Argument to dynamic structure reference must evaluate to a valid field name

Here is a code snippet that results in the error message recorded in my question. The program passes the if clause that starts

if isfield...

but baulks (North America: balks) at the next line. I hope that it is obvious to a human reader what I am trying to do. How do I make it obvious to Matlab? Have I given enough info so that someone can help me? I would be happy to give more if required.

      for fldn={'raw','scl','svd','pft'}
          if isfield(R,(fldn))
              err = X*(R.(fldn).coef - Y);
              R.(fldn).err = num2cell(mean(err.^2));
          end
      end

Here are some answers given by the Matlab debugger:

fldn
    ans='raw'
R.fldn; 
    Reference to a non-existent field
R.(fldn);
     Argument to dynamic structure reference must evaluate to a valid field name.
R.raw
 ans=coef: {[21x1 double]}

Answer 1

[Steven Lord]

Display the contents of fldn as well as the class of fldn on the line immediately after the start of the for loop.

mystruct.('abc') = 1 % valid dynamic field name
mystruct.({'def'}) = 2; % NOT a valid dynamic field name

If you want to use fldn as a dynamic field name, you need to extract the char array inside the cell.

x = {'ghi'};
mystruct.(x{1}) = 3;

[Walter Roberson]

To expand on this: when you use

for fldn={'raw','scl','svd','pft'}

then for iterates over the columns of the resulting array, and because it is a cell array, each of the columns is a cell array. Your loop variable is not going to be set to strings in turn: it is going to be set to cell arrays that happen to contain one element that is a string.

If you had had

for fldn = {'raw', 'scl'; 'svd', 'pft'}

then because for iterates over columns, the variable would first be set to {'raw'; 'svd'} and then to {'scl' ; 'pft'} .

It does not do what one might prefer, but it is consistent behavior.