Because inf is the extreme, you can use it to find the next value in your data. As an example, the code below finds the maximum value in a vector:
max_val=-inf;
data=rand(1,100);
for n=1:numel(data)
if data(n)>max_val%is bigger than current max?
max_val=data(n);%store new max
end
end
Now the -inf will be replaced by the actual larger value once it finds one. The only case where it will stay -inf is when all values in the data are -inf, in which case it is indeed the maximum value.
Minimization works the same in the reverse case.
