Max Balandat

Are you using any priors / constraints on the respective models? In general a single-task MTGP should reduce to a regular GP but it's possible that it's parameterized somewhat differently...

Thanks for flagging this @saitcakmak, @dme65 re side effects of changing gpytorch settings upon import in botorch - We had talked about this before, can you please look into finding...

> it is switched back to model.train() just to be switched back again to model.eval(). Is there a reason to do this? Did you check the timing if you don't...

Thanks. I expect this to cause some backward compatibility problems though, e.g. [here](https://github.com/pytorch/botorch/blob/3e34a4fc35d5b01f7b6e256a92cc4b351d361cbe/test/acquisition/test_analytic.py#L402) in BoTorch. Not sure if it's worth the trouble of changing this in a way that deprecates...

I don't see any obvious issues with this. One downside of passing `**kwargs` through is that it's hard to know if some of them get silently ignored, but that's a...

You can't just assign the attribute for the prior, you need to use the [`register_prior()`](https://github.com/cornellius-gp/gpytorch/blob/master/gpytorch/module.py#L203) method on the GPyTorch module so that gpytorch knows to use this in the computations...

Technically, the `noise_covar` here can be arbitrary, right? I.e. in the general case this would be `K + Sigma` where `Sigma` is p.d. (either non-uniform noise levels, or potentially even...

cc @SebastianAment, @sdaulton, @dme65 for any potential ideas here

@gpleiss any thoughts on this?

If you only have a 100 data points per task (and the locations are shared across all tasks) you can also use https://github.com/cornellius-gp/gpytorch/blob/master/examples/03_Multitask_Exact_GPs/Multitask_GP_Regression.ipynb, which uses the Kronecker structure of the...