Work on vi visualized
1 unresolved thread
1 unresolved thread
Merge request reports
Activity
Filter activity
@parras does this look reasonable? If you like it then I think it can be merged.
19 19 ############################################################################### 20 20 # Variational Inference (VI) 21 21 # 22 # This script demonstrates how MGVI and GeoVI work for an inference problem 23 # with only two real quantities of interest. This enables us to plot the 24 # posterior probability density as two-dimensional plot. The approximate 25 # posterior samples are contrasted with the maximum-a-posterior (MAP) solution 26 # together with samples drawn with the Laplace method. This method uses the 27 # local curvature at the MAP solution as inverse covariance of a Gaussian 28 # probability density. 22 # This script demonstrates how MGVI, GeoVI, MeanfieldVI and FullCovarianceVI 23 # work for an inference problem with only two real quantities of interest. This 24 # enables us to plot the posterior probability density as two-dimensional plot. 25 # The approximate posterior samples are contrasted with the maximum-a-posterior 26 # (MAP) solution together with samples drawn with the Laplace method. This changed this line in version 4 of the diff