RABBIT (Reconstruction of Attosecond Beating By Interference of Two-photon transitions) is an interferometric technique used for detecting attosecond-timescale phenomena such as charge migration and photoionization in various molecules. This technique looks at the interference pattern created by two laser fields (one ultraviolet and one infared) in electrons during the process of photoionization. The resulting interference pattern is a sine wave, and in some cases by looking at the phase of this wave we can determine the timings for electronic dynamics.

RABBIT has been used to study multiphoton photoionization dynamics experimentally and theoretically in the past (notably on CF₄ by Hans Jakob’s research group in 2021). The framework developed by our group for RABBIT requires calculations for the bound state quantum chemistry, scattering calculations for continuum states, and time-dependent perturbation for multiphoton dynamics. The underlying calculations necessary can take up to months to complete due to their computational complexity, even when running in parallel on high-performance clusters (HPCs).

This summer, I focused on optimizing the electron scattering calculations in CF₄, which were performed using the EPolyScat software. I started by testing values of different in-program variables (like the maximum angular momentum used for wave functions or the maximum asymptotic energy) to identify the lowest values that still yielded converged Dipole Matrix Elements and Cross Sectional Areas and minimized runtime. Then, I tested the convergence and timings for some variables in other molecules (like N₂ and C₃H₆O), comparing my results, and I looked at how performing calculations in various combinations of series/parallel jobs on HPCs like BEOCAT and NERSC affected queue and run timings.

Graphs with results from some of the variable testing completed. Each column is for the different values tested for each variable, and in the 1st and 2nd rows are the real and imaginary parts of the dipole matrix element for some determined continuum state. Red digits represent convergence as the variable value increases. HFacWave is the radial density of the grid used in calculations, EMax is the maximum asymptotic energy in eV, LMaxk is the maximum angular momentum used in the asymptotic expansion of the homogeneous solution, and LMax is the maximum angular momentum used for wave functions. We can see how some variables affected convergence with little impact on the calculation time, while others did the opposite or affected both. LMax barely converged, even with values up to 100, which raised some alarms considering how past papers had used an LMax of 40-60 for CF₄.

The same tests for LMax as Graphic 1 performed on N₂ and C₃H₆O, including the cross sectional area calculations. We see convergence to 4 significant figures by LMax 100 for both, which occurs in more time for the larger molecule as expected.

Acknowledgments
Thank you to Dr. Loren Greenman and Muhammad Sakhi for providing amazing guidance to my research this summer, and to Kim Coy and Dr. Bret Flanders for organizing and running the REU program. I’m also very grateful to everyone in Dr. Greenman’s group for their support with my work and for helping me feel welcome, and everyone else in the REU program this summer who helped me with any small problems I encountered. Thanks as well to Kansas State University for hosting and to the National Science Foundation for funding the program. This material is based upon work supported by the National Science Foundation under Grant No. #2244539. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the National Science Foundation.

Final Presentation