how to solve matrix?

1 次查看（过去 30 天）

Ali Mukhtar 2019-9-19

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/481222-how-to-solve-matrix

评论： Guillaume 2019-9-21

if i have matrix M=[1 0 1; 0 0 1; 1 1 1] and T=5 apply a condition on this matrix

if nnz(M)>= T perform task

and make elseif nnz(M)< T should not leave this condition until i make all elements in nnz(M) matrix ==0 but im not able to hold in this condition

回答（1 个）

Fabio Freschi 2019-9-19

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/481222-how-to-solve-matrix#answer_392679

在 MATLAB Online 中打开

If I understand correctly if nnz(M)< T the matrix is the null matrix

M = [1 0 1; 0 0 1; 1 1 1];
T = 5;
 
if nnz(M)>= T
    fprintf('My task\n')
else  % not necessary: elseif nnz(M)< T
    M(M ~= 0) = 0;
end

11 个评论
显示 9更早的评论隐藏 9更早的评论

Guillaume 2019-9-21

在 MATLAB Online 中打开

I don't know what this talk of goto is about. There are no goto statements in matlab. Very few modern languages have a goto. goto has been proven a bad idea long long ago.

As Fabio wrote in a comment in his answer,

if somecondition
    %do something
elseif opposite_of_somecondition
    %do something else
end

the opposite_of_somecondition is pointless. So, if you write if nnz(M) >= T, you know that if you enter the else that nnz(M) is <T, there's no point testing again.

if nnz(M) >= T
    %some code goes here otherwise the if is pointless
else   %no point in checking nnz(M) < T. If we get here, it's guaranteed
    %do something else    
end

Your double for loop is unnecessary. The same code without a loop:

w(w == 0.99) = 1;  %replace 0.99 by 1
w(w < 0.99) = 0;   %replace values less than 0.99 by 0
%values greater than 0.99 are unchanged

and if w is guaranteed to never be greater to 0.99 to start with, it's even simpler, just one line:

w = w == 0.99; %set w to 1 when equal to 0.99, set to 0 otherwise

In any case, what you are trying to achieve is very unclear. You need to explain in more details and give us the big picture.

Walter Roberson 2019-9-21

在 MATLAB Online 中打开

w == 0.99

There might not be any locations in w that are bit-wise identical to the internal representation of 0.99, as 1/10 is not exactly representable in binary floating point, so 0.99 is not exactly representable. There is also round-off difficulties during calculations. You should rarely compare floating point numbers exactly to a literal number that is not an integer (and a good deal of the time, you should not compare exactly to an integer either.)