Search

Using publicly available numerical relativity (NR) simulations for non-spinning eccentric binary black hole (BBH) mergers, Ref \cite{Islam:2024rhm} demonstrated that the eccentricity-induced modulations in the amplitudes and frequencies of different spherical harmonic modes are mutually consistent and can be modeled using a single time series modulation. We extend the validity of the results to all non-precessing binaries by using 83 NR simulations from the SXS, RIT, and MAYA catalogs for aligned-spin eccentric BBH mergers with mass ratios ranging from $1:1$ to $1:4$. Based on these phenomenological relations, we provide a framework named gwNRXHME to compute multi-modal eccentric non-precessing waveforms using two inputs: quadrupolar eccentric waveforms, and the corresponding multi-modal quasi-circular non-precessing waveforms. Furthermore, we compute an overall degree of departure in SXS, RIT, and MAYA NR data from these relations and find that SXS NR simulations generally adhere to these relations more strictly than RIT and MAYA data. We also show that these relations can offer a cost-effective way to filter out noisy higher-order spherical harmonic modes extracted from NR data. Our framework is publicly available through the gwModels package., Comment: 9 pages, 6 figures

Next-generation ground-based gravitational-wave (GW) detectors are expected to detect millions of binary black hole mergers during their operation period. A small fraction ($\sim 0.1 - 1\%$) of them will be strongly lensed by intervening galaxies and clusters, producing multiple copies of the GW signals. The expected number of lensed events and the distribution of the time delay between lensed images will depend on the mass distribution of the lenses at different redshifts. Warm dark matter or fuzzy dark matter models predict lower abundances of small mass dark matter halos as compared to the standard cold dark matter. This will result in a reduction in the number of strongly lensed GW events, especially at small time delays. Using the number of lensed events and the lensing time delay distribution, we can put a lower bound on the mass of the warm/fuzzy dark matter particle from a catalog of lensed GW events. The expected bounds from GW strong lensing from next-generation detectors are significantly better than the current constraints., Comment: 8 pages, 7 figures

As Large Language Models (LLMs) achieve remarkable performance across various NLP tasks, their reliability becomes essential for widespread adoption. This paper focuses on Abstention Ability (AA), a critical yet under explored aspect of reliability - the ability of LLMs to refrain from answering questions when they are uncertain or when definitive answer is not possible, while maintaining question-answering (QA) task performance. While previous works have focused on understanding the recollection abilities of LLMs or their ability to identify imponderable/unanswerable questions, we believe there is a need for an effective AA evaluation method. Therefore, we propose a black-box evaluation methodology to examine and understand the AA of LLMs across a variety of multiple-choice QA tasks. We measure AA by rewarding models for abstaining from answering when their predictions are incorrect or when the questions are inherently unanswerable. We investigate three strategies, Strict Prompting, Verbal Confidence Thresholding, and Chain-of-Thought (CoT), to understand their impact on abstention across different LLMs. Our findings reveal that while even state-of-the-art LLMs like GPT-4 struggle with abstention, strategic prompting such as CoT, can significantly enhance this ability. Furthermore, we demonstrate that improving AA also leads to better overall QA task performance, underscoring the importance of evaluating AA in LLMs., Comment: 5 pages (5th page contains References) and 2 figures

For navigation of robots, image segmentation is an important component to determining a terrain's traversability. For safe and efficient navigation, it is key to assess the uncertainty of the predicted segments. Current uncertainty estimation methods are limited to a specific choice of model architecture, are costly in terms of training time, require large memory for inference (ensembles), or involve complex model architectures (energy-based, hyperbolic, masking). In this paper, we propose a simple, light-weight module that can be connected to any pretrained image segmentation model, regardless of its architecture, with marginal additional computation cost because it reuses the model's backbone. Our module is based on maximum separation of the segmentation classes by respective prototype vectors. This optimizes the probability that out-of-distribution segments are projected in between the prototype vectors. The uncertainty value in the classification label is obtained from the distance to the nearest prototype. We demonstrate the effectiveness of our module for terrain segmentation., Comment: 10 pages

This paper explores using GPT-3.5 and GPT-4 to automate the data annotation process with automatic prompting techniques. The main aim of this paper is to reuse human annotation guidelines along with some annotated data to design automatic prompts for LLMs, focusing on the semantic proximity annotation task. Automatic prompts are compared to customized prompts. We further implement the prompting strategies into an open-source text annotation tool, enabling easy online use via the OpenAI API. Our study reveals the crucial role of accurate prompt design and suggests that prompting GPT-4 with human-like instructions is not straightforwardly possible for the semantic proximity task. We show that small modifications to the human guidelines already improve the performance, suggesting possible ways for future research., Comment: 12 pages

We present a simple variable quantization approach that quantizes different layers of a large language model (LLM) at different bit levels. Specifically, we quantize the most important layers to higher bit precision and less important layers to lower bits to achieve floating point quantization levels. We propose two effective strategies to measure the importance of layers within LLMs: the first measures the importance of a layer based on how different its output embeddings are from the input embeddings (the higher the better); the second estimates the importance of a layer using the number of layer weights that are much larger than average (the smaller the better). We show that quantizing different layers at varying bits according to our importance scores results in minimal performance drop with a far more compressed model size. Finally, we present several practical key takeaways from our variable layer-wise quantization experiments: (a) LLM performance under variable quantization remains close to the original model until 25-50% of layers are moved in lower quantization using our proposed ordering but only until 5-10% if moved using no specific ordering; (b) Quantizing LLMs to lower bits performs substantially better than pruning unless extreme quantization (2-bit) is used; and (c) Layer-wise quantization to lower bits works better in the case of larger LLMs with more layers compared to smaller LLMs with fewer layers. The code used to run the experiments is available at: https://github.com/RazvanDu/LayerwiseQuant., Comment: submitted to EMNLP, 15 pages, 10 figures, 4 tables

Instruction finetuning (IFT) is critical for aligning Large Language Models (LLMs) to follow instructions. While many effective IFT datasets have been introduced recently, they predominantly focus on high-resource languages like English. To better align LLMs across a broad spectrum of languages and tasks, we propose a fully synthetic, novel taxonomy (Evol) guided Multilingual, Multi-turn instruction finetuning dataset, called M2Lingual. It is constructed by first selecting a diverse set of seed examples and then utilizing the proposed Evol taxonomy to convert these seeds into complex and challenging multi-turn instructions. We demonstrate the effectiveness of M2Lingual by training LLMs of varying sizes and showcasing the enhanced performance across a diverse set of languages. We contribute the 2 step Evol taxonomy with the guided generation code: https://github.com/ServiceNow/M2Lingual, as well as the first fully synthetic, general and task-oriented, multi-turn, multilingual dataset built with Evol - M2Lingual: https://huggingface.co/datasets/ServiceNow-AI/ M2Lingual - containing 182K total IFT pairs, covering 70 languages and 17+ NLP tasks., Comment: 39 pages

Despite rapid progress in large language models (LLMs), their performance on a vast majority of languages remain unsatisfactory. In this paper, we study building language-specific LLMs by adapting monolingual and multilingual LLMs. We conduct systematic experiments on how design choices (base model selection, vocabulary extension, and continued fine-tuning) impact the adapted LLM, both in terms of efficiency (how many tokens are needed to encode the same amount of information) and end task performance. We find that (1) the initial performance before the adaptation is not always indicative of the final performance. (2) Efficiency can easily improved with simple vocabulary extension and continued fine-tuning in most LLMs we study, and (3) The optimal adaptation method is highly language-dependent, and the simplest approach works well across various experimental settings. Adapting English-centric models can yield better results than adapting multilingual models despite their worse initial performance on low-resource languages. Together, our work lays foundations on efficiently building language-specific LLMs by adapting existing LLMs., Comment: 15 pages, 6 figures, 11 tables

Popular parameter-efficient fine-tuning (PEFT) methods, such as LoRA and its variants, freeze pre-trained model weights \(W\) and inject learnable matrices \(\Delta W\). These \(\Delta W\) matrices are structured for efficient parameterization, often using techniques like low-rank approximations or scaling vectors. However, these methods typically show a performance gap compared to full fine-tuning. Although recent PEFT methods have narrowed this gap, they do so at the cost of additional learnable parameters. We propose SVFT, a simple approach that fundamentally differs from existing methods: the structure imposed on \(\Delta W\) depends on the specific weight matrix \(W\). Specifically, SVFT updates \(W\) as a sparse combination of outer products of its singular vectors, training only the coefficients (scales) of these sparse combinations. This approach allows fine-grained control over expressivity through the number of coefficients. Extensive experiments on language and vision benchmarks show that SVFT recovers up to 96% of full fine-tuning performance while training only 0.006 to 0.25% of parameters, outperforming existing methods that only recover up to 85% performance using 0.03 to 0.8% of the trainable parameter budget., Comment: 17 pages, 5 figures, 14 tables

We present a detailed exposition of a statistical method for estimating cosmological parameters from the observation of a large number of strongly lensed binary-black-hole (BBH) mergers observable by third-generation gravitational-wave (GW) detectors. This method, first presented in $\href{https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.130.261401}{[1]}$, compares the observed number of strongly lensed GW events and their time delay distribution (between lensed images) with observed events to infer cosmological parameters. We show that the precision of the estimation of the cosmological parameters does not have a strong dependance on the assumed BBH redshift distribution model. An incorrect model for the distribution of lens properties and redshifts can lead to a biased cosmological inference. However, Bayesian model selection can assist in selecting the right model from a set of available parametric models for the distribution of lens redshift and properties. We also present a way to incorporate the effect of contamination in the data due to the limited efficiency of lensing identification methods, so that it will not bias the cosmological inference., Comment: 17 pages, 5 figures

Nearly all template-based gravitational wave (GW) searches only include the quasi-circular quadrupolar modes of the signals in their templates. Including additional degrees of freedom in the GW templates corresponding to higher-order harmonics, orbital precession, or eccentricity is challenging because: ($i$) the size of template banks and the matched-filtering cost increases significantly with the number of degrees of freedom, $(ii)$ if these additional degrees are not included properly, the search can lose sensitivity overall (due to an increase in the rate of background triggers). Here, we focus on including aligned-spin higher harmonics in GW search templates. We use a new mode-by-mode filtering approach, where we separately filter GW strain data with three harmonics [namely $(\ell, |m|)=(2,2)$, $(3,3)$ and $(4,4)$]. This results in an increase in the matched-filtering cost by only a factor of $3$ compared to that of a $(2,2)$-only search. We develop computationally cheap trigger-ranking statistics to optimally combine the different signal-to-noise ratios (SNR) timeseries from different harmonics, which ensure only physically-allowed combinations of the different harmonics are triggered on. We use an empirical template-dependent background model in our ranking statistic to account for non-Gaussian transients. In addition, we develop a tool called band eraser which specifically excises narrow time-varying noisy bands in time-frequency space (without having to excise entire time chunks in the data). New GW candidate events that we detect using our $\texttt{IAS-HM}$ search pipeline and the details of our template banks are discussed in accompanying papers: Wadekar et al. [1] and [2] respectively. Apart from higher harmonics, we expect our methodology to also be useful for cheap and optimal searches including orbital precession and eccentricity in GW waveforms., Comment: 11+4 pages, 5 figures. Version appearing in PRD. The IAS-HM pipeline is available at https://github.com/JayWadekar/gwIAS-HM

We introduce an algorithm to marginalize the likelihood for a gravitational wave signal from a quasi-circular binary merger over its extrinsic parameters, accounting for the effects of higher harmonics and spin-induced precession. The algorithm takes as input the matched-filtering time series of individual waveform harmonics against the data in all operational detectors, and the covariances of the harmonics. The outputs are the Gaussian likelihood marginalized over extrinsic parameters describing the merger time, location and orientation, along with samples from the conditional posterior of these parameters. Our algorithm exploits the waveform's known analytical dependence on extrinsic parameters to efficiently marginalize over them using a single waveform evaluation. Our current implementation achieves a 10% precision on the marginalized likelihood within $\approx 50$ ms on a single CPU core and is publicly available through the package `cogwheel`. We discuss applications of this tool for gravitational wave searches involving higher modes or precession, efficient and robust parameter estimation, and generation of sky localization maps in low latency for electromagnetic followup of gravitational-wave alerts. The inclusion of higher modes can improve the distance measurement, providing an advantage over existing low-latency localization methods., Comment: 19 pages, 7 figures

Direct Preference Optimization (DPO) is an effective technique that leverages pairwise preference data (usually one chosen and rejected response pair per user prompt) to align LLMs to human preferences. In practice, multiple responses can exist for a given prompt with varying quality relative to each other. With availability of such quality ratings for multiple responses, we propose utilizing these responses to create multiple preference pairs for a given prompt. Our work focuses on systematically using the constructed multiple preference pair in DPO training via curriculum learning methodology. In particular, we order these multiple pairs of preference data from easy to hard (emulating curriculum training) according to various criteria. We show detailed comparisons of our proposed approach to the standard single-pair DPO setting. Our method, which we call Curry-DPO consistently shows increased performance gains on MTbench, Vicuna, WizardLM, and the UltraFeedback test set, highlighting its effectiveness. More specifically, Curry-DPO achieves a score of 7.43 on MT-bench with Zephy-7B model outperforming majority of existing LLMs with similar parameter size. Curry-DPO also achieves the highest adjusted win rates on Vicuna, WizardLM, and UltraFeedback test datasets (90.7%, 87.1%, and 87.9% respectively) in our experiments, with notable gains of upto 7.5% when compared to standard DPO technique., Comment: Work in progress

In this study, we experimentally examine the behavior of a free falling rigid sphere impacting normally onto, and penetrating into a quiescent liquid pool. Parameters, which are varied include the impact velocity, the density, and the diameter of the sphere. Observations of the sphere trajectory in time are made using two orthogonally placed high-speed cameras, yielding the velocity and acceleration vector through repeated differentiation of the time resolved trajectories. Upon penetration, the sphere goes through three very distinct phases of penetration, denoted as the submersion, deceleration and settling phase, each clearly identifiable through either features seen in trajectory direction or in changes of velocity. These phases exist for all impact Reynolds numbers and density ratios investigated, and their respective duration remain astoundingly constant in terms of dimensionless time. The motion of the sphere is analysed using a scalar force balance for each of instantaneous drag and lift, yielding quantitative estimates of the drag and lift coefficients throughout the trajectory. The variation of these forces can be phenomenologically explained by unsteady wake behavior arising from strong deceleration and through transient asymmetry, leading to variations in trajectory curvature. Despite the large trajectory randomness observed in repetitive experiments, there exist strong commonalities in motion behavior.

This review provides a conceptual and technical survey of methods for parameter estimation of gravitational wave signals in ground-based interferometers such as LIGO and Virgo. We introduce the framework of Bayesian inference and provide an overview of models for the generation and detection of gravitational waves from compact binary mergers, focusing on the essential features that are observable in the signals. Within the traditional likelihood-based paradigm, we describe various approaches for enhancing the efficiency and robustness of parameter inference. This includes techniques for accelerating likelihood evaluations, such as heterodyne/relative binning, reduced-order quadrature, multibanding and interpolation. We also cover methods to simplify the analysis to improve convergence, via reparametrization, importance sampling and marginalization. We end with a discussion of recent developments in the application of likelihood-free (simulation-based) inference methods to gravitational wave data analysis., Comment: 28 pages, 3 figures. Accepted for publication in Annual Review of Nuclear and Particle Science. v2: GW200115 reanalyzed in Figure 2, to fix an issue with corrupted high-frequency data; plus minor edits

A vital aspect of Indian Classical Music (ICM) is Raga, which serves as a melodic framework for compositions and improvisations alike. Raga Recognition is an important music information retrieval task in ICM as it can aid numerous downstream applications ranging from music recommendations to organizing huge music collections. In this work, we propose a deep learning based approach to Raga recognition. Our approach employs efficient pre possessing and learns temporal sequences in music data using Long Short Term Memory based Recurrent Neural Networks (LSTM-RNN). We train and test the network on smaller sequences sampled from the original audio while the final inference is performed on the audio as a whole. Our method achieves an accuracy of 88.1% and 97 % during inference on the Comp Music Carnatic dataset and its 10 Raga subset respectively making it the state-of-the-art for the Raga recognition task. Our approach also enables sequence ranking which aids us in retrieving melodic patterns from a given music data base that are closely related to the presented query sequence.

Nearly all of the previous gravitational wave (GW) searches in the LIGO-Virgo data included GW waveforms with only the dominant quadrupole mode, i.e., omitting higher-order harmonics which are predicted by general relativity. Based on the techniques developed in Wadekar et al. [1,2], we improve the IAS pipeline by ($i$) introducing higher harmonics in the GW templates, ($ii$) downweighting noise transients ('glitches') to improve the search sensitivity to high-mass and high-redshift binary black hole (BBH) mergers. We find 14 new BBH mergers with $0.53\leq p_{\rm astro}\leq 0.88$ on running our pipeline over the public LIGO-Virgo data from the O3 run (we use the detection threshold as $p_{\rm astro}>0.5$ following the approach of other pipelines). We also broadly recover the high-significance events from earlier catalogs, except some which were either vetoed or fell below our SNR threshold for trigger collection. A few notable properties of our new candidate events are as follows. At $>95$\% credibility, 4 candidates have total masses in the IMBH range (i.e., above 100 $M_\odot$), and 9 candidates have $z>0.5$. 9 candidates have median mass of the primary BH falling roughly within the pair instability mass gap, with the highest primary mass being $300_{+60}^{-120} M_\odot$. 5 candidates have median mass ratio $q < 0.5$. Under a prior uniform in effective spin $\chi_{\rm eff}$, 6 candidates have $\chi_{\rm eff} > 0$ at $>95\%$ credibility. We also find that including higher harmonics in our search raises the significance of a few previously reported marginal events (e.g., GW190711_030756). While our new candidate events have modest false alarm rates ($\gtrsim 1.6 $/yr), a population inference study including these can better inform the parameter space of BHs corresponding to the pair instability mass gap, high redshifts, positive effective spins and asymmetric mass ratios., Comment: 15+7 pages, 6+5 figures. Our event catalog (including sub-threshold events) and the samples from our PE runs are available at https://github.com/JayWadekar/GW_higher_harmonics_search

Efficient and accurate product relevance assessment is critical for user experiences and business success. Training a proficient relevance assessment model requires high-quality query-product pairs, often obtained through negative sampling strategies. Unfortunately, current methods introduce pooling bias by mistakenly sampling false negatives, diminishing performance and business impact. To address this, we present Bias-mitigating Hard Negative Sampling (BHNS), a novel negative sampling strategy tailored to identify and adjust for false negatives, building upon our original False Negative Estimation algorithm. Our experiments in the Instacart search setting confirm BHNS as effective for practical e-commerce use. Furthermore, comparative analyses on public dataset showcase its domain-agnostic potential for diverse applications., Comment: Submitted to WWW'24 Industry Track

We report the detection of 6 new candidate binary black hole (BBH) merger signals in the publicly released data from the second half of the third observing run (O3b) of advanced LIGO and advanced Virgo. The LIGO-Virgo-KAGRA (LVK) collaboration reported 35 compact binary coalescences (CBCs) in their analysis of the O3b data [1], with 30 BBH mergers having coincidence in the Hanford and Livingston detectors. We confirm 17 of these for a total of 23 detections in our analysis of the Hanford-Livingston coincident O3b data. We identify candidates using a search pipeline employing aligned-spin quadrupole-only waveforms. Our pipeline is similar to the one used in our O3a coincident analysis [2], except for a few improvements in the veto procedure and the ranking statistic, and we continue to use an astrophysical probability of one half as our detection threshold, following the approach of the LVK catalogs. Most of the new candidates reported in this work are placed in the upper and lower-mass gap of the black hole (BH) mass distribution. We also identify a possible neutron star-black hole (NSBH) merger. We expect these events to help inform the black hole mass and spin distributions inferred in a full population analysis., Comment: 16 pages, 12 figures

Searches for gravitational wave events use models, or templates, for the signals of interest. The templates used in current searches in the LIGO-Virgo-Kagra (LVK) data model the dominant quadrupole mode $(\ell,m)=(2,2)$ of the signals, and omit sub-dominant higher-order modes (HM) such as $(\ell,m)=(3,3)$, $(4,4)$, which are predicted by general relativity. Hence, these searches could lose sensitivity to black hole mergers in interesting parts of parameter space, such as systems with high-masses and asymmetric mass ratios. We develop a new strategy to include HM in template banks that exploits the natural connection between the modes. We use a combination of post-Newtonian formulae and machine learning tools to model aligned-spin $(3,3)$, $(4,4)$ waveforms corresponding to a given $(2,2)$ waveform. Each of these modes can be individually filtered against the data to yield separate timeseries of signal-to-noise ratios (SNR), which can be combined in a relatively inexpensive way to marginalize over extrinsic parameters of the signals. This leads to a HM search pipeline whose matched-filtering cost is just $\approx 3\times$ that of a quadrupole-only search (in contrast to being $\approx\! 100 \times$, as in previously proposed HM search methods). Our method is effectual and is generally applicable for template banks constructed with either stochastic or geometric placement techniques. Additionally, we discuss compression of $(2,2)$-only geometric-placement template banks using machine learning algorithms., Comment: 12+2 pages, 7+1 figures. The template bank described here will be publicly available at https://github.com/JayWadekar/GW_higher_harmonics_search

A diverse set of Internet of Things (IoT) devices are becoming an integrated part of daily lives, and playing an increasingly vital role in various industry, enterprise and agricultural settings. The current IoT ecosystem relies on several IoT management platforms to manage and operate a large number of IoT devices, their data, and their connectivity. Considering their key role, these platforms must be properly secured against cyber attacks. In this work, we first explore the core operations/features of leading platforms to design a framework to perform a systematic security evaluation of these platforms. Subsequently, we use our framework to analyze a representative set of 52 IoT management platforms, including 42 web-hosted and 10 locally-deployable platforms. We discover a number of high severity unauthorized access vulnerabilities in 9/52 evaluated IoT management platforms, which could be abused to perform attacks such as remote IoT SIM deactivation, IoT SIM overcharging and IoT device data forgery. More seriously, we also uncover instances of broken authentication in 13/52 platforms, including complete account takeover on 8/52 platforms along with remote code execution on 2/52 platforms. In effect, 17/52 platforms were affected by vulnerabilities that could lead to platform-wide attacks. Overall, vulnerabilities were uncovered in 33 platforms, out of which 28 platforms responded to our responsible disclosure. We were also assigned 11 CVEs and awarded bounty for our findings., Comment: Accepted in IEEE Internet of Things Journal in June 2023

We present IMRPhenomXODE, a new phenomenological frequency-domain waveform approximant for gravitational wave (GW) signals from precessing binary black holes (BBHs) with generic spin configurations. We build upon the success of IMRPhenomXPHM [G. Pratten et al., Phys. Rev. D 103, 104056 (2021), which is one of the most widely adopted waveform approximants in GW data analyses that include spin precession, and introduce two additional significant improvements. First, we employ an efficient technique to numerically solve the (next-to)$^4$-leading-order post-Newtonian precession equations, which allows us to accurately determine the evolution of the orientation of the orbital angular momentum $\boldsymbol{\hat{L}}_{\rm N}$ even in cases with complicated precession dynamics, such as transitional precession. Second, we recalibrate the phase of GW modes in the frame coprecessing with $\boldsymbol{\hat{L}}_{\rm N}$ against SEOBNRv4PHM [S. Ossokine et al., Phys. Rev. D 102, 044055 (2020)] to capture effects due to precession such as variations in the spin components aligned with $\boldsymbol{\hat{L}}_{\rm N}$. By incorporating these new features, IMRPhenomXODE achieves matches with SEOBNRv4PHM that are better than 99% for most BBHs with mass ratios $q \geq 1/6$ and with arbitrary spin configurations. In contrast, the mismatch between IMRPhenomXPHM and SEOBNRv4PHM often exceeds 10% for a BBH with $q\lesssim 1/2$ and large in-plane or antialigned spin components. Our implementation is also computationally efficient, with waveform evaluation times that can even be shorter than those of IMRPhenomXPHM for BBH signals with long durations and hence high frequency resolutions. The accuracy and efficiency of IMRPhenomXODE position it as a valuable tool for GW event searches, parameter estimation analyses, and the inference of underlying population properties., Comment: 24 pages, 14 figures, 3 tables. Published in PRD

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

Localization for autonomous robots in prior maps is crucial for their functionality. This paper offers a solution to this problem for indoor environments called InstaLoc, which operates on an individual lidar scan to localize it within a prior map. We draw on inspiration from how humans navigate and position themselves by recognizing the layout of distinctive objects and structures. Mimicking the human approach, InstaLoc identifies and matches object instances in the scene with those from a prior map. As far as we know, this is the first method to use panoptic segmentation directly inferring on 3D lidar scans for indoor localization. InstaLoc operates through two networks based on spatially sparse tensors to directly infer dense 3D lidar point clouds. The first network is a panoptic segmentation network that produces object instances and their semantic classes. The second smaller network produces a descriptor for each object instance. A consensus based matching algorithm then matches the instances to the prior map and estimates a six degrees of freedom (DoF) pose for the input cloud in the prior map. The significance of InstaLoc is that it has two efficient networks. It requires only one to two hours of training on a mobile GPU and runs in real-time at 1 Hz. Our method achieves between two and four times more detections when localizing, as compared to baseline methods, and achieves higher precision on these detections.