HYP Smearing routine with one excluded direction at all levels

[vishalrao009]

I am writing a new routine to exclude one of the four directions(links in that direction won't be smeared) in HYP smearing method. There is one assertion at third level "assert(!(temp==temp_chk));" which fails at smear count 11, but works fine upto 10 smear steps. I would like to know if something like this is expected; I tried to check the functioning of original HYP smearing method of SIMULATeQCD and it works fine even at smear count of 211.

Any comments; as to where I might be going wrong.

[clarkedavida]

It sounds like you managed to figure this out? May I close this issue? (Sorry I didn't respond to your original post, but I wasn't involved in the HYP smearing.)

[vishalrao009]

Hi, It seems like the parameters (alpha_i) are probably not optimised for two level smearing; It works fine for roughly all smear steps but fails for just one configuration out of 500. anyway I would like to get the code seen from one of experts if possible. Thanks

[clarkedavida]

Rasmus, I think you oversaw this getting merged into the main. Any advice?

[RasmusNL]

Hi David

Sadly I dont think I have anything smart to say. The code was a port from (as far as I remember) MILC, and was ported by Dan Hoying. I mainly checked for consistency with known results.

Regards Rasmus

[vishalrao009]

Hi David,

I just checked that even smear_all function (original code) crashes due to assertion failure in some of the gauge configurations (in third level) for smaller temperature. Seems like the assertion failure occur more often for smaller temperature configurations. For highest configuration it did not occur at all (maybe cause had 500 for highest and 1800 for smallest temperature). any comments? could you assign Dan Hoying for looking into it?

Thanks Vishal

[clarkedavida]

It seems that Dr. Hoying is not active developing anymore, taking a break for health reasons, so I'm not sure I can reach out to her. I'm sorry we can't be of more help Vishal; you unfortunately may have to look into the code somewhat deeply, trying to understand the purpose of the assertion and whether it makes sense for your situation.

EDIT: Another option might be to reach out to one of the MILC developers, since Rasmus says it's a port from MILC, which you can find here: https://github.com/milc-qcd.