Is there a simple way to tell FindMinimum to constrain certain coordinates when doing multi-dimensional optimization?
For example, I might be working in 3D, with {x,y,z} vectors. But only x and z are allowed to change. y must be fixed at y = 0.
I could re-write the problem to be 2-dimensional and only tell FindMinimum about x and z, but that is a pain to implement. My actual problem is many-dimensional, with a variable number of dimensions. The coordinates which need to be fixed also vary.
It's stunningly boring to find out that all those vitals you measure are mostly not indicating nothing you wouldn't know. In the end, my ECG with 90% mean prediction bands looks like this, over couple years of daily measurements:
So, only place to see some slight variation is in the T wave. Maybe it's good not to have that "interesting" life, but still... ;)
Quantified self and big dada has let me down... :)