Dear David,
thanks for pointing me to the respective code line!
I'll try that.
Best,
Anselm
> On Wed, Jan 01, 2025, PD Dr. Anselm Horn via AMBER wrote:
>>
>> I have a very large system with a cuboic solvent box that largely extends
>> in z direction (> 1500 A) in a rod-like manner.
>> When trying to run a first minimization with positional restraints via
>> pmemd.cuda, I obtain the following error:
>>
>> | ERROR: c must be in the range of 0.10000E+01 to 0.10000E+04!
>
> This is just saying that the box length along the z-direction (called "c")
> must be less than 1000 Ang. You can try changing this with the "boxhi"
> parameter, which is set at about line 142 in
> $AMBERHOME/src/pmemd/src/mdin_ewald_dat.F90. This change may also affect
> the needed value of gridhi.
>
> I'd recommend trying short initial calculations in the CPU version. If
> things
> look OK there, make sure the GPU version is giving results that are the
> same
> as on the CPU. There may be changes needed in the GPU code as well to
> accommodate such a large box size, but I don't think so (see below).
>
> FWIW, I've run rod-like simulations with a box length in the long direction
> of 710 Ang. without any problems (at least that I noticed!). So twice that
> length is probably going to be OK. The upper limits in pmemd were
> established 20 years ago or so, and probably assumed that a box length more
> than 1000 Ang. was likely to be an input error.
>
> ...good luck...dac
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jan 07 2025 - 08:30:02 PST
