MIDAS Seminar Series Presents: Michigan Data Science Fellows: Arya Farahi and Qianying Lin

Arya Farahi: A Quest for the Dark Sector of the Universe: The Role of Galaxy Clusters in the Era of Precision Cosmology

Abstract: Unexpected experimental discoveries constitute a significant force behind advances in scientific understanding. Recent empirical studies revealed that the energy content of the Universe requires a “dark sector” comprising 95% of the energy density of the Universe. Today, despite the overwhelming empirical evidence for the existence of the dark sector, which itself consists of dark matter and dark energy, its physical origin is not understood. A primary role of modern cosmology is to determine the physical origin of these two components. To pursue this quest, we started to collect data at an unprecedented rate with ever-increasing quality. Analysis and modeling these datasets introduce computational and theoretical challenges. These demand scalable, fast, and innovative forms of information processing and theoretical modeling that enable process automation, enhanced insight, and precise and accurate constraints. In this talk, I focus on population statistics of galaxy clusters, the most massive gravitationally bound objects in our Universe, as a primary probe of the dark sector. I will discuss how I use complex numerical simulations to establish new observables to test theoretical models, assess the accuracy and precision of inference pipelines, and train machine learning motivated inference models.

Qianying Lin: Integration of Epidemic Time Series and Genetic Sequences

Abstract: The explosion of epidemiology-related data (i.e., epidemic, genetic, geographic data, etc.) opens up a wide door for methods and frameworks that integrate data of infectious diseases from multiple sources. The term “phylodynamics” indicates the diversity of both epidemiological patterns and phylogenetic evolutions, as well as their interactions. This talk first shows a primary result of inferring the epidemic trends by integrating two types of data––numeric epidemic time series and genetic sequences. In the sense of population genetics, this talk presents how the forward-time epidemic time series and the backward-time sequences are consistent in a stochastic setting. Eventually, the talk proposes a future approach to further incorporate asynchronous sampled genetic sequences to establish a framework that can update the inferences correspondingly.