combine for loop and ~isinf,but there is an error in every loops except the result in the first loop

I use ~isinf and a for loop to make a code,in this code ,i want to let the un-Inf value to show me in each loop,however,i found that there is a error here,but i can't find where do i wrong

c=[ 3 6 1 Inf]
fg=c(~isinf(c))
for k=1:4
   ad=c(~isinf(c(k))) 
end

So in fact ad should show me as the same as each element in fg,like this

fg =  3 6 1 
ad =  3
ad =  6
ad =  1
ad =  []<-----maybe won't be shown

However,the window show me

fg =  3 6 1 
ad =  3
ad =  3
ad =  3
ad =  []<-----maybe won't be shown

Where i am wrong with the code?

Answer 1

[Stephen23]

v = [3,6,1,Inf];
for k = 1:numel(v)
    n = v(k);
    z = n(~isinf(n)) 
end

displays:

z =  3
z =  6
z =  1
z = []

"i can't find where do i wrong"

Your code uses a scalar logical index:

~isinf(c(k))

Logical indexing is positional (it is not like subscript or linear indexing), so when you use a scalar logical index it acts only on the first element of the array, i.e. you either select the first element or you do not. That is exactly what your example output shows.

You cannot use a scalar logical index to work like a subscript and also expect it to select different elements of an array (i.e. not the first element). For that you would need to create a non-scalar logical index (normally these are the same size as the array being indexed) or use subscript/linear indexing first (which is what my answer does).

[yang-En Hsiao]

If i want to mean the non-inf value,(3+6+1)/3=3.3333,what can i do with the code,because i found that if i write the code like this

v = [3,6,1,Inf];
for k = 1:numel(v)
    n = v(k);
    z = n(~isinf(n)) 
end
meann=mean(z)

or

v = [3,6,1,Inf];
for k = 1:numel(v)
    n = v(k);
    z = n(~isinf(n)) 
    meann=mean(z)
end

The result of both code are wrong.

[Stephen23]

"The result of both code are wrong."

They are "wrong" because you did nothing in your code to actually store those values in an output vector: you would need to use indexing or concatenation to make that work...

But personally I would not bother, because that approach is very inefficient. It is much simpler and more efficient to use logical indexing and avoid the loop entirely:

>> v = [3,6,1,Inf];
>> m = mean(v(~isinf(v)))
m =  3.3333

Read more here:

https://www.mathworks.com/help/matlab/matlab_prog/vectorization.html

And also this:

https://www.mathworks.com/matlabcentral/answers/228557-experts-of-matlab-how-did-you-learn-any-advice-for-beginner-intermediate-users

[yang-En Hsiao]

Why do i write code as this? because in fact i need to solve the mean question in the code,in that code, there is no a vector like

c = [3,6,1,Inf];

to let me use the

mean(c(~isinf(c)))

directly,the elements like 3,6,1,Inf, in that code is the result from every loops.that's why i ask how should i mean the z instead of c,because as we know,the z is the result of the every loop,but c is not .

[Stephen23]

"...the elements like 3,6,1,Inf, in that code is the result from every loops..."

Then the simplest solution is to store all of those values inside the loop, and then use the trivial logical indexing after the loop:

y = nan(1,N); % preallocate!
for k = 1:N
    y(k) = ... your value
end
m = mean(y(~isinf(y)))

Trying to select the non-inf values inside the loop will just make your code more complex, and also less efficient. Preallocation is also important to learn about:

https://www.mathworks.com/help/matlab/matlab_prog/preallocating-arrays.html

All beginners need to learn how to preallocate their arrays! It improves efficiency.

[Walter Roberson]

ad = [];
for k = ...
    ...
    if isfinite(z)
       ad(end+1) = z;
    end
end
mean(ad)

or

for k = ...
   ...
   ad(end+1) = z;
end
mean(ad(isfinite(ad)))

[yang-En Hsiao]

you mean like this ?

v = [3,6,1,Inf ]; 
z=nan(1,length(k))
for k = 1:length(v) 
n = v(k)
z(k) = n(~isinf(n )) 
end 
meann=mean(z(~isinf(z )))

but the error show to me

In an assignment  A(I) = B, the number of
elements in B and I must be the same.

Error in Untitled2 (line 6)
z(k) = n(~isinf(n ))

[Stephen23]

"you mean like this ?"

No. You told us that you generate "the result from every loop", so that is what my code requires.

Also your length(k) does not make much sense for array preallocation, because k is a scalar.

You should do something like this:

v = [3,6,1,Inf];
N = numel(v); % or however you define the number of loops.
y = nan(1,N); % preallocate!
for k = 1:N
    y(k) = v(k); % your value (calculated...)
end
m = mean(y(~isinf(y)))

Where, clearly, in your actual code you generate that value, rather than selecting elements of the vector v like we are using here for these examples.