Multiwfn forum

hehe

Member

Registered: 2025-10-30

Posts: 6

Weird Explosion of atoms in IRC

Hi Tian lu,
              I encountered a weird error when i was doing IRC on my system....

First run:
Step size                    =   0.265 sqrt(amu)*bohr
I was playing around so I did 10 steps first.
%oldchk=TS.chk
%chk=IRC.chk
#p irc=(rcfc,maxpoints=10,stepsize=10,phase=(123,51),reverse) geom=allcheck external='./orca.sh'

Second run:
Step size                    =   0.265 sqrt(amu)*bohr
I extended but hit an error in HPC correction step in ~20 steps
%oldchk=IRC.chk
%chk=IRC_2.chk
#p irc=(restart,maxpoints=50) geom=allcheck external='./orca.sh'

Third run
Step size                    =   0.100 sqrt(amu)*bohr
I read that you can keep consistent step size if N<0
StepSize=N
Step size along the reaction path, in units of 0.01 Bohr. If N<0, then the step size is taken in units of 0.01 amu1/2-Bohr. The default is 10.

%oldchk=IRC_2.chk
%chk=IRC_3.chk
#p irc=(calcfc,downhill,stepsize=-10,maxpoints=50) geom=allcheck external='./orca.sh'

Last run:
Step size                    =   0.100 sqrt(amu)*bohr
%oldchk=IRC_3.chk
%chk=IRC_4.chk
#p irc=(restart,downhill,maxpoints=100,stepsize=-10,recorrect=never) geom=allcheck external='./orca.sh'

I get a weird error in my last step (run 4) where my atoms essentially explode, they are scattered everywhere
grep 'CHANGE IN THE REACTION COORDINATE' IRC_2.log
   CHANGE IN THE REACTION COORDINATE =    0.26530
   CHANGE IN THE REACTION COORDINATE =    0.26527
   CHANGE IN THE REACTION COORDINATE =    0.26532
   CHANGE IN THE REACTION COORDINATE =    0.26529
   CHANGE IN THE REACTION COORDINATE =    0.26523
[s4638218@bun091 run]$ grep IRC_3.log
   CHANGE IN THE REACTION COORDINATE =    0.09985
   CHANGE IN THE REACTION COORDINATE =    0.09997
   CHANGE IN THE REACTION COORDINATE =    0.09996
   CHANGE IN THE REACTION COORDINATE =    0.09995
   CHANGE IN THE REACTION COORDINATE =    0.09993
   CHANGE IN THE REACTION COORDINATE =    0.09990
   CHANGE IN THE REACTION COORDINATE =    0.09986
   CHANGE IN THE REACTION COORDINATE =    0.09981
   CHANGE IN THE REACTION COORDINATE =    0.09980
   CHANGE IN THE REACTION COORDINATE =    0.09984
   CHANGE IN THE REACTION COORDINATE =    0.09989
   CHANGE IN THE REACTION COORDINATE =    0.09993
   CHANGE IN THE REACTION COORDINATE =    0.09996
   CHANGE IN THE REACTION COORDINATE =    0.09997
   CHANGE IN THE REACTION COORDINATE =    0.09997
   CHANGE IN THE REACTION COORDINATE =    0.09997
   CHANGE IN THE REACTION COORDINATE =    0.09997
   CHANGE IN THE REACTION COORDINATE =    0.09995
   CHANGE IN THE REACTION COORDINATE =    0.09980
grep 'CHANGE IN THE REACTION COORDINATE' IRC_4.log
   CHANGE IN THE REACTION COORDINATE =  318.74307
   CHANGE IN THE REACTION COORDINATE =    0.00000
-------------------------------
I have previously tried
Step size                    =   0.4386 sqrt(amu)*bohr
%oldchk=IRC_2.chk
%chk=IRC_3.chk
#p irc=(calcfc,downhill,maxpoints=100) geom=allcheck external='./orca.sh'

%oldchk=IRC_3.chk
%chk=IRC_4.chk
#p irc=(restart,downhill,maxpoints=100,recorrect=never) geom=allcheck external='./orca.sh'

grep 'CHANGE IN THE REACTION COORDINATE' IRC_3.log
   CHANGE IN THE REACTION COORDINATE =    0.43712
   CHANGE IN THE REACTION COORDINATE =    0.43010
grep 'CHANGE IN THE REACTION COORDINATE' IRC_4.log
   CHANGE IN THE REACTION COORDINATE =    0.42702
   CHANGE IN THE REACTION COORDINATE =    0.41369
   CHANGE IN THE REACTION COORDINATE =    0.13986
   CHANGE IN THE REACTION COORDINATE =    0.01068

This worked out fine...
I was just experimenting with a smaller step size and am not sure why sudden "explosion" occurred..?
Did my chk file just became corrupted?

Appreciate any advice

Kind regards,
hehe

sobereva

Tian Lu (Multiwfn developer)

From: Athens, Greece

Registered: 2017-09-11

Posts: 2,198

Website

Re: Weird Explosion of atoms in IRC

Perhaps it is just a bug, or the algorithm is not robust enough in rare case... (especially "downhill" is not a popular option)

hehe

Member

Registered: 2025-10-30

Posts: 6

Re: Weird Explosion of atoms in IRC

Hi Tian lu,

Thanks for the reply. Honestly, im unaware but i just decided to take the geometry and redo downhill and it worked fine. I reckon it is some issue when using chk to read the geometry which worked fine in the initial stages but for some reason, fails in this particular restart. "Restarting" completely with geometry written (without geom=allcheck) worked as expected

I do have a followup question:
I am not entirely aware of the algorithm but is there a difference in optimising when you are somewhat close to local minimum vs using downhill....?

Thanks,
hehe

Last edited by hehe (2026-02-20 11:51:25)