Switching track variable precision to float32
Summary
This MR introduces the following changes
- Switching the default precision of the track variables from
float16
tofloat32
. - Adding check against
inf
as scaling/shift value.
Relates to the following issues
- Closes #206 (closed)