Compare two numeric cell arrays

I have two huge cells, a dictionary and a set of possible hands in scrabble. I want to compare the two and find what arrays match

myAletters = ['D','I','R','N','E','E','P'];
myAletters = upper(myAletters);
myAletters = int8(myAletters);
myAletters = myAletters-64;
hand_count=0;
for i=1:length(myAletters)
      hand_interim_combination=nchoosek(myAletters,i);
      for j=1:length(hand_interim_combination(:,1))
          hand_interim_permutation=perms(hand_interim_combination(j,:));
           for k=1:length(hand_interim_permutation(:,1))
               hand_count=hand_count+1;
               hand{hand_count}=hand_interim_permutation(k,:);
           end
      end
end
end
load('myDict.mat');
word_plays=ismember(hand,myDict);

Ismember won't work because it has to be strings, going one by one won't work because it takes too long. Wat do?

[Andrei Bobrov]

Please attach your file myDict.mat.

[Stephen23]

"going one by one won't work because it takes too long. Wat do?"

Stop ignoring mlint messages in the Editor. There is already one message in your code telling you that the way you have written it is going to be slow. The message even tells you how to solve this issue:

Also note that the following code does not create an array of separate characters, because [] is a concatenation operator, not a list operators as many beginners think:

 >> myAletters = ['D','I','R','N','E','E','P']
 myAletters =
 DIRNEEP

That code is therefore equivalent to simply writing this:

 >> myAletters = 'DIRNEEP'
 myAletters =
 DIRNEEP

Answer 1

[Stephen23]

Once I got rid of that unnecessary conversion to numeric value then this task becomes a lot easier. I also preallocated the cell array hand to significantly speed the code up.

myAletters = 'DIRNEEP';
hand = cell(1,floor(exp(1)*factorial(numel(myAletters)))-1);
hand_count=0;
for i=1:length(myAletters)
      hand_interim_combination=nchoosek(myAletters,i);
      for j=1:length(hand_interim_combination(:,1))
          hand_interim_permutation=perms(hand_interim_combination(j,:));
           for k=1:length(hand_interim_permutation(:,1))
               hand_count=hand_count+1;
               hand{hand_count}=hand_interim_permutation(k,:);
           end
      end
end
%load('myDict.mat');
myDict = {'IRE','PEER','XYZ'};
word_plays = ismember(hand,myDict);

and checking the output:

>> hand(word_plays) 
ans = 
  'IRE'    'IRE'    'PEER'    'PEER'

I also tidied up the formatting (crtl+I), replaced unpredictable length with reliable size, and renamed variables to make it more readable (for me):

vec = 'DIRNEEP';
hand = cell(1,floor(exp(1)*factorial(numel(vec)))-1);
itr = 0;
for m = 1:numel(vec)
    cmb = nchoosek(vec,m);
    for n = 1:size(cmb,1)
        prm = perms(cmb(n,:));
        for k = 1:size(prm,1)
            itr = itr+1;
            hand{itr} = prm(k,:);
        end
    end
end

Perhaps the simplest solution would be to use cellstr instead of the innermost loop:

vec = 'DIRNEEP';
hand = cell(1,2^numel(vec)-1);
k = 0;
for m = 1:numel(vec)
    cmb = nchoosek(vec,m);
    for n = 1:size(cmb,1)
        k = k+1;
        hand{k} = cellstr(perms(cmb(n,:)));
    end
end
hand = vertcat(hand{:});

Note that these are all just improvements on your algorithm. With a large dictionary a more efficient algorithm might use the small number of available letters to iteratively select smaller sets of the dictionary, until matches are achieved.

[Russell Marki]

Thanks for your help.