[gmx-developers] proposed change to g_rdf with -nopbc

[Yang Ye]

My system has an extensible rod-like structure which may elongate more 
than half the diagonal distance. It is then desirable for g_rdf to count 
this distance with -nopbc. I believer that g_rdf actually get this 
distance but because it is out of the cut-off distance, it doesn't get 
counted. To be consistent, -nopbc shall have a (possible) longest 
cut-off distance for RDF calculation. Because max_cutoff2() doesn't 
return such long distance, I add the code to get the diagonal length.

On 5/3/2007 2:05 PM, David van der Spoel wrote:
> Yang Ye wrote:
>> Hi,
>>
>> I would like to add a fix to g_rdf with -nopbc. Instead of taking the 
>> largest distance with PBC, the diagonal length shall be taken as the 
>> cut-off distance.
>>
>> inside do_rdf(): ("red" marks the change)
>>
>>  /* initialize some handy things */
>>  copy_mat(box,box_pbc);
>>  if (bXY) {
>>    check_box_c(box);
>>    /* Make sure the z-height does not influence the cut-off */
>>    box_pbc[ZZ][ZZ] = 2*max(box[XX][XX],box[YY][YY]);
>>  }
>>
>>   if(bPBC) {
>>     rmax2   = 0.99*0.99*max_cutoff2(box_pbc);
>>   } else {
>>   rmax2   = 
>> 0.99*0.99*(box[XX][XX]*box[XX][XX]+box[YY][YY]*box[YY][YY]+box[ZZ][ZZ]*box[ZZ][ZZ]); 
>>
>>   }
>>
> I don't understand... The purpose of this fix was not to have PBC at all.
> If the simulation is with PBC than distance will be less than one 
> box-diagonal anyway. However if you don't have pbc (infinite system) 
> you still would like to be able to handle that in some way.