[AMBER] Energy different between igb=0 and igb=6 during vacuum minimization with sander

From: Jingxiao Bao <m15295783798.163.com>
Date: Fri, 19 Jul 2019 21:52:44 +0800 (CST)

Dear amber users and developers:


Recently, I am using sander (in ambertools18) to do minimization of my complex in vacuum. According to the manual, setting ˇ°igb=6ˇ± is logically equivalent to ˇ°igb=0, eedmeth=4ˇ±. However, the results I got are different.


I attached the topology (COM.top) and initial coordinate files (COM.crd) of my complex here, as well as the out files from sander with different settings.


As you can see in the attachments, the complex total energy of step 1 in ˇ°COM_min_sander_igb6_cut10.outˇ± (where I set igb=6 and cut=10.0) and ˇ°COM_min_sander_igb0_eedmeth4_cut10.outˇ± (where I set igb=0 and cut=10.0 plus eedmeth=4) are both -1.3701E+03. However, starting from step 2, the energy became different.


I also tried increasing the cut to 9999.0, and I got identical results between igb=0 (COM_min_sander_igb0_eedmeth4_cut9999.out) and igb=6 (COM_min_sander_igb6_cut9999.out). So it seems that the difference may related to the treatment of nonbonded pairlist between igb=0 and igb=6.


I want to know if I made some mistakes in my inputs, or the difference is made intended, or this is a bug?


PS: Due to some network problems here, I cannot download ambertools19 right now, so I can only do these tests in ambertools18. My ambertools18 was compiled with gnu 6.4.3 and all the tests were passed. I also ran very short MD (only 10 steps) with sander at 1K with ig set to the same number, and there were no energy difference between igb=0 and igb=6 even with cut=10.0.


Thank you.

Jingxiao Bao

AMBER mailing list

Received on Fri Jul 19 2019 - 07:00:02 PDT
Custom Search