********> bugfix.29
Correction Author: David A. Pearlman
Correction Date: 1/8/92

Programs: gibbs

Severity: slight

Problem:  The Gibbs manual makes the following claim:

"  10.11  ITRSLU    If ITRSLU > 0, then during periodic boundary conditions
           simulations, SOLUTE atoms which exit the primary image will
           be translated back into this cube. Normally, (ITRSLU=0)
           this is only done for SOLVENT atoms."

    In fact, only by setting ITRSLU=1 will you get the behavior described.
    All values of ITRSLU.NE.1 will result in the same behavior as 
    ITRSLU=0, i.e. the solute is never translated back into the primary
    image box.
     
Affects: If ITRSLU is set to a value >0 but .NE. 1, the user would expect
    the solute to be translated back into the primary image box if it 
    drifted out, but this would not occur.

    This would only cause problems if part of the solute eventually
    drifted outside of the region defined by the central image box
    and the 26 adjoining image boxes. That is, if the solute drifts
    more than one image box length away from the central image box.
    In this case, the non-bonded list would not be calculated correctly.

    Obviously, there is no possible problem if periodic boundary conditions
    are not being imposed.

Cause: The documentation is incorrect. The value of ITRSLU is compared to 1, 
    rather than checking to simply determine if ITRSLU > 0.

Fix: To routine MDREAD in src/gibbs/giba.f and doc/gibbs.me

---------------------giba.f:
*** OLD giba.f
--- NEW giba.f
***************
*** 1288,1293
          write(6,'(/2x,a,i3,a)') 'IDSX0 (',idsx0,') must be >= -1.'
          inerr = 1
        endif
        if (ioleps.ne. 0 .and. ioleps.ne. 1) then
          write(6,'(/2x,a,i3,a)') 'IOLEPS (',ioleps,') must be 0 or 1.'
          inerr = 1

--- 1288,1297 -----
          write(6,'(/2x,a,i3,a)') 'IDSX0 (',idsx0,') must be >= -1.'
          inerr = 1
        endif
+       if (itrslu.ne.0 .and. itrslu.ne.1) then
+         write(6,'(/2x,a,i3,a)') 'ITRSLU (',itrslu,') must be 0 or 1.'
+         inerr = 1
+       endif
        if (ioleps.ne. 0 .and. ioleps.ne. 1) then
          write(6,'(/2x,a,i3,a)') 'IOLEPS (',ioleps,') must be 0 or 1.'
          inerr = 1

---------------------gibbs.me:

*** OLD gibbs.me
--- NEW gibbs.me
***************
*** 1091,1100
  
          = -1: results in r*(state where one atom vanished) = 0.0
  
! 10.11  ITRSLU    If ITRSLU > 0, then during periodic boundary conditions
!          simulations, SOLUTE atoms which exit the primary image will
!          be translated back into this cube. Normally, (ITRSLU=0)
!          this is only done for SOLVENT atoms.
  
  10.12  IOLEPS    Controls how parameter mixing is performed for non-bonded
           interactions.

--- 1091,1109 -----
  
          = -1: results in r*(state where one atom vanished) = 0.0
  
! 10.11  ITRSLU    During a periodic boundary conditions simulation,
!          controls whether SOLUTE molecules which exit the primary
!          image box will be translated back into the central box. SOLVENT
!          molecules which exit the central image box are always translated
!          back into the box. A molecule is considered to have floated out
!          of the central box if the first atom of the molecule exits the box.
! 
!         = 1  Both SOLUTE and SOLVENT molecules which exit the primary image 
!              box will be translated back into the box. (Recommended for most
!              systems).
! 
!         = 0  Only SOLVENT molecules will be translated back into the primary 
!              image box.  SOLUTE molecules are not translated.
  
  10.12  IOLEPS    Controls how parameter mixing is performed for non-bonded
           interactions.

---------------------

Temporary workarounds: Simply be sure to set the value of ITRSLU to 1 if
     you want solute molecules translated back into the primary image box.

Routines affected:

     GIBBS:  Routine RUNMD in file .../amber4/src/gibbs/giba.f
             documentation file    .../amber4/doc/gibbs.me