Search

The definition and investigation of general classes of nonparametric priors has recently been an active research line in Bayesian statistics. Among the various proposals, the Gibbs-type family, which includes the Dirichlet process as a special case, stands out as the most tractable class of nonparametric priors for exchangeable sequences of observations. This is the consequence of a key simplifying assumption on the learning mechanism, which, however, has justification except that of ensuring mathematical tractability. In this paper, we remove such an assumption and investigate a general class of random probability measures going beyond the Gibbs-type framework. More specifically, we present a nonparametric hierarchical structure based on transformations of completely random measures, which extends the popular hierarchical Dirichlet process. This class of priors preserves a good degree of tractability, given that we are able to determine the fundamental quantities for Bayesian inference. In particular, we derive the induced partition structure and the prediction rules and characterize the posterior distribution. These theoretical results are also crucial to devise both a marginal and a conditional algorithm for posterior inference. An illustration concerning prediction in genomic sequencing is also provided.

The mean density of a random closed set with integer Hausdorff dimension is a crucial notion in stochastic geometry, in fact it is a fundamental tool in a large variety of applied problems, such as image analysis, medicine, computer vision, etc. Hence the estimation of the mean density is a problem of interest both from a theoretical and computational standpoint. Nowadays different kinds of estimators are available in the literature, in particular here we focus on a kernel–type estimator, which may be considered as a generalization of the traditional kernel density estimator of random variables to the case of random closed sets. The aim of the present paper is to provide asymptotic properties of such an estimator in the context of Boolean models, which are a broad class of random closed sets. More precisely we are able to prove large and moderate deviation principles, which allow us to derive the strong consistency of the estimator of the mean density as well as asymptotic confidence intervals. Finally we underline the connection of our theoretical findings with classical literature concerning density estimation of random variables.

The West Sardinian offshore represents the east passive continental margin of the Ligure-Provencal oceanic basin. Since the Upper Oligocene the westward subduction of the Tethyan ocean below the Europe plate originated the opening of the Ligure-Provencal-Balearic back-arc basin. It was related to the separation from Europe plate and the anticlockwise rotation of the Corso-Sardinian block time. The rift sequence fills the graben and onlaps the north-sud horst of the basement (fig. 1.1A). After the end of the drift phase, during the MSC strongly affected the West Sardinian margin originating the erosional truncation (MES) on the slope and the evaporate deposition (Messinian trilogy of Rehault et al., 1984) on the deep basin (fig. 1.2). As in the main Mediterranean Sea, in the studied area the Lower Pliocene Trubi marls represents the quick drop of the sea level. The overlaying Upper Pliocene-Quaternary sequence is thick on the deep basin, while on the slope the sea bottom currents take a main role with erosion and reduced deposition on some limited throughs. Seismic facies of MSC unconformities and units are described.

13 pages, 12 figures, 5 tables, Sample disturbance is still a key issue in offshore investigations, especially when logistic and financial limitations do not allow the use of drilling equipment. This paper focuses on the comparison between the disturbance induced by a conventional free-fall piston corer (FF) and a modified piston corer (AD) equipped with a velocity control (Angel Descent method). Twin core samples were retrieved in two successions of pelitic sediments with a prevailing non-clayey fraction and a non-negligible sandy fraction. Comparison was based on different acquisition, physical and mechanical parameters ranging from accelerometer data to magnetic susceptibility logs and geotechnical parameters from laboratory investigations, including oedometer compression tests and cyclic simple shear tests. Accelerometer data highlighted the sharp reduction in velocity obtained for AD samples. Magnetic susceptibility logs, characterized by a pattern of peaks induced by several volcaniclastic levels present in the succession, indicated that the AD method significantly reduces core shortening. Among geotechnical investigations, cyclic shear tests provided small-strain shear moduli always higher in AD samples, whilst the response of oedometer compression tests was equivocal. In fact, methods for assessing sample disturbance have demonstrated to bear limited effectiveness when applied to soils with relatively low clay content and significant overconsolidation as it is the case of the studied sediments, R. Urgeles and J. Llopart acknowledge funding from project INSIGHT (CTM2015-70155-R) jointly funded by the Spanish Ministerio de Economía y Competitividad and the European Regional Development Fund. The research was funded by grants awarded to FB, AC and PT from the National Research Program Flagship Projects RITMARE – The Italian Research for the Sea - coordinated by the Italian National Research Council

15 pages, 8 figures, 1 table, supplementary data http://dx.doi.org/10.1016/j.margeo.2019.106064, The Western Mediterranean basin has been formed by Miocene back-arc extension and is underlain by a thin and young lithosphere. This young lithosphere is warm, as testified by an overall elevated offshore heat flow. Heat flow within the Western Mediterranean is, however, highly variable and existing data are unevenly distributed and poorly studied in the central part of the Liguro-Provençal and Algero-Balearic basins. This central part is floored by a young oceanic crust, bordered by different continental margins, cut by transform faults, and filled by up to 8 km of sediments. We present a total of 148 new heat flow data collected during the MedSalt and WestMedFlux cruises in 2015 and 2016 and aligned along seven regional profiles that show an important heat flow variability on the basin-scale, but also locally on the margins. A new heat flow map for the Western Mediterranean outlines the following regional features: (1) a higher average heat flow in the Algero-Balearic basin compared to the Liguro-Provençal basin (94 ± 13 mW/m2 and 78 ± 16 mW/m2, respectively), and (2) a regional thermal asymmetry in both basins, but with opposed heat flow trends. Up to 20% of this heat flow difference can be explained by sediment blanketing, but age and heterogeneity of ocean crust due to an asymmetric and polyphased opening of the basins are believed to have given the major thermal imprint. Estimates of the age of the oceanic crust based on the new heat flow suggest a considerably younger West Algerian basin (16–23 Ma) compared to the East Algerian basin and the West Sardinia oceanic floor (31–37 Ma). On the margins and ocean-continent transitions of the Western Mediterranean the new heat flow data point out the existence of two types of local anomalies (length scale 5–30 km): (1) locally increased heat flow up to 153 mW/m2 on the Gulf of Lion margin results from thermal refraction of large salt diapirs, and (2) the co-existing of both low (110 mW/m2) heat flow areas on the South Balearic margin suggests a heat redistribution system. We suspect the lateral heat advection is resulting from a regional fluid circulation in the sediments associated to the widespread Plio-Pleistocene volcanism on the South Balearic margin, This work was also supported by the “Laboratoire d'Excellence” LabexMER (ANR-10-LABX-19) and co-funded by a grant from the French government under the program “Investissements d'Avenir”, With the funding support of the ‘Severo Ochoa Centre of Excellence’ accreditation (CEX2019-000928-S), of the Spanish Research Agency (AEI)

This chapter begins by discussing the advantages of a diachronic approach to buildings; it turns, then, to distinguishing major approaches to the study of their long lives. To explain the dynamics of the long construction times, Trachtenberg identifies willingness on the part of medieval builders not only to redesign continuously, but also to offer up solutions as new challenges arose on a work site. The conventional approach, which this edited volume revisits and challenges, focuses on a building’s moment of inception, with particular attention given to the architect, patron, social conditions, or even technical aspects that brought the building into existence. A diachronic approach does more than look at a building over many synchronic moments. Those moments—what might be termed a building’s temporal parts—are means to an end for the scholar of a building’s long life. Any dissimilarity in a building’s temporal parts is an instance of change; conversely, any similarity is a case of continuity.

Feature models are popular in machine learning and they have been recently used to solve many unsupervised learning problems. In these models every observation is endowed with a finite set of features, usually selected from an infinite collection $(F_{j})_{j\geq 1}$. Every observation can display feature $F_{j}$ with an unknown probability $p_{j}$. A statistical problem inherent to these models is how to estimate, given an initial sample, the conditional expected number of hitherto unseen features that will be displayed in a future observation. This problem is usually referred to as the missing mass problem. In this work we prove that, using a suitable multiplicative loss function and without imposing any assumptions on the parameters $p_{j}$, there does not exist any universally consistent estimator for the missing mass. In the second part of the paper, we focus on a special class of heavy-tailed probabilities $(p_{j})_{j\geq 1}$, which are common in many real applications, and we show that, within this restricted class of probabilities, the nonparametric estimator of the missing mass suggested by Ayed et al. (2017) is strongly consistent. As a byproduct result, we will derive concentration inequalities for the missing mass and the number of features observed with a specified frequency in a sample of size $n$.

27 pages, 14 figures, 4 tables, supporting information https://doi.org/10.1111/bre.12331, Using a combination of geophysical and geotechnical data from Storfjorden Trough Mouth Fan off southern Svalbard, we investigate the hydrogeology of the continental margin and how this is affected by Quaternary glacial advances and retreats over the continental shelf. The geotechnical results show that plumites, deposited during the deglaciation, have high porosities, permeabilities and compressibilities with respect to glacigenic debris flows and tills. These results together with margin stratigraphic models obtained from seismic reflection data were used as input for numerical finite element models to understand focusing of interstitial fluids on glaciated continental margins. The modelled evolution of the Storfjorden TMF shows that tills formed on the shelf following the onset of glacial sedimentation (ca. 1.5 Ma) acted as aquitards and therefore played a significant role in decreasing the vertical fluid flow towards the sea floor and diverting it towards the slope. The model shows that high overpressure ratios (up to λ ca. 0.6) developed below the shelf edge and on the middle slope. A more detailed model for the last 220 kyrs accounting for ice loading during glacial maxima shows that the formation of these aquitards on the shelf focused fluid flow towards the most permeable plumite sediments on the slope. The less permeable glacigenic debris flows that were deposited during glacial maxima on the slope hinder fluid evacuation from plumites allowing high overpressure ratios (up to λ ca. 0.7) to develop in the shallowest plumite layers. These high overpressures likely persist to the Present and are a critical precondition for submarine slope failure, This study is funded by the ‘Ministerio Economia y Competitividad’ through grants DEGLABAR, (CTM2010‐17386), CORIBAR‐ES (CTM2011‐14807‐E) and SVAIS (POL2006‐07390). UNESCO and IUGS are also acknowledged for funding through projects IGCP‐585 and IGCP‐640. The Italian contribution was supported by PNRA project 2013/B2.08 VALFLU

The problem of maximizing cell type discovery under budget constraints is a fundamental challenge for the collection and analysis of single-cell RNA-sequencing (scRNA-seq) data. In this paper, we introduce a simple, computationally efficient, and scalable Bayesian nonparametric sequential approach to optimize the budget allocation when designing a large scale experiment for the collection of scRNA-seq data for the purpose of, but not limited to, creating cell atlases. Our approach relies on the following tools: i) a hierarchical Pitman-Yor prior that recapitulates biological assumptions regarding cellular differentiation, and ii) a Thompson sampling multi-armed bandit strategy that balances exploitation and exploration to prioritize experiments across a sequence of trials. Posterior inference is performed by using a sequential Monte Carlo approach, which allows us to fully exploit the sequential nature of our species sampling problem. We empirically show that our approach outperforms state-of-the-art methods and achieves near-Oracle performance on simulated and scRNA-seq data alike. HPY-TS code is available at https://github.com/fedfer/HPYsinglecell.

EGU General Assembly 2019 in Viena, Austria , 7-12 april, 2019, During the Messinian salinity crisis (MSC), between 5.97 and 5.33 Ma, the Mediterranean Sea became disconnectedfromtheworld’soceansandafastandcontinuousevaporationgenerateditspartialdesiccation,resultingin the deposition of more than one million cubic kilometres of salt. The extent of the evaporative drawdown phases associated to the MSC remain poorly constrained, particularly in the central Mediterranean. In this study we investigate the geomorphology of the Malta Escarpment to provide answers to these questions. We use multibeam echosounder, seismic reflection data, and gravity cores to identify more than two hundred linear to arcuate scars and more than two hundred canyons. The largest of the canyons – Noto, Cumecs, and Heron Canyons – are up to 100 km long and 39 km wide. We also identify concave breaks of slopes and terraces, the most extensive of which are located in the northern Malta Escarpment, and which are 70 km long and 2478 m deep, and 25 km long and 2545 m deep. Along most of the base of the northern Malta Escarpment, the western Ionian Basin seafloor is characterised by a series of ridges that are up to 8 km long, have an average height of 30 m and are spaced 2-3 km apart. Seismic reflection profiles from the western Ionian Basin show the occurrence of lenses of intermediate amplitude and sub-parallel reflectors, which are up to 0.2 s (TWTT) thick and 8 km wide, downslope of the large canyonmouthsandbeneaththeevaporitesequence.Gravitycoresfromthelargesubmarinecanyonsfeaturesandy turbidites and debrites. We propose that during the MSC, base level fall, fluvial erosion formed a dense network of canyons across the Malta Escarpment whilst coastal erosion developed extensive palaeoshorelines and shore platforms. We carry out an isostatic restoration of the palaeoshorelines and shore platforms on the northern Malta Escarpment to infer an evaporative drawdown of 1800 – 2000 m in the eastern Mediterranean Sea during the MSC. We interpret the occurrence of pre-evaporite sedimentary lobes in the western Ionian Basin as suggesting that either evaporative drawdown and canyon formation occurred predominantly before salt deposition, or that only the latest salt deposition at the basin margin occurred after the formation of the sedimentary lobes. After the MSC, the drivers of geomorphic evolution of the Malta Escarpment included: (i) canyon erosion by submarine gravity flows, with the most recent activity taking place

Sample disturbance is still a key point in offshore investigations, especially when logistic and financial limitations do not allow the use of drilling equipment controlled on board. In fact many geotechnical, geological and geophysical analyses require knowledge of physical and mechanical properties, e.g. undrained shear strength, strength parameters in effective stresses, stiffness and damping parameters. These parameters are affected by soil disturbance induced by the sampling procedure, which in most offshore projects is particularly significant. This presentation focuses on the comparison between the disturbance induced by a conventional free fall piston corer (FF) and a modified piston corer (AD) equipped with a velocity control (Angel Descent method). Twin core samples were retrieved in two successions of pelitic sediments with a prevailing nonclayey fraction and a non-negligible sandy fraction. Seafloor sampling was performed in 2014 during the SAOS (Stability Assessment of an Open Slope) cruise onboard R/V Urania (CNR) within the national research project RITMARE. The cruise was jointly organized by the IAMC (now ISMAR) and IGAG institutes of the CNR and by OGS, in the South Tyrrhenian Sea (Southeastern Italy). At three of the ten coring locations during the SAOS cruise, sampling was duplicated using a standard piston corer and a piston corer employing the Angel Descent® method described in Magagnoli (2017). Two pair of cores were collected in the unfailed sedimentary section just up-slope of the Licosa Landslide crown scarp in about 250 m water depth (Sammartini et al., 2018). One pair of cores was collected from an intra-slope basin in 673 m water depth. Comparison was based on different acquisition, physical and mechanical parameters ranging from accelerometer data to magnetic susceptibility logs and geotechnical parameters from laboratory investigations, including oedometer compression tests and cyclic simple shear tests. Accelerometer data highlighted the sharp reduction in velocity obtained for AD samples. Magnetic susceptibility logs, characterized by a pattern of peaks induced by volcaniclastic levels interspersed in the succession, indicated that the AD method significantly reduces core shortening. Among geotechnical investigations, cyclic shear tests provided small-strain shear moduli always higher in AD samples, whilst the response of oedometer compression tests was non-univocal, being influenced by the prevailing non-clayey fraction, as it has been pointed out in the very recent technical literature

Presented at EGU 2019.

Information about predation risk is of fundamental value in biological communities. Because many prey species have shared predators, eavesdropping on other species' alarms is a widely recognized mechanism underlying the formation of mixed-species groups. However, information transfer may vary both across and within groups because some species provide higher-quality information about predators than others. We tested this phenomenon in Amazonian understory mixed-species flocks of birds in which two sentinel species-the bluish-slate antshrike (Thamnomanes schistogynus) and the dusky-throated antshrike (Thamnomanes ardesiacus)-occupy different habitats and provide alarm calls that are used by eavesdropping flock mates. In a playback experiment, two associate species responded significantly more strongly to alarm calls from the same sentinel species, reflecting the greater reliability of information about predator threats that could affect survival and habitat choice. Our work provides evidence of a repeated asymmetry across space in the available information about threats.

Presented in the session "Subaqueous mass movements and their consequences: from scientific knowledge to geohazard assessment" at IAS Rome 2019 (http://iasroma2019.org/).Submarine landslide deposits often appear "acoustically transparent" when imaged in sub-bottom profiles (very high-frequency single-channel seismic reflection images). These blank zones are sometimes attributed to complete disaggregation of the sediments from sliding processes (petrophysical homogenisation) or to residual gas trapped within the failed sediments (seismic attenuation). This study uses full-waveform elastic seismic modelling to explore an alternative possibility -- that strong deformation from sliding (stratal disruption) can cause this apparent low-reflectivity seismic response, despite metre-scale structure and reflectivity remaining within the slide deposit.We model a simplified submarine landslide deposit scenario from an example from the Black Sea. The geometry of the seafloor and top and base of the slide deposit is based on a chirp sub-bottom profile acquired from an AUV. We limit the component sediments to two distinct lithologies, with elastic properties taken from MSCL measurements of P-wave velocity and density from a core that penetrates the slide. The failed zone is modelled as a statistical medium with structure with a lateral scale length of ~2 metres. The full-waveform modelling uses a pseudo-spectral approach for computational efficiency, and the signal processing follows a similar workflow to the original profile. The resulting sub-bottom profile image shows a reduction in RMS amplitude in the failed zone by a factor of ~2, despite the magnitude of impedance contrasts within the slide being unchanged with respect to the impedance contrasts in the unfailed sediments.This study looks to resolve a long-standing curiosity in subacqueous mass-movement research -- why do we often acquire cores which are well-stratified and show no evidence of sediment re-mobilisation from apparently "transparent" submarine landslide deposits? Our hypothesis is that apparent low-reflectivity in sub-bottom profiles can be generated by stratal disruption at metre-scale (e.g., slumping), even when strong impedance constrasts remain after sliding. Cores typically have a diameter of around 10 cm, meaning metre-scale structure can appear undeformed and well-stratified at core scale. This is the situation in the Black Sea example, where we observe a completely transparent mass-transport deposit in the sub-bottom profile, but cores penetrating the slide mass show little evidence of sediment re-mobilisation.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has not been fixed in the paper.

The estimation of the mean density of random closed sets in $$\mathbb {R}^d$$Rd with integer Hausdorff dimension $$n

The prediction of future outcomes of a random phenomenon is typically based on a certain number of “analogous” observations from the past. When observations are generated by multiple samples, a natural notion of analogy is partial exchangeability and the problem of prediction can be effectively addressed in a Bayesian nonparametric setting. Instead of confining ourselves to the prediction of a single future experimental outcome, as in most treatments of the subject, we aim at predicting features of an unobserved additional sample of any size. We first provide a structural property of prediction rules induced by partially exchangeable arrays, without assuming any specific nonparametric prior. Then we focus on a general class of hierarchical random probability measures and devise a simulation algorithm to forecast the outcome of m future observations, for any m≥1. The theoretical result and the algorithm are illustrated by means of a real dataset, which also highlights the “borrowing strength” behavior across samples induced by the hierarchical specification.

Vectors of hierarchical random probability measures are popular tools in Bayesian nonparametrics. They may be used as priors whenever partial exchangeability is assumed at the level of either the observations or of some latent variables involved in the model. The first contribution in this direction can be found in Teh et al. (2006), who introduced the hierarchical Dirichlet process. Recently, Camerlenghi et al. (2018) have developed a general distribution theory for hierarchical processes, which includes the derivation of the partition structure, the posterior distribution and the prediction rules. The present paper is a review of these theoretical findings for vectors of hierarchies of Pitman--Yor processes.

Reconstruction of Paleomagnetic Secular Variation (PSV) of the geomagnetic field is fundamental both to assess geodynamo models and to obtain age constraints for rocks, sediments and archaeological material. We present refined age-calibrated Holocene PSV and relative paleointensity (RPI) stack curves derived from Arctic marine sediments (Northwestern Barents Sea). The Holocene sections of four sedimentary cores were correlated on the basis of paleomagnetic trends and age models, and stacked. The resultant composite PSV and RPI Holocene records (NBS stack) and the reconstructed Holocene Virtual Geomagnetic Pole (VGP) path were evaluated in comparison with the most recent paleomagnetic stack curves and geomagnetic field models. The data indicate that during the Holocene time, the VGPs moved within the superficial projection of the inner core tangent cylinder, with the exception of short time intervals around 5600 and 3200 cal yr BP when VGPs extended to lower latitudes. These deviations might reflect regional geomagnetic features, such as persistent geomagnetic flux lobes at core-mantle boundary. Our data confirm that the large VGP shift observed around 5600 cal yr BP is the result of an increased radial magnetic field at the core-mantle boundary over North America, whilst the VGP shift around 3200 cal yr BP represents a major swing to middle latitudes toward the Middle East and might be associated to a regional high paleointensity peak, known as Levantine Iron Age Anomaly (LIAA).

The Barents Sea Ice Sheet was a marine-based ice sheet, i.e., it rested on the Barents Sea floor during the Last Glacial Maximum (21 ky BP). The Bjørnøyrenna Ice Stream was the largest ice stream draining the Barents Sea Ice Sheet and is regarded as an analogue for contemporary ice streams in West Antarctica. Here, the retreat of the Bjørnøyrenna Ice Stream is simulated by means of two numerical ice sheet models and results assessed against geological data. We investigate the sensitivity of the ice stream to changes in ocean temperature and the impact of grounding-line physics on ice stream retreat. Our results suggest that the role played by sub-shelf melting depends on how the grounding-line physics is represented in the models. When an analytic constraint on the ice flux across the grounding line is applied, the retreat of Bjørnøyrenna Ice Stream is primarily driven by internal ice dynamics rather than by oceanic forcing. This suggests that implementations of grounding-line physics need to be carefully assessed when evaluating and predicting the response of contemporary marine-based ice sheets and individual ice streams to ongoing and future ocean warming.

8th International Symposium on Submarine Mass Movements and Their Consequences (ISSMMTC), 7-9 May 2018, Victoria, British Columbia, Canada, Using a combination of geophysical and geotechnical data from the Storfjorden Trough Mouth Fan, off southern Svalbard, we investigate the role of glacial advances and retreats over the continental shelf on the hydrogeology of the continental margin. The results of compressibility and permeability tests are used together with margin stratigraphic models from seismic data, as input for numerical finite element models to understand focusing of interstitial fluids in glaciated continental margins. The modeled evolution of the Storfjorden TMF from 2.7 to 0.2 Ma shows that onset of glacial sedimentation (~1.5 Ma) had a significant role in developing aquicludes (tills) on the shelf that decreased the vertical fluid flow towards the sea floor and diverted it towards the slope. This model shows that prior to 220 ka, high overpressure (~0.6) develops below the shelf edge and in the middle slope. A more detailed, high resolution model for the last 220 kyrs accounting for ice loading during Glacial Maxima shows that the less permeable glacigenic debris flows deposited during glacial maxima on the slope hinder fluid evacuation from the plumites. This effect in combination with the fluid flow focusing from the shelf allows high overpressures (~0.7) to develop in the shallower-most plumite layers. These high overpressures likely persist to the Present and are critical in determining the onset of submarine slope failure. The safety factor of the upper continental slope is reduced by60% due to the combination of high sedimentation rates, ice loading and focusing of fluids during Glacial Maxima with values of the factor of safety reaching 1.2 during the LGM and beginning of the last deglaciation

no abstract

Discrete random structures are important tools in Bayesian nonparametrics and the resulting models have proven effective in density estimation, clustering, topic modeling and prediction, among others. In this paper, we consider nested processes and study the dependence structures they induce. Dependence ranges between homogeneity, corresponding to full exchangeability, and maximum heterogeneity, corresponding to (unconditional) independence across samples. The popular nested Dirichlet process is shown to degenerate to the fully exchangeable case when there are ties across samples at the observed or latent level. To overcome this drawback, inherent to nesting general discrete random measures, we introduce a novel class of latent nested processes. These are obtained by adding common and group-specific completely random measures and, then, normalising to yield dependent random probability measures. We provide results on the partition distributions induced by latent nested processes, and develop an Markov Chain Monte Carlo sampler for Bayesian inferences. A test for distributional homogeneity across groups is obtained as a by product. The results and their inferential implications are showcased on synthetic and real data.