Dovi Poznanski

Seismic Stability at LIGO

The incredible precision required to measure gravitational waves depends on a set of passive and active isolation systems that decouple the interferometers from ground motion.

With Prof. Brian Lantz and his group, we are using data-science and machine-learning tools to study and fine-tune the feedback system, with the goal of improving the uptime and range of these instruments.

Variability and Similarity

The luminosity of stars changes with time. Most stars vary on timescales we cannot measure directly, but modern surveys, especially with telescopes on spacecraft like the Transiting Exoplanet Survey Satellite (TESS) and the Kepler Space Telescope, let us measure stellar variability with extreme precision and ever-growing samples.

A light curve is a time series of flux measurements. Asking how similar two light curves are is surprisingly difficult, but central to classification, discovery, and anomaly detection. At TESS-L8 you can see our latest contribution to the subject.

Anomaly Detection

As data sets grow in size and complexity, we can no longer rely on manually noticing something new. Anomaly detection is a useful but fundamentally ill-posed scientific problem, because there are many unknown ways to be abnormal.

In a series of works we explored tools for finding unusual galaxy spectra, rare quasars, infrared stellar spectra from the Apache Point Observatory Galactic Evolution Experiment (APOGEE), unexpected sources in LIGO data, and technosignatures in Breakthrough Listen observations.

Dust and the Interstellar Medium

The Sloan Digital Sky Survey (SDSS) has gathered millions of spectra of stars, galaxies, and quasars. These objects are observed through the gas and dust that permeate our own Milky Way, which can itself be studied through the imprint it leaves on the spectra.

In multiple studies we used large spectral samples to recover the signature of that gas and study its properties. We measured the correlation between sodium absorption and dust extinction, studied the mysterious diffuse interstellar bands, and constrained the impact of these small features on future cosmology measurements that rely on weak and noisy statistical correlations of spectral features from distant sources.

Core-Collapse Supernovae

The most massive stars end their lives in fireworks: supernova explosions that are a key ingredient of galactic and stellar evolution. They dynamically affect their surroundings, enrich the interstellar medium with elements heavier than hydrogen, and provide raw material for the next generation of stars and planets.

There is a large zoo of supernova types, and modern surveys are increasing the diversity we see. Some are brighter, some evolve faster, and some have very different spectra. We think we understand, in broad terms, which stars explode as which supernovae, but that mapping remains highly incomplete.

Cosmography with Type II Supernovae

The universe is expanding and accelerating, propelled by a mysterious dark-energy component that accounts for about 70% of the energy budget. The best tracer we have of this accelerated expansion is distance measurements to Type Ia supernovae, which are exploding white dwarfs.

Type II supernovae, the explosions that follow the collapse of massive stars, are less bright but far more numerous. They are less precise distance indicators, but they may help control different systematic uncertainties and provide useful complementarity.

Strange Supernovae

There are many unusual supernovae of many unusual kinds. One of my contributions was SN 2002bj, published in Science, where we showed that it was an extremely fast-evolving supernova with a very peculiar chemical composition, and to date, very few analogs, if any.