Freiburg 2024 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
A: Fachverband Atomphysik
A 10: Interaction with Strong or Short Laser Pulses I (joint session A/MO)
A 10.4: Talk
Tuesday, March 12, 2024, 12:00–12:15, HS 1010
Retrieval of the time-dependent bond length in a molecule from photoelectron momentum distributions using deep learning — •Nikolay Shvetsov-Shilovskiy and Manfred Lein — Leibniz Universität Hannover
We apply a convolutional neural network (CNN) to photoelectron momentum distributions produced by strong-field ionization in order to retrieve the time-varying bond length in the dissociating two-dimensional H2+ molecule. We consider the pump-probe scheme and treat the motion of the atomic nuclei either classically, semiclassically, or quantum mechanically. In all these cases, the CNN trained on momentum distributions with fixed internuclear distances [1] predicts the time-dependent bond length with a good accuracy. We investigate whether the neural network can also simultaneously retrieve both the internuclear distance and the velocity with which it increases. Therefore, our results show that deep learning can be used not only for static, but also for dynamic molecular imaging.
[1] N. I. Shvetsov-Shilovski and M. Lein, Phys. Rev. A 105 L021102 (2022).
Keywords: strong-field ionization; deep learning; photoelectron momentum distributions; time-dependent internuclear distance