Nested FOR loops and creating text file (3d-layers) too slow.

Hello everyone,

I am pretty new with Matlab. I am facing a problem concering nested for-loops and creating a text file. It is too slow. The text file contains data for 3d-layers.

If the number of layers and rows and columns in each layer are too big, it takes lots of hours to execute the code.

If anyone here can help me regarding to optimize my code or telling me an alternate to solve this problem, it would be very helpful.

Thank you very much.

clc
clear all
b=100;
l=150;
d=0.1;
dn=1.0;
nnR=b/dn+1;
nnC=l/dn+1;
nL=d/0.1+1;
R=1;
C=1;
L=0;
x=0;
y=0;
z=0;
space='          ';
add=0;
nline=sprintf('Start\ntextfileindex\n$    nod               row               column               layer                        \n');
% fileID=fopen('node.txt','w');
% fprintf(fileID,'%s', q);
for b = 1:nL
  
for i = 1:nnR
    dummieR=R
    for i = 1:nnC
        abc=b;
        abc=num2str(abc);
        if b<10
            
        abc=strcat('0',abc)
        else
            abc=b;
        end
        dummieC=C
        numC=num2str(dummieC)
        dummiex=x
        dummiey=y
        dummiez=z
    numz = sprintf('%.4f',dummiez)   
    numx = sprintf('%.4f',dummiex)
    numy = sprintf('%.4f',dummiey)
    
%     numx = sprintf('%.4f',dummiex)
    if C<100 && C>9
        numC=strcat('0',numC)
    end
    if C<10
        numC=strcat('00',numC)
    end
        
        numR=num2str(dummieR)
    if R<100 && R>9
        numR=strcat('0',numR)
    end
    if R<10
        numR=strcat('00',numR)
    end
%     dummiex=x
%     numx = sprintf('%.4f',dummiex)
%     numx=num2str(dummiex)
   if dummiex<100 && dummiex>9.9999
    def= "\n"+abc+numR+numC+"         "+numx
   elseif dummiex<10
     def= "\n"+abc+numR+numC+"          "+numx
   else
      def= "\n"+abc+numR+numC+"        "+numx
   end
   if dummiey<100 && dummiey>9.9999
    def= def+"         "+numy
   elseif dummiey<10
     def= def+"          "+numy
   else
      def= def+"        "+numy
   end
     
 if nL<10
 def= def+"           "+numz
 else
  def= def+"          "+numz 
 end
    C=C+1;
    nline= strcat(nline,def)
    nline= sprintf(nline)
    x=x+dn;
    end
    R=R+1;
    C=1;
    y=y+dn;
    x=0;
end
z=z+0.1;
x=0;
y=0;
C=1;
R=1;
end
nline=sprintf(nline+ "\n" + "END")
fileID=fopen('textfilem.txt','wt');
fprintf(fileID, nline);

[dpb]

Hard to tell w/o way more time than have what it is you're actually trying to do???

Tell us the desired result and probably post a (short) example of what the output should look like.

All the string manipulation is probably the time culprit and probably unneeded...but, need to know what the underlying problem is.

[Klavers Christoph]

I just want to write some numbers in a text file in a specific format like the one shown below.

01001001 0.0000 0.0000 0.0000

01001002 5.0000 0.0000 0.0000

01001003 10.0000 0.0000 0.0000

01001004 15.0000 0.0000 0.0000

01001005 20.0000 0.0000 0.0000

01001006 25.0000 0.0000 0.0000

01001007 30.0000 0.0000 0.0000

01001008 35.0000 0.0000 0.0000

01001009 40.0000 0.0000 0.0000

01001010 45.0000 0.0000 0.0000

01001011 50.0000 0.0000 0.0000

Answer 1

[dpb]

So, as suspected, just generate the data to print and use the desired formatting string:

% generate some sample data...
A=1001000+[1:11];
B=0:5:50;
C=[A.' B.' zeros(numel(A),2)];
% define the desired format for output file...
fmt=['%08d' repmat('%12.4f',1,3) '\n'];
% Illustrate what will get to command line...
>> fprintf(1,fmt,C.')
01001001      0.0000      0.0000      0.0000
01001002      5.0000      0.0000      0.0000
01001003     10.0000      0.0000      0.0000
01001004     15.0000      0.0000      0.0000
01001005     20.0000      0.0000      0.0000
01001006     25.0000      0.0000      0.0000
01001007     30.0000      0.0000      0.0000
01001008     35.0000      0.0000      0.0000
01001009     40.0000      0.0000      0.0000
01001010     45.0000      0.0000      0.0000
01001011     50.0000      0.0000      0.0000
>>

NB1: Internal Matlab storage order is column major so must transpose C to output by row for sequential file.

NB2: The first column is possible to be written as part of formatting statement as well if the values are just 1:N. Include the literal string '0100' in the format string as '0100%05d' to handle i from 1:99999

[Klavers Christoph]

Hello,

thanks for your answer, it really helped me. The other thing is, the first entry of the first coloumn should be in a specific format. The first entry consists of 8 numbers (i.e 01001001) .The first two numbers (i.e 01) represent layer numbers and next three elements represent row numbers (i.e 001) and then the remaining three elements represent columns (i.e 001). I want to change these elements independent of each other.

01001001

01001002

. . . . .

. . . . .

after 11 entries, it would look like this

01001011

and after this entry, I want to increase the row number and coloumn numbers start increasing again from 001 like this

01002001

01002002

01002003

.........

..........

and so on

The entry data is (01001001). The least significant 12 numbers represent the location of the points: For example 001001 which means row 1, column 1.

The two numbers at the beginning represent the layers, because more layers of points get added: 01 is layer 1, 02 is layer 2

After the first coloumn, the next three columns in the text file just represent the coordinates. (x,y,z)

An example for layer 2, row 4, column 5 would be:

02004005 x_coordinates y_coordinates z_coordinates

[dpb]

Sure, just write the variables as needed with the proper format string for them.

I built sample vectors lay, row, col that contained made up layer, row, column values for a hypothetical mesh/grid as follows when catenated:

>> C=[lay row col B.' zeros(numel(A),2)]
C =
     1     1     1     0     0     0
     1     1     2     5     0     0
     1     1     3    10     0     0
     1     2     1    15     0     0
     1     2     2    20     0     0
     2     3     1    25     0     0
     2     3     2    30     0     0
     2     3     3    35     0     0
     2     4     1    40     0     0
     2     4     2    45     0     0
     3     1     1    50     0     0
     
fmt=['%02d%03d%03d' repmat('%12.4f',1,3) '\n'];

>> fprintf(1,fmt,C.')
01001001      0.0000      0.0000      0.0000
01001002      5.0000      0.0000      0.0000
01001003     10.0000      0.0000      0.0000
01002001     15.0000      0.0000      0.0000
01002002     20.0000      0.0000      0.0000
02003001     25.0000      0.0000      0.0000
02003002     30.0000      0.0000      0.0000
02003003     35.0000      0.0000      0.0000
02004001     40.0000      0.0000      0.0000
02004002     45.0000      0.0000      0.0000
03001001     50.0000      0.0000      0.0000
>> 

You don't have to build the full array; you can iterate and simply pass the current value in a loop if that is more convenient--the key thing is you can write numeric values under whatever formatting you wish using the format string itself.

fprintf(fid,fmt,lay(i),row(j),col(k),x(i,j,k),y(i,j,k),z(i,j,k))

or for whatever is convenient way to address the variables.

There's full documentation at links in the fprintf documentation that explains all the ins and outs.

[Klavers Christoph]

Hello,

thank you so much, it helped me a lot! :)