Cannot compute number of steps

Hey,

Im trying to integrate using the trapezoid method, and works just fine when working with constant boundaries. However i'm having a hard time coming around this issue:

Error using : (line 38)
Cannot compute the number of steps from 0 to 5*sin((xx*pi)/10) by (5*sin((xx*pi)/10))/round(50*sin((xx*pi)/10)).

Error in Lab2 (line 24)
y=[0:hy:ymax(xx)] ;

n=100;

hx=10/n;

x=[0:hx:10];

ymax= @ (xx)5.*sin(pi.*xx./10);

syms xx

hy=ymax(xx)/round(ymax(xx)/hx);

y=[0:hy:ymax(xx)] ;

[xx,yy]=ndgrid(x,y);

mat = exp(-0.25.*((xx-8).^2 + (yy-0).^2)) .*cos((xx-8-yy)/6); %trapets

b{1}=x; b{2}=y;

T=trapets(b,mat,2);

F = @(x,y)exp(-0.25.*((x-8).^2 + (y-0).^2)).*cos((x-8-y)/6); %integral

Refvalue=integral2(F,0,10,0,ymax) ;

disp(T)

What to do?

[John D'Errico]

So, you found some function that we don't have and cannot see, called trapets. Now you have a problem with it. How can we know what is the problem?

Contact the source of the code. Ask them.

[Philip Baczek]

Its not a problem with the function. The function just uses the built in function trapz. My problem lies getting y=[0:yh:ymax(xx)] to work, which is the steplength

[Philip Baczek]

function T = trapets(x,mat,N)

mat = trapz(x{N},mat,N) ;

if N==1

      T=mat;
      return;
  end

T = trapets(x,mat,N-1) ;

PS: sorry for the format, not quite sure how to paste code properly

Answer 1

[Walter Roberson]

Look at your code,

ymax= @ (xx)5.*sin(pi.*xx./10);

Anonymous function. Okay so far.

syms xx

Symbolic declaration. Okay so far.

hy=ymax(xx)/round(ymax(xx)/hx);

Applying the anonymous function to the symbolic value -- dubious. Expecting to be able to round() the result of the symbolic calculation: very dubious. These are not impossible but are unlikely (there are some uses for passing a symbolic value to an anonymous function.)

y=[0:hy:ymax(xx)] ;

Clearly broken. hy is symbolic, ymax(xx) is symbolic, and you cannot use the colon operator with symbolic bounds.

Unfortunately there is a shortage of comments about the purpose of that section of code...

[Philip Baczek]

The syms was just an act of desperation to try something different. It becomes obvious to me now that you mention that fact. The problem I cant come around is how to use a function as ymax(which depends on x) as an upper boundary to my trapezoid method since i in the previous section used similar code with constant boundary and it worked. Example. y=0:hy:10, x=0:hx:10

After further inspection i notice that my y variable just is a 0by1 matrix and contains no values. Where as x is 1x101(depending on n) My best guess being this is my main concern.