QMMM convergence/optimization issues in CHCL3 solvent

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Tue Nov 23 2021 - 10:26:50 CST

An organic molecule of 28 CHON atoms could be optimized, with all positive
frequency analysis, with latest ORCA 5 at level
PBE0 RIJCOSX D4 def2-TZVP Opt Freq tightscf defgrid2 CPCM (CHLOROFORM)

With QMMM 100 cycles in CHCl3, starting from an MD quasi equilibrated .pdb
in that solvent, at level
PBE0 RIJCOSX D4 def2-TZVP defgrid2 enGrad CPCM
one of the tree rings of the molecule, and only that, underwent rapidly
extreme distortion, even against the sp3 sp3 sp protocols.

Adding optimization (is that permitted with QMMM?), i.e., level
# qmConfigLine "! MORead"
qmConfigLine "! PBE0 RIJCOSX D4 def2-TZVP Opt tightscf defgrid2 CPCM"
# qmConfigLine "%%moinp \"........gbw\""
qmConfigLine "%%maxcore 5000"
qmConfigLine "%%scf Maxiter 500 end"
qmConfigLine "%%geom Maxiter 500 end"
qmConfigLine "%%pal nproc 47 end"
qmConfigLine "%%output Printlevel Mini Print\[ P_Mulliken \] 1
Print\[P_AtCharges_M\] 1 end"

rapid scf convergence was observed. However, at the ORCA GEOMETRY
Warning: the length of the step is outside the trust region - taking
restricted step instead
The input lambda is .... -862.826127
   iter: 1 x= -888.315690 g= 0.656919 f(x)= 16.744573
iter: 101 x=-1512.546611 g= 0.000257 f(x)= 0.000000
[file orca_gstep/orca_gstep.cpp, line 790]: Error (ORCA_GSTEP): The lambda
equations have not converged

Does that suggest any adjustment? After a couple of hours of thinking, I am
short of ideas

Thanks for advice
francesco pietran MD

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:12 CST