Will Meierjurgen Farr

icon indicating a website link http://www.birmingham.ac.uk/staff/profiles/physics/farr-will.aspx icon indicating an email [email protected]

icon company Stony Brook University icon location Stony Brook, NY, USA icon location Associate Professor, Department of Physics and Astronomy, Stony Brook University and group leader for Gravitational Wave Astronomy at Flatiron CCA

Results 7 comments of Will Meierjurgen Farr

Error message, maybe related to special character in URL

Can confirm: definitely due to the '%2B' in the URL (which corresponds to the '+' characters in the filename).

Errors in Sampling When Parameter Bounds Depend on Parameters

Great---thanks! FWIW, I think it's definitely worth supporting this natively in the backend, even if it adds some complication. It's *super* useful. (I come from the Stan world, and all...

Dense Adaptation Fragile

I can confirm that this behavior persists into the present day. Just did a ~10 dimensional sampling run using PyStan 2.18.0.0 and found *crazy* jumps in log-posterior (i.e. changes in...

Dense Adaptation Fragile

Oh---great. Thanks! Will On Wed, Dec 5, 2018 at 11:55 AM bgoodri wrote: > You need to get the develop version of Stan from GitHub if you want to do...

LKJCholesky returns a Union

Just discovered the same thing, and it's absolutely *killing* performance of Zygote compared to ReverseDiff in a simple model I just wrote. I'm assuming that the problem is somehow in...

Gradients Through Broadcasts Slow Due to Type Inference Failing?

Yeah, I know that much of the time is spent in `log_logistic_unit`---but why? In my profiling, it seems that the comparison (`x > zero(x)`) and the math in `log_dNdx` are...

Gradients Through Broadcasts Slow Due to Type Inference Failing?

Wow! That's amazing---and the performance looks great. Thank you! Just to make sure I understand: you defined an array version of `log_logistic_unit` and `log_dNdx` to sort of "break up" the...