Search

Generating high-resolution simulations is key for advancing our understanding of one of the universe's most violent events: Black Hole mergers. However, generating Black Hole simulations is limited by prohibitive computational costs and scalability issues, reducing the simulation's fidelity and resolution achievable within reasonable time frames and resources. In this work, we introduce a novel method that circumvents these limitations by applying a super-resolution technique without directly needing high-resolution labels, leveraging the Hamiltonian and momentum constraints-fundamental equations in general relativity that govern the dynamics of spacetime. We demonstrate that our method achieves a reduction in constraint violation by one to two orders of magnitude and generalizes effectively to out-of-distribution simulations., Comment: Code available at https://github.com/ThomasHelfer/TorchGRTL and data at https://huggingface.co/datasets/thelfer/BinaryBlackHole

Scaling laws for large language models (LLMs) have provided useful guidance in training ever larger models for predictable performance gains. Time series forecasting shares a similar sequential structure to language, and is amenable to large-scale transformer architectures. Here we show that foundational decoder-only time series transformer models exhibit analogous scaling-behavior to LLMs, with architectural details (aspect ratio and number of heads) having a minimal effect over broad ranges. We assemble a large corpus of heterogenous time series data on which to train, and establish for the first time power-law scaling with parameter count, dataset size, and training compute, spanning five orders of magnitude., Comment: 4 main pages (16 total), 4 figures; Accepted for oral presentation in Time Series in the Age of Large Models (TSALM) Workshop at Neurips 2024

We present a framework for the efficient computation of optimal Bayesian decisions under intractable likelihoods, by learning a surrogate model for the expected utility (or its distribution) as a function of the action and data spaces. We leverage recent advances in simulation-based inference and Bayesian optimization to develop active learning schemes to choose where in parameter and action spaces to simulate. This allows us to learn the optimal action in as few simulations as possible. The resulting framework is extremely simulation efficient, typically requiring fewer model calls than the associated posterior inference task alone, and a factor of $100-1000$ more efficient than Monte-Carlo based methods. Our framework opens up new capabilities for performing Bayesian decision making, particularly in the previously challenging regime where likelihoods are intractable, and simulations expensive., Comment: 12 pages, 4 figures

We consider state and parameter estimation for compartmental models having both time-varying and time-invariant parameters. Though the described Bayesian computational framework is general, we look at a specific application to the susceptible-infectious-removed (SIR) model which describes a basic mechanism for the spread of infectious diseases through a system of coupled nonlinear differential equations. The SIR model consists of three states, namely, the three compartments, and two parameters which control the coupling among the states. The deterministic SIR model with time-invariant parameters has shown to be overly simplistic for modelling the complex long-term dynamics of diseases transmission. Recognizing that certain model parameters will naturally vary in time due to seasonal trends, non-pharmaceutical interventions, and other random effects, the estimation procedure must systematically permit these time-varying effects to be captured, without unduly introducing artificial dynamics into the system. To this end, we leverage the robustness of the Markov Chain Monte Carlo (MCMC) algorithm for the estimation of time-invariant parameters alongside nonlinear filters for the joint estimation of the system state and time-varying parameters. We demonstrate performance of the framework by first considering a series of examples using synthetic data, followed by an exposition on public health data collected in the province of Ontario.

Ranked data is commonly used in research across many fields of study including medicine, biology, psychology, and economics. One common statistic used for analyzing ranked data is Kendall's {\tau} coefficient, a non-parametric measure of rank correlation which describes the strength of the association between two monotonic continuous or ordinal variables. While the mathematics involved in calculating Kendall's {\tau} is well-established, there are relatively few graphing methods available to visualize the results. Here, we describe a visualization method and provide an interactive app for graphing Kendall's {\tau} which uses a series of rigid Euclidean transformations along a Cartesian plane to map rank pairs into discrete quadrants. The resulting graph provides a visualization of rank correlation which helps display the proportion of concordant and discordant pairs. Moreover, this method highlights other key features of the data which are not represented by Kendall's {\tau} alone but may nevertheless be meaningful, such as the relationship between discrete pairs of observations. We demonstrate the effectiveness of our approach through several examples and compare our results to other visualization methods.

We investigate the possibility of improving the accuracy of the phenomenological waveform model, IMRPhenomD, by jointly optimizing all the calibration coefficients at once, given a set of numerical relativity (NR) waveforms. When IMRPhenomD was first calibrated to NR waveforms, different parts (i.e., the inspiral, merger, and ringdown) of the waveform were calibrated separately. Using ripple, a library of waveform models compatible with automatic differentiation, we can, for the first time, perform gradient-based optimization on all the waveform coefficients at the same time. This joint optimization process allows us to capture previously ignored correlations between separate parts of the waveform. We found that after recalibration, the median mismatch between the model and NR waveforms decreases by 50%. We further explore how different regions of the source parameter space respond to the optimization procedure. We find that the degree of improvement correlates with the spins of the source. This work shows a promising avenue to help understand and treat systematic error in waveform models., Comment: 11 pages, 9 figures

We report results on the first matched-filtering search for binaries with compact objects having large tidal deformabilities in the LIGO-Virgo gravitational wave (GW) data. The tidal deformability of a body is quantified by the ``Love number" $\Lambda \propto \hskip 1pt (r/m)^5$, where $r/m$ is the body's (inverse) compactness. Due to its strong dependence on compactness, the $\Lambda$ of larger-sized compact objects can easily be many orders of magnitude greater than those of black holes and neutron stars, leaving phase shifts which are sufficiently large for these binaries to be missed by binary black hole (BBH) templated searches. In this paper, we conduct a search using inspiral-only waveforms with zero spins but finite tides, with the search space covering chirp masses $3 M_\odot < \mathcal{M} < 15 M_\odot$ and effective tidal deformabilities $10^2 \lesssim \tilde{\Lambda} \lesssim 10^6$. We find no statistically significant GW candidates. This null detection implies an upper limit on the merger rate of such binaries in the range $[1-300] \hskip 2pt \text{Gpc}^{-3} \text{year}^{-1}$, depending on $\mathcal{M}$ and $\tilde{\Lambda}$. While our constraints are model agnostic, we discuss the implications on beyond the Standard Model scenarios that give rise to boson stars and superradiant clouds. Using inspiral-only waveforms we recover many of the BBH signals which were previously identified with full inspiral-merger-ringdown templates. We also constrain the Love number of black holes to $\Lambda \lesssim 10^3$ at the 90\% credible interval. Our work is the first-ever dedicated template-based search for compact objects that are not only black holes and neutron stars. Additionally, our work demonstrates a novel way of finding new physics in GW data, widening the scope of potential discovery to previously unexplored parameter space., Comment: 42+11 pages, 12 figures

We present a lightweight, flexible, and high-performance framework for inferring the properties of gravitational-wave events. By combining likelihood heterodyning, automatically-differentiable and accelerator-compatible waveforms, and gradient-based Markov chain Monte Carlo (MCMC) sampling enhanced by normalizing flows, we achieve full Bayesian parameter estimation for real events like GW150914 and GW170817 within a minute of sampling time. Our framework does not require pretraining or explicit reparameterizations and can be generalized to handle higher dimensional problems. We present the details of our implementation and discuss trade-offs and future developments in the context of other proposed strategies for real-time parameter estimation. Our code for running the analysis is publicly available on GitHub https://github.com/kazewong/jim.

We propose the use of automatic differentiation through the programming framework jax for accelerating a variety of analysis tasks throughout gravitational wave (GW) science. Firstly, we demonstrate that complete waveforms which cover the inspiral, merger, and ringdown of binary black holes (i.e. IMRPhenomD) can be written in jax and demonstrate that the serial evaluation speed of the waveform (and its derivative) is similar to the lalsuite implementation in C. Moreover, jax allows for GPU-accelerated waveform calls which can be over an order of magnitude faster than serial evaluation on a CPU. We then focus on three applications where efficient and differentiable waveforms are essential. Firstly, we demonstrate how gradient descent can be used to optimize the $\sim 200$ coefficients that are used to calibrate the waveform model. In particular, we demonstrate that the typical match with numerical relativity waveforms can be improved by more than 50% without any additional overhead. Secondly, we show that Fisher forecasting calculations can be sped up by $\sim 100\times$ (on a CPU) with no loss in accuracy. This increased speed makes population forecasting substantially simpler. Finally, we show that gradient-based samplers like Hamiltonian Monte Carlo lead to significantly reduced autocorrelation values when compared to traditional Monte Carlo methods. Since differentiable waveforms have substantial advantages for a variety of tasks throughout GW science, we propose that waveform developers use jax to build new waveforms moving forward. Our waveform code, ripple, can be found at https://github.com/tedwards2412/ripple, and will continue to be updated with new waveforms as they are implemented., Comment: 12 pages, 6 figures, code can be found at https://github.com/tedwards2412/ripple

We present highly accurate, dimensionally-reduced gravitational waveforms for binary inspirals whose components have large spin-induced quadrupole moments. The spin-induced quadrupole of a body first appears in the phase of a waveform at the early inspiral stage of the binary coalescence, making it a relatively clean probe of the internal structure of the body. However, for objects with large quadrupolar deviations from Kerr, searches using binary black hole (BBH) models would be ineffective. In order to perform a computationally-feasible search, we present two dimensionally-reduced models which are derived from the original six-dimensional post-Newtonian waveform for such systems. Our dimensional reduction method is guided by power counting in the post-Newtonian expansion, suitable reparameterizations of the source physics, and truncating terms in the phase that are small in most physically well-motivated regions of parameter space. In addition, we note that large quadrupolar deviations cause the frequency at which a binary system reaches its minimum binding energy to be reduced substantially. This minimum signals the end of the inspiral regime and provides a natural cutoff for the PN waveform. We provide accurate analytic estimates for these frequency cutoffs. Finally, we perform injection studies to test the effectualness of the dimensionally reduced waveforms. We find that over $80\%$ of the injections have an effectualness of $\varepsilon > 0.999$, significantly higher than is typically required for standard BBH banks, for systems with component spins of $|\chi_i| \lesssim 0.6$ and dimensionless quadrupole of $\kappa_i \lesssim 10^3$. Importantly, these waveforms represent an essential first step towards enabling an effective search for astrophysical objects with large quadrupoles., Comment: 18 pages, 10 figures

A nonlinear partial differential equation (PDE) based compartmental model of COVID-19 provides a continuous trace of infection over space and time. Finer resolutions in the spatial discretization, the inclusion of additional model compartments and model stratifications based on clinically relevant categories contribute to an increase in the number of unknowns to the order of millions. We adopt a parallel scalable solver allowing faster solutions for these high fidelity models. The solver combines domain decomposition and algebraic multigrid preconditioners at multiple levels to achieve the desired strong and weak scalability. As a numerical illustration of this general methodology, a five-compartment susceptible-exposed-infected-recovered-deceased (SEIRD) model of COVID-19 is used to demonstrate the scalability and effectiveness of the proposed solver for a large geographical domain (Southern Ontario). It is possible to predict the infections up to three months for a system size of 92 million (using 1780 processes) within 7 hours saving months of computational effort needed for the conventional solvers.

We explore the assumption, widely used in many astrophysical calculations, that the stellar initial mass function (IMF) is universal across all galaxies. By considering both a canonical Salpeter-like IMF and a non-universal IMF, we are able to compare the effect of different IMFs on multiple observables and derived quantities in astrophysics. Specifically, we consider a non-universal IMF which varies as a function of the local star formation rate, and explore the effects on the star formation rate density (SFRD), the extragalactic background light, the supernova (both core-collapse and thermonuclear) rates, and the diffuse supernova neutrino background. Our most interesting result is that our adopted varying IMF leads to much greater uncertainty on the SFRD at $z \approx 2-4$ than is usually assumed. Indeed, we find a SFRD (inferred using observed galaxy luminosity distributions) that is a factor of $\gtrsim 3$ lower than canonical results obtained using a universal Salpeter-like IMF. Secondly, the non-universal IMF we explore implies a reduction in the supernova core-collapse rate of a factor of $\sim2$, compared against a universal IMF. The other potential tracers are only slightly affected by changes to the properties of the IMF. We find that currently available data do not provide a clear preference for universal or non-universal IMF. However, improvements to measurements of the star formation rate and core-collapse supernova rate at redshifts $z \gtrsim 2$ may offer the best prospects for discernment., Comment: 15 pages, 11 figures, 1 appendix

Teacher emotional factors influence the classroom environment. The purpose of the study was to examine the association of teacher emotional exhaustion and teacher efficacy with student office discipline referrals (ODRs), in-school suspensions (ISSs), and out-of-school suspensions (OSSs) using multilevel logistic regression models. The sample included 105 teachers and 1,663 students from nine elementary schools in the United States. Higher teacher emotional exhaustion was associated with increased use of ODR and ISS but not OSS. For students with teachers experiencing burnout, the odds of receiving an ISS increased by a factor of 1.74 (d=0.31). Greater teacher efficacy was also associated with lower use of OSS but not ODR or ISS. The results suggest that improving teacher efficacy and reducing teacher emotional exhaustion may support the reduced use of exclusionary discipline practices. [This is the online version of an article published in "School Psychology Review." For the final published version of this article, see EJ1279381.]

The most sensitive search pipelines for gravitational waves from compact binary mergers use matched filters to extract signals from the noisy data stream coming from gravitational wave detectors. Matched-filter searches require banks of template waveforms covering the physical parameter space of the binary system. Unfortunately, template bank construction can be a time-consuming task. Here we present a new method for efficiently generating template banks that utilizes automatic differentiation to calculate the parameter space metric. Principally, we demonstrate that automatic differentiation enables accurate computation of the metric for waveforms currently used in search pipelines, whilst being computationally cheap. Additionally, by combining random template placement and a Monte Carlo method for evaluating the fraction of the parameter space that is currently covered, we show that search-ready template banks for frequency-domain waveforms can be rapidly generated. Finally, we argue that differentiable waveforms offer a pathway to accelerating stochastic placement algorithms. We implement all our methods into an easy-to-use Python package based on the jax framework, diffbank, to allow the community to easily take advantage of differentiable waveforms for future searches., Comment: 15 pages, 6 figures. Comments welcome! Code can be found at https://github.com/adam-coogan/diffbank. Updated to match PRD version

Astroparticle physics is undergoing a profound transformation, due to a series of extraordinary new results, such as the discovery of high-energy cosmic neutrinos with IceCube, the direct detection of gravitational waves with LIGO and Virgo, and many others. This white paper is the result of a collaborative effort that involved hundreds of theoretical astroparticle physicists and cosmologists, under the coordination of the European Consortium for Astroparticle Theory (EuCAPT). Addressed to the whole astroparticle physics community, it explores upcoming theoretical opportunities and challenges for our field of research, with particular emphasis on the possible synergies among different subfields, and the prospects for solving the most fundamental open questions with multi-messenger observations., Comment: White paper of the European Consortium for Astroparticle Theory (EuCAPT). 135 authors, 400 endorsers, 133 pages, 1382 references

Paleo-detectors are a proposed experimental technique to search for dark matter by reading out the damage tracks caused by nuclear recoils in small samples of natural minerals. Unlike a conventional real-time direct detection experiment, paleo-detectors have been accumulating these tracks for up to a billion years. These long integration times offer a unique possibility: by reading out paleo-detectors of different ages, one can explore the time-variation of signals on megayear to gigayear timescales. We investigate two examples of dark matter substructure that could give rise to such time-varying signals. First, a dark disk through which the Earth would pass every $\sim$45 Myr, and second, a dark matter subhalo that the Earth encountered during the past gigayear. We demonstrate that paleo-detectors are sensitive to these examples under a wide variety of experimental scenarios, even in the presence of substantial background uncertainties. This paper shows that paleo-detectors may hold the key to unraveling our Galactic history., Comment: 26 pages, 16 figures, 2 tables, code available at https://github.com/sbaum90/paleoSens and https://github.com/sbaum90/paleoSpec. Published version

Paleo-detectors are a proposed experimental technique to search for dark matter (DM). In lieu of the conventional approach of operating a tonne-scale real-time detector to search for DM-induced nuclear recoils, paleo-detectors take advantage of small samples of naturally occurring rocks on Earth that have been deep underground ($\gtrsim 5$ km), accumulating nuclear damage tracks from recoiling nuclei for $\mathcal{O}(1)$ Gyr. Modern microscopy techniques promise the capability to read out nuclear damage tracks with nanometer resolution in macroscopic samples. Thanks to their $\mathcal{O}(1)$ Gyr integration times, paleo-detectors could constitute nuclear recoil detectors with keV recoil energy thresholds and 100 kilotonne-yr exposures. This combination would allow paleo-detectors to probe DM-nucleon cross sections orders of magnitude below existing upper limits from conventional direct detection experiments. In this article, we use improved background modeling and a new spectral analysis technique to update the sensitivity forecast for paleo-detectors. We demonstrate the robustness of the sensitivity forecast to the (lack of) ancillary measurements of the age of the samples and the parameters controlling the backgrounds, systematic mismodeling of the spectral shape of the backgrounds, and the radiopurity of the mineral samples. Specifically, we demonstrate that even if the uranium concentration in paleo-detector samples is $10^{-8}$ (per weight), many orders of magnitude larger than what we expect in the most radiopure samples obtained from ultra basic rock or marine evaporite deposits, paleo-detectors could still probe DM-nucleon cross sections below current limits. For DM masses $\lesssim 10$ GeV/$c^2$, the sensitivity of paleo-detectors could still reach down all the way to the conventional neutrino floor in a Xe-based direct detection experiment., Comment: Invited contribution to Instruments "Special Issue Innovative Experimental Techniques for Direct Dark Matter Detection)". 30 pages, 5 figures, 1 table. Code available at https://github.com/sbaum90/paleoSpec and https://github.com/sbaum90/paleoSens

The most promising indirect search for the existence of axion dark matter uses radio telescopes to look for narrow spectral lines generated from the resonant conversion of axions in the magnetospheres of neutron stars. Unfortunately, a large list of theoretical uncertainties has prevented this search strategy from being fully accepted as robust. In this work we attempt to address major outstanding questions related to the role and impact of the plasma, including: $(i)$ does refraction and reflection of radio photons in the magnetosphere induce strong inhomogeneities in the flux, $(ii)$ can refraction induce premature axion-photon de-phasing, $(iii)$ to what extent do photon-plasma interactions induce a broadening of the spectral line, $(iv)$ does the flux have a strong time dependence, and $(v)$ can radio photons sourced by axions be absorbed by the plasma. We present an end-to-end analysis pipeline based on ray-tracing that exploits a state-of-the-art auto-differentiation algorithm to propagate photons from the conversion surface to asymptotically large distances. Adopting a charge symmetric Goldreich-Julian model for the magnetosphere, we show that for reasonable parameters one should expect a strong anisotropy of the signal, refraction induced axion-photon de-phasing, significant line-broadening, a variable time-dependence of the flux, and, for large enough magnetic fields, anisotropic absorption. Our simulation code is flexible enough to serve as the basis for follow-up studies with a large range of magnetosphere models., Comment: v3: Fix "?" citation. v2: Accepted for publication PRD. Convergence improved, plots updated, minor bugs and typos corrected. No major changes to results/conclusions. v1: 9 pages + Appendix

In recent years, there have been significant advances in multi-messenger astronomy due to the discovery of the first, and so far only confirmed, gravitational wave event with a simultaneous electromagnetic (EM) counterpart, as well as improvements in numerical simulations, gravitational wave (GW) detectors, and transient astronomy. This has led to the exciting possibility of performing joint analyses of the GW and EM data, providing additional constraints on fundamental properties of the binary progenitor and merger remnant. Here, we present a new Bayesian framework that allows inference of these properties, while taking into account the systematic modeling uncertainties that arise when mapping from GW binary progenitor properties to photometric light curves. We extend the relative binning method presented in Zackay et al. (2018) to include extrinsic GW parameters for fast analysis of the GW signal. The focus of our EM framework is on light curves arising from r-process nucleosynthesis in the ejected material during and after merger, the so called kilonova, and particularly on black hole - neutron star systems. As a case study, we examine the recent detection of GW190425, where the primary object is consistent with being either a black hole (BH) or a neutron star (NS). We show quantitatively how improved mapping between binary progenitor and outflow properties, and/or an increase in EM data quantity and quality are required in order to break degeneracies in the fundamental source parameters., Comment: 22 pages, 12 figures, comments are welcome

The QCD axion is expected to form dense structures known as axion miniclusters if the Peccei-Quinn symmetry is broken after inflation. Miniclusters that have survived until today will interact with neutron stars (NSs) in the Milky Way to produce transient radio signals from axion-photon conversion in the NS magnetosphere. We quantify the properties of these encounters and find that they occur frequently ($\mathcal{O}(1-100)\,\mathrm{day}^{-1}$); last between a day and a few months; are spatially clustered towards the Galactic center; and can reach observable fluxes. These radio transients are within reach of current generation telescopes and therefore offer a promising pathway to discovering QCD axion dark matter., Comment: 11 pages, 5 figures. Code is available online at https://github.com/bradkav/axion-miniclusters/ (archived at https://doi.org/10.5281/zenodo.4006128 ). Joint submission with "Stellar Disruption of Axion Miniclusters in the Milky Way" in astro-ph (arXiv:2011.05377). Updated to match published version

Axion miniclusters are dense bound structures of dark matter axions that are predicted to form in the post-inflationary Peccei-Quinn symmetry breaking scenario. Although dense, miniclusters can easily be perturbed or even become unbound by interactions with baryonic objects such as stars. Here, we characterize the spatial distribution and properties of miniclusters in the Milky Way (MW) today after undergoing these stellar interactions throughout their lifetime. We do this by performing a suite of Monte Carlo simulations which track the miniclusters' structure and, in particular, accounts for partial disruption and mass loss through successive interactions. We consider two density profiles - Navarro-Frenk-White (NFW) and Power-law (PL) - for the individual miniclusters in order to bracket the uncertainties on the minicluster population today due to their uncertain formation history. For our fiducial analysis at the Solar position, we find a survival probability of 99% for miniclusters with PL profiles and 46% for those with NFW profiles. Our work extends previous estimates of this local survival probability to the entire MW. We find that towards the Galactic center, the survival probabilities drop drastically. Although we present results for a particular initial halo mass function, our simulations can be easily recast to different models using the provided data and code. Finally, we comment on the impact of our results on lensing, direct, and indirect detection., Comment: 31 pages, 18 figures. Code available online at https://github.com/bradkav/axion-miniclusters/ (archived at https://doi.org/10.5281/zenodo.4006128 ). Joint submission w/ "Transient Radio Signatures from Neutron Star Encounters with QCD Axion Miniclusters", arXiv:2011.05378 . v2: Minor changes and references added. v3: Updated to match version published in PRD

We provide an updated assessment of the power of the Cherenkov Telescope Array (CTA) to search for thermally produced dark matter at the TeV scale, via the associated gamma-ray signal from pair-annihilating dark matter particles in the region around the Galactic centre. We find that CTA will open a new window of discovery potential, significantly extending the range of robustly testable models given a standard cuspy profile of the dark matter density distribution. Importantly, even for a cored profile, the projected sensitivity of CTA will be sufficient to probe various well-motivated models of thermally produced dark matter at the TeV scale. This is due to CTA's unprecedented sensitivity, angular and energy resolutions, and the planned observational strategy. The survey of the inner Galaxy will cover a much larger region than corresponding previous observational campaigns with imaging atmospheric Cherenkov telescopes. CTA will map with unprecedented precision the large-scale diffuse emission in high-energy gamma rays, constituting a background for dark matter searches for which we adopt state-of-the-art models based on current data. Throughout our analysis, we use up-to-date event reconstruction Monte Carlo tools developed by the CTA consortium, and pay special attention to quantifying the level of instrumental systematic uncertainties, as well as background template systematic errors, required to probe thermally produced dark matter at these energies. "Full likelihood tables complementing our analysis are provided here [ https://doi.org/10.5281/zenodo.4057987 ]", Comment: 68 pages (including references) and 26 figures; text identical to the version published in JCAP

We study whether binary black hole template banks can be used to search for the gravitational waves emitted by general binary coalescences. To recover binary signals from noisy data, matched-filtering techniques are typically required. This is especially true for low-mass systems, with total mass $M \lesssim 10 \, M_\odot$, which can inspiral in the LIGO and Virgo frequency bands for thousands of cycles. In this paper, we focus on the detectability of low-mass binary systems whose individual components can have large spin-induced quadrupole moments and small compactness. The quadrupole contributes to the phase evolution of the waveform whereas the compactness affects the merger frequency of the binary. We find that binary black hole templates (with dimensionless quadrupole $\kappa=1$) cannot be reliably used to search for objects with large quadrupoles ($\kappa\gtrsim 20$) over a wide range of parameter space. This is especially true if the general object is highly spinning and has a larger mass than its binary companion. A binary that consists of objects with small compactness could merge in the LIGO and Virgo frequency bands, thereby reducing its accumulated signal-to-noise ratio during the inspiraling regime. Template banks which include these more general waveforms must therefore be constructed. These extended banks would allow us to realistically search for the existence of new astrophysical and beyond the Standard Model compact objects., Comment: 16 pages, 5 figures. Additional figure added. Updated to match published version

In this paper, we introduce a novel form of value function, $Q(s, s')$, that expresses the utility of transitioning from a state $s$ to a neighboring state $s'$ and then acting optimally thereafter. In order to derive an optimal policy, we develop a forward dynamics model that learns to make next-state predictions that maximize this value. This formulation decouples actions from values while still learning off-policy. We highlight the benefits of this approach in terms of value function transfer, learning within redundant action spaces, and learning off-policy from state observations generated by sub-optimal or completely random policies. Code and videos are available at http://sites.google.com/view/qss-paper., Comment: Accepted into ICML 2020

Axion dark matter can resonantly convert into photons in the magnetospheres of neutron stars (NSs). It has recently been shown that radio observations of nearby NSs can therefore provide a highly sensitive probe of the axion parameter space. Here we extend existing calculations by performing the first three-dimensional computation of the photon flux, taking into account the isotropic phase-space distribution of axions and the structure of the NS magnetosphere. In particular, we study the overall magnitude of the flux and its possible time variation. We find that overall signal strength is robust to our more realistic analysis. In addition, we find that the variance of the signal with respect to the NS rotation is washed out by the additional trajectories in our treatment. Nevertheless, we show that SKA observations towards J0806.4-4123 are sensitive to $g_{a\gamma\gamma}\sim 3\times10^{-13}\mathrm{\,GeV^{-1}}$ at $m_a\sim 7\times10^{-6}\mathrm{\,eV}$, even when accounting for Doppler broadening. Finally, we provide the necessary code to calculate the photon flux for any given NS system., Comment: 12 pages, 6 figures

Paleo-detectors are a proposed experimental technique in which one would search for traces of recoiling nuclei in ancient minerals. Natural minerals on Earth are as old as $\mathcal{O}(1)\,$Gyr and, in many minerals, the damage tracks left by recoiling nuclei are also preserved for timescales long compared to 1 Gyr once created. Thus, even reading out relatively small target samples of order 100 g, paleo-detectors would allow one to search for very rare events thanks to the large exposure, $\varepsilon \sim 100\,{\rm g}\,{\rm Gyr} = 10^5\,{\rm t}\,{\rm yr}$. Here, we explore the potential of paleo-detectors to measure nuclear recoils induced by neutrinos from galactic core collapse supernovae. We find that they would not only allow for a direct measurement of the average core collapse supernova rate in the Milky Way, but would also contain information about the time-dependence of the local supernova rate over the past $\sim$1 Gyr. Since the supernova rate is thought to be directly proportional to the star formation rate, such a measurement would provide a determination of the local star formation history. We investigate the sensitivity of paleo-detectors to both a smooth time evolution and an enhancement of the core collapse supernova rate on relatively short timescales, as would be expected for a starburst period in the local group., Comment: 15 pages, 5 figures. code available at https://github.com/tedwards2412/SN-paleology (archived at https://doi.org/10.5281/zenodo.3066206 ) v2: matches the published version

Imitation by observation is an approach for learning from expert demonstrations that lack action information, such as videos. Recent approaches to this problem can be placed into two broad categories: training dynamics models that aim to predict the actions taken between states, and learning rewards or features for computing them for Reinforcement Learning (RL). In this paper, we introduce a novel approach that learns values, rather than rewards, directly from observations. We show that by using values, we can significantly speed up RL by removing the need to bootstrap action-values, as compared to sparse-reward specifications., Comment: Accepted into the Workshop on Self-Supervised Learning at ICML 2019

We propose a multi-messenger probe of QCD axion Dark Matter based on observations of black hole-neutron star binary inspirals. It is suggested that a dense Dark Matter spike may grow around intermediate mass black holes ($10^{3}-10^{5} \mathrm{\,M_{\odot}}$). The presence of such a spike produces two unique effects: a distinct phase shift in the gravitational wave strain during the inspiral and an enhancement of the radio emission due to the resonant axion-photon conversion occurring in the neutron star magnetosphere throughout the inspiral and merger. Remarkably, the observation of the gravitational wave signal can be used to infer the Dark Matter density and, consequently, to predict the radio emission. We study the projected reach of the LISA interferometer and next-generation radio telescopes such as the Square Kilometre Array. Given a sufficiently nearby system, such observations will potentially allow for the detection of QCD axion Dark Matter in the mass range $10^{-7}\,\mathrm{eV}$ to $10^{-5}\,\mathrm{eV}$., Comment: 5 pages, 3 figures. Appendix added with additional figures. Updated to published version