Abstract
By using exponential activation functions with a neural network (NN) method we show that it is possible to fit potentials to a sum-of-products form. The sum-of-products form is desirable because it reduces the cost of doing the quadratures required for quantum dynamics calculations. It also greatly facilitates the use of the multiconfiguration time dependent Hartree method. Unlike potfit product representation algorithm, the new NN approach does not require using a grid of points. It also produces sum-of-products potentials with fewer terms. As the number of dimensions is increased, we expect the advantages of the exponential NN idea to become more significant.
References
56
Referenced
165
{'volume-title': 'Photodissociation Dynamics', 'year': '1995', 'key': '2023071503542386100_c1'}
/ Photodissociation Dynamics (1995){'volume-title': 'Dynamics of Molecules and Chemical Reactions', 'year': '1996', 'author': 'Wyatt', 'key': '2023071503542386100_c2'}
/ Dynamics of Molecules and Chemical Reactions by Wyatt (1996)10.1063/1.474210
/ J. Chem. Phys. (1997)10.1016/S0009-2614(01)01381-1
/ Chem. Phys. Lett. (2002)10.1063/1.1804174
/ J. Chem. Phys. (2004)10.1063/1.2001654
/ J. Chem. Phys. (2005)10.1063/1.1857472
/ J. Chem. Phys. (2005)10.1063/1.1759627
/ J. Chem. Phys. (2004)10.1063/1.478750
/ J. Chem. Phys. (1999)10.1007/s002140050379
/ Theor. Chem. Acc. (1998)10.1016/S0010-4655(98)00152-0
/ Comput. Phys. Commun. (1999)10.1023/A:1019188517934
/ J. Math. Chem. (1999)10.1021/jp010450t
/ J. Phys. Chem. A (2001)10.1063/1.2336223
/ J. Chem. Phys. (2006)10.1063/1.471513
/ J. Chem. Phys. (1996)10.1063/1.476977
/ J. Chem. Phys. (1998)10.1063/1.469292
/ J. Chem. Phys. (1995){'key': '2023071503542386100_c18', 'first-page': '115', 'volume': '5', 'year': '1986', 'journal-title': 'Comput. Phys. Rep.'}
/ Comput. Phys. Rep. (1986)10.1063/1.465576
/ J. Chem. Phys. (1993)10.1016/0167-7977(86)90005-5
/ Comput. Phys. Rep. (1986)10.1146/annurev.physchem.40.1.469
/ Annu. Rev. Phys. Chem. (1989){'key': '2023071503542386100_c22', 'first-page': '1', 'volume': '9', 'year': '1990', 'journal-title': 'Annu. Rev. Phys. Chem.'}
/ Annu. Rev. Phys. Chem. (1990)10.1002/0470845015
/ Encyclopedia of Computational Chemistry by Schleyer (1998){'key': '2023071503542386100_c24', 'first-page': '263', 'volume': '114', 'year': '2000', 'journal-title': 'Adv. Chem. Phys.'}
/ Adv. Chem. Phys. (2000)10.1021/j100319a003
/ J. Phys. Chem. (1988)10.1002/0470845015
/ Encyclopedia of Computational Chemistry by Schleyer (1998)10.1021/ar00028a007
/ Acc. Chem. Res. (1993){'volume-title': 'Numerical Recipes in Fortran 77: The Art of Scientific Computing', 'year': '1992', 'key': '2023071503542386100_c28'}
/ Numerical Recipes in Fortran 77: The Art of Scientific Computing (1992)10.1063/1.467273
/ J. Chem. Phys. (1994)10.1063/1.463916
/ J. Chem. Phys. (1992)10.1063/1.1407277
/ J. Chem. Phys. (2001)10.1063/1.1348274
/ J. Chem. Phys. (2001)10.1063/1.460317
/ J. Chem. Phys. (1991)10.1063/1.473908
/ J. Chem. Phys. (1997)10.1063/1.463332
/ J. Chem. Phys. (1992)10.1016/S0370-1573(99)00047-2
/ Phys. Rep. (2000)10.1007/BF01449770
/ Math. Ann. (1906)10.1063/1.2192499
/ J. Chem. Phys. (2006)10.1063/1.2171246
/ J. Chem. Phys. (2006)10.1063/1.2085167
/ J. Chem. Phys. (2005)10.1016/j.chemphys.2004.06.006
/ Chem. Phys. (2004){'volume-title': 'Neural Network Learning: Theoretical Foundations', 'year': '1999', 'key': '2023071503542386100_c42'}
/ Neural Network Learning: Theoretical Foundations (1999)10.1146/annurev.physchem.45.1.439
/ Annu. Rev. Phys. Chem. (1994)10.1021/jp055253z
/ J. Phys. Chem. A (2006)10.1063/1.2185638
/ J. Chem. Phys. (2006)10.1016/j.cplett.2004.07.076
/ Chem. Phys. Lett. (2004)10.1063/1.477550
/ J. Chem. Phys. (1998)10.1063/1.1850458
/ J. Chem. Phys. (2005)10.1016/0893-6080(89)90003-8
/ Neural Networks (1989)10.1016/S0925-2312(98)00111-8
/ Neurocomputing (1999){'key': '2023071503542386100_c51', 'first-page': '45', 'volume': '11', 'year': '1998', 'journal-title': 'Appl. Math. Lett.'}
/ Appl. Math. Lett. (1998)10.1016/S0893-6080(97)00118-4
/ Neural Networks (1998)10.1109/72.761726
/ IEEE Trans. Neural Netw. (1999)10.1016/S0893-6080(03)00189-8
/ Neural Networks (2004)10.1063/1.479534
/ J. Chem. Phys. (1999){'year': '2002', 'key': '2023071503542386100_c56'}
(2002)
Dates
Type | When |
---|---|
Created | 18 years, 9 months ago (Nov. 16, 2006, 6:06 p.m.) |
Deposited | 1 year, 6 months ago (Feb. 8, 2024, 4:30 a.m.) |
Indexed | 2 weeks, 3 days ago (Aug. 7, 2025, 4:53 p.m.) |
Issued | 18 years, 9 months ago (Nov. 16, 2006) |
Published | 18 years, 9 months ago (Nov. 16, 2006) |
Published Online | 18 years, 9 months ago (Nov. 16, 2006) |
Published Print | 18 years, 9 months ago (Nov. 21, 2006) |
@article{Manzhos_2006, title={Using neural networks to represent potential surfaces as sums of products}, volume={125}, ISSN={1089-7690}, url={http://dx.doi.org/10.1063/1.2387950}, DOI={10.1063/1.2387950}, number={19}, journal={The Journal of Chemical Physics}, publisher={AIP Publishing}, author={Manzhos, Sergei and Carrington, Tucker}, year={2006}, month=nov }