Problem with arrayfun and parameter dimensions

[John W. Eaton]

On  7-Oct-2011, Christoph Mahnke wrote:

| 
| Hello everybody,
| 
| I want to avoid  for-loops when applying a function on a matrix of
| values, so my idea is to choose arrayfun.
| A problem arises, as the function expects input parameters with
| different dimensions, here as a simplified example:
| 
| function result = dummyfunction( x, parameter1, parameter2)
|   result = x * parameter1 - sum(x ./ parameter2 );
| endfunction
| 
| where parameter1 is  a simple number and parameter2 is a (fixed-length
| 1xN) vector.
| I want to apply arrayfun on a array  X of values, but i'm not sure how
| to create a suitable array for the vector parameter. Some simple code to
| illustrate my problem:
| 
| %
| %  for a single value
| %
| 
| x1 = rand()
| parameter1 = 7
| parameter2 = 1:10
| result1 = dummyfunction(x1,parameter1,parameter2)   % (works well)
| 
| %
| % this time a matrix
| % 
| X=rand(3,4)        
| 
| p1array= parameter1 * ones( rows(X), columns(X));  
| p2array= ???
| 
| R = arrayfun(@dummyfunction, X, p1array, p2array)
| 
| 
| Maybe you can give me a hint how i have to create the parametermatrix
| p2array to use in arrayfun.

Do you want to apply the function for each value of X with the
parameters fixed for each value of X?  If so, then I think you want
something like

  X = ...;
  parameter1 = ...;
  parameter2 = ...;
  arrayfun (@(X) dummyfunction (X, parameter1, parameter2), X);

When the anonymous function

  @(X) dummyfunction (X, parameter1, parameter2)

is constructed, parameter1 and parameter2 are set from the context in
which the anonymous function is constructed.

But if I understand correctly, your particular problem would probably
solved more efficiently by writing

  X * (parameter1 - sum (1 ./ parameter2))

jwe