Re: [ESPResSo] Implementation of full DPD algorithm in Espresso ?

[Marek Maly]

Dear Christoph,

thank you very much for your complex response.

My problem is that I am really under big time press so
I already found temporary solution.

I simply modified already existing  "Soft-sphere interaction" resp.

"soft_sphere.h" as cheapest way as  possible.

Namely functions:

MDINLINE double soft_force_r(double a, double n, double r )
MDINLINE double soft_energy_r(double a, double n, double r)

For those who are interested I attached modified header file.


I kept there all the original parameters (to prevent me from errors  
regarding mismatch of func. prototypes in different files).

I call it in script like this:

inter 0 0 soft-sphere $a_00 2 $Rc 0

where $a_00 has now meaning of the repulsive DPD parameter, second  
parameter (here 2) can be chosen arbitrary (is not used inside redefined  
functions anymore),
$Rc is DPD cutoff and should be set to 1 (Rc is in DPD unit of the length)  
and the last parameter (original r_offset) should be 0 for DPD  
redefinition.

if I combine it with

thermostat dpd $T $gamma $Rc

if works perfectly ( I compared to Energies, pressure from my own old DPD  
software ).

Anyway thank you for the detailed description of adding of the new  
functions into existing Espresso code !
I will use it surely in the future.

Anyway thanks to all developers for this very nice software. I am new here  
but anyway it seems that I already
fell in love with Espresso :))

Best wishes,

   Marek






Dne Mon, 29 Mar 2010 10:45:40 +0200 Christoph Junghans  
<address@hidden> napsal/-a:

> Am 27.03.2010 16:18, schrieb Marek Maly:
> > The soft conservative DPD force/potential is unfortunately not
> > implemented. Am I right ?
> You are right. I thought about implementing it one day...
>
> > If I am right I would like to know how to use/modify Espresso to be able
> > to carry out full DPD
> > simulation (including DPD conservative force).
> Implementing the conservative DPD part is not difficult. Just use a  
> simple non-bonded interaction (like ljgen) as template.
>
> > One way is probably to use DPD thermostat together with tabulated DPD
> > potential/force but this is
> > a little bit uncomfortable solution.
> And slow....
>
> > The second way is to implement conservative DPD force/potential directly
> > into Espresso code like Leonnard-Jones, Morse etc.
> >
> > Espresso developers probably has no interest here ? but I can try it by
> > my self (I have experience with C/C++ however I am pretty new in TCL).
> > I can learn from implementations of the other potentials in Espresso but
> > I would be grateful for some info, about the list of all files which
> > I have to create or modify in order to implement successfully this new
> > feature.
> >
> > Maybe the easiest way could be also just proper redefinition of some of
> > the already existing potential (Soft-sphere for example), just to  
> > prevent
> > me from the problems with compilation of Espresso with new code files.
> > But again I need to know all the files (.h,.c,.tcl) which I have to
> > modify .
> I would call the conservative DPD part nb_harmonic, because it is  
> basically like a spring, but non-bonded.
>
> How to create a new non-bonded interaction?
>
> -Add new interaction parameters to IA_parameters struct in  
> interaction_data.h
> -Add these parameters to initialize_ia_params and copy_ia_params in  
> interaction_data.c
> -Add interaction cutoff to calc_maximal_cutoff in interaction_data.c
> -Add a check if interataction is on to checkIfInteraction in  
> interaction_data.c
> -Create a new .c,.h file with the following 4 functions (see ljgen.c as  
> example):
> int printYYYIAToResult(Tcl_Interp *interp, int i, int j);
>
> int YYY_parser(Tcl_Interp * interp,
>                 int part_type_a, int part_type_b,
>                 int argc, char ** argv);
>
> void add_YYY_pair_force(Particle *p1, Particle *p2,
>                          IA_parameters,*ia_params,double d[3],
>                          double dist, double force[3]);
>
> double YYY_pair_energy(Particle *p1, Particle *p2,
>                           IA_parameters *ia_params, double d[3],
>                           double dist);
>
> where YYY stands for the new interaction name (e.g. ljgen).
>
> -Add printYYYIAToResult to printNonbondedIAToResult in interaction_data.c
> -Add YYY_parser to inter_parse_non_bonded in interaction_data.c
> -Add add_YYY_pair_force to calc_non_bonded_pair_force_parts in  
> pressure.h (!!!)
> -Add YYY_pair_energy to calc_non_bonded_pair_energy in energy.h
> -Add the new .h, .c file to Espresso_bin_SOURCES in Makefile-am.am and  
> run autogen.sh
> -Compile and TEST
>
> -Put all new code in '#ifdef YYY' blocks
> -Add YYY to compilation_callback in config.c
> -Add YYY to config-sample.h
> -Create a patch using
> cvs diff -u (CVS VERSION) or
> diff -Naur original_Espresso_dir your_Espresso_dir (release version)
> -Send it to us
>
> I think that's it. If you need any help just ask again.
> If you send us the patch and we are happy to include the code in our  
> next release.
>
> Cheers,
>
> Christoph
> > Thank you very much in advance for your help !
> >
> > Best wishes,
> >
> > Marek


--
Tato zpráva byla vytvořena převratným poštovním klientem Opery:  
http://www.opera.com/mail/

soft_sphere.h
Description: Binary data