Regression Test Failure about maestroex HOT 6 CLOSED

I've tried running the double_bubble test with DEBUG=TRUE, and located the crash to here

This is within a function that is not tiled. It seems like there is an error with utrans - when I try printing it, the program crashes at a corner (i=64, j=-1). utrans is passed as an argument and is calculated by functions called just before VelPred. I'll try exploring them to see if I can work out why it might be crashing. My guess would be that something is not being initialised correctly and because the DEBUG mode compiler is stricter it's throwing an error

from maestroex.

I think I've fixed this with ea05efe and 8d1a646 - a couple of MultiFabs were not being initialised properly. double_bubble now runs for me without crashing with DEBUG=TRUE

from maestroex.

I'll re-run the entire regression suite now and report back.

from maestroex.

Regression failed because some other commit to amrex breaks compilation. Looking into the other issue now.

from maestroex.

Everything works now except for wdconvect
https://ccse.lbl.gov/pub/RegressionTesting/MAESTROeX/

from maestroex.

Regression test fixed; wdconvect benchmark is being updated. The tilling implementation changes the order in which random numbers are generated in the velocity initialization, so we will keep the new result.

from maestroex.