********> bugfix.48
Correction Author: David A. Pearlman
Correction Date: 10/12/95

Programs: Sander

Severity: moderate

Problem: 

       Attempts to change the relative weights of non-bonded
       (van der Waals) interactions in a system using the VDW, 
       NB or ALL weight change cards do not work correctly.
       On some systems this could cause a crash due to an
       attempt to use an undefined variable. On other systems,
       results would simply not be as expected.
  
       This bug is new to Amber 4.1. This bug does not exist in
       Amber 4.0.

Affects: 

       The non-bonded interactions will not be scaled as
       expected.

Cause: 

       A statement assigning the number of non-bonded parameters
       was incorrectly placed in a routine above that which performs
       the weight changes. As a result, the non-bonded parameters
       were not all changed correctly.

Fix:  Make the following changes to force.f and forcemp.f. The
      correction amounts to moving a single line defining the 
      variable NTTYP to an 'earlier' position.

--------------------------------------------------------------------
Changes to force.f in amber41/src/sander:
--------------------------------------------------------------------
*** OLD force.f
--- NEW force.f
***************
*** 151,156
        BELLY = IBELLY.GT.0
        NOCRST = .false.
        KBOND = NBONA-MBONA
  #ifdef MPI
        if (mpi_orig) then
  c     =========================== AMBER/MPI ===========================

--- 151,157 -----
        BELLY = IBELLY.GT.0
        NOCRST = .false.
        KBOND = NBONA-MBONA
+       NTTYP = NTYPES*(NTYPES+1)/2
  #ifdef MPI
        if (mpi_orig) then
  c     =========================== AMBER/MPI ===========================
***************
*** 371,377
           end if
        END IF
  C
-       NTTYP = NTYPES*(NTYPES+1)/2
        emtot = 0.0d0
  C
  C ----------------------------------------------------------------

--- 372,377 -----
           end if
        END IF
  C
        emtot = 0.0d0
  C
  C ----------------------------------------------------------------
--------------------------------------------------------------------
Changes to forcemp.f in amber41/src/sander:
--------------------------------------------------------------------
*** OLD forcemp.f
--- NEW forcemp.f
***************
*** 150,155
        BELLY = IBELLY.GT.0
        NOCRST = .false.
        KBOND = NBONA-MBONA
  #ifdef MPI
  c     =========================== AMBER/MPI ===========================
  c

--- 150,156 -----
        BELLY = IBELLY.GT.0
        NOCRST = .false.
        KBOND = NBONA-MBONA
+       NTTYP = NTYPES*(NTYPES+1)/2
  #ifdef MPI
  c     =========================== AMBER/MPI ===========================
  c
***************
*** 413,419
           end if
        END IF
  C
-       NTTYP = NTYPES*(NTYPES+1)/2
        emtot = 0.0d0
  C
  C ----------------------------------------------------------------

--- 414,419 -----
           end if
        END IF
  C
        emtot = 0.0d0
  C
  C ----------------------------------------------------------------
------------------------------------------------------------------------------

Temporary workarounds: 

     None. 

Routines affected:

     SANDER files ...amber4/src/sander/force.f        
                  ...amber4/src/sander/forcemp.f        

--