By now, tens of gravitational-wave (GW) events have been detected by the LIGO and Virgo detectors. These GWs have all been emitted by compact binary coalescence, for which we have excellent predictive models. However, there might be other sources for
which we do not have reliable models. Some are expected to exist but to be very rare (e.g., supernovae), while others may be totally unanticipated. So far, no unmodeled sources have been discovered, but the lack of models makes the search for such sources much more difficult and less sensitive. We present here a search for unmodeled GW signals using semi-supervised machine learning. We apply deep learning and outlier detection algorithms to labeled spectrograms of GW strain data, and then search for spectrograms with anomalous patterns in public LIGO data. We searched $sim 13%$ of the coincident data from the first two observing runs. No candidates of GW signals were detected in the data analyzed. We evaluate the sensitivity of the search using simulated signals, we show that this search can detect spectrograms containing unusual or unexpected GW patterns, and we report the waveforms and amplitudes for which a $50%$ detection rate is achieved.
In this work we apply and expand on a recently introduced outlier detection algorithm that is based on an unsupervised random forest. We use the algorithm to calculate a similarity measure for stellar spectra from the Apache Point Observatory Galacti
c Evolution Experiment (APOGEE). We show that the similarity measure traces non-trivial physical properties and contains information about complex structures in the data. We use it for visualization and clustering of the dataset, and discuss its ability to find groups of highly similar objects, including spectroscopic twins. Using the similarity matrix to search the dataset for objects allows us to find objects that are impossible to find using their best fitting model parameters. This includes extreme objects for which the models fail, and rare objects that are outside the scope of the model. We use the similarity measure to detect outliers in the dataset, and find a number of previously unknown Be-type stars, spectroscopic binaries, carbon rich stars, young stars, and a few that we cannot interpret. Our work further demonstrates the potential for scientific discovery when combining machine learning methods with modern survey data.
In this work we present a uniform analysis of the temperature evolution and bolometric luminosity of a sample of 29 type-II supernovae (SNe), by fitting a black body model to their multi-band photometry. Our sample includes only SNe with high quality
multi-band data and relatively well sampled time coverage. Most of the SNe in our sample were detected less than a week after explosion so their light curves cover the evolution both before and after recombination starts playing a role. We use this sample to study the signature of hydrogen recombination, which is expected to appear once the observed temperature drops to $approx 7,000$K. Theory predicts that before recombination starts affecting the light curve, both the luminosity and the temperature should drop relatively fast, following a power-law in time. Once the recombination front reaches inner parts of the outflow, it sets the observed temperature to be nearly constant, and slows the decline of the luminosity (or even leads to a re-brightening). We compare our data to analytic studies and find strong evidence for the signature of recombination. We also find that the onset of the optical plateau in a given filter, is effectively the time at which the black body peak reaches the central wavelength of the filter, as it cools, and it does not correspond to the time at which recombination starts affecting the emission.
Post starburst E+A galaxies are thought to have experienced a significant starburst that was quenched abruptly. Their disturbed, bulge-dominated morphologies suggest that they are merger remnants. We present ESI/Keck observations of SDSS J132401.63+4
54620.6, a post starburst galaxy at redshift z = 0.125, with a starburst that started 400 Myr ago, and other properties, like star formation rate (SFR) consistent with what is measured in ultra luminous infrared galaxies (ULRIGs). The galaxy shows both zero velocity narrow lines, and blueshifted broader Balmer and forbidden emission lines (FWHM=1350 +- 240 km/s). The narrow component is consistent with LINER-like emission, and the broader component with Seyfert-like emission, both photoionized by an active galactic nucleus (AGN) whose properties we measure and model. The velocity dispersion of the broad component exceeds the escape velocity, and we estimate the mass outflow rate to be in the range 4-120 Mo/yr. This is the first reported case of AGN-driven outflows, traced by ionized gas, in post starburst E+A galaxies. We show, by ways of a simple model, that the observed AGN-driven winds can consistently evolve a ULIRG into the observed galaxy. Our findings reinforce the evolutionary scenario where the more massive ULIRGs are quenched by negative AGN feedback, evolve first to post starburst galaxies, and later become typical red and dead ellipticals.
In recent years, the autocorrelation of the hydrogen Lyman-{alpha} forest has been used to observe the baryon acoustic peak at redshift 2 < z < 3.5 using tens of thousands of QSO spectra from the BOSS survey. However, the interstellar medium of the M
ilky-Way introduces absorption lines into the spectrum of any extragalactic source. These lines, while weak and undetectable in a single BOSS spectrum, could potentially bias the cosmological signal. In order to examine this, we generate absorption line maps by stacking over a million spectra of galaxies and QSOs. We find that the systematics introduced are too small to affect the current accuracy of the baryon acoustic peak, but might be relevant to future surveys such as the Dark Energy Spectroscopic Instrument (DESI). We outline a method to account for this with future datasets.
We study a sample of 11 Type II supernovae (SNe) discovered by the OGLE-IV survey. All objects have well sampled I-band light curves, and at least one spectrum. We find that 2 or 3 of the 11 SNe have a declining light curve, and spectra consistent wi
th other SNe II-L, while the rest have plateaus that can be as short as 70d, unlike the 100d typically found in nearby galaxies. The OGLE SNe are also brighter, and show that magnitude limited surveys find SNe that are different than usually found in nearby galaxies. We discuss this sample in the context of understanding Type II SNe as a class and their suggested use as standard candles.
We study a sample of 23 Type II Plateau supernovae (SNe II-P), all observed with the same set of instruments. Analysis of their photometric evolution confirms that their typical plateau duration is 100 days with little scatter, showing a tendency to
get shorter for more energetic SNe. The rise time from explosion to plateau does not seem to correlate with luminosity. We analyze their spectra, measuring typical ejecta velocities, and confirm that they follow a well behaved power-law decline. We find indications of high-velocity material in the spectra of six of our SNe. We test different dust extinction correction methods by asking the following -- does the uniformity of the sample increase after the application of a given method? A reasonably behaved underlying distribution should become tighter after correction. No method we tested made a significant improvement.
Three lines of evidence indicate that in the most common type of core collapse supernovae, the energy deposited in the ejecta by the exploding core is approximately proportional to the progenitor mass cubed. This results stems from an observed unifor
mity of light curve plateau duration, a correlation between mass and ejecta velocity, and the known correlation between luminosity and velocity. This result ties in analytical and numerical models together with observations, providing us with clues as to the mechanism via which the explosion of the core deposits a small fraction of its energy into the hurled envelope.
Dust extinction and reddening are ubiquitous in astronomical observations and are often a major source of systematic uncertainty. We present here a study of the correlation between extinction in the Milky Way and the equivalent width of the NaI D abs
orption doublet. Our sample includes more than 100 high resolution spectra from the KECK telescopes and nearly a million low resolution spectra from the Sloan Digital Sky Survey (SDSS). We measure the correlation to unprecedented precision, constrain its shape, and derive an empirical relation between these quantities with a dispersion of order 0.15 magnitude in E(B-V). From the shape of the curve of growth we further show that a typical sight line through the Galaxy, as seen within the SDSS footprint, crosses about three dust clouds. We provide a brief guide on how to best estimate extinction to extragalactic sources such as supernovae, using the NaI D absorption feature, under a variety of circumstances.
Dust extinction is generally the least tractable systematic uncertainty in astronomy, and particularly in supernova science. Often in the past, studies have used the equivalent width of Na I D absorption measured from low-resolution spectra as proxie
s for extinction, based on tentative correlations that were drawn from limited data sets. We show here, based on 443 low-resolution spectra of 172 Type Ia supernovae for which we have measured the dust extinction as well as the equivalent width of Na I D, that the two barely correlate. We briefly examine the causes for this large scatter that effectively prevents one from inferring extinction using this method.
