# Inf problem

33 views (last 30 days)
shab shekan on 1 Aug 2011
I'm working on video, and have alot of computations in my processes. in one part i divide a number on another one, but I receive wrong result.
for example:
1.000000000000000e-03/ 1.000000000000000e-03+ 0
ans
=Inf
how can i handle this?
I use format short, but it doesn't help me.

Arnaud Miege on 2 Aug 2011
You probably want (note the brackets):
m12 = deltaR12 / (min(deltaRF(ii, jj, tt), deltaRF_V(ii, jj, tt)) + epsilon);
Your code, as is divides deltaR12 by min(..,..), and then adds epsilon to the result of the division, which I suspect is not what you want.
HTH,
Arnaud
Walter Roberson on 2 Aug 2011
Ever since Mr. D. Adams (RIP), suppliers have been demanding a share of intellectual property rights for everything that is powered by a Nice Hot Cup Of Tea.

Walter Roberson on 1 Aug 2011
Please copy the exact code that reproduces this on your system. It appears that the denominator on the division is becoming 0, which could perhaps happen due to the exact spacing you use.
Also, I suggest you try
format short g
shab shekan on 2 Aug 2011
epsilon=.001;
c=0;
format short;
for tt=1 : nframes-1
for ii=1: height -1
for jj=1: width -1
M12 = M(ii*3-2:ii*3, jj*3-2:jj*3, tt) + M_V(ii*3-2:ii*3, jj*3-2:jj*3, tt);
deltaR12 = det(M12) / det(M12(1:2 , 1:2)) * norm(M12(1:2 , 1:2), 'fro');
if isnan(deltaR12)
deltaR12 =0.001;
end
m12 = deltaR12 / min(deltaRF(ii, jj, tt), deltaRF_V(ii, jj, tt)) + epsilon;
c = c+ m12;
end
end
end
c=1/c*N;
-----
denominator never can be zero, because of 'epsilon' is constant in denominator. My problem is in this line "m12 = deltaR12 / min(deltaRF(ii, jj, tt), deltaRF_V(ii, jj, tt)) + epsilon;"
and I used "format short g" before, and it doesn't usefull for my problem.
by the way, thank you so much walter.