A problem with a matrix

Question

Artem Smirnov 2016-11-17

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/312756-a-problem-with-a-matrix

评论： Artem Smirnov 2016-11-17

Hello. I am doing research connected with magnetic observatory data. I have 50 observatories and each of them produces 12 directories a year (for every month) with 30 files inside. That is why I need a loop. I downloaded data from the files successfully and counted a "MA-value" in a Class.

    for k=1:1:12
meas=CCmpNew(1000,rect,0); %MA-value in a class
a = find(meas<0.4); 
for i=1:kmax    %kmax is the number of days in a month
b = a>(i-1)*1440 & a<=i*1440; 
n(i)=sum(b)/1440; 
for j=1:kmax*1440 
if j>(i-1)*1440 && j<=i*1440 
if meas(j)>0.55 
n(i)=0; 
end 
end 
end 
end
[N,I]=sort(n,'descend');
for d=1:nnz(N)       % It seems to me that the problem is in this loop.
kdu_I(d)=I(d);
kdu_N(d)=N(d);
K_(d)=k;
end
C=[K_' kdu_I' kdu_n'];
dlmwrite('path',C,'-append');
  end

The problem is that when I write C matrix (actually, append it to file), if on some iteration there were less elements in C than on the previous iteration, the programme completes C up to the dimension of it on the previous iteration. How can I fix it? Could you please help me with this? I know my mistake is a silly one, but I'm beginner...

2 个评论
显示无隐藏无

Guillaume 2016-11-17

编辑：Guillaume 2016-11-17

To make it easier for others (and yourself) to follow your code, I would recommend that you use

proper indenting for loops, ifs, etc.
meaningful variable names, so that we don't have to go back through the whole code to remember what was a, b, c, d, n (and N!), etc. If k is the directory index, why not call the variable directoryindex. Immediately, it's clear on any line where it's used, what it contains.
comments that explain what the purpose of each loop is.

John D'Errico 2016-11-17

So many times when I want to be able to upvote a comment. Guillaume's comment is spot on. The code posted is quite difficult for anyone to read. And in order for us to debug that code, we need to completely understand what the OP is doing, and why. Since all we have to go on is the code, that can be nearly impossible to do.

Writing code that you can actually read is incredibly valuable as a programmer. It allows you to return to that code next month or next year when possibly you find a problem.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Guillaume 2016-11-17

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/312756-a-problem-with-a-matrix#answer_243709

编辑：Guillaume 2016-11-17

在 MATLAB Online 中打开

The problem is not with C, it's with the kdu_I, kdu_N and K_ and potentially n (seriously, what useless names! See my comments to the question) which are never initialised and just grow as required (but never shrink) on each iteration of the loops.

n = zero(1, kmax);  
for i = ...
%...
[N, I] = ...
kdu_I = zeros(1, nnz(N));
kdu_N = zeros(1, nnz(N));
K_ = zeros(1, nnz(N));
for d = ...

would fix the problem.

However, I'm fairly certain that none of the loops are necessary but because of the lack of comments, indentation, and poor variable names, I've not tried to understand what exactly you are doing (see comment to question). Certainly, the whole d loop + variable initialisation above + assignment to C can be replaced by the much simpler:

C = [repmat(k, nnz(N), 1), nonzeros(N), I(1:nnz(N)).'];

If you make the rest of the code clearer you'll probably get suggestions on how to simplify it.