# Empirical source terms with PDEPE

5 views (last 30 days)
Matthew Hunt on 7 May 2020
Commented: Bjorn Gustavsson on 11 May 2020
Suppose I'm solving an equation:
With initial condition and boundary conditions and
Now I want to import the source term, as data, so I will have a Nx2 array [t Q] with data. Can I just treat that as I would a constant?
The code I have is:
function sol=temp_pde(t,R,X,source)
m=1; %Sets the geometry to cylindrical
global theta kappa h Q;
theta=X(1); kappa=X(2); h=X(3);
r=linspace(0,R,800);
t=source(:,1);
Q=source(:,2);
sol=pdepe(m,pdefun,icfun,bcfun,r,t);
end
function [c,f,s] = pdefun(r,t,u,dudx)
global Q theta kappa;
c = theta;
f = kappa*dudx;
s = Q;
end
function u0 = icfun(r)
u0 = 0;
end
function [pl,ql,pr,qr] = bcfun(xl,ul,xr,ur,t)
global h kappa;
pl = 0;
ql = 1;
pr = h*ur;
qr = kappa;
end
Where I obtain my t parameter from my data in source. Would this work?
##### 2 CommentsShowHide 1 older comment
Matthew Hunt on 7 May 2020
\dot{Q} is purely a function of time, so is homogeneous.

Bjorn Gustavsson on 7 May 2020
To the best of my understanding you should be able to define your pde-function something like this:
function [c,f,s] = pde_interpS(x,t,u,dudx,tQ,Q,theta,kappa)
c = theta;
f = kappa*dudx;
s = interp1(tQ,Q,t,'pchip');
end
Then you can convert that function to a function of t, x, u and dudx with the standard @()-trick. My preference is to avoid globals as much as possible, but you can use that too.
HTH
Bjorn Gustavsson on 11 May 2020
Seems like your pde-function that should return your c, f and s either doesn't return all three of those variables, or some of them are not set to column vectors - either one or more of them are empty or of the wrong size - perhaps one of them becomes a row-vector. If you set a debug-stop in the pde-function you can check the sizes of the outputs there.
HTH