Wednesday, November 30, 2016

ANI-1: An extensible neural network potential with DFT accuracy at force field computational cost

Justin S. Smith, Olexandr Isayev, Adrian E. Roitberg (2016)
Contributed by Jan Jensen

This paper basically presents a neural network force field, which the authors call a neural network potential (NNP).  The authors heavily modify the Behler-Parinello symmetry functions (also used in this CCH) to improve the transferability and train it against 13.8 million ωB97X/6-31G(d) energies computed for CHON-containing molecules with 8 or less non-hydrogen atoms. This huge training set made it possible to parameterise a neural net with three hidden layers with a total of 320 nodes and 124,033 optimisable parameters.  Deep learning indeed.  

What makes this work particularly exiting is that the NNP appears to be transferable to larger molecules. For example, the figure above shows that the NNP can reproduce the relative ωB97X/6-31G(d) energies of retinol conformers with en RMSE of 0.6 kcal/mol.  For comparison the corresponding value for DFTB (not clear if it's DFTB2 or DFTB3) is 1.2 kcal/mol, although ωB97X/6-31G(d) is not the definitive reference by which to judge DFTB accuracy.

I think this work holds a lot of promise. One of the key challenges is to reduce the size of the training set to a point where high level calculations can be used to compute the energies. Alternatively, perhaps approaches like ∆-machine learning can be used to correct the NNP using a smaller representative training set.

This work is licensed under a Creative Commons Attribution 4.0

Wednesday, November 16, 2016

Calculation of NMR Spin–Spin Coupling Constants in Strychnine

Helgaker, T.; Jaszuński, M.; Świder, P. J. Org. Chem. 2016
Contributed by Steven Bachrach
Reposted from Computational Organic Chemistry with permission

Helgaker, Jaszunski, and Swider1 have examined the use of B3LYP with four different basis sets to compute the spin-spin coupling constants in strychnine 1.

They used previously optimized coordinates of the two major conformations of strychnine, shown in Figure 1.

Conformer A

Conformer B
Figure 1. Confrmations of strychnine 1.

They tested four basis sets designed for NMR computations: pcJ-0,2 pcJ-1,2 6-31G-J,3 and 6-311G-J.3 pCJ-0 and 6-31G-J are relatively small basis sets, while the other two are considerably larger.

All four basis sets provide values of the 122 J(C-H) with a root mean square deviation of less than 0.6 Hz. J(HH) and J(CC) coupling constants are also well predicted, especially with the larger pcJ-1 basis set. They also examined the four Ramsey terms in the coupling model. The Fermi contact term dominates, and if the large pcJ-1 basis set is used to calculate it, and the smaller pcJ-0 basis set is used for the other three terms, the RMS error only increases from 0.18 to 0.20 Hz. Taking this to the extreme, they omitted calculating any of the non-Fermi contact terms, with again only small increases in the RMS – even with the small pcJ-0 basis set. Considering the computational costs, one should seriously consider whether the non-Fermi contact terms and a small basis set might be satisfactory for your own problem(s) at hand.


1) Helgaker, T.; Jaszuński, M.; Świder, P., "Calculation of NMR Spin–Spin Coupling Constants in Strychnine." J. Org. Chem. 2016, ASAP, DOI: 10.1021/acs.joc.6b02157.
2) Jensen, F., "The Basis Set Convergence of Spin−Spin Coupling Constants Calculated by Density Functional Methods." J. Chem. Theor. Comput. 2006, 2, 1360-1369, DOI: 10.1021/ct600166u.
3) Kjær, H.; Sauer, S. P. A., "Pople Style Basis Sets for the Calculation of NMR Spin–Spin Coupling Constants: the 6-31G-J and 6-311G-J Basis Sets." J. Chem. Theor. Comput. 2011, 7, 4070-4076, DOI: 10.1021/ct200546q.


Strychnine 1: InChI=1S/C21H22N2O2/c24-18-10-16-19-13-9-17-21(6-7-22(17)11-12(13)5-8-25-16)14-3-1-2-4-15(14)23(18)20(19)21/h1-5,13,16-17,19-20H,6-11H2/t13-,16-,17-,19-,20-,21+/m0/s1

This work is licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License.