How to introduce equipment error in the weight matrix of the data misfit as uncertainty?
Hi jcbarreto,
There are several way you can handle it. One can be at the Survey level:
survey.std = 0.05 # standard-deviation expressed as percent of the data. Can be either a scalar or a vector of size "number of data"
survey.eps = 0.01 #floor value, expressed in the same unit as the data. either scalar or vector too
This syntax of the Survey class might however change with the arrival of the simulation class. One other way that I think will still work would be to attribute it at the creation of the DataMisfit object:
dmis = DataMisfit.l2_DataMisfit(survey, std=0.05, eps=1e-4) #std and eps can be either scalar or vectors
And if really you need something more specific. You can act on the weight matrix directly, for example, what setting a std and an eps does under the hood is:
from SimPEG import Utils
std = 0.05
eps = 1e-4
dmis.W = Utils.sdiag(1/(abs(survey.dobs)*std+eps))
I hope this helps.
3 Likes