Isolate first non-zero integer of each element of an array

Lets say,

A is some 2D array such that A = [m n] with both large values (e10) and tiny decimal values (e-15)

For Example,

A = [0.00663270674527115 36798861787.4757 0.0165559157141383

0.00845305563147772 0.000298646998074807 2561194549424.91]

I need an array such that,

B = [6 3 1

8 2 2]

Methods I've tried

Large multiplier

• B = A * 10^10; %multiply by large number to remove leading zeros
• B = fix(B./10.^fix(log10(B))); %isolate first value of each element
• I got a higher than expected amount of 1's with this method - perhaps matlab has a value limit

Format as Scientific first

• B = sprintf('%0.5g',A); %convert each element to scientific notation (no leading zeros)
• B = fix(B./10.^fix(log10(B))); %isolate first value of each element
• This creates a string - not an array. Its slow and not a great solution
• I've also tried format short eng, but to my understanding that doesn't seem to effect the array itself just the displayed answer
• fprint seemed promising but got an error calling in array A

B = fprintf(A,'%1.5s',0.1)

Answer 1

[Bruno Luong]

A = [0.00663270674527115	36798861787.4757	   0.0165559157141383
0.00845305563147772	0.000298646998074807	   2561194549424.91]

Then

AA = abs(A); % in case A negative
B = floor(AA./(10.^floor(log10(AA))))

returns

B =

     6     3     1
     8     2     2

B contains NaN at the place where A is 0 (normal not avoid this special case).

[Adam Nasinski]

@Bruno Luong

Awesome solution! Not only does it work - it works well. Quick and efficient. Thank you!

Answer 2

[Ameer Hamza]

Are values stored in a .txt file? Because MATLAB floating-point types cannot contain values with such precision. The following assumes that the values are present in a .txt file

str = fileread('data.txt');
str = strrep(str, '0', '');
str = strrep(str, '.', '');

out = regexp(str, '(\d)\d*', 'tokens');
out = [out{:}];
out = cellfun(@str2double, out);
out = reshape(out, 3, []).';

data.txt is attached.

Result

>> out
out =
     6     3     1
     8     2     2

[Adam Nasinski]

Unfortuanetly, it's not a .txt file.