Re: AMBER: subroutine ephi !!

From: Tom Darden <>
Date: Fri, 19 Oct 2007 08:06:33 -0400 (EDT)

dear sampath
it may take a while to sort out the ephi code. there are strange codings
to signal improper dihedrals and other situations i don't immediately
recall. this was written in the mid-80's when dihedrals were a major
fraction of the simulation run time and tricks were used to better
vectorize it. it was cleaned up a bit more recently but now has a lot of
other complexities like les and evb added on top

If you like, you can take a look at the dihedral code in
amoeba_valence (in sander version 9--if you have 8 or earlier i'll send
you the amoeba valence code and a sample prmtop) look at the "torsions"
section. i tried to lay it out clearly. its a little different
algorithm with respect to how the dihedral derivs are defined--i can try
to help with any confusions.

if you want to stick with ephi i can try to help--it takes anyone a
little time to see what's happening there. for starters the coords are
stored in x as a 3N dimensional array i.e. x(1) is xcoord of atom
1, x(2) is ycoord of atom 1, x(3) is zcoord of atom 1, x(4) is
xcoord of atom 2, x(5) is ycoord of atom 2 etc ---note that atom offsets
are multiples of 3 in ip, jp, kp and lp. forces come back in f with the
same offsets

email if you have more queries
tom darden
  On Fri, 19 Oct 2007, Sampath Koppole wrote:

> Hello,
> For some test, I am trying to do MD using only the
> dihedral angles phi, psi and omega. I m writing a
> FORTRAN code to do this and want to reuse the
> "subroutine ephi" present in "ene.f" (in the sander
> directory).
> subroutine
> ephi(nphiin,ip,jp,kp,lp,icp,cg,iac,x,xx,ix,f,dvdl, &
> ep,enbp,eelp,dcharge,qpsander)
> Could someone tell me what are the different
> variables. Which is the array thro' which we give the
> coordinates to ephi to return us the dihedral angle
> and the forces ?
> Unfortunately, the source is poorly documented.
> Thanks a lot for your help.
> With best regards,
> Sampath.
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Oct 21 2007 - 06:07:47 PDT
Custom Search