Remove specific rows of matrix from memory not just delete them

I'm writting an excel using Matlab. The problem is that when I'm trying to remove certain rows I get the result shown in the jpg. I don't want the empty rows.

for c = 1:1000
    u=H(c,1)
    if (u~=0)
        Hf(c,1)=H(c,1);  
    else
        Hf=NaN;  
    end  
end  

Answer 1

[Image Analyst]

Try this instead of all that code you have:

rowsToTransfer = (H(:, 1) ~= 0);
Hf = H(rowsToTransfer, :);

If you just want the first thousand rows, you can do this:

rowsToTransfer = (H(1 : 1000, 1) ~= 0);
Hf = H(rowsToTransfer, :);

Answer 2

[dpb]

Use MATLAB <logical addressing> and vectorized operations--

Hf=Hf(HF~=0);

for the case you show that Hf is a vector. If it were an array but to remove the rows based on the one column is almost as easy--

Hf=Hf(HF(:,1)~=0,:);

The above is easier for the novice to see if write

ix=(HF(:,1)~=0);        % the logical T/F vector of column 1 not zero
Hf=Hf(ix,:);            % keep those rows, all columns

[Image Analyst]

It appears H is one matrix, and Hf is a second, output matrix. At least that's what I got from his code attempt.

[oscillator]

Thank you for your answer. I know it's a different question but can you tell me how to import more than one matrixes in the same excel for example Hf and H ? I use the following command for each one :

xlswrite('file.xlsx',Hf(:,1))

Is the fact that they now have different number of rows going to be a problem?

[oscillator]

I'm trying to accept this answer as well. Thanks for your help.

[dpb]

"...H is one matrix, and Hf is a second, output matrix...."

Yes, that's what OP wrote,but given his problem description there's no need to not do the operation in place, I didn't see any reason to not write the optimal solution of not duplicating data needlessly. Of course, it is always possible may want/need to keep the original as well; in that case then, indeed, would need to have another LHS variable or another solution.

[dpb]

It's more inefficient to write multiple times to the same workbook/worksheet if can avoid it, but if it can't be helped, then use the optional 'Range' named parameter to set the location in the worksheet at which to write the additional data that doesn't overwrite the first. These are all illustrated in the documentation.(*)

Unless you are using a truly by now ancient version of MATLAB, xlsread and xlswrite have been deprecated for ages; use writematrix and friends, instead. They are much improved in both features and speed.

(*) The best way in a case such as this would be to create the data as either a 2D array and augment the shorter column with NaN to the length of the longer; then writematrix() the array. Or, create a cell array that is 2x1 with each cell containing the given column as a vector of doubles. Then use writecell instead.