Shannon fano code error

function [sorted1 sorted2] = shannon_fano(keyword)

probabilities_calculation = zeros(size(keyword));

for i = 1:length(keyword)  %find the probabilities of the symbols/occurence of each letter

probabilities_calculation(i) = (sum(keyword==keyword(i))/length(keyword));

end

[sorted1 sorted2] = shannon_split(probabilities_calculation);

function [first_half second_half] = shannon_split(input_symbols)

   add_prob = 0;

   for k = 1:length(input_symbols)

   symbol1 = zeros(size(input_symbols));

   if(input_symbols > 1)

       add_prob = add_prob + input_symbols(k);

       symbol1(k) = input_symbols(k);

   if(sum(input_symbols)- add_prob <= 0.5)

       first_half = zeros(size(symbol1(symbol1~=0)));
       symbol2 = input_symbols(1:end) - symbol1;  %now for the second half we consider the remaining elements
       updated = symbol2(symbol2~=0);
       second_half = ones(size(updated));

       shannon_split(first_half);
       shannon_split(second_half);
   end
   end
   end

This is my code for shannon fano but i am receiving an error in it and is unable to solve it :(

This is the error:

Error in ==> shannon_fano>shannon_split at 15
 add_prob = 0;

??? Output argument "first_half" (and maybe others) not assigned during call to
"D:\Matlab\shannon_fano.m (shannon_split)".

Error in ==> shannon_fano at 11
[sorted1 sorted2] = shannon_split(probabilities_calculation);

[Maheen]

oh got it so if i use while instead of if then would it be fine? as i am getting error in that too

Answer 1

[Thorsten]

if(sum(input_symbols)- add_prob <= 0.5) if false, first_half and second_half are not computed in shannon_split, so the output arguments are not computed.

[Maheen]

oh got it so if i use while instead of if then would it be fine? as i am getting error in that too

[Thorsten]

You have to ensure that for all input_symbols the values of both output parameters of your function are defined. So for this is only the case if at least once in your for loop the condition

 (sum(input_symbols)- add_prob <= 0.5) i

becomes true.

[Maheen]

function mid = division(probabilities)

add_prob = 0;

for k = 1:length(probabilities) %go through the entire set

    add_prob = add_prob + probabilities(k); %add single probability at a time

    mid(k) = probabilities(k); %save the no of added probabilities

    if(sum(probabilities)-add_prob <= 0.5)

        break;
    end
   end

This is a code which i have improved for the division of probabilities so now where i can use the recursion technique so that i could repeat this task until we have only one element left