Daniel Andrés Díaz-pachón

Daniel Andrés Díaz-Pachón (November 13, 1978) is a Research Assistant Professor in the Division of Biostatistics at the University of Miami. His main research topic is fine-tuning, and his key research tool is active information. He explores the role of information in complex systems, usually of biological interest, using tools from probability, statistics, machine learning, and information theory, with a great deal of inspiration from the philosophy of science —a final amalgamation that came after digging in different fields throughout his career.

He was born in Bogotá, Colombia, and influenced by his grandmother, an educator, learned addition, subtraction, and reading before going to kindergarten. In 1989, when he was 10, prompted by the petroleum bonanza, his family migrated to Yopal, Casanare, a small town in the Colombian plains, where he lived until the end of high school at the Braulio González public school.

The move to Yopal and the petroleum bonanza made his family very rich. In 1996, motivated by the family business in Engineering, Díaz returned to Bogotá, to Universidad de los Andes, the primer private university in Colombia, to a honors program in Engineering. However, two years into the program, due to the Colombian violence, his family was displaced by the FARC guerrilla, and the family lost all its wealth. Díaz then left los Andes to start anew at Universidad Nacional de Colombia, the primer public university in the country, where he studied Statistics and completed two “minors”, one in Mathematics and one in Biostatistics.

For his undergraduate research on experimental designs, which was later published as a scientific article, he received his first NSF travel award to present the results in a talk at the IX Latin American Congress of Probability and Mathematical Statistics, in Punta del Este, Uruguay. This event paved the way to go to the prestigious Instituto de Matemática e Estatística at Universidade de São Paulo, Brazil, where he obtained his doctoral degree in 2009, under the guidance of Serguei Popov, for his research in percolation and large deviations of the so-called stable marriage of Poisson and Lebesgue.

In 2010 he migrated to the US, and the next year became a Postdoctoral Associate in the Division of Biostatistics at the University of Miami, Florida. There, under the mentorship of Sunil Rao, he was introduced to machine learning, working on high-dimensional bump-hunting algorithms. In 2016, now as Research Assistant Professor at the University of Miami, he investigated the No-Free-Lunch Theorems. The latter led to a collaboration with Robert J. Marks II (Baylor University), who earlier had proposed active information as a measure of the informational difference in reaching a target between a search algorithm and blind chance. With Marks, Diaz developed a generalization of active information, based on maximum entropy. This extension of active information became the base of all his subsequent research in population genetics, bump hunting, COVID-19 prevalence estimation, and fine-tuning.

In population genetics, he has used active information to measure the amount of information that non-neutral models add with respect to their neutral versions. In bump-hunting he has used it to detect the presence of modes in high-dimensional datasets. And in COVID-19 estimation of prevalence, he has used it to correct over-estimation of prevalence due to large representation of symptomatic individuals in convenience samples.

As for fine-tuning, using maximum entropy and Bayes theory, together with Marks and Ola Hössjer (Stockholm University), he developed a general framework to measure whether cosmological tuning is either fine or coarse. This solved the so-called normalization objection, according to which tuning cannot be measured when a constant of nature could take an infinite number of values. Since then, noticing that search problems in computer science are tuning problems, he has been working on developing a generalization of fine-tuning theory and measurement applicable throughout all the sciences.

GOOGLE SCHOLAR LINK