CCL: IRC calculation using GAMESS

 Sent to CCL by: "Tae-Rae Kim" [eigenskim^^^gmail.com]
 
 Problem 1):
 If I set up my IRC calculation using a small number for NPOINT, the entire IRC
 is not found. When I try to restart it using the restart information provided by
 the program in the *.dat file, I get the message:
 RADIUS IN CIRCLE OPTIMIZATION 0.0067834 DEVIATES SIGNIFICANTLY FROM CONSTRAIN
 CONDITION 0.0050000
  IT IS POSSIBLE THAT THE NEXT IRC POINT IS CLOSE TO A MINIMUM
 I find it to be extremely unlikely that the system is near a minimum at exactly
 the point which corresponds to my setting of NPOINT. Does anybody know the fix
 for this problem? I guess what I am asking is how does one resart an IRC
 calculation correctly in GAMESS?
 
 I do not rely on restart options. It can be very nervous sometimes.
 Instead, I copy required informations (groups punched in .dat or .irc file)
 and paste it into input file.
 Some options should be added to let GAMESS read the groups.
 
 Problem 2)
 When I ran into problem 1, I increased NPOINT and let the calculation run
 longer. Then eventually the calculation stopped with the same message (but this
 time at a point smaller than my setting of NPOINT):
 RADIUS IN CIRCLE OPTIMIZATION 0.0067834 DEVIATES SIGNIFICANTLY FROM CONSTRAIN
 CONDITION 0.0050000
  IT IS POSSIBLE THAT THE NEXT IRC POINT IS CLOSE TO A MINIMUM
 This time it may be true that the system is near a minimum, I dont know. When I
 use the MacMolPlt-program and look at the potential enenrgy surface for the
 IRC-calculation the system appears NOT to be near a minimum. The potential
 energy surface is not curved at all for the last points in the IRC-calculation
 which is in my mind what should happen near the minimum. However, I can be wrong
 about this. Does anybody know how I should continue to reach the true minimum? I
 don't think I can restart using the information provided by the program because
 then I will run into problem 1.
 
 I think that the curve needs not be rolled up.
 In principle, dynamic path should oscillate around minimum.
 But the program exits when the system reaches near local minimum,
 before oscillation.
 --
 Kim, Tae-Rae
 Dycube Lab. (Prof. S.Shin)
 Department of Chemistry
 Seoul Natl. Univ., South Korea