Problem in QMMM energy and gradient, when calculating fused cyclopropane

Questions
#1

Dear QCHEM users

Hello. I encountered problems in QM/MM energy and gradient
calculations when handling a pair of ketoamides containing
fused cyclopropanes. I’ll greatly appreciate if anyone could share
previous experiences with a similar situation and hopefully advise
the way to handle the situation. FYI, please use the links below
to access the QChem input and the MM parameter files.
I am using QChem v.5.4.1

  1. QChem input
  2. MM parameter

As shown below, vdW and Coulomb energy
terms resulted in nan or inf, indicating some sort of missenses
took place in the calculation.

================================
------ MM Energies ------
** Ebond: 65.825928531134**
** Eangle: 5.296243443813**
** EUreyBrad: 0.000000000000**
** Eimptors: 0.237187444071**
** Etorsion: 12.756236279087**
** Evdw: -nan**
** Ecoulomb: inf**
** Etot: -nan Kcal/mol ( -nan hartree)**
** -------------------------**
================================

Upon investigation of the rest of output, I recognized the gradients
of carbon atoms comprising the cyclopropane are reported to be “-nan.”

My wild guess is that the 1-4 vdW and 1-4 Coulomb energy failed to exclude
self-interactions (i.e. C1-C2-C3-C1) that emerges from cyclopropanes.
However, as I have no access to the source codes, this is just a speculation
not having any solid ground. :slight_smile:

All the best,
Jiyong

#2

Thank you for reporting this, you are correct in your diagnosis. Q-Chem is trying to evaluate 1-4 vdW terms in 3-member rings resulting in self-interaction. There isn’t much one can do to work around this issue at the moment except to disable 1-4 vdW altogether. I am submitting this issue internally, and it will be fixed in a future release.

1 Like