arXiv subjects

J. de Jong

Publications and source records attributed to J. de Jong.

9 recordsLinked to original sources

arXiv2024

Multinomial belief networks for healthcare data

Healthcare data from patient or population cohorts are often characterized by sparsity, high missingness and relatively small sample sizes. In addition, being able to quantify uncertainty is often important in a medical context. To address these analytical requirements we propose a deep generative Bayesian model for multinomial count data. We develop a collapsed Gibbs sampling procedure that takes advantage of a series of augmentation relations, inspired by the Zhou$\unicode{x2013}$Cong$\unicode{x2013}$Chen model. We visualise the model's ability to identify coherent substructures in the data using a dataset of handwritten digits. We then apply it to a large experimental dataset of DNA mutations in cancer and show that we can identify biologically meaningful clusters of mutational signatures in a fully data-driven way.

stat.ML↗

arXiv2022

Consistent lensing and clustering in a low-$S_8$ Universe with BOSS, DES Year 3, HSC Year 1 and KiDS-1000

We evaluate the consistency between lensing and clustering probes of large-scale structure based on measurements of projected galaxy clustering from BOSS combined with overlapping galaxy-galaxy lensing from three surveys: DES Y3, HSC Y1, and KiDS-1000. An intra-lensing-survey study finds good agreement between these lensing data. We model the observations using the Dark Emulator and fit the data at two fixed cosmologies: Planck, with $S_8=0.83$, and a Lensing cosmology with $S_8=0.76$. For a joint analysis limited to scales with $R>5.25h^{-1}$Mpc, we find that both cosmologies provide an acceptable fit to the data. Full utilisation of the small-scale clustering and lensing measurements is hindered by uncertainty in the impact of baryon feedback and assembly bias, which we account for with a reasoned theoretical error budget. We incorporate a systematic scaling parameter for each redshift bin, $A$, that decouples the lensing and clustering to capture any inconsistency. When a wide range of scales ($0.15<R<60h^{-1}$Mpc) are incorporated, we find different results for the consistency of clustering and lensing between the two cosmologies. Limiting the analysis to the bins for which the impact of the selection of the lens sample is expected to be minimal, for the low-$S_8$ Lensing cosmology, the measurements are consistent with $A$=1; $A=0.91\pm0.04$ using DES+KiDS and $A=0.97\pm0.06$ using HSC. For the Planck cosmology case, we find a discrepancy: $A=0.79\pm0.03$ using DES+KiDS and $A=0.84\pm0.05$ using HSC. We demonstrate that a kSZ-based estimate for baryonic effects alleviates some of the discrepancy in the Planck cosmology. This analysis demonstrates the statistical power of these small-scale measurements, but also indicates that caution is still warranted given current uncertainties in modelling baryonic effects, assembly bias, and selection effects in the foreground sample.

astro-ph.CO↗

arXiv2019

The frequency of dust lanes in edge-on spiral galaxies identified by Galaxy Zoo in KiDS imaging of GAMA targets

Dust lanes bisect the plane of a typical edge-on spiral galaxy as a dark optical absorption feature. Their appearance is linked to the gravitational stability of spiral disks; the fraction of edge-on galaxies that displays a dust lane is a direct indicator of the typical vertical balance between gravity and turbulence; a balance struck between the energy input from star-formation and the gravitational pull into the plane of the disk. Based on morphological classifications by the Galaxy~Zoo project on the Kilo-Degree Survey (KiDS) imaging data in the Galaxy and Mass Assembly (GAMA) fields, we explore the relation of dust lanes to the galaxy characteristics, most of which were determined using the {\sc magphys} spectral energy distribution fitting tool: stellar mass, total and specific star-formation rates, and several parameters describing the cold dust component. We find that the fraction of dust lanes does depend on the stellar mass of the galaxy; they start to appear at $M^* \sim 10^9 M_\odot$. A dust lane also implies strongly a dust mass of at least $10^5 M_\odot$, but otherwise does not correlate with cold dust mass parameters of the {\sc magphys} spectral energy distribution analysis, nor is there a link with star-formation rate, specific or total. Dust lane identification does not depend on disk ellipticity (disk thickness) or Sersic profile but correlates with bulge morphology; a round bulge favors dust lane votes. The central component along the line of sight that produces the dust lane is not associated with either one of the components fit by {\sc magphys}, the cold diffuse component or the localized, heated component in HII regions, but a mix of these two.

astro-ph.GA↗

arXiv2017

The asymptotic volume of diagonal subpolytopes of symmetric stochastic matrices

The asymptotic volume of the polytope of symmetric stochastic matrices can be determined by asymptotic enumeration techniques as in the case of the Birkhoff polytope. These methods can be extended to polytopes of symmetric stochastic matrices with given diagonal, if this diagonal varies not too wildly. To this end, the asymptotic number of symmetric matrices with natural entries, zero diagonal and varying row sums is determined and a third order correction factor to this is examined.

math.CO↗

arXiv2010

Magnetic field measurements and wind-line variability of OB-type stars

Context. The first magnetic fields in O- and B-type stars that do not belong to the Bp-star class, have been discovered. The cyclic UV wind-line variability, which has been observed in a significant fraction of early-type stars, is likely to be related to such magnetic fields. Aims. We attempt to improve our understanding of massive-star magnetic fields, and observe twenty-five carefully-selected, OB-type stars. Methods. Of these stars we obtain 136 magnetic field strength measurements. We present the UV wind-line variability of all selected targets and summarise spectropolarimetric observations acquired using the MUSICOS spectropolarimeter, mounted at the TBL, Pic du Midi, between December 1998 and November 2004. From the average Stokes I and V line profiles, derived using the LSD method, we measure the magnetic field strengths, radial velocities, and first moment of the line profiles. Results. No significant magnetic field is detected in any OB-type star that we observed. Typical 1σ errors are between 15 and 200 G. A possible magnetic-field detection for the O9V star 10 Lac remains uncertain, because the field measurements depend critically on the fringe- effect correction in the Stokes V spectra. We find excess emission in UV-wind lines, centred about the rest wavelength, to be a new indirect indicator of the presence of a magnetic field in early B-type stars. The most promising candidates to host magnetic fields are the B-type stars δ Cet and 6 Cep, and a number of O stars. Conclusions. Although some O and B stars have strong dipolar field, which cause periodic variability in the UV wind-lines, such strong fields are not widespread. If the variability observed in the UV wind-lines of OB stars is generally caused by surface magnetic fields, these fields are either weak (<~few hundred G) or localised.

astro-ph.SR↗

arXiv2010

The Photodetector Array Camera and Spectrometer (PACS) on the Herschel Space Observatory

The Photodetector Array Camera and Spectrometer (PACS) is one of the three science instruments on ESA's far infrared and submillimetre observatory. It employs two Ge:Ga photoconductor arrays (stressed and unstressed) with 16x25 pixels, each, and two filled silicon bolometer arrays with 16x32 and 32x64 pixels, respectively, to perform integral-field spectroscopy and imaging photometry in the 60-210μ m wavelength regime. In photometry mode, it simultaneously images two bands, 60-85μ m or 85-125μ\m and 125-210μ m, over a field of view of ~1.75'x3.5', with close to Nyquist beam sampling in each band. In spectroscopy mode, it images a field of 47"x47", resolved into 5x5 pixels, with an instantaneous spectral coverage of ~1500km/s and a spectral resolution of ~175km/s. We summarise the design of the instrument, describe observing modes, calibration, and data analysis methods, and present our current assessment of the in-orbit performance of the instrument based on the Performance Verification tests. PACS is fully operational, and the achieved performance is close to or better than the pre-launch predictions.

astro-ph.IM↗

arXiv2010

Herschel-PACS spectroscopy of IR-bright galaxies at high redshift

We present Herschel-PACS observations of rest-frame mid-infrared and far-infrared spectral line emissions from two lensed, ultra-luminous infrared galaxies at high redshift: MIPS J142824.0+352619 (MIPS J1428), a starburst-dominated system at z = 1.3, and IRAS F10214+4724 (F10214), a source at z = 2.3 hosting both star-formation and a luminous AGN. We have detected [OI]63 micron and [OIII]52 micron in MIPS J1428, and tentatively [OIII]52 micron in F10214. Together with the recent ZEUS-CSO [CII]158 micron detection in MIPS J1428 we can for the first time combine [OI], [CII] and far-IR (FIR) continuum measurements for photo-dissociation (PDR) modeling of an ultra-luminous (L_IR > 10^12 L_sun) star forming galaxy at the peak epoch of cosmic star formation. We find that MIPS J1428, contrary to average local ULIRGs, does not show a deficit in [OI] relative to FIR. The combination of far-UV flux G_0 and gas density n (derived from the PDR models), as well as the star formation efficiency (derived from CO and FIR) is similar to normal or starburst galaxies, despite the high infrared luminosity of this system. In contrast, F10214 has stringent upper limits on [OIV] and [SIII], and an [OIII]/FIR ratio at least an order of magnitude lower than local starbursts or AGN, similar to local ULIRGs.

astro-ph.CO↗

arXiv2010

The effects of star formation on the low-metallicity ISM: NGC4214 mapped with Herschel/PACS spectroscopy

We present Herschel/PACS spectroscopic maps of the dwarf galaxy NC4214 observed in 6 far infrared fine-structure lines: [C II] 158mu, [O III] 88mu, [O I] 63mu, [O I] 146mu, [N II] 122mu, and [N II] 205mu. The maps are sampled to the full telescope spatial resolution and reveal unprecedented detail on ~ 150 pc size scales. We detect [C II] emission over the whole mapped area, [O III] being the most luminous FIR line. The ratio of [O III]/[C II] peaks at about 2 toward the sites of massive star formation, higher than ratios seen in dusty starburst galaxies. The [C II]/CO ratios are 20 000 to 70 000 toward the 2 massive clusters, which are at least an order of magnitude larger than spiral or dusty starbursts, and cannot be reconciled with single-slab PDR models. Toward the 2 massive star-forming regions, we find that L[CII] is 0.5 to 0.8% of the LTIR . All of the lines together contribute up to 2% of LTIR . These extreme findings are a consequence of the lower metallicity and young, massive-star formation commonly found in dwarf galaxies. These conditions promote large-scale photodissociation into the molecular reservoir, which is evident in the FIR line ratios. This illustrates the necessity to move to multiphase models applicable to star-forming clusters or galaxies as a whole.

astro-ph.CO↗

arXiv2006

GRB 051028: an intrinsically faint GRB at high redshift?

We present multiwavelength observations of the gamma-ray burst GRB 051028 detected by HETE-2 in order to derive its afterglow emission parameters and to determine the reason for its optical faintness when compared to other events. Observations were taken in the optical (2.0m Himalayan Chandra Telescope, 1.34m Tautenburg, 4.2m William Herschel Telescope) and in X-rays (Swift/XRT) between 2.7 hours and 10 days after the onset of the event. The data can be interpreted by collimated emission in a jet with a typical value of $p$ = 2.4 which is moving in an homogeneous interstellar medium and with a cooling frequency nu_{c} still above the X-rays at 0.5 days after the burst onset. GRB 051028 can be classified as a ``gray'' or ``potentially dark'' GRB. On the basis of the combined optical and Swift/XRT data, we conclude that the reason for the optical dimness is not extra absorption in the host galaxy, but rather the GRB taking place at high-redshift.We also notice the very striking similarity with the optical lightcurve of GRB 050730, a burst with a spectroscopic redshift of 3.967, although GRB 051028 is about 3 mag fainter. We suggest that the bump could be explained by multiple energy injection episodes and that the burst is intrinsically faint when compared to the average afterglows detected since 1997. The non-detection of the host galaxy down to R = 25.1 is also consistent with the burst arising at high redshift, compatible with the published pseudo-z of 3.7 +/- 1.8.

astro-ph↗