********> bugfix.65 Correction Authors: David A. Pearlman and Thomas Fox Correction Date: 02/26/96 Programs: GIBBS Severity: Slight Problem: Certain numbers written to the restart file may be overflow the format (i.e. print as '*'s) when calculating free energy components or derivates (IPERAT > 0) and using TI (IDIFRG=1). While the overflows will have absolutely NO affect on the simulation or the information calculated, they could cause problems if one attempts to restart from the resulting restart file. Note that the possible overflows result from writing an unused variable which is calculated using an uninitialized variable. Whether an overflow occurs on a particular computer will depend on what value that computer assigns to an uninitialized variable, which could vary from run to run. Affects: Nothing, except that one may encounter difficulties when trying to run a restart simulation (IREST=1) using the restart file. Cause: A series of variables which are not used when IPERAT>0 and IDIFRG = 1 are nonetheless written to the restart file. These variables have, in this case, been calculated using an unitialized variable. Fix: Make the following changes to routine PEPHI in gibb.f: ------------------------------------------------------------------------------ *** OLD gibb.f --- NEW gibb.f *************** *** 4818,4827 **** IF (IDIFRG.EQ.1) THEN IF (KPERT.EQ.1) THEN ECNTF1 = -EKEEP(IMM) * DLF * FACT1 ! ECNTF1 = -EKEEP(IMM) * DLR * FACT1 ELSE ECNTF1 = (+EKEEP(IMM)+ESINKP(IMM)) * DLF * FACT1 ! ECNTF1 = (+EKEEP(IMM)+ESINKP(IMM)) * DLR * FACT1 END IF END IF IF (IAPER(IIAT).GT.0) THEN --- 4818,4827 ---- IF (IDIFRG.EQ.1) THEN IF (KPERT.EQ.1) THEN ECNTF1 = -EKEEP(IMM) * DLF * FACT1 ! ECNTR1 = -EKEEP(IMM) * DLR * FACT1 ELSE ECNTF1 = (+EKEEP(IMM)+ESINKP(IMM)) * DLF * FACT1 ! ECNTR1 = (+EKEEP(IMM)+ESINKP(IMM)) * DLR * FACT1 END IF END IF IF (IAPER(IIAT).GT.0) THEN ------------------------------------------------------------------------------ Temporary workarounds: If this problem is encountered (no bug reports related to this problem have been received), one could manually change all the overflowed fields to 0.0000 in the restart file. (Be sure THIS is the problem before attempting this type of change!) Routines affected: GIBBS Routine PEHPI in file ...amber4/src/gibbs/gibb.f --