Dear all,
Amber is right, the issue has been resolved as an input error:
The two simulations mistakenly used the same ig-value and thus correctly
produced the same trajectory; for actual different ig-values different
trajectories are generated, as expected.
Many thanks to Adrian Roitberg for his support here and apologies for
any confusion caused.
All the best,
Anselm
Am 04.02.2022 um 19:53 schrieb Carlos Simmerling:
> not that I see, but if you set nstlim smaller and then ntpr=1, you can more
> easily check differences in energies and velocities.
>
> On Fri, Feb 4, 2022 at 12:38 PM Dr. Anselm Horn <anselm.horn.fau.de> wrote:
>
>> Dear Carlos,
>>
>>> correcting my typo: if you set tempi>0 (not temp0>0)
>>> then the seed should
>>> affect the velocity assignment.
>>
>> yes, that's what I thought.
>>
>> I had the following input for this test:
>>
>> &cntrl
>> imin = 0,
>> nstlim = 100000,
>> dt = 0.002
>> ntt = 1,
>> temp0 = 310.0,
>> tempi = 050.0,
>> ntp = 1,
>> ntb = 2,
>> ntc = 2,
>> ntf = 2,
>> ntr = 1,
>> restraint_wt = 5.0,
>> restraintmask =':1-100',
>> ntwx =10000,
>> ntwe =10000,
>> ntwr =500000,
>> ntpr =10000,
>> ioutfm = 1,
>> ig = 112233,
>> &end
>>
>> Is there obvious mistake?
>>
>> Best regards,
>>
>> Anselm
>>
>>
>>
>>
>> Am 04.02.2022 um 18:21 schrieb Carlos Simmerling:
>>>
>>> On Fri, Feb 4, 2022 at 12:11 PM Carlos Simmerling <
>>> carlos.simmerling.gmail.com> wrote:
>>>
>>>> ntt=1 doesn't use the random number generator, so won't be affected by
>> the
>>>> seed (ntt=3 is). if you set temp0>0, then the seed should affect the
>>>> initial velocity assignment.
>>>>
>>>> On Fri, Feb 4, 2022 at 12:01 PM Dr. Anselm Horn <anselm.horn.fau.de>
>>>> wrote:
>>>>
>>>>> Dear Carlos,
>>>>>
>>>>> I did not read in any velocities, since I used a structure from a
>>>>> minimization. The thermostat was NTT=1.
>>>>>
>>>>> Do I miss something?
>>>>>
>>>>> Best regards,
>>>>>
>>>>> Anselm
>>>>>
>>>>>
>>>>> Am 04.02.2022 um 16:00 schrieb Carlos Simmerling:
>>>>>> which thermostat did you use? also, did you read velocities from the
>>>>> input
>>>>>> coordinates? these are two places that can be affected by the seed.
>>>>>>
>>>>>> On Fri, Feb 4, 2022 at 9:09 AM Dr. Anselm Horn <anselm.horn.fau.de>
>>>>> wrote:
>>>>>>
>>>>>>> Dear Amber experts,
>>>>>>>
>>>>>>> I have a short question regarding the random number generator for
>>>>>>> initial velocities (TEMPi) on GPUs:
>>>>>>>
>>>>>>> I ran two MD simulations in a periodic solvent box starting from a
>>>>>>> minimized structure, with weak-temperature coupling (NTT=1) and an
>>>>>>> initial temperature of 50 K (TEMPi=50.0); for each simulation, I
>> used
>>>>>>> an explicit, but different random number seed (ig=NNN). I used the
>>>>>>> executable pmemd.cuda_DPFP (Amber20, patch12) on a A40-GPU, since my
>>>>>>> original intention was to test whether system relaxation might be
>> save
>>>>>>> on GPUs.
>>>>>>>
>>>>>>> The resulting trajectories of the two runs, however, were identical,
>>>>>>> although different random seeds were used.
>>>>>>>
>>>>>>> Is the ig-seed not used on GPUs, or for NTT=1 in general?
>>>>>>>
>>>>>>> Maybe I misunderstood the section 21.6.7 in the Amber manual about
>>>>>>> temperature regulation...
>>>>>>> I did not check ig=-1 up to now, though.
>>>>>>>
>>>>>>> Best regards,
>>>>>>>
>>>>>>> Anselm
>>>>>>>
>>>>>>> Bioinformatik | NHR.FAU
>>>>>>> Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)
>>>>>>> Germany
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> AMBER mailing list
>>>>>>> AMBER.ambermd.org
>>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>>>
>>>>>> _______________________________________________
>>>>>> AMBER mailing list
>>>>>> AMBER.ambermd.org
>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>
>>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Feb 09 2022 - 08:00:02 PST
