BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Yi Mao

Yi Mao appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
astro-ph.COastro-ph.GAComputation and Languageastro-ph.IMMachine LearningArtificial Intelligencehep-phastro-phComputer Visiongr-qchep-thHuman-Computer InteractionInformation Retrieval

Catalog footprint

What is connected

34works
13topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

34 published item(s)

preprint2022arXiv

A Token-level Reference-free Hallucination Detection Benchmark for Free-form Text Generation

Large pretrained generative models like GPT-3 often suffer from hallucinating non-existent or incorrect content, which undermines their potential merits in real applications. Existing work usually attempts to detect these hallucinations based on a corresponding oracle reference at a sentence or document level. However ground-truth references may not be readily available for many free-form text generation applications, and sentence- or document-level detection may fail to provide the fine-grained signals that would prevent fallacious content in real time. As a first step to addressing these issues, we propose a novel token-level, reference-free hallucination detection task and an associated annotated dataset named HaDes (HAllucination DEtection dataSet). To create this dataset, we first perturb a large number of text segments extracted from English language Wikipedia, and then verify these with crowd-sourced annotations. To mitigate label imbalance during annotation, we utilize an iterative model-in-loop strategy. We conduct comprehensive data analyses and create multiple baseline models.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

An End-to-End Dialogue Summarization System for Sales Calls

Summarizing sales calls is a routine task performed manually by salespeople. We present a production system which combines generative models fine-tuned for customer-agent setting, with a human-in-the-loop user experience for an interactive summary curation process. We address challenging aspects of dialogue summarization task in a real-world setting including long input dialogues, content validation, lack of labeled data and quality evaluation. We show how GPT-3 can be leveraged as an offline data labeler to handle training data scarcity and accommodate privacy constraints in an industrial setting. Experiments show significant improvements by our models in tackling the summarization and content validation tasks on public datasets.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Estimation of HII Bubble Size Distribution from 21cm Power Spectrum with Artificial Neural Networks

The bubble size distribution of ionized hydrogen regions probes the information about the morphology of \HII\ bubbles during the reionization. Conventionally, the \HII\ bubble size distribution can be derived from the tomographic imaging data of the redshifted 21~cm signal from the epoch of reionization, which, however, is observationally challenging even for the upcoming large radio interferometer arrays. Given that these interferometers promise to measure the 21~cm power spectrum accurately, we propose a new method, which is based on the artificial neural networks (ANN), to reconstruct the \HII\ bubble size distribution from the 21~cm power spectrum. We demonstrate that the reconstruction from the 21~cm power spectrum can be almost as accurate as directly measured from the imaging data with the fractional error $\lesssim 10\%$, even with thermal noise at the sensitivity level of the Square Kilometre Array. Nevertheless, the reconstruction implicitly exploits the modelling in reionization simulations, and hence the recovered \HII\ bubble size distribution is not an independent summary statistic from the power spectrum, and should be used only as the indicator for understanding \HII\ bubble morphology and its evolution.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Implicit Likelihood Inference of Reionization Parameters from the 21 cm Power Spectrum

The first measurements of the 21 cm brightness temperature power spectrum from the epoch of reionization will very likely be achieved in the near future by radio interferometric array experiments such as the Hydrogen Epoch of Reionization Array (HERA) and the Square Kilometre Array (SKA). Standard MCMC analyses use an explicit likelihood approximation to infer the reionization parameters from the 21 cm power spectrum. In this paper, we present a new Bayesian inference of the reionization parameters where the likelihood is implicitly defined through forward simulations using density estimation likelihood-free inference (DELFI). Realistic effects including thermal noise and foreground avoidance are also applied to the mock observations from the HERA and SKA. We demonstrate that this method recovers accurate posterior distributions for the reionization parameters, and outperforms the standard MCMC analysis in terms of the location and size of credible parameter regions. With the minutes-level processing time once the network is trained, this technique is a promising approach for the scientific interpretation of future 21 cm power spectrum observation data. Our code 21cmDELFI-PS is publicly available at this link.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Knowledge-Grounded Dialogue Generation with a Unified Knowledge Representation

Knowledge-grounded dialogue systems are challenging to build due to the lack of training data and heterogeneous knowledge sources. Existing systems perform poorly on unseen topics due to limited topics covered in the training data. In addition, heterogeneous knowledge sources make it challenging for systems to generalize to other tasks because knowledge sources in different knowledge representations require different knowledge encoders. To address these challenges, we present PLUG, a language model that homogenizes different knowledge sources to a unified knowledge representation for knowledge-grounded dialogue generation tasks. PLUG is pre-trained on a dialogue generation task conditioned on a unified essential knowledge representation. It can generalize to different downstream knowledge-grounded dialogue generation tasks with a few training examples. The empirical evaluation on two benchmarks shows that our model generalizes well across different knowledge-grounded tasks. It can achieve comparable performance with state-of-the-art methods under a fully-supervised setting and significantly outperforms other methods in zero-shot and few-shot settings.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

OmniTab: Pretraining with Natural and Synthetic Data for Few-shot Table-based Question Answering

The information in tables can be an important complement to text, making table-based question answering (QA) systems of great value. The intrinsic complexity of handling tables often adds an extra burden to both model design and data annotation. In this paper, we aim to develop a simple table-based QA model with minimal annotation effort. Motivated by the fact that table-based QA requires both alignment between questions and tables and the ability to perform complicated reasoning over multiple table elements, we propose an omnivorous pretraining approach that consumes both natural and synthetic data to endow models with these respective abilities. Specifically, given freely available tables, we leverage retrieval to pair them with relevant natural sentences for mask-based pretraining, and synthesize NL questions by converting SQL sampled from tables for pretraining with a QA loss. We perform extensive experiments in both few-shot and full settings, and the results clearly demonstrate the superiority of our model OmniTab, with the best multitasking approach achieving an absolute gain of 16.2% and 2.7% in 128-shot and full settings respectively, also establishing a new state-of-the-art on WikiTableQuestions. Detailed ablations and analyses reveal different characteristics of natural and synthetic data, shedding light on future directions in omnivorous pretraining. Code, pretraining data, and pretrained models are available at https://github.com/jzbjyb/OmniTab.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Theoretical Models of the Atomic Hydrogen Content in Dark Matter Halos

Atomic hydrogen (H I) gas, mostly residing in dark matter halos after cosmic reionization, is the fuel for star formation. Its relation with properties of host halo is the key to understand the cosmic H I distribution. In this work, we propose a flexible, empirical model of H I-halo relation. In this model, while the H I mass depends primarily on the mass of host halo, there is also secondary dependence on other halo properties. We apply our model to the observation data of the Arecibo Fast Legacy ALFA Survey (ALFALFA), and find it can successfully fit to the cosmic H I abundance ($Ω_{\rm HI}$), average H I-halo mass relation $\langle M_{\rm HI}|M_{\rm h}\rangle$, and the H I clustering. The bestfit of the ALFALFA data rejects with high confidence level the model with no secondary halo dependence of H I mass and the model with secondary dependence on halo spin parameter ($λ$), and shows strong dependence on halo formation time ($a_{1/2}$) and halo concentration ($c_{\rm vir}$). In attempt to explain these findings from the perspective of hydrodynamical simulations, the IllustrisTNG simulation confirms the dependence of H I mass on secondary halo parameters. However, the IllustrisTNG results show strong dependence on $λ$ and weak dependence on $c_{\rm vir}$ and $a_{1/2}$, and also predict a much larger value of H I clustering on large scales than observations. This discrepancy between the simulation and observation calls for improvements in understanding the H I-halo relation from both theoretical and observational sides.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Antisymmetric Cross-correlation between H I and CO Line Intensity Maps as a New Probe of Cosmic Reionization

Intensity mapping of the H I 21 cm line and the CO 2.61 mm line from the epoch of reionization has emerged as powerful, complementary, probes of the high-redshift Universe. However, both maps and their cross-correlation are dominated by foregrounds. We propose a new analysis by which the signal is unbiased by foregrounds, i.e. it can be measured without foreground mitigation. We construct the antisymmetric part of two-point cross-correlation between intensity maps of the H I 21 cm line and the CO 2.61 mm line, arising because the statistical fluctuations of two fields have different evolution in time. We show that the sign of this new signal can distinguish model-independently whether inside-out reionization happens during some interval of time. More importantly, within the framework of the excursion set model of reionization, we demonstrate that the slope of the dipole of H I-CO cross-power spectrum at large scales is linear to the rate of change of global neutral fraction of hydrogen in a manner independent of reionization parameters, until the slope levels out near the end of reionization, but this trend might possibly depend on the framework of reionization modelling. The H I-CO dipole may be a smoking-gun probe for the speed of reionization, or "standard speedometer" for cosmic reionization. Observations of this new signal will unveil the global reionization history from the midpoint to near the completion of reionization.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Robust Intensity Mapping Analysis against Foregrounds for the Epoch of Reionization

Intensity mapping of the HI 21 cm line and the CO 2.61 mm line from the epoch of reionization has emerged as powerful, complementary, probes of the high-redshift Universe. However, both maps and their cross-correlation are dominated by foregrounds. We propose a new analysis by which the signal is unbiased by foregrounds, i.e. it can be measured without foreground mitigation. We construct the antisymmetric part of the HI-CO cross-correlation, arising because the statistical fluctuations of two fields have different evolution in time. We show that the sign of this new signal can distinguish model-independently whether inside-out reionization happens during some interval of time.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Simulation-Based Inference of Reionization Parameters From 3D Tomographic 21 cm Lightcone Images

Tomographic three-dimensional 21 cm images from the epoch of reionization contain a wealth of information about the reionization of the intergalactic medium by astrophysical sources. Conventional power spectrum analysis cannot exploit the full information in the 21 cm data because the 21 cm signal is highly non-Gaussian due to reionization patchiness. We perform a Bayesian inference of the reionization parameters where the likelihood is implicitly defined through forward simulations using density estimation likelihood-free inference (DELFI). We adopt a trained 3D Convolutional Neural Network (CNN) to compress the 3D image data into informative summaries (DELFI-3D CNN). We show that this method recovers accurate posterior distributions for the reionization parameters. Our approach outperforms earlier analysis based on two-dimensional 21 cm images. In contrast, an MCMC analysis of the 3D lightcone-based 21 cm power spectrum alone and using a standard explicit likelihood approximation results in less accurate credible parameter regions than inferred by the DELFI-3D CNN, both in terms of the location and shape of the contours. Our proof-of-concept study implies that the DELFI-3D CNN can effectively exploit more information in the 3D 21 cm images than a 2D CNN or power spectrum analysis. This technique can be readily extended to include realistic effects and is therefore a promising approach for the scientific interpretation of future 21 cm observation data.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Conditional Self-Attention for Query-based Summarization

Self-attention mechanisms have achieved great success on a variety of NLP tasks due to its flexibility of capturing dependency between arbitrary positions in a sequence. For problems such as query-based summarization (Qsumm) and knowledge graph reasoning where each input sequence is associated with an extra query, explicitly modeling such conditional contextual dependencies can lead to a more accurate solution, which however cannot be captured by existing self-attention mechanisms. In this paper, we propose \textit{conditional self-attention} (CSA), a neural network module designed for conditional dependency modeling. CSA works by adjusting the pairwise attention between input tokens in a self-attention module with the matching score of the inputs to the given query. Thereby, the contextual dependencies modeled by CSA will be highly relevant to the query. We further studied variants of CSA defined by different types of attention. Experiments on Debatepedia and HotpotQA benchmark datasets show CSA consistently outperforms vanilla Transformer and previous models for the Qsumm problem.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Exploiting Structured Knowledge in Text via Graph-Guided Representation Learning

In this work, we aim at equipping pre-trained language models with structured knowledge. We present two self-supervised tasks learning over raw text with the guidance from knowledge graphs. Building upon entity-level masked language models, our first contribution is an entity masking scheme that exploits relational knowledge underlying the text. This is fulfilled by using a linked knowledge graph to select informative entities and then masking their mentions. In addition we use knowledge graphs to obtain distractors for the masked entities, and propose a novel distractor-suppressed ranking objective which is optimized jointly with masked language model. In contrast to existing paradigms, our approach uses knowledge graphs implicitly, only during pre-training, to inject language models with structured knowledge via learning from raw text. It is more efficient than retrieval-based methods that perform entity linking and integration during finetuning and inference, and generalizes more effectively than the methods that directly learn from concatenated graph triples. Experiments show that our proposed model achieves improved performance on five benchmark datasets, including question answering and knowledge base completion tasks.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Ly$α$ forest power spectrum as an emerging window into the epoch of reionization and cosmic dawn

Conventional wisdom was that thermal relics from the epoch of reionization (EOR) would vanish swiftly. Recently, however, it was shown that these relics can survive to lower redshifts ($z \sim 2$) than previously thought, due to gas at mean density being heated to $T \sim 3 \times 10^4$ K by reionization, which is inhomogeneous, and shocks. Given the high sensitivities of upcoming Ly$α$ forest surveys, this effect will be a novel broadband systematic for cosmological application. From the astrophysical point of view, however, the imprint of inhomogeneous reionization can shed light on the EOR and cosmic dawn. We utilize a hybrid method -- which includes two different simulation codes capable of handling the huge dynamical range -- to show the impact of patchy reionization on the Ly$α$ forest and its dependence on different astrophysical scenarios. We found statistically significant deviations in the 1D Ly$α$ power spectrum at $k = 0.14$ cMpc$^{-1}$ that range from $\sim 1\%$ at $z = 2$ up to almost $\sim 20\%$ at $z = 4$. The deviations in the 3D Ly$α$ power spectrum, at the same wavenumber, are large and range from a few per cent at $z = 2$ up to $\sim 50\%$ at $z = 4$, although these deviations ignore the effect of He II reionization and AGN feedback at $z<4$. By exploiting different $k$-dependence of power spectrum among various astrophysical scenarios, the effect of patchy reionization on the Ly$α$ forest power spectrum can open a new window into cosmic reionization and possibly cosmic dawn.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

The Breakdown Scale of HI Bias Linearity

The 21 cm intensity mapping experiments promise to obtain the large-scale distribution of HI gas at the post-reionization epoch. In order to reveal the underlying matter density fluctuations from the HI mapping, it is important to understand how HI gas traces the matter density distribution. Both nonlinear halo clustering and nonlinear effects modulating HI gas in halos may determine the scale below which the HI bias deviates from linearity. We employ three approaches to generate the mock HI density from a large-scale N-body simulation at low redshifts, and demonstrate that the assumption of HI linearity is valid at the scale corresponding to the first peak of baryon acoustic oscillations, but breaks down at $k \gtrsim 0.1\,h\, {\rm Mpc}^{-1}$. The nonlinear effects of halo clustering and HI content modulation counteract each other at small scales, and their competition results in a model-dependent "sweet-spot" redshift near $z$=1 where the HI bias is scale-independent down to small scales. We also find that the linear HI bias scales approximately linearly with redshift for $z\le 3$.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Testing the scale-dependent hemispherical asymmetry with the 21-cm power spectrum from the epoch of reionization

Hemispherical power asymmetry has emerged as a new challenge to cosmology in early universe. While the cosmic microwave background (CMB) measurements indicated the asymmetry amplitude $A \simeq 0.07$ at the CMB scale $k_{\rm CMB}\simeq 0.0045\,{\rm Mpc}^{-1}$, the high-redshift quasar observations found no significant deviation from statistical isotropy. This conflict can be reconciled in some scale-dependent asymmetry models. We put forward a new parameterization of scale-dependent asymmetric power spectrum, inspired by a multi-speed inflation model. The 21-cm power spectrum from the epoch of reionization can be used to constrain the scale-dependent hemispherical asymmetry. We demonstrate that an optimum, multi-frequency observation by the Square Kilometre Array (SKA) Phase 2 can impose a constraint on the amplitude of the power asymmetry anomaly at the level of $ΔA \simeq 0.2$ at $0.056 \lesssim k_{\rm 21cm} \lesssim 0.15 \,{\rm Mpc}^{-1}$. This limit may be further improved by an order of magnitude as $ΔA \simeq 0.01$ with a cosmic variance limited experiment such as the Omniscope.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

The impact of inhomogeneous subgrid clumping on cosmic reionization

Cosmic reionization was driven by the imbalance between early sources and sinks of ionizing radiation, both of which were dominated by small-scale structure and are thus usually treated in cosmological reionization simulations by subgrid modelling. The recombination rate of intergalactic hydrogen is customarily boosted by a subgrid clumping factor, ${\left<n^2\right>/\left<n\right>^2}$, which corrects for unresolved fluctuations in gas density ${n}$ on scales below the grid-spacing of coarse-grained simulations. We investigate in detail the impact of this inhomogeneous subgrid clumping on reionization and its observables, as follows: (1) Previous attempts generally underestimated the clumping factor because of insufficient mass resolution. We perform a high-resolution $N$-body simulation that resolves haloes down to the pre-reionization Jeans mass to derive the time-dependent, spatially-varying local clumping factor and a fitting formula for its correlation with local overdensity. (2) We then perform a large-scale $N$-body and radiative transfer simulation that accounts for this inhomogeneous subgrid clumping by applying this clumping factor-overdensity correlation. Boosting recombination significantly slows the expansion of ionized regions, which delays completion of reionization and suppresses 21 cm power spectra on large scales in the later stages of reionization. (3) We also consider a simplified prescription in which the globally-averaged, time-evolving clumping factor from the same high-resolution $N$-body simulation is applied uniformly to all cells in the reionization simulation, instead. Observables computed with this model agree fairly well with those from the inhomogeneous clumping model, e.g. predicting 21 cm power spectra to within 20% error, suggesting it may be a useful approximation.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

The Impact of Nonlinear Structure Formation on the Power Spectrum of Transverse Momentum Fluctuations and the Kinetic Sunyaev-Zel'dovich Effect

Cosmological transverse momentum fields, whose directions are perpendicular to Fourier wave vectors, induce temperature anisotropies in the cosmic microwave background via the kinetic Sunyaev-Zeldovich (kSZ) effect. The transverse momentum power spectrum contains the four-point function of density and velocity fields, $\langleδδv v\rangle$. In the post-reionization epoch, nonlinear effects dominate in the power spectrum. We use perturbation theory and cosmological $N$-body simulations to calculate this nonlinearity. We derive the next-to-leading order expression for the power spectrum with a particular emphasis on the connected term that has been ignored in the literature. While the contribution from the connected term on small scales ($k>0.1\,h\,\rm{Mpc}^{-1}$) is subdominant relative to the unconnected term, we find that its contribution to the kSZ power spectrum at $\ell = 3000$ at $z<6$ can be as large as ten percent of the unconnected term, which would reduce the allowed contribution from the reionization epoch ($z>6$) by twenty percent. The power spectrum of transverse momentum on large scales is expected to scale as $k^2$ as a consequence of momentum conservation. We show that both the leading and the next-to-leading order terms satisfy this scaling. In particular, we find that both of the unconnected and connected terms are necessary to reproduce $k^2$.

0 citations0 reviewsNo code linkedOpen access
preprint2015arXiv

The Linear Perturbation Theory of Reionization in Position-Space: Cosmological Radiative Transfer Along the Light-Cone

The linear perturbation theory of inhomogeneous reionization (LPTR) has been developed as an analytical tool for predicting the global ionized fraction and large-scale power spectrum of ionized density fluctuations during reionization. In the original formulation of the LPTR, the ionization balance and radiative transfer equations are linearized and solved in Fourier space. However, the LPTR's approximation to the full solution of the radiative transfer equation is not straightforward to interpret, since the latter is most intuitively conceptualized in position space. To bridge the gap between the LPTR and the language of numerical radiative transfer, we present a new, equivalent, position-space formulation of the LPTR that clarifies the approximations it makes and facilitates its interpretation. We offer a comparison between the LPTR and the excursion-set model of reionization (ESMR), and demonstrate the built-in capability of the LPTR to explore a wide range of reionization scenarios, and to go beyond the ESMR in exploring scenarios involving X-rays.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Cosmologically probing ultra-light particle dark matter using 21 cm signals

There can arise ubiquitous ultra-light scalar fields in the Universe, such as the pseudo-Goldstone bosons from the spontaneous breaking of an approximate symmetry, which can make a partial contribution to the dark matter and affect the large scale structure of the Universe. While the properties of those ultra-light dark matter are heavily model dependent and can vary in a wide range, we develop a model-independent analysis to forecast the constraints on their mass and abundance using futuristic but realistic 21 cm observables as well as CMB fluctuations, including CMB lensing measurements. Avoiding the highly nonlinear regime, the 21 cm emission line spectra are most sensitive to the ultra-light dark matter with mass m ~10^{-26} eV for which the precision attainable on mass and abundance bounds can be of order of a few percent.

0 citations0 reviewsNo code linkedOpen access
preprint2014arXiv

Light cone effect on the reionization 21-cm signal II: Evolution, anisotropies and observational implications

Measurements of the HI 21-cm power spectra from the reionization epoch will be influenced by the evolution of the signal along the line-of-sight direction of any observed volume. We use numerical as well as semi-numerical simulations of reionization in a cubic volume of 607 Mpc across to study this so-called light cone effect on the HI 21-cm power spectrum. We find that the light cone effect has the largest impact at two different stages of reionization: one when reionization is $\sim 20\%$ and other when it is $\sim 80\%$ completed. We find a factor of $\sim 4$ amplification of the power spectrum at the largest scale available in our simulations. We do not find any significant anisotropy in the 21-cm power spectrum due to the light cone effect. We argue that for the power spectrum to become anisotropic, the light cone effect would have to make the ionized bubbles significantly elongated or compressed along the line-of-sight, which would require extreme reionization scenarios. We also calculate the two-point correlation functions parallel and perpendicular to the line-of-sight and find them to differ. Finally, we calculate an optimum frequency bandwidth below which the light cone effect can be neglected when extracting power spectra from observations. We find that if one is willing to accept a $10 \%$ error due to the light cone effect, the optimum frequency bandwidth for $k= 0.056 \, \rm{Mpc}^{-1}$ is $\sim 7.5$ MHz. For $k = 0.15$ and $0.41 \, \rm{Mpc}^{-1}$ the optimum bandwidth is $\sim 11$ and $\sim 16$ MHz respectively.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Primordial Non-Gaussianity Estimation using 21 cm Tomography from the Epoch of Reionization

Measuring the small primordial nonGaussianity (PNG) predicted by cosmic inflation theories may help diagnose them. The detectability of PNG by its imprint on the 21cm power spectrum from the epoch of reionization is reassessed here in terms of $f_{NL}$, the local nonlinearity parameter. We find that an optimum, multi-frequency observation by SKA can achieve $Δf_{NL} \sim 3$ (comparable to recent Planck CMB limits), while a cosmic-variance-limited array of this size like Omniscope can even detect $Δf_{NL} \sim 0.2$. This substantially revises the methods and results of previous work.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Probing reionization with LOFAR using 21-cm redshift space distortions

One of the most promising ways to study the epoch of reionization (EoR) is through radio observations of the redshifted 21-cm line emission from neutral hydrogen. These observations are complicated by the fact that the mapping of redshifts to line-of-sight positions is distorted by the peculiar velocities of the gas. Such distortions can be a source of error if they are not properly understood, but they also encode information about cosmology and astrophysics. We study the effects of redshift space distortions on the power spectrum of 21-cm radiation from the EoR using large scale $N$-body and radiative transfer simulations. We quantify the anisotropy introduced in the 21-cm power spectrum by redshift space distortions and show how it evolves as reionization progresses and how it relates to the underlying physics. We go on to study the effects of redshift space distortions on LOFAR observations, taking instrument noise and foreground subtraction into account. We find that LOFAR should be able to directly observe the power spectrum anisotropy due to redshift space distortions at spatial scales around $k \sim 0.1$ Mpc$^{-1}$ after $\gtrsim$ 1000 hours of integration time. At larger scales, sample errors become a limiting factor, while at smaller scales detector noise and foregrounds make the extraction of the signal problematic. Finally, we show how the astrophysical information contained in the evolution of the anisotropy of the 21-cm power spectrum can be extracted from LOFAR observations, and how it can be used to distinguish between different reionization scenarios.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Simulating cosmic reionization: How large a volume is large enough?

We present the largest-volume (425 Mpc/h=607 Mpc on a side) full radiative transfer simulation of cosmic reionization to date. We show that there is significant additional power in density fluctuations at very large scales. We systematically investigate the effects this additional power has on the progress, duration and features of reionization, as well as on selected reionization observables. We find that comoving simulation volume of ~100 Mpc/h per side is sufficient for deriving a convergent mean reionization history, but that the reionization patchiness is significantly underestimated. We use jackknife splitting to quantify the convergence of reionization properties with simulation volume for both mean-density and variable-density sub-regions. We find that sub-volumes of ~100 Mpc/h per side or larger yield convergent reionization histories, except for the earliest times, but smaller volumes of ~50 Mpc/h or less are not well converged at any redshift. Reionization history milestones show significant scatter between the sub-volumes, of Delta z=0.6-1 for ~50 Mpc/h volumes, decreasing to Delta z=0.3-0.5 for ~100 Mpc/h volumes, and $Δz$~0.1 for ~200 Mpc/h volumes. If we only consider mean-density sub-regions the scatter decreases, but remains at Delta z~0.1-0.2 for the different size sub-volumes. Consequently, many potential reionization observables like 21-cm rms, 21-cm PDF skewness and kurtosis all show good convergence for volumes of ~200 Mpc/h, but retain considerable scatter for smaller volumes. In contrast, the three-dimensional 21-cm power spectra at large scales (k<0.25 h/Mpc) do not fully converge for any sub-volume size. These additional large-scale fluctuations significantly enhance the 21-cm fluctuations, which should improve the prospects of detection considerably, given the lower foregrounds and greater interferometer sensitivity at higher frequencies. (abridged)

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

The scale-dependent signature of primordial non-Gaussianity in the large-scale structure of cosmic reionization

(ABRIDGED)The rise of cosmic structure depends upon the statistical distribution of initial density fluctuations generated by inflation. While the simplest models predict an almost perfectly Gaussian distribution, more-general models predict a level of primordial non-Gaussianity (PNG) that observations might yet be sensitive enough to detect. Recent Planck Collaboration measurements of the CMB temperature anisotropy bispectrum significantly tighten the observational limits, but they are still far from the PNG level predicted by the simplest models of inflation. Probing levels below CMB sensitivities will require other methods, such as searching for the statistical imprint of PNG on galactic halo clustering. During the epoch of reionization (EoR), the first stars and galaxies released radiation into the intergalactic medium (IGM) that created ionized patches whose large-scale geometry and evolution reflected the underlying abundance and large-scale clustering of the star-forming galaxies. This statistical connection between ionized patches in the IGM and galactic halos suggests that observing reionization may be another way to constrain PNG. We employ the linear perturbation theory of reionization and semi-analytic models based on the excursion-set formalism to model the effects of PNG on the EoR. We quantify the effects of PNG on the large-scale structure of reionization by deriving the ionized density bias, i.e. ratio of ionized atomic to total matter overdensities in Fourier space, at small wavenumber. Just as previous studies found that PNG creates a scale-dependent signature in the halo bias, so, too, we find a scale-dependent signature in the ionized density bias. Our results, which differ significantly from previous attempts in the literature to characterize this PNG signature, will be applied elsewhere to predict its observable consequences, e.g. in the cosmic 21cm background.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Will Nonlinear Peculiar Velocity and Inhomogeneous Reionization Spoil 21cm Cosmology from the Epoch of Reionization?

The 21cm background from the epoch of reionization is a promising cosmological probe: line-of-sight velocity fluctuations distort redshift, so brightness fluctuations in Fourier space depend upon angle, which linear theory shows can separate cosmological from astrophysical information. Nonlinear fluctuations in ionization, density and velocity change this, however. The validity and accuracy of the separation scheme are tested here for the first time, by detailed reionization simulations. The scheme works reasonably well early in reionization (< 40% ionized), but not late (> 80% ionized).

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Detecting the Rise and Fall of the First Stars by Their Impact on Cosmic Reionization

The intergalactic medium was reionized before redshift z~6, most likely by starlight which escaped from early galaxies. The very first stars formed when hydrogen molecules (H2) cooled gas inside the smallest galaxies, minihalos of mass between 10^5 and 10^8 solar masses. Although the very first stars began forming inside these minihalos before redshift z~40, their contribution has, to date, been ignored in large-scale simulations of this cosmic reionization. Here we report results from the first reionization simulations to include these first stars and the radiative feedback that limited their formation, in a volume large enough to follow the crucial spatial variations that influenced the process and its observability. We show that, while minihalo stars stopped far short of fully ionizing the universe, reionization began much earlier with minihalo sources than without, and was greatly extended, which boosts the intergalactic electron-scattering optical depth and the large-angle polarization fluctuations of the cosmic microwave background significantly. Although within current WMAP uncertainties, this boost should be readily detectable by Planck. If reionization ended as late as z_ov<~7, as suggested by other observations, Planck will thereby see the signature of the first stars at high redshift, currently undetectable by other probes.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Domain Knowledge Uncertainty and Probabilistic Parameter Constraints

Incorporating domain knowledge into the modeling process is an effective way to improve learning accuracy. However, as it is provided by humans, domain knowledge can only be specified with some degree of uncertainty. We propose to explicitly model such uncertainty through probabilistic constraints over the parameter space. In contrast to hard parameter constraints, our approach is effective also when the domain knowledge is inaccurate and generally results in superior modeling accuracy. We focus on generative and conditional modeling where the parameters are assigned a Dirichlet or Gaussian prior and demonstrate the framework with experiments on both synthetic and real-world data.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Light cone effect on the reionization 21-cm power spectrum

Observations of redshifted 21-cm radiation from neutral hydrogen during the epoch of reionization (EoR) are considered to constitute the most promising tool to probe that epoch. One of the major goals of the first generation of low frequency radio telescopes is to measure the 3D 21-cm power spectrum. However, the 21-cm signal could evolve substantially along the line of sight (LOS) direction of an observed 3D volume, since the received signal from different planes transverse to the LOS originated from different look-back times and could therefore be statistically different. Using numerical simulations we investigate this so-called light cone effect on the spherically averaged 3D 21-cm power spectrum. For this version of the power spectrum, we find that the effect mostly `averages out' and observe a smaller change in the power spectrum compared to the amount of evolution in the mean 21-cm signal and its rms variations along the LOS direction. Nevertheless, changes up to 50% at large scales are possible. In general the power is enhanced/suppressed at large/small scales when the effect is included. The cross-over mode below/above which the power is enhanced/suppressed moves toward larger scales as reionization proceeds. When considering the 3D power spectrum we find it to be anisotropic at the late stages of reionization and on large scales. The effect is dominated by the evolution of the ionized fraction of hydrogen during reionization and including peculiar velocities hardly changes these conclusions. We present simple analytical models which explain qualitatively all the features we see in the simulations.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Redshift Space Distortion of the 21cm Background from the Epoch of Reionization I: Methodology Re-examined

The peculiar velocity of the intergalactic gas responsible for the cosmic 21cm background from the epoch of reionization and beyond introduces an anisotropy in the three-dimensional power spectrum of brightness temperature fluctuations. Measurement of this anisotropy by future 21cm surveys is a promising tool for separating cosmology from 21cm astrophysics. However, previous attempts to model the signal have often neglected peculiar velocity or only approximated it crudely. This paper re-examines the effects of peculiar velocity on the 21cm signal in detail, improving upon past treatment and addressing several issues for the first time. (1) We show that properly accounting for finite optical depth eliminates the unphysical divergence of 21cm brightness temperature in overdense regions of the IGM found by previous work that employed the usual optically-thin approximation. (2) The approximation made previously to circumvent the diverging brightness temperature problem by capping velocity gradient can misestimate the power spectrum on all scales. (3) The observed power spectrum in redshift-space remains finite even in the optically-thin approximation if one properly accounts for the redshift-space distortion. However, results that take full account of finite optical depth show that this approximation is only accurate in the limit of high spin temperature. (4) The linear theory for redshift-space distortion results in ~30% error in the observationally relevant wavenumber range, at the 50% ionized epoch. (5) We describe and test two numerical schemes to calculate the 21cm signal from reionization simulations to incorporate peculiar velocity effects in the optically-thin approximation accurately. One is particle-based, the other grid-based, and while the former is most accurate, we demonstrate that the latter is computationally more efficient and can achieve sufficient accuracy. [Abridged]

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Simulating Cosmic Reionization and the Radiation Backgrounds from the Epoch of Reionization

Large-scale reionization simulations are described which combine the results of cosmological N-body simulations that model the evolving density and velocity fields and identify the galactic halo sources, with ray-tracing radiative transfer calculations which model the nonequilibrium ionization of the intergalactic medium. These simulations have been used to predict some of the signature effects of reionization on cosmic radiation backgrounds, including the CMB, near-IR, and redshifted 21cm backgrounds. We summarize some of our recent progress in this work, and address the question of whether observations of such signature effects can be used to distinguish the relative contributions of galaxies of different masses to reionization.

0 citations0 reviewsNo code linkedOpen access
preprint2012arXiv

Statistical Translation, Heat Kernels and Expected Distances

High dimensional structured data such as text and images is often poorly understood and misrepresented in statistical modeling. The standard histogram representation suffers from high variance and performs poorly in general. We explore novel connections between statistical translation, heat kernels on manifolds and graphs, and expected distances. These connections provide a new framework for unsupervised metric learning for text documents. Experiments indicate that the resulting distances are generally superior to their more standard counterparts.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Can 21-cm observations discriminate between high-mass and low-mass galaxies as reionization sources?

The prospect of detecting the first galaxies by observing their impact on the intergalactic medium as they reionized it during the first billion years leads us to ask whether such indirect observations are capable of diagnosing which types of galaxies were most responsible for reionization. We attempt to answer this by considering a set of large-scale radiative transfer simulations of reionization in sufficiently large volumes to make statistically meaningful predictions of observable signatures, while also directly resolving all atomically-cooling halos down to 10^8 M_solar. We focus here on predictions of the 21-cm background, to see if upcoming observations are capable of distinguishing a universe ionized primarily by high-mass halos from one in which both high-mass and low-mass halos are responsible, and to see how these results depend upon the uncertain source efficiencies. We find that 21-cm fluctuation power spectra observed by the first generation EoR/21-cm radio interferometer arrays should be able to distinguish the case of reionization by high-mass halos alone from that by both high- and low-mass halos, together. Some reionization scenarios yield very similar power spectra and rms evolution and thus can only be discriminated by their different mean reionization history and 21-cm PDF distributions. We find that the skewness of the 21-cm PDF distribution smoothed over LOFAR-like window shows a clear feature correlated with the rise of the rms due to patchiness. Measurements of the mean photoionization rates are sensitive to the average density of the regions being studied and therefore could be strongly skewed in certain cases. (abridged)

0 citations0 reviewsNo code linkedOpen access
preprint2010arXiv

Linguistic Geometries for Unsupervised Dimensionality Reduction

Text documents are complex high dimensional objects. To effectively visualize such data it is important to reduce its dimensionality and visualize the low dimensional embedding as a 2-D or 3-D scatter plot. In this paper we explore dimensionality reduction methods that draw upon domain knowledge in order to achieve a better low dimensional embedding and visualization of documents. We consider the use of geometries specified manually by an expert, geometries derived automatically from corpus statistics, and geometries computed from linguistic resources.

0 citations0 reviewsNo code linkedOpen access
preprint2009arXiv

Inflationary Potential from 21 cm Tomography and Planck

Three-dimensional neutral hydrogen mapping using the redshifted 21 cm line has recently emerged as a promising cosmological probe. Within the framework of slow-roll reconstruction, we analyze how well the inflationary potential can be reconstructed by combining data from 21 cm experiments and cosmic microwave background data from the Planck satellite. We consider inflationary models classified according to the amplitude of their tensor component, and show that 21 cm measurements can significantly improve constraints on the slow-roll parameters and determine the shape of the inflationary potential.

0 citations0 reviewsNo code linkedOpen access