Showing papers by "Xie Chen published in 2021"

GWTC-2: Compact Binary Coalescences Observed by LIGO and Virgo During the First Half of the Third Observing Run

Abstract: We report on gravitational wave discoveries from compact binary coalescences detected by Advanced LIGO and Advanced Virgo in the first half of the third observing run (O3a) between 1 April 2019 15:00 UTC and 1 October 2019 15:00. By imposing a false-alarm-rate threshold of two per year in each of the four search pipelines that constitute our search, we present 39 candidate gravitational wave events. At this threshold, we expect a contamination fraction of less than 10%. Of these, 26 candidate events were reported previously in near real-time through GCN Notices and Circulars; 13 are reported here for the first time. The catalog contains events whose sources are black hole binary mergers up to a redshift of ~0.8, as well as events whose components could not be unambiguously identified as black holes or neutron stars. For the latter group, we are unable to determine the nature based on estimates of the component masses and spins from gravitational wave data alone. The range of candidate events which are unambiguously identified as binary black holes (both objects $\geq 3~M_\odot$) is increased compared to GWTC-1, with total masses from $\sim 14~M_\odot$ for GW190924_021846 to $\sim 150~M_\odot$ for GW190521. For the first time, this catalog includes binary systems with significantly asymmetric mass ratios, which had not been observed in data taken before April 2019. We also find that 11 of the 39 events detected since April 2019 have positive effective inspiral spins under our default prior (at 90% credibility), while none exhibit negative effective inspiral spin. Given the increased sensitivity of Advanced LIGO and Advanced Virgo, the detection of 39 candidate events in ~26 weeks of data (~1.5 per week) is consistent with GWTC-1.

Developing Real-Time Streaming Transformer Transducer for Speech Recognition on Large-Scale Dataset

Abstract: Recently, Transformer based end-to-end models have achieved great success in many areas including speech recognition. However, compared to LSTM models, the heavy computational cost of the Transformer during inference is a key issue to prevent their applications. In this work, we explored the potential of Transformer Transducer (T-T) models for the fist pass decoding with low latency and fast speed on a large-scale dataset. We combine the idea of Transformer- XL and chunk-wise streaming processing to design a streamable Transformer Transducer model. We demonstrate that T-T outperforms the hybrid model, RNN Transducer (RNN-T), and streamable Transformer attention-based encoder-decoder model in the streaming scenario. Furthermore, the runtime cost and latency can be optimized with a relatively small look-ahead.

Observation of gravitational waves from two neutron star-black hole coalescences

Abstract: We report the observation of gravitational waves from two compact binary coalescences in LIGO's and Virgo's third observing run with properties consistent with neutron star-black hole (NSBH) binaries. The two events are named GW200105_162426 and GW200115_042309, abbreviated as GW200105 and GW200115; the first was observed by LIGO Livingston and Virgo, and the second by all three LIGO-Virgo detectors. The source of GW200105 has component masses $8.9^{+1.2}_{-1.5}\,M_\odot$ and $1.9^{+0.3}_{-0.2}\,M_\odot$, whereas the source of GW200115 has component masses $5.7^{+1.8}_{-2.1}\,M_\odot$ and $1.5^{+0.7}_{-0.3}\,M_\odot$ (all measurements quoted at the 90% credible level). The probability that the secondary's mass is below the maximal mass of a neutron star is 89%-96% and 87%-98%, respectively, for GW200105 and GW200115, with the ranges arising from different astrophysical assumptions. The source luminosity distances are $280^{+110}_{-110}$ Mpc and $300^{+150}_{-100}$ Mpc, respectively. The magnitude of the primary spin of GW200105 is less than 0.23 at the 90% credible level, and its orientation is unconstrained. For GW200115, the primary spin has a negative spin projection onto the orbital angular momentum at 88% probability. We are unable to constrain spin or tidal deformation of the secondary component for either event. We infer a NSBH merger rate density of $45^{+75}_{-33}\,\mathrm{Gpc}^{-3} \mathrm{yr}^{-1}$ when assuming GW200105 and GW200115 are representative of the NSBH population, or $130^{+112}_{-69}\,\mathrm{Gpc}^{-3} \mathrm{yr}^{-1}$ under the assumption of a broader distribution of component masses.

Internal Language Model Training for Domain-Adaptive End-to-End Speech Recognition

Abstract: The efficacy of external language model (LM) integration with existing end-to-end (E2E) automatic speech recognition (ASR) systems can be improved significantly using the internal language model estimation (ILME) method. In this method, the internal LM score is subtracted from the score obtained by interpolating the E2E score with the external LM score, during inference. To improve the ILME-based inference, we propose an internal LM training (ILMT) method to minimize an additional internal LM loss by updating only the E2E model components that affect the internal LM estimation. ILMT encourages the E2E model to form a standalone LM inside its existing components, without sacrificing ASR accuracy. After ILMT, the more modular E2E model with matched training and inference criteria enables a more thorough elimination of the source-domain internal LM, and therefore leads to a more effective integration of the target-domain external LM. Experimented with 30K-hour trained recurrent neural network transducer and attention-based encoder-decoder models, ILMT with ILME-based inference achieves up to 31.5% and 11.4% relative word error rate reductions from standard E2E training with Shallow Fusion on out-of-domain LibriSpeech and in-domain Microsoft production test sets, respectively.

Internal Language Model Estimation for Domain-Adaptive End-to-End Speech Recognition

Abstract: The external language models (LM) integration remains a challenging task for end-to-end (E2E) automatic speech recognition (ASR) which has no clear division between acoustic and language models. In this work, we propose an internal LM estimation (ILME) method to facilitate a more effective integration of the external LM with all pre-existing E2E models with no additional model training, including the most popular recurrent neural network transducer (RNN-T) and attention-based encoder-decoder (AED) models. Trained with audio-transcript pairs, an E2E model implicitly learns an internal LM that characterizes the training data in the source domain. With ILME, the internal LM scores of an E2E model are estimated and subtracted from the log-linear interpolation between the scores of the E2E model and the external LM. The internal LM scores are approximated as the output of an E2E model when eliminating its acoustic components. ILME can alleviate the domain mismatch between training and testing, or improve the multi-domain E2E ASR. Experimented with 30K-hour trained RNN-T and AED models, ILME achieves up to 15.5% and 6.8% relative word error rate reductions from Shallow Fusion on out-of-domain LibriSpeech and in-domain Microsoft production test sets, respectively.

Internal Language Model Training for Domain-Adaptive End-To-End Speech Recognition

Abstract: The efficacy of external language model (LM) integration with existing end-to-end (E2E) automatic speech recognition (ASR) systems can be improved significantly using the internal language model estimation (ILME) method [1]. In this method, the internal LM score is subtracted from the score obtained by interpolating the E2E score with the external LM score, during inference. To improve the ILME-based inference, we propose an internal LM training (ILMT) method to minimize an additional internal LM loss by updating only the E2E model components that affect the internal LM estimation. ILMT encourages the E2E model to form a standalone LM inside its existing components, without sacrificing ASR accuracy. After ILMT, the more modular E2E model with matched training and inference criteria enables a more thorough elimination of the source-domain internal LM, and therefore leads to a more effective integration of the target-domain external LM. Experimented with 30K-hour trained recurrent neural network transducer and attention-based encoder- decoder models, ILMT with ILME-based inference achieves up to 31.5% and 11.4% relative word error rate reductions from standard E2E training with Shallow Fusion on out-of-domain LibriSpeech and in-domain Microsoft production test sets, respectively.

GWTC-2.1: Deep Extended Catalog of Compact Binary Coalescences Observed by LIGO and Virgo During the First Half of the Third Observing Run

Abstract: The second gravitational-wave transient catalog, GWTC-2, reported on 39 compact binary coalescences observed by the Advanced LIGO and Advanced Virgo detectors between 1 April 2019 15:00 UTC and 1 October 2019 15:00 UTC. Here, we present GWTC-2.1, which reports on a deeper list of candidate events observed over the same period. We analyze the final version of the strain data over this period, which is now publicly released. We employ three matched-filter search pipelines for candidate identification, and estimate the probability of astrophysical origin for each candidate event. While GWTC-2 used a false alarm rate threshold of 2 per year, we include in GWTC-2.1, 1201 candidates that pass a false alarm rate threshold of 2 per day. We calculate the source properties of a subset of 44 high-significance candidates that have a probability of astrophysical origin greater than 0.5, using the default priors. Of these candidates, 36 have been reported in GWTC-2. If the 8 additional high-significance candidates presented here are astrophysical, the mass range of candidate events that are unambiguously identified as binary black holes (both objects $\geq 3M_\odot$) is increased compared to GWTC-2, with total masses from $\sim 14M_\odot$ for GW190924_021846 to $\sim 184M_\odot$ for GW190426_190642. The primary components of two new candidate events (GW190403_051519 and GW190426_190642) fall in the mass gap predicted by pair-instability supernova theory. We also expand the population of binaries with significantly asymmetric mass ratios reported in GWTC-2 by an additional two events ($q \lt 0.61$ and $q \lt 0.62$ at $90\%$ credibility for GW190403_051519 and GW190917_114630 respectively), and find that 2 of the 8 new events have effective inspiral spins $\chi_\mathrm{eff} > 0$ (at $90\%$ credibility), while no binary is consistent with $\chi_\mathrm{eff} \lt 0$ at the same significance.

Searches for continuous gravitational waves from young supernova remnants in the early third observing run of Advanced LIGO and Virgo

Abstract: We present results of three wide-band directed searches for continuous gravitational waves from 15 young supernova remnants in the first half of the third Advanced LIGO and Virgo observing run. We use three search pipelines with distinct signal models and methods of identifying noise artifacts. Without ephemerides of these sources, the searches are conducted over a frequency band spanning from 10~Hz to 2~kHz. We find no evidence of continuous gravitational radiation from these sources. We set upper limits on the intrinsic signal strain at 95% confidence level in sample sub-bands, estimate the sensitivity in the full band, and derive the corresponding constraints on the fiducial neutron star ellipticity and $r$-mode amplitude. The best 95% confidence constraints placed on the signal strain are $7.7\times 10^{-26}$ and $7.8\times 10^{-26}$ near 200~Hz for the supernova remnants G39.2--0.3 and G65.7+1.2, respectively. The most stringent constraints on the ellipticity and $r$-mode amplitude reach $\lesssim 10^{-7}$ and $ \lesssim 10^{-5}$, respectively, at frequencies above $\sim 400$~Hz for the closest supernova remnant G266.2--1.2/Vela Jr.

Search for continuous gravitational waves from 20 accreting millisecond X-ray pulsars in O3 LIGO data

Abstract: Results are presented of searches for continuous gravitational waves from 20 accreting millisecond X-ray pulsars with accurately measured spin frequencies and orbital parameters, using data from the third observing run of the Advanced LIGO and Advanced Virgo detectors. The search algorithm uses a hidden Markov model, where the transition probabilities allow the frequency to wander according to an unbiased random walk, while the $\mathcal{J}$-statistic maximum-likelihood matched filter tracks the binary orbital phase. Three narrow sub-bands are searched for each target, centered on harmonics of the measured spin frequency. The search yields 16 candidates, consistent with a false alarm probability of 30% per sub-band and target searched. These candidates, along with one candidate from an additional target-of-opportunity search done for SAX J1808.4$-$3658, which was in outburst during one month of the observing run, cannot be confidently associated with a known noise source. Additional follow-up does not provide convincing evidence that any are a true astrophysical signal. When all candidates are assumed non-astrophysical, upper limits are set on the maximum wave strain detectable at 95% confidence, $h_0^{95\%}$. The strictest constraint is $h_0^{95\%} = 4.7\times 10^{-26}$ from IGR J17062$-$6143. Constraints on the detectable wave strain from each target lead to constraints on neutron star ellipticity and $r$-mode amplitude, the strictest of which are $\epsilon^{95\%} = 3.1\times 10^{-7}$ and $\alpha^{95\%} = 1.8\times 10^{-5}$ respectively. This analysis is the most comprehensive and sensitive search of continuous gravitational waves from accreting millisecond X-ray pulsars to date.

Constraints on dark photon dark matter using data from LIGO's and Virgo's third observing run

Abstract: We present a search for dark photon dark matter that could couple to gravitational-wave interferometers using data from Advanced LIGO and Virgo's third observing run. To perform this analysis, we use two methods, one based on cross-correlation of the strain channels in the two nearly aligned LIGO detectors, and one that looks for excess power in the strain channels of the LIGO and Virgo detectors. The excess power method optimizes the Fourier Transform coherence time as a function of frequency, to account for the expected signal width due to Doppler modulations. We do not find any evidence of dark photon dark matter with a mass between $m_{\rm A} \sim 10^{-14}-10^{-11}$ eV/$c^2$, which corresponds to frequencies between 10-2000 Hz, and therefore provide upper limits on the square of the minimum coupling of dark photons to baryons, i.e. $U(1)_{\rm B}$ dark matter. For the cross-correlation method, the best median constraint on the squared coupling is $\sim1.31\times10^{-47}$ at $m_{\rm A}\sim4.2\times10^{-13}$ eV/$c^2$; for the other analysis, the best constraint is $\sim 1.2\times 10^{-47}$ at $m_{\rm A}\sim 5.7\times 10^{-13}$ eV/$c^2$. These limits improve upon those obtained in direct dark matter detection experiments by a factor of $\sim100$ for $m_{\rm A}\sim [2-4]\times 10^{-13}$ eV/$c^2$.

Constraints from LIGO O3 data on gravitational-wave emission due to r-modes in the glitching pulsar PSR J0537-6910

Abstract: We present a search for continuous gravitational-wave emission due to r-modes in the pulsar PSR J0537-6910 using data from the LIGO-Virgo Collaboration observing run O3. PSR J0537-6910 is a young energetic X-ray pulsar and is the most frequent glitcher known. The inter-glitch braking index of the pulsar suggests that gravitational-wave emission due to r-mode oscillations may play an important role in the spin evolution of this pulsar. Theoretical models confirm this possibility and predict emission at a level that can be probed by ground-based detectors. In order to explore this scenario, we search for r-mode emission in the epochs between glitches by using a contemporaneous timing ephemeris obtained from NICER data. We do not detect any signals in the theoretically expected band of 86-97 Hz, and report upper limits on the amplitude of the gravitational waves. Our results improve on previous amplitude upper limits from r-modes in J0537-6910 by a factor of up to 3 and place stringent constraints on theoretical models for r-mode driven spin-down in PSR J0537-6910, especially for higher frequencies at which our results reach below the spin-down limit defined by energy conservation.

Screw dislocations in the X-cube fracton model

Abstract: The X-cube model, a prototypical gapped fracton model, has been shown to have a foliation structure. That is, inside the 3+1D model, there are hidden layers of 2+1D gapped topological states. A screw dislocation in a 3+1D lattice can often reveal nontrivial features associated with a layered structure. In this paper, we study the X-cube model on lattices with screw dislocations. In particular, we find that a screw dislocation results in a finite change in the logarithm of the ground state degeneracy of the model. Part of the change can be traced back to the effect of screw dislocations in a simple stack of 2+1D topological states, hence corroborating the foliation structure in the model. The other part of the change comes from the induced motion of fractons or sub-dimensional excitations along the dislocation, a feature absent in the stack of 2+1D layers.

Search of the Early O3 LIGO Data for Continuous Gravitational Waves from the Cassiopeia A and Vela Jr. Supernova Remnants

Abstract: We present directed searches for continuous gravitational waves from the neutron stars in the Cassiopeia A (Cas A) and Vela Jr. supernova remnants. We carry out the searches in the LIGO data from the first six months of the third Advanced LIGO and Virgo observing run, using the Weave semi-coherent method, which sums matched-filter detection-statistic values over many time segments spanning the observation period. No gravitational wave signal is detected in the search band of 20--976 Hz for assumed source ages greater than 300 years for Cas A and greater than 700 years for Vela Jr. Estimates from simulated continuous wave signals indicate we achieve the most sensitive results to date across the explored parameter space volume, probing to strain magnitudes as low as ~$6.3\times10^{-26}$ for Cas A and ~$5.6\times10^{-26}$ for Vela Jr. at frequencies near 166 Hz at 95% efficiency.

Improving RNN-T for Domain Scaling Using Semi-Supervised Training with Neural TTS

Abstract: Recurrent neural network transducer (RNN-T) has shown to be comparable with conventional hybrid model for speech recognition. However, there is still a challenge in out-of-domain scenarios with context or words different from training data. In this paper, we explore the semi-supervised training which optimizes RNN-T jointly with neural text-to-speech (TTS) to better generalize to new domains using domain-specific text data. We apply the method to two tasks: one with out-of-domain context and the other with significant out-of-vocabulary (OOV) words. The results show that the proposed method significantly improves the recognition accuracy in both tasks, resulting in 61.4% and 53.8% relative word error rate (WER) reductions respectively, from a well-trained RNN-T with 65 thousand hours of training data. We do further study on the semi-supervised training methodology: 1) which modules of RNN-T model to be updated; 2) the impact of using different neural TTS models; 3) the performance of using text with different relevancy to target domain. Finally, we compare several RNN-T customization methods, and conclude that semi-supervised training with neural TTS is comparable and complementary with Internal Language Model Estimation (ILME) or biasing.

All-sky, all-frequency directional search for persistent gravitational-waves from Advanced LIGO's and Advanced Virgo's first three observing runs

Abstract: We present the first results from an all-sky all-frequency (ASAF) search for an anisotropic stochastic gravitational-wave background using the data from the first three observing runs of the Advanced LIGO and Advanced Virgo detectors. Upper limit maps on broadband anisotropies of a persistent stochastic background were published for all observing runs of the LIGO-Virgo detectors. However, a broadband analysis is likely to miss narrowband signals as the signal-to-noise ratio of a narrowband signal can be significantly reduced when combined with detector output from other frequencies. Data folding and the computationally efficient analysis pipeline, {\tt PyStoch}, enable us to perform the radiometer map-making at every frequency bin. We perform the search at 3072 {\tt{HEALPix}} equal area pixels uniformly tiling the sky and in every frequency bin of width $1/32$~Hz in the range $20-1726$~Hz, except for bins that are likely to contain instrumental artefacts and hence are notched. We do not find any statistically significant evidence for the existence of narrowband gravitational-wave signals in the analyzed frequency bins. Therefore, we place $95\%$ confidence upper limits on the gravitational-wave strain for each pixel-frequency pair, the limits are in the range $(0.030 - 9.6) \times10^{-24}$. In addition, we outline a method to identify candidate pixel-frequency pairs that could be followed up by a more sensitive (and potentially computationally expensive) search, e.g., a matched-filtering-based analysis, to look for fainter nearly monochromatic coherent signals. The ASAF analysis is inherently independent of models describing any spectral or spatial distribution of power. We demonstrate that the ASAF results can be appropriately combined over frequencies and sky directions to successfully recover the broadband directional and isotropic results.

All-sky search for long-duration gravitational-wave bursts in the third Advanced LIGO and Advanced Virgo run

Abstract: After the detection of gravitational waves from compact binary coalescences, the search for transient gravitational-wave signals with less well-defined waveforms for which matched filtering is not well suited is one of the frontiers for gravitational-wave astronomy. Broadly classified into “short” ≲ 1 s and “long” ≳ 1 s duration signals, these signals are expected from a variety of astrophysical processes, including non-axisymmetric deformations in magnetars or eccentric binary black hole coalescences. In this work, we present a search for long-duration gravitational-wave transients from Advanced LIGO and Advanced Virgo’s third observing run from April 2019 to March 2020. For this search, we use minimal assumptions for the sky location, event time, waveform morphology, and duration of the source. The search covers the range of 2–500 s in duration and a frequency band of 24–2048 Hz. We find no significant triggers within this parameter space; we report sensitivity limits on the signal strength of gravitational waves characterized by the root-sum-square amplitude h rss as a function of waveform morphology. These h rss limits improve upon the results from the second observing run by an average factor of 1.8.

Search for lensing signatures in the gravitational-wave observations from the first half of LIGO-Virgo's third observing run

Abstract: We search for signatures of gravitational lensing in the gravitational-wave signals from compact binary coalescences detected by Advanced LIGO and Advanced Virgo during O3a, the first half of their third observing run. We study: 1) the expected rate of lensing at current detector sensitivity and the implications of a non-observation of strong lensing or a stochastic gravitational-wave background on the merger-rate density at high redshift; 2) how the interpretation of individual high-mass events would change if they were found to be lensed; 3) the possibility of multiple images due to strong lensing by galaxies or galaxy clusters; and 4) possible wave-optics effects due to point-mass microlenses. Several pairs of signals in the multiple-image analysis show similar parameters and, in this sense, are nominally consistent with the strong lensing hypothesis. However, taking into account population priors, selection effects, and the prior odds against lensing, these events do not provide sufficient evidence for lensing. Overall, we find no compelling evidence for lensing in the observed gravitational-wave signals from any of these analyses.

Minimum Word Error Rate Training with Language Model Fusion for End-to-End Speech Recognition

Abstract: Integrating external language models (LMs) into end-to-end (E2E) models remains a challenging task for domain-adaptive speech recognition. Recently, internal language model estimation (ILME)-based LM fusion has shown significant word error rate (WER) reduction from Shallow Fusion by subtracting a weighted internal LM score from an interpolation of E2E model and external LM scores during beam search. However, on different test sets, the optimal LM interpolation weights vary over a wide range and have to be tuned extensively on well-matched validation sets. In this work, we perform LM fusion in the minimum WER (MWER) training of an E2E model to obviate the need for LM weights tuning during inference. Besides MWER training with Shallow Fusion (MWER-SF), we propose a novel MWER training with ILME (MWER-ILME) where the ILME-based fusion is conducted to generate N-best hypotheses and their posteriors. Additional gradient is induced when internal LM is engaged in MWER-ILME loss computation. During inference, LM weights pre-determined in MWER training enable robust LM integrations on test sets from different domains. Experimented with 30K-hour trained transformer transducers, MWER-ILME achieves on average 8.8% and 5.8% relative WER reductions from MWER and MWER-SF training, respectively, on 6 different test sets

Factorized Neural Transducer for Efficient Language Model Adaptation.

Abstract: In recent years, end-to-end (E2E) based automatic speech recognition (ASR) systems have achieved great success due to their simplicity and promising performance. Neural Transducer based models are increasingly popular in streaming E2E based ASR systems and have been reported to outperform the traditional hybrid system in some scenarios. However, the joint optimization of acoustic model, lexicon and language model in neural Transducer also brings about challenges to utilize pure text for language model adaptation. This drawback might prevent their potential applications in practice. In order to address this issue, in this paper, we propose a novel model, factorized neural Transducer, by factorizing the blank and vocabulary prediction, and adopting a standalone language model for the vocabulary prediction. It is expected that this factorization can transfer the improvement of the standalone language model to the Transducer for speech recognition, which allows various language model adaptation techniques to be applied. We demonstrate that the proposed factorized neural Transducer yields 15% to 20% WER improvements when out-of-domain text data is used for language model adaptation, at the cost of a minor degradation in WER on a general test set.

Internal Language Model Adaptation with Text-Only Data for End-to-End Speech Recognition.

Abstract: Text-only adaptation of an end-to-end (E2E) model remains a challenging task for automatic speech recognition (ASR). Language model (LM) fusion-based approaches require an additional external LM during inference, significantly increasing the computation cost. To overcome this, we propose an internal LM adaptation (ILMA) of the E2E model using text-only data. Trained with audio-transcript pairs, an E2E model implicitly learns an internal LM that characterizes the token sequence probability which is approximated by the E2E model output after zeroing out the encoder contribution. During ILMA, we fine-tune the internal LM, i.e., the E2E components excluding the encoder, to minimize a cross-entropy loss. To make ILMA effective, it is essential to train the E2E model with an internal LM loss besides the standard E2E loss. Furthermore, we propose to regularize ILMA by minimizing the Kullback-Leibler divergence between the output distributions of the adapted and unadapted internal LMs. ILMA is the most effective when we update only the last linear layer of the joint network. ILMA enables a fast text-only adaptation of the E2E model without increasing the run-time computational cost. Experimented with 30K-hour trained transformer transducer models, ILMA achieves up to 34.9% relative word error rate reduction from the unadapted baseline.

Environmental Noise in Advanced LIGO Detectors.

Abstract: The sensitivity of the Advanced LIGO detectors to gravitational waves can be affected by environmental disturbances external to the detectors themselves. Since the transition from the former initial LIGO phase, many improvements have been made to the equipment and techniques used to investigate these environmental effects. These methods have aided in tracking down and mitigating noise sources throughout the first three observing runs of the advanced detector era, keeping the ambient contribution of environmental noise below the background noise levels of the detectors. In this paper we describe the methods used and how they have led to the mitigation of noise sources, the role that environmental monitoring has played in the validation of gravitational wave events, and plans for future observing runs.

Constraints on the cosmic expansion history from GWTC-3.

Abstract: We use 47 gravitational-wave sources from the Third LIGO-Virgo-KAGRA Gravitational-Wave Transient Catalog (GWTC-3) to estimate the Hubble parameter $H(z)$, including its current value, the Hubble constant $H_0$. Each gravitational-wave (GW) signal provides the luminosity distance to the source and we estimate the corresponding redshift using two methods: the redshifted masses and a galaxy catalog. Using the binary black hole (BBH) redshifted masses, we simultaneously infer the source mass distribution and $H(z)$. The source mass distribution displays a peak around $34\, {\rm M_\odot}$, followed by a drop-off. Assuming this mass scale does not evolve with redshift results in a $H(z)$ measurement, yielding $H_0=68^{+12}_{-7} {\rm km\,s^{-1}\,Mpc^{-1}}$ ($68\%$ credible interval) when combined with the $H_0$ measurement from GW170817 and its electromagnetic counterpart. This represents an improvement of 17% with respect to the $H_0$ estimate from GWTC-1. The second method associates each GW event with its probable host galaxy in the catalog GLADE+, statistically marginalizing over the redshifts of each event's potential hosts. Assuming a fixed BBH population, we estimate a value of $H_0=68^{+8}_{-6} {\rm km\,s^{-1}\,Mpc^{-1}}$ with the galaxy catalog method, an improvement of 42% with respect to our GWTC-1 result and 20% with respect to recent $H_0$ studies using GWTC-2 events. However, we show that this result is strongly impacted by assumptions about the BBH source mass distribution; the only event which is not strongly impacted by such assumptions (and is thus informative about $H_0$) is the well-localized event GW190814.

Approaching the motional ground state of a 10 kg object

Abstract: The motion of a mechanical object -- even a human-sized object -- should be governed by the rules of quantum mechanics. Coaxing them into a quantum state is, however, difficult: the thermal environment effectively masks any quantum signature of the object's motion. Indeed, it also masks effects of proposed modifications of quantum mechanics at large mass scales. We prepare the center-of-mass motion of a $10$ kg mechanical oscillator in a state with an average phonon occupation of $10.8$. The reduction in oscillator temperature, from room temperature to $77$ nK, represents a 100-fold improvement in the reduction of temperature of a solid-state mechanical oscillator -- commensurate with a 11 orders-of-magnitude suppression of quantum back-action by feedback -- and a 10 orders-of-magnitude increase in the mass of an object prepared close to its motional ground state.

All-sky search for continuous gravitational waves from isolated neutron stars in the early O3 LIGO data

Abstract: We report on an all-sky search for continuous gravitational waves in the frequency band 20-2000 Hz and with a frequency time derivative in the range of $[-1.0, +0.1]\times10^{-8}$ Hz/s. Such a signal could be produced by a nearby, spinning and slightly non-axisymmetric isolated neutron star in our galaxy. This search uses the LIGO data from the first six months of Advanced LIGO's and Advanced Virgo's third observational run, O3. No periodic gravitational wave signals are observed, and 95% confidence-level (CL) frequentist upper limits are placed on their strengths. The lowest upper limits on worst-case (linearly polarized) strain amplitude $h_0$ are $~1.7\times10^{-25}$ near 200 Hz. For a circularly polarized source (most favorable orientation), the lowest upper limits are $\sim6.3\times10^{-26}$. These strict frequentist upper limits refer to all sky locations and the entire range of frequency derivative values. For a population-averaged ensemble of sky locations and stellar orientations, the lowest 95% CL upper limits on the strain amplitude are $\sim1.\times10^{-25}$. These upper limits improve upon our previously published all-sky results, with the greatest improvement (factor of $\sim$2) seen at higher frequencies, in part because quantum squeezing has dramatically improved the detector noise level relative to the second observational run, O2. These limits are the most constraining to date over most of the parameter space searched.

Microscopic characterization of Ising conformal field theory in Rydberg chains

Abstract: Rydberg chains provide an appealing platform for probing conformal field theories (CFTs) that capture universal behavior in a myriad of physical settings. Focusing on a Rydberg chain at the Ising transition separating charge density wave and disordered phases, we establish a detailed link between microscopics and low-energy physics emerging at criticality. We first construct lattice incarnations of primary fields in the underlying Ising CFT including chiral fermions -- a nontrivial task given that the Rydberg chain Hamiltonian does not admit an exact fermionization. With this dictionary in hand, we compute correlations of microscopic Rydberg operators, paying special attention to finite, open chains of immediate experimental relevance. We further develop a method to quantify how second-neighbor Rydberg interactions tune the sign and strength of four-fermion couplings in the Ising CFT. Finally, we determine how the Ising fields evolve when four-fermion couplings drive an instability to Ising tricriticality. Our results pave the way to a thorough experimental characterization of Ising criticality in Rydberg arrays, and can also be exploited to design novel higher-dimensional phases based on coupled critical chains.

Memory-Efficient Pipeline-Parallel DNN Training

Abstract: Many state-of-the-art results in domains such as NLP and computer vision have been obtained by scaling up the number of parameters in existing models. However, the weight parameters and intermediate outputs of these large models often do not fit in the main memory of a single accelerator device; this means that it is necessary to use multiple accelerators to train large models, which is challenging to do in a time-efficient way. In this work, we propose PipeDream-2BW, a system that performs memory-efficient pipeline parallelism, a hybrid form of parallelism that combines data and model parallelism with input pipelining. Our system uses a novel pipelining and weight gradient coalescing strategy, combined with the double buffering of weights, to ensure high throughput, low memory footprint, and weight update semantics similar to data parallelism. In addition, PipeDream-2BW automatically partitions the model over the available hardware resources, while being cognizant of constraints such as compute capabilities, memory capacities, and interconnect topologies, and determines when to employ existing memory-savings techniques, such as activation recomputation, that trade off extra computation for lower memory footprint. PipeDream-2BW is able to accelerate the training of large language models with up to 2.5 billion parameters by up to 6.9x compared to optimized baselines.

Searches for continuous gravitational waves from young supernova remnants in the early third observing run of Advanced LIGO and Virgo

Abstract: We present results of three wide-band directed searches for continuous gravitational waves from 15 young supernova remnants in the first half of the third Advanced LIGO and Virgo observing run. We use three search pipelines with distinct signal models and methods of identifying noise artifacts. Without ephemerides of these sources, the searches are conducted over a frequency band spanning from 10~Hz to 2~kHz. We find no evidence of continuous gravitational radiation from these sources. We set upper limits on the intrinsic signal strain at 95\% confidence level in sample sub-bands, estimate the sensitivity in the full band, and derive the corresponding constraints on the fiducial neutron star ellipticity and $r$-mode amplitude. The best 95\% confidence constraints placed on the signal strain are $7.7\times 10^{-26}$ and $7.8\times 10^{-26}$ near 200~Hz for the supernova remnants G39.2--0.3 and G65.7+1.2, respectively. The most stringent constraints on the ellipticity and $r$-mode amplitude reach $\lesssim 10^{-7}$ and $ \lesssim 10^{-5}$, respectively, at frequencies above $\sim 400$~Hz for the closest supernova remnant G266.2--1.2/Vela Jr.

Constraints from LIGO O3 data on gravitational-wave emission due to r-modes in the glitching pulsar PSR J0537-6910

Abstract: We present a search for continuous gravitational-wave emission due to r-modes in the pulsar PSR J0537-6910 using data from the LIGO-Virgo Collaboration observing run O3. PSR J0537-6910 is a young energetic X-ray pulsar and is the most frequent glitcher known. The inter-glitch braking index of the pulsar suggests that gravitational-wave emission due to r-mode oscillations may play an important role in the spin evolution of this pulsar. Theoretical models confirm this possibility and predict emission at a level that can be probed by ground-based detectors. In order to explore this scenario, we search for r-mode emission in the epochs between glitches by using a contemporaneous timing ephemeris obtained from NICER data. We do not detect any signals in the theoretically expected band of 86-97 Hz, and report upper limits on the amplitude of the gravitational waves. Our results improve on previous amplitude upper limits from r-modes in J0537-6910 by a factor of up to 3 and place stringent constraints on theoretical models for r-mode driven spin-down in PSR J0537-6910, especially for higher frequencies at which our results reach below the spin-down limit defined by energy conservation.