Syms, fit and integral in a standalone app

Hi,

I have the following code that I want to compile in a standalone app :

syms z;
K_fit = fit(app.gdist,gK,'linearinterp');
func1= @(app,z)(K_fit(z).*(((z.^1-app.gdist(i,1)))));
gpot1(counter,1)=2*app.cz*integral(@(z)func1(app,z), app.gdist(i,1),max(app.gdist),'ArrayValued',true);

Can you tell how to rewrite it so that it doesn't use syms and is deployabe in a standalone app?

Thanks.

Answer 1

[Walter Roberson]

That code never uses z in its sym form, so the syms z is not needed.

The definition of func1 uses z as an input parameter, so that "shadows" the previous definition of z as a symbol.

Meanwhile, your integral call is a numeric call, and you apply the numeric call to the anonymous function @(z)func1(app,z) which again "shadows" z .

[Walter Roberson]

Meanwhile, the replacement for the fit() is

[sortg, sortorder] = sort(app.gdist);
sortgK = gK(sortorder);
K_fit = @(z) interp1(sortg, sortgK, z, 'spline');

func1= @(app,z)(K_fit(z).*(((z.^1-app.gdist(i,1)))));
gpot1(counter,1)=2*app.cz*integral(@(z)func1(app,z), app.gdist(i,1),max(app.gdist),'ArrayValued',true);

Using 'spline' for the interp1 reduces the issue that the default 'linear' interpolation has a discontinuus derivative at every interpolation point. The documentation for integral() does not prohibit such functions, but if you are going to use linear interpolation then you really should add all of the interpolation points as 'Waypoints'

[Simon]

It works. Thank you.