eval vz feval, different results

Hi. I would like to ask, when using feval and eval, the result is different? I thing, that I'm using same approach. Thank you!

   %F = some symbolic function with this symbolic: n1i,n2i,n3i,n4i,n5i,n6i,n7i
    f = matlabFunction(F); % in my case: f =  @(x)f(x(1),x(2),x(3),x(4),x(5),x(6),x(7))
    Xa = [0.1000    0.1000    0.1000    0.1000    0.1000    0.1000    0.1000] %for example some real value

%approach1--------------
fa=feval(f,Xa)
   -0,00430563198070900 + 0,00000000000000i
   -0,00358093388632469 + 0,00000000000000i
   -0,00313919755488547 + 0,00000000000000i
   -0,00338161000029060 + 0,00000000000000i
   -0,00799484623495874 + 0,00000000000000i
   -0,00277875928862729 + 0,00000000000000i
   -0,00410906741467898 + 2,22648608470585e-06i

%approach2-------------
fa_0=subs(F,{n1i,n2i,n3i,n4i,n5i,n6i,n7i},{Xa(1),Xa(2),Xa(3),Xa(4),Xa(5),Xa(6),Xa(7)})
fa=eval(fa_0)
   -0,00430563198070900
   -0,00358093388632500
   -0,00313919755488500
   -0,00338161000029100
   -0,00799484623495900
   -0,00277875928862700
   -0,00409570849817100

[Stephen23]

Why is eval required at all? Surely subs and double would be enough?

Answer 1

[Rik]

The reason would depend on a slightly different method of calculation and rounding. I don't think the exact difference is really relevant, because you should use neither. Use this instead:

fa=f(Xa);

Stephen Cobeldick is very pasionate about this topic (and with good reason), so I have taken the liberty of adapting some of his thoughts on this topic: Introspective programing (eval and other functions like it) is slow, and the MATLAB documentation warns specifically against it. Using eval is slow, makes debugging difficult, and removes all code helper tools (code checking, syntax hints, code completion, variable highlighting, etc.). A longer discussion can be found here. If you are not an expert and you think you need eval, there are probably better solutions to your problem.