i'm dealing with a least squares(SCIPY) problem. usually a least sq problem has the X data and Y data given and the general form of the function is given such as linear/quadratic/log and based on the x and y data, we optimize the parameters of the curve equation-this is CURVE FITTING
but what i wanna do is given the Y data and the exact form of the function, estimate what the X data would be...the x-data is multidimensional, i.e there are 40 parameters of which only 4 are realized in every experiment. I've shown an example below-
X data Y data function
a,b,c,d 0.4 a+b+c+d
a,c,d,e 0.2 a+c+d+e
c,d,e,k 0.7 c+d+e+k
..................................
..................................
..................................
so:
a+b+c+d = 0.4
a+c+d+e = 0.2
c+d+e+k = 0.7
...........
............
..........
ASSUME WE HAVE LARGE AMOUNTS OF SUCH DATA
so the function is exactly known, it is simply the sum of the x data. there are many parameters to be estimated,(40), in each experiment only 4 of these parameters are realized, and each x data has the corresponding y value IS THIS EVEN POSSIBLE??
NOT IMPORTANT I AM WORKING ON A QUESTION RELATED TO DNA-TRANSCRIPTION FACTOR BINDING
