arXiv subjects

Joshua Meyers

Publications and source records attributed to Joshua Meyers.

17 recordsLinked to original sources

arXiv2026

An overview of stray light findings and interpretation during on-sky commissioning of LSSTCam

Wide-field telescopes are intrinsically difficult to shield from unwanted stray and scattered light, while the search to identify sources of contaminating light is frequently a challenging task. The Vera C.~Rubin Observatory, which achieved its first photon with the LSST Camera (LSSTCam) on April 15, 2025, will initiate a revolutionary era for the study of dark matter, dark energy, the transient sky, the Solar System, and the Milky Way. LSSTCam will provide near seeing-limited images of the sky in six bands ($u,g,r,i,z,y$) over a $3.^\circ 5$-diameter field of view, and over the course of a decade, it will execute the Legacy Survey of Space and Time (LSST). This work provides an overview of the dedicated stray and scattered light test campaign that has been undertaken since the start of Rubin commissioning. In particular, we highlight the processes used to characterize, model, and mitigate stray light present in LSSTCam images. The Rubin commissioning team created a series of testing and analysis tools to track stray light artifacts from their initial discovery through reproduction with timely observations, simulation using ray tracing to identify opto-mechanical origins, and finally devising corrective actions. The complex stray light features encountered by Rubin provide a wealth of experience for the future wide-field and extremely wide-field observatories. This work covers the many stages of a long journey that started with conceiving an innovative and challenging optical design, followed by the engineering and system engineering efforts to build it, to finally delivering an optimized and revolutionary cutting-edge facility.

astro-ph.IM↗

arXiv2025

On fine-tuning Boltz-2 for protein-protein affinity prediction

Accurate prediction of protein-protein binding affinity is vital for understanding molecular interactions and designing therapeutics. We adapt Boltz-2, a state-of-the-art structure-based protein-ligand affinity predictor, for protein-protein affinity regression and evaluate it on two datasets, TCR3d and PPB-affinity. Despite high structural accuracy, Boltz-2-PPI underperforms relative to sequence-based alternatives in both small- and larger-scale data regimes. Combining embeddings from Boltz-2-PPI with sequence-based embeddings yields complementary improvements, particularly for weaker sequence models, suggesting different signals are learned by sequence- and structure-based models. Our results echo known biases associated with training with structural data and suggest that current structure-based representations are not primed for performant affinity prediction.

cs.LG↗

arXiv2025

Presenting Profunctors

Motivated by problems in categorical database theory, we introduce and compare two notions of presentation for profunctors, uncurried and curried, which arise intuitively from thinking of profunctors either as functors C^op x D -> Set or C^op -> Set^D. Although the Cartesian closure of Cat means these two perspectives can be used interchangeably at the semantic level, a surprising amount of subtlety is revealed when looking through the lens of syntax. Indeed, we prove that finite uncurried presentations are strictly more expressive than finite curried presentations, hence the two notions do not induce the same class of finitely presentable profunctors. Moreover, an explicit construction for the composite of two curried presentations shows that the class of finitely curried presentable profunctors is closed under composition, in contrast with the larger class of finitely uncurried presentable profunctors, which is not. This shows that curried profunctor presentations are more appropriate for computational tasks that use profunctor composition. We package our results on curried profunctor presentations into a double equivalence from a syntactic double category into the double category of profunctors. Finally, we study the relationship between curried and uncurried presentations, leading to the introduction of curryable presentations. These constitute a subcategory of uncurried presentations which is equivalent to the category of curried presentations, therefore acting as a bridge between the two syntactic choices.

math.CT↗

arXiv2024

'Weather' in the LSST Camera: Investigating Patterns in Differenced Flat Images

During electro-optical testing of the camera for the upcoming Vera C. Rubin Observatory Legacy Survey of Space and Time, a unique low-signal pattern was found in differenced pairs of flat images used to create photon transfer curves, with peak-to-peak variations of a factor of 10^-3. A turbulent pattern of this amplitude was apparent in many differenced flat-fielded images. The pattern changes from image to image and shares similarities with atmospheric 'weather' turbulence patterns. We applied several strategies to determine the source of the turbulent pattern and found that it is representative of the mixing of the air and index of refraction variations caused by the internal camera purge system displacing air, which we are sensitive to due to our flat field project setup. Characterizing this changing environment with 2-D correlation functions of the 'weather' patterns provides evidence that the images reflect the changes in the camera environment due to the internal camera purge system. Simulations of the full optical system using the galsim and batoid codes show that the weather pattern affects the dispersion of the camera point-spread function at only the one part in 10^-4 level

astro-ph.IM↗

arXiv2022

Fast Left Kan Extensions Using The Chase

We show how computation of left Kan extensions can be reduced to computation of free models of cartesian (finite-limit) theories. We discuss how the standard and parallel chase compute weakly free models of regular theories and free models of cartesian theories, and compare the concept of "free model" with a similar concept from database theory known as "universal model". We prove that, as algorithms for computing finite free models of cartesian theories, the standard and parallel chase are complete under fairness assumptions. Finally, we describe an optimized implementation of the parallel chase specialized to left Kan extensions that achieves an order of magnitude improvement in our performance benchmarks compared to the next fastest left Kan extension algorithm we are aware of.

cs.DB↗

arXiv2021

The LSST Dark Energy Science Collaboration (DESC) Science Requirements Document

The Large Synoptic Survey Telescope (LSST) Dark Energy Science Collaboration (DESC) will use five cosmological probes: galaxy clusters, large scale structure, supernovae, strong lensing, and weak lensing. This Science Requirements Document (SRD) quantifies the expected dark energy constraining power of these probes individually and together, with conservative assumptions about analysis methodology and follow-up observational resources based on our current understanding and the expected evolution within the field in the coming years. We then define requirements on analysis pipelines that will enable us to achieve our goal of carrying out a dark energy analysis consistent with the Dark Energy Task Force definition of a Stage IV dark energy experiment. This is achieved through a forecasting process that incorporates the flowdown to detailed requirements on multiple sources of systematic uncertainty. Future versions of this document will include evolution in our software capabilities and analysis plans along with updates to the LSST survey strategy.

astro-ph.CO↗

arXiv2020

Molecular representation learning with language models and domain-relevant auxiliary tasks

We apply a Transformer architecture, specifically BERT, to learn flexible and high quality molecular representations for drug discovery problems. We study the impact of using different combinations of self-supervised tasks for pre-training, and present our results for the established Virtual Screening and QSAR benchmarks. We show that: i) The selection of appropriate self-supervised task(s) for pre-training has a significant impact on performance in subsequent downstream tasks such as Virtual Screening. ii) Using auxiliary tasks with more domain relevance for Chemistry, such as learning to predict calculated molecular properties, increases the fidelity of our learnt representations. iii) Finally, we show that molecular representations learnt by our model `MolBert' improve upon the current state of the art on the benchmark datasets.

cs.LG↗

arXiv2020

Optimising LSST Observing Strategy for Weak Lensing Systematics

The LSST survey will provide unprecedented statistical power for measurements of dark energy. Consequently, controlling systematic uncertainties is becoming more important than ever. The LSST observing strategy will affect the statistical uncertainty and systematics control for many science cases; here, we focus on weak lensing systematics. The fact that the LSST observing strategy involves hundreds of visits to the same sky area provides new opportunities for systematics mitigation. We explore these opportunities by testing how different dithering strategies (pointing offsets and rotational angle of the camera in different exposures) affect additive weak lensing shear systematics on a baseline operational simulation, using the $ρ-$statistics formalism. Some dithering strategies improve systematics control at the end of the survey by a factor of up to $\sim 3-4$ better than others. We find that a random translational dithering strategy, applied with random rotational dithering at every filter change, is the most effective of those strategies tested in this work at averaging down systematics. Adopting this dithering algorithm, we explore the effect of varying the area of the survey footprint, exposure time, number of exposures in a visit, and exposure to the Galactic plane. We find that any change that increases the average number of exposures (in filters relevant to weak lensing) reduces the additive shear systematics. Some ways to achieve this increase may not be favorable for the weak lensing statistical constraining power or for other probes, and we explore the relative trade-offs between these options given constraints on the overall survey parameters.

astro-ph.IM↗

arXiv2020

Synthesizing Property & Casualty Ratemaking Datasets using Generative Adversarial Networks

Due to confidentiality issues, it can be difficult to access or share interesting datasets for methodological development in actuarial science, or other fields where personal data are important. We show how to design three different types of generative adversarial networks (GANs) that can build a synthetic insurance dataset from a confidential original dataset. The goal is to obtain synthetic data that no longer contains sensitive information but still has the same structure as the original dataset and retains the multivariate relationships. In order to adequately model the specific characteristics of insurance data, we use GAN architectures adapted for multi-categorical data: a Wassertein GAN with gradient penalty (MC-WGAN-GP), a conditional tabular GAN (CTGAN) and a Mixed Numerical and Categorical Differentially Private GAN (MNCDP-GAN). For transparency, the approaches are illustrated using a public dataset, the French motor third party liability data. We compare the three different GANs on various aspects: ability to reproduce the original data structure and predictive models, privacy, and ease of use. We find that the MC-WGAN-GP synthesizes the best data, the CTGAN is the easiest to use, and the MNCDP-GAN guarantees differential privacy.

stat.ML↗

arXiv2018

An Overview of the LSST Image Processing Pipelines

The Large Synoptic Survey Telescope (LSST) is an ambitious astronomical survey with a similarly ambitious Data Management component. Data Management for LSST includes processing on both nightly and yearly cadences to generate transient alerts, deep catalogs of the static sky, and forced photometry light-curves for billions of objects at hundreds of epochs, spanning at least a decade. The algorithms running in these pipelines are individually sophisticated and interact in subtle ways. This paper provides an overview of those pipelines, focusing more on those interactions than the details of any individual algorithm.

astro-ph.IM↗

arXiv2016

Isophote Shapes of Early-Type Galaxies in Massive Clusters at $z\sim1$ and 0

We compare the isophote shape parameter $a_{4}$ of early-type galaxies (ETGs) between $z\sim1$ and 0 as a proxy for dynamics to investigate the epoch at which the dynamical properties of ETGs are established, using cluster ETG samples with stellar masses of $\log(M_{*}/M_{\odot})\geq10.5$ which have spectroscopic redshifts. We have 130 ETGs from the Hubble Space Telescope Cluster Supernova Survey for $z\sim1$ and 355 ETGs from the Sloan Digital Sky Survey for $z\sim0$. We have developed an isophote shape analysis method which can be used for high-redshift galaxies and has been carefully compared with published results. We have applied the same method for both the $z\sim1$ and $0$ samples. We find similar dependence of the $a_{4}$ parameter on the mass and size at $z\sim1$ and 0; the main population of ETGs changes from disky to boxy at a critical stellar mass of $\log(M_{*}/M_{\odot})\sim11.5$ with the massive end dominated by boxy. The disky ETG fraction decreases with increasing stellar mass both at $z\sim1$ and $0$, and is consistent between these redshifts in all stellar mass bins when the Eddington bias is taken into account. Although uncertainties are large, the results suggest that the isophote shapes and probably dynamical properties of ETGs in massive clusters are already in place at $z>1$ and do not significantly evolve in $z<1$, despite significant size evolution in the same galaxy population. The constant disky fraction favors less violent processes than mergers as a main cause of the size and morphological evolution of intermediate mass ETGs in $z<1$.

astro-ph.GA↗

arXiv2016

Probabilistic Cosmological Mass Mapping from Weak Lensing Shear

We infer gravitational lensing shear and convergence fields from galaxy ellipticity catalogs under a spatial process prior for the lensing potential. We demonstrate the performance of our algorithm with simulated Gaussian-distributed cosmological lensing shear maps and a reconstruction of the mass distribution of the merging galaxy cluster Abell 781 using galaxy ellipticities measured with the Deep Lens Survey. Given interim posterior samples of lensing shear or convergence fields on the sky, we describe an algorithm to infer cosmological parameters via lens field marginalization. In the most general formulation of our algorithm we make no assumptions about weak shear or Gaussian distributed shape noise or shears. Because we require solutions and matrix determinants of a linear system of dimension that scales with the number of galaxies, we expect our algorithm to require parallel high-performance computing resources for application to ongoing wide field lensing surveys.

astro-ph.CO↗

arXiv2014

Hierarchical probabilistic inference of cosmic shear

Point estimators for the shearing of galaxy images induced by gravitational lensing involve a complex inverse problem in the presence of noise, pixelization, and model uncertainties. We present a probabilistic forward modeling approach to gravitational lensing inference that has the potential to mitigate the biased inferences in most common point estimators and is practical for upcoming lensing surveys. The first part of our statistical framework requires specification of a likelihood function for the pixel data in an imaging survey given parameterized models for the galaxies in the images. We derive the lensing shear posterior by marginalizing over all intrinsic galaxy properties that contribute to the pixel data (i.e., not limited to galaxy ellipticities) and learn the distributions for the intrinsic galaxy properties via hierarchical inference with a suitably flexible conditional probabilitiy distribution specification. We use importance sampling to separate the modeling of small imaging areas from the global shear inference, thereby rendering our algorithm computationally tractable for large surveys. With simple numerical examples we demonstrate the improvements in accuracy from our importance sampling approach, as well as the significance of the conditional distribution specification for the intrinsic galaxy properties when the data are generated from an unknown number of distinct galaxy populations with different morphological characteristics.

astro-ph.CO↗

arXiv2009

Subaru FOCAS Spectroscopic Observations of High-Redshift Supernovae

We present spectra of high-redshift supernovae (SNe) that were taken with the Subaru low resolution optical spectrograph, FOCAS. These SNe were found in SN surveys with Suprime-Cam on Subaru, the CFH12k camera on the Canada-France-Hawaii Telescope (CFHT), and the Advanced Camera for Surveys (ACS) on the Hubble Space Telescope (HST). These SN surveys specifically targeted z>1 Type Ia supernovae (SNe Ia). From the spectra of 39 candidates, we obtain redshifts for 32 candidates and spectroscopically identify 7 active candidates as probable SNe Ia, including one at z=1.35, which is the most distant SN Ia to be spectroscopically confirmed with a ground-based telescope. An additional 4 candidates are identified as likely SNe Ia from the spectrophotometric properties of their host galaxies. Seven candidates are not SNe Ia, either being SNe of another type or active galactic nuclei. When SNe Ia are observed within a week of maximum light, we find that we can spectroscopically identify most of them up to z=1.1. Beyond this redshift, very few candidates were spectroscopically identified as SNe Ia. The current generation of super red-sensitive, fringe-free CCDs will push this redshift limit higher.

astro-ph.CO↗

arXiv2009

Galaxy Clusters in Formation: Determining the Age of the Red-Sequence in Optical and X-ray Clusters at z~1 with HST

Using deep two-band imaging from the Hubble Space Telescope, we measure the color-magnitude relations (CMR) of E/S0 galaxies in a set of 9 optically-selected clusters principally from the Red-Sequence Cluster Survey (RCS) at 0.9 < z < 1.23. We find that the mean scatter in the CMR in the observed frame of this set of clusters is 0.049 +/- 0.008, as compared to 0.031 +/- 0.007 in a similarly imaged and identically analyzed X-ray sample at similar redshifts. Single-burst stellar population models of the CMR scatter suggest that the E/S0 population in these RCS clusters truncated their star-formation at z~1.6, some 0.9 Gyrs later than their X-ray E/S0 counterparts which were truncated at z~2.1. The notion that this is a manifestation of the differing evolutionary states of the two populations of cluster galaxies is supported by comparison of the fraction of bulge-dominated galaxies found in the two samples which shows that optically-selected clusters contain a smaller fraction of E/S0 galaxies at the their cores

astro-ph.CO↗

arXiv2009

The XMM Cluster Survey: Galaxy Morphologies and the Color-Magnitude Relation in XMMXCS J2215.9-1738 at z=1.46

We present a study of the morphological fractions and color-magnitude relation in the most distant X-ray selected galaxy cluster currently known, XMMXCS J2215.9-1738 at z=1.46, using a combination of optical imaging data obtained with the Hubble Space Telescope Advanced Camera for Surveys, and infrared data from the Multi-Object Infrared Camera and Spectrograph, mounted on the 8.2m Subaru telescope. We find that the morphological mix of the cluster galaxy population is similar to clusters at z~1: approximately ~62% of the galaxies identified as likely cluster members are ellipticals or S0s; and ~38% are spirals or irregulars. We measure the color-magnitude relations for the early type galaxies, finding that the slope in the z_850-J relation is consistent with that measured in the Coma cluster, some ~9 Gyr earlier, although the uncertainty is large. In contrast, the measured intrinsic scatter about the color-magnitude relation is more than three times the value measured in Coma, after conversion to rest frame U-V. From comparison with stellar population synthesis models, the intrinsic scatter measurements imply mean luminosity weighted ages for the early type galaxies in J2215.9-1738 of ~3 Gyr, corresponding to the major epoch of star formation coming to an end at z_f = 3-5. We find that the cluster exhibits evidence of the `downsizing' phenomenon: the fraction of faint cluster members on the red sequence expressed using the Dwarf-to-Giant Ratio (DGR) is 0.32+/-0.18. This is consistent with extrapolation of the redshift evolution of the DGR seen in cluster samples at z < 1. In contrast to observations of some other z > 1 clusters, we find a lack of very bright galaxies within the cluster.

astro-ph.CO↗

arXiv2008

Clusters of Galaxies in the First Half of the Universe from the IRAC Shallow Survey

We have identified 335 galaxy cluster and group candidates, 106 of which are at z > 1, using a 4.5 um selected sample of objects from a 7.25 deg^2 region in the Spitzer Infrared Array Camera (IRAC) Shallow Survey. Clusters were identified as 3-dimensional overdensities using a wavelet algorithm, based on photometric redshift probability distributions derived from IRAC and NOAO Deep Wide-Field Survey data. We estimate only ~10% of the detections are spurious. To date 12 of the z > 1 candidates have been confirmed spectroscopically, at redshifts from 1.06 to 1.41. Velocity dispersions of ~750 km/s for two of these argue for total cluster masses well above 10^14 M_sun, as does the mass estimated from the rest frame near infrared stellar luminosity. Although not selected to contain a red sequence, some evidence for red sequences is present in the spectroscopically confirmed clusters, and brighter galaxies are systematically redder than the mean galaxy color in clusters at all redshifts. The mean I - [3.6] color for cluster galaxies up to z ~ 1 is well matched by a passively evolving model in which stars are formed in a 0.1 Gyr burst starting at redshift z_f = 3. At z > 1, a wider range of formation histories is needed, but higher formation redshifts (i.e. z_f > 3) are favored for most clusters.

astro-ph↗