Normalize unit vector to single didgit integers
10 views (last 30 days)
I am trying to normalize a vector to integers. For example, I would like [ .4 .-4 .8] to be [ 1 -1 2]. However, if it is not a perfect for example, [.123 .123 .5] would be [123 123 500] but I would like it to be rounded to a single didgit integer. I.E instead of [123 123 500] it would be [1 1 5].
Ive seen v=x/norm(x) but this does not work even for obvious cases.
the cyclist on 15 Sep 2021
Does the following do what you want?
f = @(t) round(t/abs(min(t)));
f([ 0.4 -0.4 0.8])
f([ 0.123 0.123 .5])
It gives a different result for your second example, but I think it is a better normalization because it is closer to the original ratio.
Stephen23 on 16 Sep 2021
Edited: Stephen23 on 16 Sep 2021
fun = @(V) round(V./10.^floor(log10(abs(V))));
Note that this scales each element independently, to give the result you asked for. The ratio of values is better represented by the cyclist's answer.