Content

• We separate out the expected information Gain from GBOED to create the IES (iterated Emulate-Sample) algorithms. These simply sample the posterior without doing information gain.
• We find that these have almost identical performance to GBOED except in 1D when EIG gives some benefit

MISC

• We also run the missing adam and lm experiments
• Add family groups to the suitability tables, method taxonomy, and to help organize the plotting
• Remove score toggles