Problem with ploting involving function handle

Hi all, I encounter the following error when I want to plot a function involving function handle with symsum, I cannot figure out the error. My code is as follows. In my matlab 2023b, I receive the following error:

Warning: Error updating FunctionLine.

The following error was reported evaluating the function in FunctionLine update: Undefined function 'symsum' for input arguments of type 'double'.

 N= 600

N = 600

  sym ii
  COMx_k = @(k) symsum(1.*exp(-2*pi*1i*(k/N)*ii), ii,1,N)

COMx_k = function_handle with value:
    @(k)symsum(1.*exp(-2*pi*1i*(k/N)*ii),ii,1,N)

  %k=0:0.1:300
  fplot(COMx_k)
  xlabel('k')
  ylabel('COMx_k')

Warning: Error in state of SceneNode.
The following error was reported evaluating the function in FunctionLine update: Unrecognized function or variable 'ii'.

Answer 1

[Matt J]

Perhaps this is what you meant?

 N= 600;
  COMx_k = @(k) real(sum(1.*exp(-2*pi*1i*(k/N).*(1:N)' ),1));
  %k=0:0.1:300
  fplot(COMx_k,[0,300])
  xlabel('k')
  ylabel('COMx_k')

[Tsz Tsun]

Thanks a lot, it works!!

[Tsz Tsun]

May I know what does ,1 mean at the end of the syntax? Thanks !

[Stephen23]

"May I know what does ,1 mean at the end of the syntax?"

The dimension to sum over:

https://www.mathworks.com/help/matlab/ref/double.sum.html#btv6ok6-1-dim

[Paul]

I'm not sure how fplot works in this instance, and that warning sounds ominous. Maybe fplot detects the problem with array inputs and then does only scalar evaluations? The doc page states that:

"The function must accept a vector input argument and return a vector output argument of the same size."

Hmm, "a vector input" is requred in the doc, but the warning is about more general "array inputs".

Furthermore, it looks like COMx_k is supposed to be a DFT-like computation, but can't say for sure w/o more information from @Tsz Tsun.

I suspect that the function is supposed to be (using scalar expansion):

N = 600;
COMx_k = @(k) reshape(real(sum(1.*exp(-2*pi*1i*(reshape(k,1,[])/N) .* (1:N)'),1)),size(k));

Can k really take on non-integer values?

%k=0:0.1:300
%fplot(COMx_k,[0,300])

If this problem is really related to the DFT, then the function should be

COMx_k = @(k) reshape(real(sum(1.*exp(-2*pi*1i*(reshape(k,1,[])/N) .* (0:N-1)'),1)),size(k));

But this is all just speculation w/o more information from the OP as to the what COMx_k is supposed to represent mathematically.

[Tsz Tsun]

Yes, in fact I am trying to do a DFT. I just notice that there is a function package called fft(X) in matlab, but I am not sure if fft(X) is doing the job of DFT.

[Walter Roberson]

fft() does a two-sided numeric discrete fourier transform.

Answer 2

[Walter Roberson]

 sym ii

That is equivalent to

 ans = sym('ii');

which creates the symbolic symbol ii but throws away the reference to the symbolic symbol.

What you probably wanted was

 N= 600
 syms ii
 COMx_k = @(k) symsum(1.*exp(-2*pi*1i*(k/N)*ii), ii,1,N)
 fplot(COMx_k)
 xlabel('k')
 ylabel('COMx_k')

However, this takes a fair amount of time to plot !!

[Walter Roberson]

@(k) symsum(1.*exp(-2*pi*1i*(k/N)*ii), ii,1,N)

That is complex-valued when k is not an integer, and is 0 when k is an integer.