********> bugfix.35
Correction Author: Julian Tirado-Rives & David A. Pearlman
Correction Date: 3/6/92  

(Note: Julian Tirado-Rives at Yale discovered and tracked down this
       bug. All I did was flesh out the bug report form and determine
       the circumstances under which the bugs would be encountered. Thanks
       Julian! -- dap).

Programs: GIBBS

Severity: moderate

Problem: 

       An attempt to run a free energy simulation where the following
       conditions are met will produce erroneous free energy results:

         1) The contributions from constraints to the free energy
            have been requested (NCORC = 1).

         2) "Old style" electrostatic mixing rules have been explicitly
            requested (IOLEPS = 1). 
 
            Note that if one leaves IOLEPS=0 (generally recommended,
            and suggested in the manual), this bug will not be encountered.


Affects: 
         The free energy reported will probably be very large. Enough
   so that it should be clear that some type of anomalous behavior is
   occurring.

Cause: 
        
         The bug is caused by a mistake in the code that calculates 
    certain 1-4 electrostatic interactions, in the case where free 
    energy contributions from constraints are being determined and IOLEPS=1.

Fix:  Make the following changes

***To routine NBNDCC in src/gibbs/machinedep.f

*** OLD machinedep.f
--- NEW machinedep.f
***************
*** 2950,2956
            E10 = (ALMU(1)*CHGL1 + ALMU(2)*CHGL0) * R6W(1)
        ELSE
            CGJP0 = CG(J+INDEL1)*ALMU(1)+CG(J+INDEL2)*ALMU(2)
!           E10 = CGIP0*CGJP0*R6W(JN)
        END IF
  C
        G1W(1) = E10/SEE

--- 2950,2956 -----
            E10 = (ALMU(1)*CHGL1 + ALMU(2)*CHGL0) * R6W(1)
        ELSE
            CGJP0 = CG(J+INDEL1)*ALMU(1)+CG(J+INDEL2)*ALMU(2)
!           E10 = CGIP0*CGJP0*R6W(1)
        END IF
  C
        G1W(1) = E10/SEE

Temporary workarounds: Use IOLEPS=0. For almost all applications, this
     option is recommended. If you must use IOLEPS=1 and NCORC=1,
     you must install this fix.

Routines affected:

     GIBBS   Routine NBNDCC in file .../amber4/src/gibbs/machinedep.f