2021
DOI: 10.1039/d0sc05610c
|View full text |Cite
|
Sign up to set email alerts
|

Automatic discovery of photoisomerization mechanisms with nanosecond machine learning photodynamics simulations

Abstract: Photochemical reactions are widely used by academia and industry to construct complex molecular architectures via mechanisms that are often inaccessible by other means.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

3
100
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
6
2

Relationship

1
7

Authors

Journals

citations
Cited by 69 publications
(103 citation statements)
references
References 49 publications
3
100
0
Order By: Relevance
“…41,[43][44][45] The coupling vectors tend to infinity at zero energy gaps, therefore it is also helpful to train the model on these vectors multiplied by the energy gap. 41,[43][44][45]140 The problem of the arbitrary phase of the coupling vectors can be addressed by either correcting the phase before the ML training 41,140 or using ML itself to select the phases that lead to best predictions during the training process. 44 For nonadiabatic dynamics involving intersystem crossing (transition between states of different spin multiplicities), one may also need to learn spin-orbit couplings.…”
Section: [H2] Nonadiabatic Dynamicsmentioning
confidence: 99%
“…41,[43][44][45] The coupling vectors tend to infinity at zero energy gaps, therefore it is also helpful to train the model on these vectors multiplied by the energy gap. 41,[43][44][45]140 The problem of the arbitrary phase of the coupling vectors can be addressed by either correcting the phase before the ML training 41,140 or using ML itself to select the phases that lead to best predictions during the training process. 44 For nonadiabatic dynamics involving intersystem crossing (transition between states of different spin multiplicities), one may also need to learn spin-orbit couplings.…”
Section: [H2] Nonadiabatic Dynamicsmentioning
confidence: 99%
“…All simulation times taken from studies of transition metal complexes, except surface hopping/machine learning studies which were applied to organic molecules. 243 , 244 MCTDH = multiconfigurational time-dependent Hartree. AIMS = ab initio multiple spawning.…”
Section: Bridging Time Scalesmentioning
confidence: 99%
“… 251 , 252 In contrast, dynamics based on ML featuring neural networks seem to handle these problematic regions better. 243 , 244 , 253 , 254 …”
Section: Bridging Time Scalesmentioning
confidence: 99%
See 1 more Smart Citation
“…3 Machine learning (ML) approaches have the potential to revolutionise force-field based simulations, aiming to provide the best of both worlds, [4][5][6] and have indeed begun to provide new insights into a range of challenging research problems. [7][8][9][10][11][12][13][14][15][16] The development of an ML potential applicable to the whole periodic table mapping nuclear coordinates to total energies and forces is, however, precluded by the curse of dimensionality. Within small chemical subspaces, models can be achieved using neural networks (NNs), 6,[17][18][19][20][21] kernel-based methods such as the Gaussian Approximation Potential (GAP) framework 22,23 or gradient-domain machine learning (GDML), 24 and linear fitting with properly chosen basis functions, 25,26 each with different data requirements and transferability.…”
Section: Introductionmentioning
confidence: 99%