Dartmouth Events

Abstract: Magnetic dipolar interactions between atomic nuclei cause the broadening of spectral linewidths and reduced coherence times in solids. Thus, it is desirable to decouple dipolar interactions, which is typically achieved by applying periodic trains of radiofrequency pulses to a system. These pulse sequences have traditionally been designed using average Hamiltonian theory. More recently, machine learning algorithms have been used to create sequences that perform comparably to sequences designed by hand. In order to learn how to create improved dipolar decoupling sequences, we use average Hamiltonian theory and Floquet theory to numerically investigate why some pulse sequences perform better than others, and why all existing pulse sequences have significant reductions in fidelity as dipolar couplings become stronger. It has been traditionally understood that both of these phenomena can be explained by growth in immediate higher order terms in the Magnus series, and so pulse sequences have traditionally been improved by being designed to cancel more terms in the Magnus series. However, our numerical results suggest that this may not actually be the case, which has important implications for the design of new dipolar decoupling sequences.

Advisor: Professor Chandrasekhar Ramanathan

Join Zoom Meeting
https://dartmouth.zoom.us/j/99268168836?pwd=V3JzWUhMWWJvd3VZMklOYjJ2U3k1Zz09

Meeting ID: 992 6816 8836
Email physics.department@dartmouth.edu for passcode