^-3}, {x, 0, 0.01}][]

Plot3D[(h[t,x])/.mag/.solCart,{x,0,10^-4},{t,0,10^-4},AxesLabel->{"x","t","h"},PlotRange->All,PlotLabel->"Bogus
solution"]
Plot3D[Erfc[x/(2*Sqrt[t/k1])]/.const,{x,0,10^-4},{t,0,10^-4},AxesLabel->{"X","t","h"},PlotRange->All,PlotLabel->"The
right one"]

(* Test the Erfc[x/(2Sqrt[t/k1])] solution *)
Release[Hold[D[h[t,x],x,x]- k1*D[h[t,x],t]]/.
h[t,x]->Erfc[x/(2*Sqrt[t/k1])]]//Simplify

I hope I'm doing something wrong, but I can find what...

--
Thanks,
Fred.

```

• Prev by Date: Re: wrong result when computing a definite integral
• Next by Date: Re: Aligning graphics on frame boundaries
• Previous by thread: Re: Ask for help: how to let mathematica output Fortran code with fewest
• Next by thread: Re: bad PDE solution...