********> bugfix.45 Correction Author: David A. Pearlman Correction Date: 7/31/92 Programs: GIBBS Severity: slight Problem: If the default value for NDMPMC is used (frequency of "dumping" to the MICSTAT file), and the simulation results in more than 100 windows being collected, the numbers of equilibration and data collection steps reported for the first window, the 101st window, the 201st window, etc. in the MICSTAT file will be incorrect. (The average numbers of equilibration/data collection steps reported at the end of the file will similarly be wrong). The simulation is correct. The correct numbers of equilibration and data collection steps will have been collected. All free energy values reported are correct. The only error is in the values reported for these step counters in MICSTAT. Affects: The reported numbers of equilibration and data collection steps every 1+N*100 windows will be in error in MICSTAT. This is only a reporting error. No part of the calculation, or any of the results, is in error. Cause: A comparison statment was incorrectly allowing values to be stored in the "101st" location in an array dimension for only 100 elements. Fix: Make the following change to routine MICST in file micst.f *** OLD micst.f --- NEW micst.f *************** *** 182,189 **** IEQUIL = 0 ICOLL = 1 C ! IF (NCOUNT.GT.MXBUF.OR.(NCOUNT.GT.NDUMP.AND.NDUMP.GT.0)) ! *GO TO 1201 RETURN C --- 182,189 ---- IEQUIL = 0 ICOLL = 1 C ! IF (NCOUNT.GE.MXBUF.OR.(NCOUNT.GT.NDUMP.AND.NDUMP.GT.0)) *GO TO 1201 RETURN C ------------------------------------------------------------------------------ (.GT. becomes .GE.) Temporary workarounds: Setting NDMPMC to 99 (or any value >0 and <100) will avoid the problem. In most cases, the error will be of negligable significance. The user should simply be aware of its origin. Routines affected: GIBBS Routine MICST in file ...amber4/src/micst.f --