icouckuy

@nknudde what's the state on this? Is it more compatible with GPflow 1.0? Or is using MCMC to marginalize hyperparameters fast and accurate enough?

These warnings should be relatively easy to fix. I'm just wondering if GPFlow 0.5.0 itself also generates warnings? In general we try to depend on the same tensorflow and numpy...

> I disagree we should automatically detect no feasible points and switch to PoF only. I'd leave that out of the library. Its possible to first use a BayesianOptimizer with...

For expensive constraints we can take a look at Gelbart's dissertation section 2.6: https://dash.harvard.edu/bitstream/handle/1/17467236/GELBART-DISSERTATION-2015.pdf?sequence=1 It summarizes a number of proposed constrained acquisitions functions over the years. At the moment we...

@yangnw The current plan is to store the (cheap) constraints in the Domain object. This can include coefficient matrices (for whitebox linear constraints) or function handles (for cheap blackbox constraints)....

Are you running master or the gpflow-1.0-v3 branch? In the latter I added some commits to set TF logging to WARN in BOPtimizer while waiting for a proper fix.