Search

Many early universe scenarios predict an enhancement of scalar perturbations at scales currently unconstrained by cosmological probes. These perturbations source gravitational waves (GWs) at second order in perturbation theory, leading to a scalar-induced gravitational wave (SIGW) background. The LISA detector, sensitive to mHz GWs, will be able to constrain curvature perturbations in a new window corresponding to scales $k \in [10^{10}, 10^{14}] \,{\rm Mpc}^{-1}$, difficult to probe otherwise. In this work, we forecast the capabilities of LISA to constrain the source of SIGWs using different approaches: i) agnostic, where the spectrum of curvature perturbations is binned in frequency space; ii) template-based, modeling the curvature power spectrum based on motivated classes of models; iii) ab initio, starting from first-principles model of inflation featuring an ultra-slow roll phase. We compare the strengths and weaknesses of each approach. We also discuss the impact on the SIGW spectrum of non-standard thermal histories affecting the kernels of SIGW emission and non-Gaussianity in the statistics of the curvature perturbations. Finally, we propose simple tests to assess whether the signal is compatible with the SIGW hypothesis. The pipeline used is built into the SIGWAY code., Comment: 95 pages, 31 figures

Learned sparse text embeddings have gained popularity due to their effectiveness in top-k retrieval and inherent interpretability. Their distributional idiosyncrasies, however, have long hindered their use in real-world retrieval systems. That changed with the recent development of approximate algorithms that leverage the distributional properties of sparse embeddings to speed up retrieval. Nonetheless, in much of the existing literature, evaluation has been limited to datasets with only a few million documents such as MSMARCO. It remains unclear how these systems behave on much larger datasets and what challenges lurk in larger scales. To bridge that gap, we investigate the behavior of state-of-the-art retrieval algorithms on massive datasets. We compare and contrast the recently-proposed Seismic and graph-based solutions adapted from dense retrieval. We extensively evaluate Splade embeddings of 138M passages from MsMarco-v2 and report indexing time and other efficiency and effectiveness metrics.

Approximate Nearest Neighbors (ANN) search is a crucial task in several applications like recommender systems and information retrieval. Current state-of-the-art ANN libraries, although being performance-oriented, often lack modularity and ease of use. This translates into them not being fully suitable for easy prototyping and testing of research ideas, an important feature to enable. We address these limitations by introducing kANNolo, a novel research-oriented ANN library written in Rust and explicitly designed to combine usability with performance effectively. kANNolo is the first ANN library that supports dense and sparse vector representations made available on top of different similarity measures, e.g., euclidean distance and inner product. Moreover, it also supports vector quantization techniques, e.g., Product Quantization, on top of the indexing strategies implemented. These functionalities are managed through Rust traits, allowing shared behaviors to be handled abstractly. This abstraction ensures flexibility and facilitates an easy integration of new components. In this work, we detail the architecture of kANNolo and demonstrate that its flexibility does not compromise performance. The experimental analysis shows that kANNolo achieves state-of-the-art performance in terms of speed-accuracy trade-off while allowing fast and easy prototyping, thus making kANNolo a valuable tool for advancing ANN research. Source code available on GitHub: https://github.com/TusKANNy/kannolo., Comment: 7 pages, 3 figures

The rise of Artificial Intelligence and Large Language Models is driving increased GPU usage in data centers for complex training and inference tasks, impacting operational costs, energy demands, and the environmental footprint of large-scale computing infrastructures. This work addresses the online scheduling problem in GPU datacenters, which involves scheduling tasks without knowledge of their future arrivals. We focus on two objectives: minimizing GPU fragmentation and reducing power consumption. GPU fragmentation occurs when partial GPU allocations hinder the efficient use of remaining resources, especially as the datacenter nears full capacity. A recent scheduling policy, Fragmentation Gradient Descent (FGD), leverages a fragmentation metric to address this issue. Reducing power consumption is also crucial due to the significant power demands of GPUs. To this end, we propose PWR, a novel scheduling policy to minimize power usage by selecting power-efficient GPU and CPU combinations. This involves a simplified model for measuring power consumption integrated into a Kubernetes score plugin. Through an extensive experimental evaluation in a simulated cluster, we show how PWR, when combined with FGD, achieves a balanced trade-off between reducing power consumption and minimizing GPU fragmentation., Comment: This work has been submitted to the IEEE for possible publication

In active systems, whose constituents have non-equilibrium dynamics at local level, fluid-fluid phase separation is widely observed. Examples include the formation of membraneless organelles within cells; the clustering of self-propelled colloidal particles in the absence of attractive forces, and some types of ecological segregation. A schematic understanding of such active phase separation was initially borrowed from what is known for the equilibrium case, in which detailed balance holds at microscopic level. However it has recently become clear that in active systems the absence of detailed balance, although it leave phase separation qualitatively unchanged in some regimes (e.g. domain growth driven by interfacial tension via Ostwald ripening), can in other regimes radically alter its phenomenology at mechanistic level. For example, microphase separation can be caused by reverse Ostwald ripening, a process that is hard to imagine from an equilibrium perspective. This and other new phenomena arise because, instead of having a single, positive interfacial tension like their equilibrium counterparts, the fluid-fluid interfaces created by active phase separation can have several distinct interfacial tensions governing different properties, some of which can be negative. These phenomena can be broadly understood by studying continuum field theories for a single conserved scalar order parameter (the fluid density), supplemented with a velocity field if momentum conservation is also present. More complex regimes arise in systems described by multiple scalar order parameters (especially with nonreciprocal interactions between these); or when an order parameter undergoes both conserved and non-conserved dynamics (such that the combination breaks detailed balance); or in systems that support orientational long-range order in one or more of the coexisting phases. In this Review [...]

In phase-separated active fluids, the Ostwald process can go into reverse leading to either microphase separation or bubbly phase separation. We show that the latter is formed of two macroscopic regions that are occupied by the homogeneous fluid and by the microphase separated one. Within the microphase separated fluid, the relative rate of the Ostwald process, coalescence, and nucleation determines whether the size distribution of mesoscopic domains is narrowly peaked or displays a broad range of sizes before attaining a cutoff independent of system-size. Our results are obtained via large-scale simulations of a minimal field theory for active phase separation and reproduced by an effective model in which the degrees of freedom are the locations and sizes of the microphase-separated domains.

Stellar-mass black-hole binaries are the most numerous gravitational-wave sources observed to date. Their properties make them suitable for observation both by ground- and space-based detectors. Starting from synthetic catalogues constructed based on observational constraints from ground-based detectors, we explore the detection rates and the characteristic parameters of the stellar-mass black-hole binaries observable by LISA during their inspiral, using signal-to-noise ratio thresholds as a detection criterion. We find that only a handful of these sources will be detectable with signal-to-noise ratio larger than 8: about 5 sources on average in 4 years of mission duration, among which only one or two are multiband ones (i.e. merging in less than 15 years). We find that detectable sources have chirp mass $10 M_\odot\lesssim \mathcal{M}_c\lesssim 100 M_\odot$, residual time-to-coalescence $4\mathrm{yr}\lesssim \tau_c\lesssim 100 \mathrm{yr}$, and redshift $z\lesssim 0.1$, much closer than those observed up to now by ground-based detectors. We also explore correlations between the number of LISA detectable sources and the parameters of the population, suggesting that a joint measurement with the stochastic signal might be informative of the population characteristics. By performing parameter estimation on a subset of sources from the catalogues, we conclude that, even if LISA measurements will not be directly informative on the population due to the low number of resolvable sources, it will characterise a few, low-redshift candidates with great precision. Furthermore, we construct for the first time the LISA waterfall plot for low chirp mass, and demonstrate that LISA will also be able to discriminate and characterize, through very precise parameter estimation, a population of binaries with higher masses, $\mathcal{M}_c\sim \mathcal{O}(10^3) M_\odot$, if it exists., Comment: 28 pages, 10 figures, 3 tables

A dedicated gamma-ray burst (GRB) afterglow observing program was performed between 2007 and 2016 with GROND, a seven-channel optical and near-infrared imager at the 2.2m telescope of the Max-Planck Society at ESO/La Silla. In this first of a series of papers, we describe the GRB observing plan, providing first readings of all so far unpublished GRB afterglow measurements and some observing statistics. In total, we observed 514 GRBs with GROND, including 434 Swift-detected GRBs, representing 81\% of the observable Swift sample. For GROND-observations within 30 min of the GRB trigger, the optical/NIR afterglow detection rate is 81\% for long- and 57\% for short-duration GRBs. We report the discovery of ten new GRB afterglows plus one candidate, along with redshift estimates (partly improved) for four GRBs and new host detections for seven GRBs. We identify the (already known) afterglow of GRB 140209A as the sixth GRB exhibiting a 2175 Angstroem dust feature. As a side result, we identified two blazars, with one at a redshift of z=3.8 (in the GRB 131209A field)., Comment: 20 pages, 31 figures, accepted for publication in Astronomy & Astrophysics

A fundamental ingredient in the unified model of active galactic nuclei (AGN) is the obscuring torus, whose innermost, hottest region dominates the near infrared (NIR) emission. Characterising the change in the torus properties and its interplay with the main AGN emission is key for our understanding of AGN physics, evolution and classification. Its covering factor ($CF$) is largely responsible for the classification of AGN on the basis of the detection of broad emission lines. It is still not clear whether the torus properties evolve over time and how they relate with the accretion parameters of the nucleus. In this work, we aim at investigating the evolution of the NIR properties with the redshift ($z$) and the bolometric luminosity ($L_{\rm bol}$) of the AGN. To this end, we assembled a large dataset of $\sim$36,000 Type 1 AGN between $0.5

Many recent studies have shown the ability of large language models (LLMs) to achieve state-of-the-art performance on many NLP tasks, such as question answering, text summarization, coding, and translation. In some cases, the results provided by LLMs are on par with those of human experts. These models' most disruptive innovation is their ability to perform tasks via zero-shot or few-shot prompting. This capability has been successfully exploited to train instructed LLMs, where reinforcement learning with human feedback is used to guide the model to follow the user's requests directly. In this paper, we investigate the ability of instructed LLMs to improve conversational search effectiveness by rewriting user questions in a conversational setting. We study which prompts provide the most informative rewritten utterances that lead to the best retrieval performance. Reproducible experiments are conducted on publicly-available TREC CAST datasets. The results show that rewriting conversational utterances with instructed LLMs achieves significant improvements of up to 25.2% in MRR, 31.7% in Precision@1, 27% in NDCG@3, and 11.5% in Recall@500 over state-of-the-art techniques.

We study the roughening of interfaces in phase-separated active suspensions on substrates. At large length and timescales, we show that the interfacial dynamics belongs to the |q|KPZ universality class discussed in Besse et al. Phys. Rev. Lett. 130, 187102 (2023). This holds despite the presence of long-ranged fluid flows. At early times, instead, the roughening exponents are the same as those in the presence of a momentum-conserving fluid. Surprisingly, when the effect of substrate friction can be ignored, the interface becomes random beyond a de Gennes-Taupin lengthscale which depends on the interfacial tension.

'Bare' active galactic nuclei (AGN) are a subclass of Type 1 AGN that show little or no intrinsic absorption. They offer an unobscured view of the central regions of the AGN and therefore serve as ideal targets to study the relativistic reflection features originating from the innermost regions of the accretion disc. We present a detailed broadband spectral analysis ($0.3 - 70$ keV) of one of the most luminous bare AGN in the local universe, RBS 1124 ($z= 0.208$) using a new, co-ordinated high signal-to-noise observation obtained by $\textit{XMM-Newton}$ and $\textit{NuSTAR}$. The source exhibits a power-law continuum with $\Gamma \sim$ 1.8 along with a soft excess below 2 keV, a weak neutral iron line and curvature at high energies ($\sim 30$ keV). The broadband spectrum, including the soft excess and the high-energy continuum, is well fit by the relativistic reflection model when the accretion disc is allowed to have densities of log$(n_{\rm e}$/cm$^{-3}$) $\gtrsim 19.2$. Our analysis therefore suggests that when high-density effects are considered, relativistic reflection remains a viable explanation for the soft excess., Comment: 14 pages, 9 figures, accepted for publication in MNRAS

Learned dense representations are a popular family of techniques for encoding queries and documents using high-dimensional embeddings, which enable retrieval by performing approximate k nearest-neighbors search (A-kNN). A popular technique for making A-kNN search efficient is based on a two-level index, where the embeddings of documents are clustered offline and, at query processing, a fixed number N of clusters closest to the query is visited exhaustively to compute the result set. In this paper, we build upon state-of-the-art for early exit A-kNN and propose an unsupervised method based on the notion of patience, which can reach competitive effectiveness with large efficiency gains. Moreover, we discuss a cascade approach where we first identify queries that find their nearest neighbor within the closest t << N clusters, and then we decide how many more to visit based on our patience approach or other state-of-the-art strategies. Reproducible experiments employing state-of-the-art dense retrieval models and publicly available resources show that our techniques improve the A-kNN efficiency with up to 5x speedups while achieving negligible effectiveness losses. All the code used is available at https://github.com/francescobusolin/faiss_pEE, Comment: 6 pages, published at CIKM 2024

Learned sparse representations form an effective and interpretable class of embeddings for text retrieval. While exact top-k retrieval over such embeddings faces efficiency challenges, a recent algorithm called Seismic has enabled remarkably fast, highly-accurate approximate retrieval. Seismic statically prunes inverted lists, organizes each list into geometrically-cohesive blocks, and augments each block with a summary vector. At query time, each inverted list associated with a query term is traversed one block at a time in an arbitrary order, with the inner product between the query and summaries determining if a block must be evaluated. When a block is deemed promising, its documents are fully evaluated with a forward index. Seismic is one to two orders of magnitude faster than state-of-the-art inverted index-based solutions and significantly outperforms the winning graph-based submissions to the BigANN 2023 Challenge. In this work, we speed up Seismic further by introducing two innovations to its query processing subroutine. First, we traverse blocks in order of importance, rather than arbitrarily. Second, we take the list of documents retrieved by Seismic and expand it to include the neighbors of each document using an offline k-regular nearest neighbor graph; the expanded list is then ranked to produce the final top-k set. Experiments on two public datasets show that our extension, named SeismicWave, can reach almost-exact accuracy levels and is up to 2.2x faster than Seismic.

In this work, we introduce Gemma 2, a new addition to the Gemma family of lightweight, state-of-the-art open models, ranging in scale from 2 billion to 27 billion parameters. In this new version, we apply several known technical modifications to the Transformer architecture, such as interleaving local-global attentions (Beltagy et al., 2020a) and group-query attention (Ainslie et al., 2023). We also train the 2B and 9B models with knowledge distillation (Hinton et al., 2015) instead of next token prediction. The resulting models deliver the best performance for their size, and even offer competitive alternatives to models that are 2-3 times bigger. We release all our models to the community.

Various scenarios of cosmic inflation enhance the amplitude of the stochastic gravitational wave background (SGWB) at frequencies detectable by the LISA detector. We develop tools for a template-based analysis of the SGWB and introduce a template databank to describe well-motivated signals from inflation, prototype their template-based searches, and forecast their reconstruction with LISA. Specifically, we classify seven templates based on their signal frequency shape, and we identify representative fundamental physics models leading to them. By running a template-based analysis, we forecast the accuracy with which LISA can reconstruct the template parameters of representative benchmark signals, with and without galactic and extragalactic foregrounds. We identify the parameter regions that can be probed by LISA within each template. Finally, we investigate how our signal reconstructions shed light on fundamental physics models of inflation: we discuss their impact for measurements of \emph{e.g.,} ~the couplings of inflationary axions to gauge fields; the graviton mass during inflation; the fluctuation seeds of primordial black holes; the consequences of excited states during inflation, and the presence of small-scale spectral features., Comment: 82 pages, 29 figures; v2 matches version published in JCAP

IRASF11119 is an ultra-luminous IR galaxy with post-merger morphology, hosting a type-1 QSO at z=0.189. Its 2013 Suzaku spectrum shows a prominent Ultra Fast Outflow (UFO) absorption feature (v_out~0.25c). In 2021, we obtained the first XMM-Newton long look of the target, coordinated with a simultaneous NuSTAR observation. The new high-quality data allow us to detect at P>99.8% c.l. multiple absorption features associated with the known UFO. Furthermore, an emission plus absorption feature at 1.1-1.3 keV reveals the presence of a blueshifted P-Cygni profile in the soft band. We associate the hard band features with blends of FeXXV and FeXXVI He$\alpha$-Ly$\alpha$ and He$\beta$-Ly$\beta$ line pairs and infer a large column (N$_H$~$10^{24}$ cm$^{-2}$) of highly ionized (log$\xi$~5) gas outflowing at v_out=0.27c. The 1 keV feature can be associated with a blend of Fe and Ne transitions, produced by a lower column (N$_H$~$10^{21}$ cm$^{-2}$) and ionization (log$\xi$~2.6) gas component outflowing at the same speed. Using a radiative-transfer disk wind model to fit the highly ionized UFO, we derive a large mass outflow rate, comparable with the mass accretion rate (M$_{out}$=4.25 M$_{Sun}$/yr, ~1.6 M$_{acc}$), and kinetic energy and momentum flux among the highest reported in the literature. We measure an extremely low high-energy cut-off (E$_c$~25 keV). Several other cases in the literature suggest that a steep X-ray continuum may be related to the formation of powerful winds. The lack of a significant momentum boost between the nuclear UFO and the different phases of the large-scale outflow, observed in IRASF11119 and in a growing number of sources with powerful UFOs, can be explained by (i) a momentum-driven expansion, (ii) an inefficient coupling of the UFO with the host ISM, or (iii) by repeated energy-driven expansion episodes with low duty-cycle, that average out on long time-scales., Comment: 21 pages, 13 figures, accepted for publication in A&A

Clustering-based nearest neighbor search is an effective method in which points are partitioned into geometric shards to form an index, with only a few shards searched during query processing to find a set of top-$k$ vectors. Even though the search efficacy is heavily influenced by the algorithm that identifies the shards to probe, it has received little attention in the literature. This work bridges that gap by studying routing in clustering-based maximum inner product search. We unpack existing routers and notice the surprising contribution of optimism. We then take a page from the sequential decision making literature and formalize that insight following the principle of ``optimism in the face of uncertainty.'' In particular, we present a framework that incorporates the moments of the distribution of inner products within each shard to estimate the maximum inner product. We then present an instance of our algorithm that uses only the first two moments to reach the same accuracy as state-of-the-art routers such as ScaNN by probing up to $50\%$ fewer points on benchmark datasets. Our algorithm is also space-efficient: we design a sketch of the second moment whose size is independent of the number of points and requires $\mathcal{O}(1)$ vectors per shard.

We present new frontiers in the modelling of the spectral energy distributions (SED) of active galaxies by introducing the radio-to-X-ray fitting capabilities of the publicly available Bayesian code AGNfitter. The new code release, called AGNfitter-rx, models the broad-band photometry covering the radio, infrared (IR), optical, ultraviolet (UV) and X-ray bands consistently, using a combination of theoretical and semi-empirical models of the AGN and host galaxy emission. This framework enables the detailed characterization of four physical components of the active nuclei: the accretion disk, the hot dusty torus, the relativistic jets/core radio emission, and the hot corona; alongside modeling three components within the host galaxy: stellar populations, cold dust, and the radio emission from the star-forming regions. Applying AGNfitter-rx to a diverse sample of 36 AGN SEDs at z<0.7 from the AGN SED ATLAS, we investigate and compare the performance of state-of-the-art torus and accretion disk emission models on fit quality and inferred physical parameters. We find that clumpy torus models that include polar winds and semi-empirical accretion disk templates including emission line features significantly increase the fit quality in 67% of the sources, by effectively reducing by $2\sigma$ fit residuals in the $1.5-5 \mu \rm m$ and $0.7 \mu \rm m$ regimes.We demonstrate that, by applying AGNfitter-rx on photometric data, we are able to estimate inclination and opening angles of the torus, consistent with spectroscopic classifications within the AGN unified model, as well as black hole mass estimates in agreement with virial estimates based on H$\alpha$. The wavelength coverage and the flexibility for the inclusion of state-of-the-art theoretical models make AGNfitter-rx a unique tool for the further development of SED modelling for AGNs in present and future radio-to-X-ray galaxy surveys., Comment: 20 pages, 10 figures, Accepted for publication by A&A

We initiate the LISA template databank for stochastic gravitational wave backgrounds sourced by cosmic strings. We include two templates, an analytical template, which enables more flexible searches, and a numerical template derived directly from large Nambu-Goto simulations of string networks. Using searches based on these templates, we forecast the parameter space within the reach of the experiment and the precision with which their parameters will be reconstructed, provided a signal is observed. The reconstruction permits probing the Hubble expansion and new relativistic DoF in the early universe. We quantify the impact that astrophysical foregrounds can have on these searches. Finally, we discuss the impact that these observations would have on our understanding of the fundamental models behind the string networks. Overall, we prove that LISA has great potential for probing cosmic string models and may reach tensions as low as $G\mu =10^{-16} - 10^{-17} $, which translates into energy scales of the order $10^{11}~\text{GeV}$., Comment: 59 pages, 15 figures

Hyperuniformity emerges generically in the coarsening regime of phase-separating fluids. Numerical studies of active and passive systems have shown that the structure factor $S(q)$ behaves as $q^\varsigma$ for $q\to 0$, with hyperuniformity exponent $\varsigma = 4$. For passive systems, this result was explained in 1991 by a qualitative scaling analysis of Tomita, exploiting isotropy at scales much larger than the coarsening length $\ell$. Here we reconsider and extend Tomita's argument to address cases of active phase separation and of non-constant mobility, again finding $\varsigma=4$. We further show that dynamical noise of variance $D$ creates a transient $\varsigma = 2$ regime for $\hat q\ll \hat{q}_\ast \sim \sqrt{D} t^{[1-(d+2)\nu]/2}$, crossing over to $\varsigma = 4$ at larger $\hat{q}$. Here, $\nu$ is the coarsening exponent, with $\ell\sim t^\nu$, and $\hat{q} \propto q \ell$ is the rescaled wavenumber. In diffusive coarsening, $\nu=1/3$, so the rescaled crossover wavevector $\hat{q}_\ast$ vanishes at large times when $d\geq 2$. The slowness of this decay suggests a natural explanation for experiments that observe a long-lived $\varsigma = 2$ scaling in phase-separating active fluids (where noise is typically large). Conversely, in $d=1$, we demonstrate that with noise the $\varsigma = 2$ regime survives as $t\to\infty$, with $\hat{q}_\ast\sim D^{5/6}$. (The structure factor is not then determined by the zero-temperature fixed point.) We confirm our analytical predictions by numerical simulations of active and passive continuum theories in the deterministic case and of Model B for the stochastic case. We also compare them with related findings for a system near an absorbing-state transition rather than undergoing phase separation. A central role is played throughout by the presence or absence of a conservation law for the centre of mass position of the order parameter field., Comment: 34 pages, 4 figures

Learned sparse representations form an attractive class of contextual embeddings for text retrieval. That is so because they are effective models of relevance and are interpretable by design. Despite their apparent compatibility with inverted indexes, however, retrieval over sparse embeddings remains challenging. That is due to the distributional differences between learned embeddings and term frequency-based lexical models of relevance such as BM25. Recognizing this challenge, a great deal of research has gone into, among other things, designing retrieval algorithms tailored to the properties of learned sparse representations, including approximate retrieval systems. In fact, this task featured prominently in the latest BigANN Challenge at NeurIPS 2023, where approximate algorithms were evaluated on a large benchmark dataset by throughput and recall. In this work, we propose a novel organization of the inverted index that enables fast yet effective approximate retrieval over learned sparse embeddings. Our approach organizes inverted lists into geometrically-cohesive blocks, each equipped with a summary vector. During query processing, we quickly determine if a block must be evaluated using the summaries. As we show experimentally, single-threaded query processing using our method, Seismic, reaches sub-millisecond per-query latency on various sparse embeddings of the MS MARCO dataset while maintaining high recall. Our results indicate that Seismic is one to two orders of magnitude faster than state-of-the-art inverted index-based solutions and further outperforms the winning (graph-based) submissions to the BigANN Challenge by a significant margin.

A critical piece of the modern information retrieval puzzle is approximate nearest neighbor search. Its objective is to return a set of $k$ data points that are closest to a query point, with its accuracy measured by the proportion of exact nearest neighbors captured in the returned set. One popular approach to this question is clustering: The indexing algorithm partitions data points into non-overlapping subsets and represents each partition by a point such as its centroid. The query processing algorithm first identifies the nearest clusters -- a process known as routing -- then performs a nearest neighbor search over those clusters only. In this work, we make a simple observation: The routing function solves a ranking problem. Its quality can therefore be assessed with a ranking metric, making the function amenable to learning-to-rank. Interestingly, ground-truth is often freely available: Given a query distribution in a top-$k$ configuration, the ground-truth is the set of clusters that contain the exact top-$k$ vectors. We develop this insight and apply it to Maximum Inner Product Search (MIPS). As we demonstrate empirically on various datasets, learning a simple linear function consistently improves the accuracy of clustering-based MIPS.

A sample of quasars has been recently assembled to investigate the non-linear relation between their monochromatic luminosities at 2500{\AA}, and 2 keV and to exploit quasars as a new class of standardized candles. The use of this technique for cosmological purposes relies on the non-evolution with redshift of the UV-optical spectral properties of quasars, as well as on the absence of possible contaminants such as dust extinction and host-galaxy contribution. We address these possible issues by analysing the spectral properties of our cosmological quasar sample. We produced composite spectra in different bins of redshift and accretion parameters (black hole mass, bolometric luminosity), to investigate any possible evolution of the spectral properties of the continuum of the composites with these parameters. We found a remarkable similarity amongst the various stacked spectra. The overall shape of the continuum does not show any statistically significant trend with the accretion parameters nor with the redshift. The composite spectrum of our quasar sample is consistent with negligible levels of both intrinsic reddening (with a colour excess E(B-V)< 0.01) and host-galaxy emission (less than 10%) in the optical. We tested whether unaccounted dust extinction could explain the discrepancy between our cosmographic fit of the Hubble-Lemaitre diagram and the concordance {\Lambda}CDM model. The average colour excess required to solve the tension should increase with redshift up to unphysically high values (E(B-V)=0.1 at z>3) that would imply that the intrinsic emission of quasars is much bluer and more luminous than ever reported in observed spectra. The similarity of quasar spectra across the parameter space excludes a significant evolution of the average continuum properties with any of the explored parameters, confirming the reliability of our sample for cosmological applications., Comment: 24 pages, 18 figures, accepted for publication on A&A

Dense retrieval techniques employ pre-trained large language models to build a high-dimensional representation of queries and passages. These representations compute the relevance of a passage w.r.t. to a query using efficient similarity measures. In this line, multi-vector representations show improved effectiveness at the expense of a one-order-of-magnitude increase in memory footprint and query latency by encoding queries and documents on a per-token level. Recently, PLAID has tackled these problems by introducing a centroid-based term representation to reduce the memory impact of multi-vector systems. By exploiting a centroid interaction mechanism, PLAID filters out non-relevant documents, thus reducing the cost of the successive ranking stages. This paper proposes ``Efficient Multi-Vector dense retrieval with Bit vectors'' (EMVB), a novel framework for efficient query processing in multi-vector dense retrieval. First, EMVB employs a highly efficient pre-filtering step of passages using optimized bit vectors. Second, the computation of the centroid interaction happens column-wise, exploiting SIMD instructions, thus reducing its latency. Third, EMVB leverages Product Quantization (PQ) to reduce the memory footprint of storing vector representations while jointly allowing for fast late interaction. Fourth, we introduce a per-document term filtering method that further improves the efficiency of the last step. Experiments on MS MARCO and LoTTE show that EMVB is up to 2.8x faster while reducing the memory footprint by 1.8x with no loss in retrieval accuracy compared to PLAID.

PDS 456 is the most luminous RQQ at z<0.3 and can be regarded as a local counterpart of the powerful QSOs shining at Cosmic Noon. It hosts a strong nuclear X-ray ultra-fast outflow, and a massive and clumpy CO(3-2) molecular outflow extending up to 5 kpc from the nucleus. We analyzed the first MUSE WFM and AO-NFM optical integral field spectroscopic observations of PDS456. The AO-NFM observations provide an unprecedented spatial resolution, reaching up to 280 pc. Our findings reveal a complex circumgalactic medium around PDS 456, extending up to a maximum projected size of ~46 kpc. This includes a reservoir of gas with a mass of ~1e7-1e8 Modot, along with eight companion galaxies, and a multi-phase outflow. WFM and NFM MUSE data reveal an outflow on a large scale (~12 kpc from the quasar) in [OIII], and on smaller scales (within 3 kpc) with higher resolution (about 280 pc) in Halpha, respectively. The [OIII] outflow mass rate is 2.3 +/- 0.2 Modot/yr which is significantly lower than those typically found in other luminous quasars. Remarkably, the Ha outflow shows a similar scale, morphology, and kinematics to the CO(3-2) molecular outflow, with the latter dominating in terms of kinetic energy and mass outflow rate by two and one orders of magnitude, respectively. Our results therefore indicate that mergers, powerful AGN activity, and feedback through AGN-driven winds will collectively contribute to shaping the host galaxy evolution of PDS 456, and likely, that of similar objects at the brightest end of the AGN luminosity function across all redshifts. Moreover, the finding that the momentum boost of the total outflow deviates from the expected energy-conserving expansion for large-scale outflows highlights the need of novel AGN-driven outflow models to comprehensively interpret these phenomena., Comment: 15 pages, 13 figures accepted for publication in Astronomy & Astrophysics

The detection of blue-shifted absorption lines likely associated with ionized Iron K-shell transitions in the X-ray spectra of many Active Galactic Nuclei (AGN) suggests the presence of a highly ionized gas outflowing with mildly relativistic velocities (0.03c-0.6c), named Ultra-Fast Outflow (UFO). Within the SUBWAYS project we characterized these winds starting from a sample of 22 radio-quiet quasars at 0.1 < z < 0.4, and compared the results with similar studies in the literature on samples of 42 local radio-quiet Seyfert galaxies and 14 high redshift radio-quiet quasars. The scope of our work is a statistical study of UFO parameters and incidence, considering key physical properties of the sources, e.g. supermassive black hole (SMBH) mass, bolometric luminosity, accretion rates and Spectral Energy Distribution, with the aim of gaining new insights into the UFO launching mechanisms. We find indications that highly luminous AGN with steeper X-ray/UV ratio, are more likely to host UFO. The presence of UFO is not significantly related to any other AGN property in our sample. These findings suggest that the UFO phenomenon may be transient. Focusing on AGN with UFO, other important results are: (1) faster UFO have larger ionization parameters and column densities; (2) X-ray radiation plays a more crucial role in driving highly ionized winds compared to UV; (3) the correlation between outflow velocity and luminosity is significantly flatter than what expected for radiatively driven winds; (4) more massive BH experience higher wind mass-losses, suppressing accretion of matter onto the BH; (5) the UFO launching radius is positively correlated with the Eddington ratio. Furthermore, our analysis suggest the involvement of multiple launching mechanisms, including radiation pressure and magneto-hydrodynamic processes, rather than pointing to a single, universally applicable mechanism., Comment: 52 pages, 154 figures and 7 tables. Accepted in Astronomy & Astrophysics, in press

We develop a tool for the analysis of stochastic gravitational wave backgrounds from cosmological first-order phase transitions with LISA: we initiate a template databank for these signals, prototype their searches, and forecast their reconstruction. The templates encompass the gravitational wave signals sourced by bubble collisions, sound waves and turbulence. Accounting for Galactic and extra-Galactic foregrounds, we forecast the region of the parameter space that LISA will reconstruct with better than $\sim 10\,\%$ accuracy, if certain experimental and theoretical uncertainties are solved by the time LISA flies. We illustrate the accuracy with which LISA can reconstruct the parameters on a few benchmark signals, both in terms of the template parameters and the phase transition ones. To show the impact of the forecasts on physics beyond the Standard Model, we map the reconstructed benchmark measurements into the parameter spaces of the singlet extension of the Standard Model and of the classically conformal invariant $U(1)_{B-L}$ model., Comment: 54 pages, 18 figures; v2 matches version published in JCAP

The Laser Interferometer Space Antenna (LISA) is the first scientific endeavour to detect and study gravitational waves from space. LISA will survey the sky for Gravitational Waves in the 0.1 mHz to 1 Hz frequency band which will enable the study of a vast number of objects ranging from Galactic binaries and stellar mass black holes in the Milky Way, to distant massive black-hole mergers and the expansion of the Universe. This definition study report, or Red Book, presents a summary of the very large body of work that has been undertaken on the LISA mission over the LISA definition phase., Comment: 155 pages, with executive summary and table of contents

The nature of the transition to collective motion in assemblies of aligning self-propelled particles remains a long-standing matter of debate. In this article, we focus on dry active matter and show that weak fluctuations suffice to generically turn second-order mean-field transitions into a `discontinuous' coexistence scenario. Our theory shows how fluctuations induce a density-dependence of the polar-field mass, even when this effect is absent at mean-field level. In turn, this dependency on density triggers a feedback loop between ordering and advection that ultimately leads to an inhomogeneous transition to collective motion and the emergence of inhomogeneous travelling bands. Importantly, we show that such a fluctuation-induced first order transition is present in both metric models, in which particles align with neighbors within a finite distance, and in `topological' ones, in which alignment is based on more complex constructions of neighbor sets. We compute analytically the noise-induced renormalization of the polar-field mass using stochastic calculus, which we further back up by a one-loop field-theoretical analysis. Finally, we confirm our analytical predictions by numerical simulations of fluctuating hydrodynamics as well as of topological particle models with either k-nearest neighbors or Voronoi alignment., Comment: 61 pages, 9 figures

A characteristic feature of quasars is the observed non-linear relationship between their monochromatic luminosities at rest-frame 2500 {\AA} and 2 keV. This relationship is evident across all redshifts and luminosities and, due to its non-linearity, can be implemented to estimate quasar distances and construct a Hubble Diagram for quasars. Historically, a significant challenge in the cosmological application of this relation has been its high observed dispersion. Recent studies have demonstrated that this dispersion can be reduced by excluding biased objects from the sample. Nevertheless, the dispersion remains considerable ($\delta \sim 0.20$ dex), especially when compared to the Phillips relation for supernovae Ia. Given the absence of a comprehensive physical model for the relation, it remains unclear how much of the remaining dispersion is intrinsically tied to the relation itself and how much can be attributed to observational factors not addressed by the sample selection and by the choice of X-ray and UV indicators. Potential contributing factors include (i) the scatter produced by using X-ray photometric results instead of spectroscopic ones, (ii) the intrinsic variability of quasars, and (iii) the inclination of the accretion disc relative to our line of sight. In this study, we thoroughly examine these three factors and quantify their individual contributions to the observed dispersion. Based on our findings, we argue that the intrinsic dispersion of the X-ray/UV luminosity relation is likely below 0.06 dex. We also discuss why high-redshift subsamples can show a significantly lower dispersion than the average one., Comment: submitted

We analyze the detection prospects for potential Primordial Black Hole Binary (PBHB) populations buried in the Stellar-Origin Black Hole Binary (SOBHB) population inferred by the LVK collaboration. We consider different PBHB population scenarios and several future Gravitational Wave (GW) detectors. To separate the PBHB component from the SOBHB one, we exploit the prediction that the PBHB merger rate does not decline as fast as the SOBHB one at high redshift. However, only a tiny fraction of PBHB events may be resolved individually, and the sub-threshold events may yield an undetectable Stochastic GW Background (SGWB). For this reason, we determine the statistical significance of the PBHB contributions in the number of resolvable events seen in future Earth-based detectors and the SGWB measured at LISA. We find that the synergy between these probes will consistently help assess whether or not a sizeable PBHB population is present., Comment: 31 pages, 8 figures

A fundamental goal of modern-day astrophysics is to understand the connection between supermassive black hole (SMBH) growth and galaxy evolution. Merging galaxies offer one of the most dramatic channels for galaxy evolution known, capable of driving inflows of gas into galactic nuclei, potentially fueling both star formation and central SMBH activity. Dual active galactic nuclei (dual AGNs) in late-stage mergers with nuclear pair separations $<10$ kpc are thus ideal candidates to study SMBH growth along the merger sequence since they coincide with the most transformative period for galaxies. However, dual AGNs can be extremely difficult to confirm and study. Hard X-ray ($>10$ keV) studies offer a relatively contamination-free tool for probing the dense obscuring environments predicted to surround the majority of dual AGN in late-stage mergers. To date, only a handful of the brightest and closest systems have been studied at these energies due to the demanding instrumental requirements involved. We demonstrate the unique capabilities of HEX-P to spatially resolve the soft and - for the first time - hard X-ray counterparts of closely-separated ($\sim2''-5''$) dual AGNs in the local Universe. By incorporating state-of-the-art physical torus models, we reproduce realistic broadband X-ray spectra expected for deeply embedded accreting SMBHs. Hard X-ray spatially resolved observations of dual AGNs - accessible only to HEX-P - will hence transform our understanding of dual AGN in the nearby Universe., Comment: Under review for Frontiers in Astronomy and Space Sciences

The hard X-ray emission in active galactic nuclei (AGN) and black hole X-ray binaries is thought to be produced by a hot cloud of electrons referred to as the corona. This emission, commonly described by a power law with a high-energy cutoff, is suggestive of Comptonization by thermal electrons. While several hypotheses have been proposed to explain the origin, geometry, and composition of the corona, we still lack a clear understanding of this fundamental component. NuSTAR has been playing a key role improving our knowledge of X-ray coronae thanks to its unprecedented sensitivity above 10 keV. However, these constraints are limited to bright, nearby sources. The High Energy X-ray Probe (HEX-P) is a probe-class mission concept combining high spatial resolution X-ray imaging and broad spectral coverage (0.2-80 keV) with a sensitivity superior to current facilities. In this paper, we highlight the major role that HEX-P will play in further advancing our insights of X-ray coronae, notably in AGN. We demonstrate how HEX-P will measure key properties and track the temporal evolution of coronae in unobscured AGN. This will allow us to determine their electron distribution and test the dominant emission mechanisms. Furthermore, we show how HEX-P will accurately estimate the coronal properties of obscured AGN in the local Universe, helping address fundamental questions about AGN unification. In addition, HEX-P will characterize coronae in a large sample of luminous quasars at cosmological redshifts for the first time and track the evolution of coronae in transient systems in real time. We also demonstrate how HEX-P will enable estimating the coronal geometry using spectral-timing techniques. HEX-P will thus be essential to understand the evolution and growth of black holes over a broad range of mass, distance, and luminosity, and will help uncover the black holes' role in shaping the Universe., Comment: Submitted to Frontiers in Astronomy and Space Sciences after reviewers reports

Ever since the discovery of the first Active Galactic Nuclei (AGN), substantial observational and theoretical effort has been invested into understanding how massive black holes have evolved across cosmic time. Circum-nuclear obscuration is now established as a crucial component, with almost every AGN observed known to display signatures of some level of obscuration in their X-ray spectra. But despite more than six decades of effort, substantial open questions remain: How does the accretion power impact the structure of the circum-nuclear obscurer? What are the dynamical properties of the obscurer? Can dense circum-nuclear obscuration exist around intrinsically weak AGN? How many intermediate mass black holes occupy the centers of dwarf galaxies? In this paper, we showcase a number of next-generation prospects attainable with the High Energy X-ray Probe (https://hexp.org) to contribute towards solving these questions in the 2030s. The uniquely broad (0.2--80 keV) and strictly simultaneous X-ray passband of HEX-P makes it ideally suited for studying the temporal co-evolution between the central engine and circum-nuclear obscurer. Improved sensitivities and reduced background will enable the development of spectroscopic models complemented by current and future multi-wavelength observations. We show that the angular resolution of HEX-P both below and above 10 keV will enable the discovery and confirmation of accreting massive black holes at both low accretion power and low black hole masses even when concealed by thick obscuration. In combination with other next-generation observations of the dusty hearts of nearby galaxies, HEX-P will hence be pivotal in paving the way towards a complete picture of black hole growth and galaxy co-evolution., Comment: To be submitted for peer review to a HEX-P Collection in Frontiers of Astronomy and Space Science. Constructive comments welcome!

Since the discovery of the cosmic X-ray background (CXB), astronomers have strived to understand the accreting supermassive black holes (SMBHs) contributing to its peak in the 10-40 keV band. Existing soft X-ray telescopes could study this population up to only 10 keV, and, while NuSTAR (focusing on 3--24 keV) made great progress, it also left significant uncertainties in characterizing the hard X-ray population, crucial for calibrating current population synthesis models. This paper presents an in-depth analysis of simulations of two extragalactic surveys (deep and wide) with the High-Energy X-ray Probe (HEX-P), each observed for 2 Ms. Applying established source detection techniques, we show that HEX-P surveys will reach a flux of $\sim$10$^{-15}$ erg s$^{-1}$ cm$^{-2}$ in the 10-40 keV band, an order of magnitude fainter than current NuSTAR surveys. With the large sample of new hard X-ray detected sources ($\sim2000$), we showcase HEX-P's ability to resolve more than 80% of the CXB up to 40 keV into individual sources. The expected precision of HEX-P's resolved background measurement will allow us to distinguish between population synthesis models of SMBH growth. HEX-P leverages accurate broadband (0.5-40 keV) spectral analysis and the combination of soft and hard X-ray colors to provide obscuration constraints even for the fainter sources, with the overall objective of measuring the Compton-thick fraction. With unprecedented sensitivity in the 10--40 keV band, HEX-P will explore the hard X-ray emission from AGN to flux limits never reached before, thus expanding the parameter space for serendipitous discoveries. Consequently, it is plausible that new models will be needed to capture the population HEX-P will unveil., Comment: 28 pages, 13 Figures, 2 tables; submitted to Frontiers in Astronomy and Space Sciences; minor edits in the text, abstract and figure labels

Constraining the primary growth channel of supermassive black holes (SMBH) remains one the most actively debated questions in the context of cosmological structure formation. Owing to the expected connection between SMBH spin parameter evolution and the accretion and merger history of individual black holes, population spin measurements offer a rare observational window into the SMBH cosmic growth. As of today, the most common method for estimating SMBH spin relies on modeling the relativistically broaden atomic profiles in the reflection spectrum observed in X-rays. In this paper, we study the observational requirements needed to confidently distinguish between the primary SMBH growth channels, based on their distinct spin-mass distributions predicted by the Horizon-AGN cosmological simulation. In doing so, we characterize outstanding limitations associated with the existing measurements and discuss the landscape of future observational campaigns, which can be planned and executed with future X-ray observatories. We focus our attention on the High-Energy X-ray Probe (HEX-P), a concept probe-class mission aimed to serve the high-energy community in the 2030s., Comment: 26 pages, 11 figures, submitted for publication in Frontiers in Astronomy and Space Sciences

Collective migration is an important component of many biological processes, including wound healing, tumorigenesis, and embryo development. Spatial agent-based models (ABMs) are often used to model collective migration, but it is challenging to thoroughly predict these models' behavior throughout parameter space due to their random and computationally intensive nature. Modelers often coarse-grain ABM rules into mean-field differential equation (DE) models. While these DE models are fast to simulate, they suffer from poor (or even ill-posed) ABM predictions in some regions of parameter space. In this work, we describe how biologically-informed neural networks (BINNs) can be trained to learn interpretable BINN-guided DE models capable of accurately predicting ABM behavior. In particular, we show that BINN-guided partial DE (PDE) simulations can 1.) forecast future spatial ABM data not seen during model training, and 2.) predict ABM data at previously-unexplored parameter values. This latter task is achieved by combining BINN-guided PDE simulations with multivariate interpolation. We demonstrate our approach using three case study ABMs of collective migration that imitate cell biology experiments and find that BINN-guided PDEs accurately forecast and predict ABM data with a one-compartment PDE when the mean-field PDE is ill-posed or requires two compartments. This work suggests that BINN-guided PDEs allow modelers to efficiently explore parameter space, which may enable data-driven tasks for ABMs, such as estimating parameters from experimental data. All code and data from our study is available at https://github.com/johnnardini/Forecasting_predicting_ABMs.

LISA, the Laser Interferometer Space Antenna, will usher in a new era in gravitational-wave astronomy. As the first anticipated space-based gravitational-wave detector, it will expand our view to the millihertz gravitational-wave sky, where a spectacular variety of interesting new sources abound: from millions of ultra-compact binaries in our Galaxy, to mergers of massive black holes at cosmological distances; from the beginnings of inspirals that will venture into the ground-based detectors' view to the death spiral of compact objects into massive black holes, and many sources in between. Central to realising LISA's discovery potential are waveform models, the theoretical and phenomenological predictions of the pattern of gravitational waves that these sources emit. This white paper is presented on behalf of the Waveform Working Group for the LISA Consortium. It provides a review of the current state of waveform models for LISA sources, and describes the significant challenges that must yet be overcome., Comment: 239 pages, 11 figures, white paper from the LISA Consortium Waveform Working Group, invited for submission to Living Reviews in Relativity, updated with comments from community