[gmx-users] g_clustsize

[Vitaly Chaban]

DvdS> But is your index a molecular index now? I think it should be in those
DvdS> versions.

I tried both cases. The result is the same as in the present version.
And the following comment is also present:
"Using molecules rather than atoms. Not reading index file index.ndx"

As for the error, it has magically disappeared after your email although the command
line is the same as earlier...
g_clustsize - mol -n index.ndx


DvdS> Vitaly Chaban wrote:
>> DvdS> Actually the text is incorrect (even the English is weird). If you use
>> DvdS> the -mol flag the program will tell you that the index file is ignored.
>> DvdS> If I recall correctly this wasn't always the case, and you could select
>> DvdS> a subset of the molecules for this analysis. No clue why that disappeared.
>> 
>> Yeah, without the possibility to select the groups to cluster there is
>> no use of the '-mol' key in most cases because in the usual liquid the distances
>> between particles are small enough and all the system appears to be
>> one huge cluster... :)
>> 
>> 
>> I just tried the same with gmx-3.3.1 but I broke with the following
>> error:
DvdS> But is your index a molecular index now? I think it should be in those
DvdS> versions.

>> 
>> Program g_clustsize, VERSION 3.3.1
>> Source code file: smalloc.c, line: 113
>> Fatal error:
>> calloc for block->index (nelem=1069438469, elsize=4, file tpxio.c, line 906)
>> -------------------------------------------------------
>> "Once Again Let Me Do This" (Urban Dance Squad)
>> : Cannot allocate memory
>> 
>> 
>> 
>> DvdS> Vitaly Chaban wrote:
>>>> RF> In this case, you topology is read from the .tpr file and all of the 
>>>> molecules there are taken into account for clustering.
>>>>
>>>>
>>>> I think you are right but the following text from the g_clustsize help 
>>>> message has confused me:
>>>>
>>>>
>>>> "
>>>>
>>>> When the -mol option is given clusters will be made out of molecules rather
>>>>
>>>> than atoms, which allows clustering of large molecules. In this case an 
>>>> index
>>>>
>>>> file would still contain atom numbers or your calculcation will die with a
>>>>
>>>> SEGV.
>>>>
>>>> "
>>>>
>>>>
>>>> So what is the sense to make index.ndx if ALL molecules are used for the 
>>>> calculation always...
>> 
>> DvdS> Actually the text is incorrect (even the English is weird). If you use
>> DvdS> the -mol flag the program will tell you that the index file is ignored.
>> DvdS> If I recall correctly this wasn't always the case, and you could select
>> DvdS> a subset of the molecules for this analysis. No clue why that disappeared.
>> 
>>>>
>>>>
>>>> Vitaly
>>>>
>>