cosmology

Portrait of a man with dark hair and a short beard and mustache wearing glasses, a brown corduroy jacket, a red and blue plaid shirt. His hands are interlaced on the table in front of him. In the lower left corner, the keyboard of a laptop peeks out. He is in front of a starry background.

What if human analysis, combined with machine learning, could advance the study of the universe? The U.S. Department of Energy awarded Fermilab scientist Brian Nord a $2.5 million Early Career Research Award to explore that possibility. Nord has envisioned a new hybrid data-analysis method to undertake the project. It integrates the strengths of artificial intelligence and interpretations of statistics in ways that could potentially advance the studies of cosmology.

A starry night sky with purple diagonal stripe from lower left to upper right corner above an observatory lit up in bright red. A shadow of a building or facility is in the lower right corner.

The Dark Energy Survey collaboration has created the largest ever maps of the distribution and shapes of galaxies, tracing both ordinary and dark matter in the universe out to a distance of over 7 billion light years. The analysis, which includes the first three years of data from the survey, is consistent with predictions from the current best model of the universe, the standard cosmological model. Nevertheless, there remain hints from DES and other experiments that matter in the current universe is a few percent less clumpy than predicted.

A tessellated image of a white cap with a blue dot on the right edge, with shadow surrounding. This image is tessellated many times. A single cap, center right, is different: It has a white dot with a white concentric circle around it on top. No blue.

DESI will capture and study the light from tens of millions of galaxies and other distant objects to better understand our universe and the properties of dark energy. The formal start of DESI’s five-year survey follows a four-month trial run of its custom instrumentation that captured 4-million spectra of galaxies — more than the combined output of all previous spectroscopic surveys. Fermilab has contributed multiple components to the international collaboration led by Berkeley Lab.