# ItoProcess for stochastic reaction-diffusion equation

I am trying to simulate a stochastic differential equation in time and space, but I'm unsure if this can be done in Mathematica. The sde that I would like to study is:

$$dN[x,t]=N[x,t](1-N[x,t])dt+\sqrt{N[x,t]}dw+\partial^2_xN[x,t]dt$$

With an exponential decreasing (in space) initial condition N[x,0]=Exp[-|x|]. Is this possible to simulate with ItoProcess?

Thank you, Best, Andrea

• it can be done, for example, by discretising spatially.
– acl
Apr 12, 2013 at 10:22
• Sorry but I don't see easily how, in particular I feel I would have no control on step sizes in time Apr 13, 2013 at 8:56
• discretize spatially, this solving a set of coupled SDEs in time. like the method of lines for PDEs.
– acl
Apr 13, 2013 at 9:15

Here's a solution following @acl's suggestion of discretizing in space. I added a diffusion coefficient d and used reflecting boundary conditions.

l = 1.0; (* length of domain *)
nx = 101; (* number of spatial cells *)
dx = l/(nx - 1); (* size of each cell *)
d = 0.1; (* diffusion coefficient *)

eqns = Join[
{\[DifferentialD]n[1][t] == (n[1][t] (1 - n[1][t])
+ d (-n[1][t] + n[2][t])/dx^2) \[DifferentialD]t
+ Sqrt[n[1][t]] \[DifferentialD]w[1][t]},
Table[
\[DifferentialD]n[i][t] == (n[i][t] (1 - n[i][t])
+ d (n[i - 1][t] - 2 n[i][t] + n[i + 1][t])/dx^2) \[DifferentialD]t
+ Sqrt[n[i][t]] \[DifferentialD]w[i][t]
, {i, 2, nx - 1}],
{\[DifferentialD]n[nx][t] == (n[nx][t] (1 - n[nx][t])
+ d (n[nx - 1][t] - n[nx][t])/dx^2) \[DifferentialD]t
+ Sqrt[n[nx][t]] \[DifferentialD]w[nx][t]}
];

noise = Table[w[i] \[Distributed] WienerProcess[0, 0.5], {i, nx}];

unks1 = Table[n[i][t], {i, nx}];
unks2 = Table[n[i], {i, nx}];

ics = Table[Exp[-(i - 1) dx], {i, nx}];

tmax = 1;
dt = 0.0001;
sol = RandomFunction[ItoProcess[
eqns, unks1, {unks2, ics}, t, noise], {0, tmax, dt}]


This give an Function::flpar message, but still runs.

ListLinePlot[sol, PlotRange -> {All, All}]


Here's an animation of n[x]:

Export["space.gif",
Table[ListLinePlot[sol["SliceData", t], PlotRange -> {0, 1}], {t,
0.0, 1.0, 0.01}]]


• can you tell me if it was N[x,y,t], then how would someone define w and n in the above program? Nov 3, 2022 at 14:20