Rare Event Searches at MIT

点击翻译按钮获取中文摘要

The nature of dark matter is one of the great mysteries of physics. Second only to this is the mystery of the universe’s observed matter-antimatter asymmetry. Axions and Majorana neutrinos, respectively, are as-yet undiscovered particles which may participate in providing the answers. Axions are a well-motivated dark matter candidate while Majorana neutrinos can produce the matter antimatter asymmetry of the universe through a process called leptogenesis. Rare-event searches are the only feasible techniques to explore these models. For this reason, the types of experiments are the highest priority for particle and nuclear physics, respectively. This award will support two efforts: the search for micro-eV axions with ABRACADABRA and DMRadio, and the neutrinoless double-beta decay (0νββ) search KamLAND-Zen. These efforts build on the expertise within the MIT group, and both experimental efforts use novel multidisciplinary techniques from deep learning to quantum sensing.The broader impacts are two-fold: the development of robust vibration isolation for cryogenic experiments and the training of the next generation of scientists. This award will bring together disparate communities from gravitational wave detection to 0νββ for a workshop on vibration isolation. Undergraduate research is a priority for the PI since it is critical for recruiting and retaining physics students. This award will sponsor a study of undergraduate research at MIT and the development of best practices to engage a diverse group of students in STEM fields.The string and QCD axion models motivate a search for Axion dark matter with masses below 10^−6 eV. The ABRACADABRA and Dark Matter Radio experiments have joined forces to probe this mass-range as the DMRadio experiment. This is a staged program that builds on the success of ABRA-10 cm, expands to DMRadio-50L to test the resonant readout, then moves to higher frequencies at the m^3-scale to search down to the QCD axion. Neutrinoless double beta decay is sought in the KamLAND-Zen experiment. Deep learning techniques will be used for training and benchmarking the search algorithm. The tuning of the KLZ detector is non-trivial because many of the Monte Carlo (MC) inputs have the same gross effect and the experiments are tuning on minor deviations of the light curves in various dimensions. The group is exploring generative adversarial networks (GAN) as a way to tackle this problem. The main use of GANs is the generation of artificial images. The adversarial network is good at picking up many small features to increase the accuracy of the classifier for candidate rare events.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.