lensfit

lensfit
Name	lensfit
Type	Algorithm / Software
Industry	Astrophysics / Cosmology / Image Analysis
Founded	2007
Developer	University of British Columbia (initial), subsequent collaborations
Latest release	2012 (major revision)

lensfit

lensfit is a model-fitting algorithm and software package developed for weak gravitational lensing shear measurement in astronomical imaging. It has been used in major surveys and projects to infer cosmic shear from galaxy images, and interfaces with survey operations and cosmological inference pipelines. The method emphasizes forward-modeling, probabilistic marginalization, and simultaneous treatment of point-spread function and pixel noise.

Introduction

lensfit was introduced to provide unbiased shear estimates for weak lensing studies associated with surveys such as the Canada–France–Hawaii Telescope Lensing Survey (CFHTLenS), the Kilo-Degree Survey (KiDS), and the Dark Energy Survey (DES). It addresses challenges encountered by teams working on the Sloan Digital Sky Survey (SDSS), the Hyper Suprime-Cam Subaru Strategic Program (HSC), and the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS). Collaborations and institutions like the University of Edinburgh, University College London, the University of Oxford, the University of Cambridge, the University of Durham, and the Institute of Astronomy contributed to algorithm development and testing. lensfit interacts with catalogs and pipelines produced by the European Southern Observatory (ESO), the National Aeronautics and Space Administration (NASA), the Space Telescope Science Institute (STScI), and survey consortia including Euclid Consortium, Vera C. Rubin Observatory (LSST) teams, and the Science Verification teams of the Subary telescope.

Methodology

lensfit performs parametric model fitting to pixel data using galaxy models convolved with measured point-spread functions from stars in the field, drawing on techniques developed in the context of Maximum Likelihood Estimation and Bayesian marginalization as applied in works by groups associated with the Royal Observatory Edinburgh, the Institute for Astronomy (IfA), and the Mullard Space Science Laboratory (MSSL). The algorithm models galaxy surface brightness with exponential and de Vaucouleurs profiles and mixtures thereof, marginalizing nuisance parameters such as centroid, size, and flux following statistical principles used in analyses at Imperial College London and the Max Planck Institute for Extraterrestrial Physics. The PSF modeling step leverages interpolation schemes used in pipelines at the European Space Agency (ESA) and ESA missions like Gaia, and borrows calibration practices from instruments such as the Hubble Space Telescope (HST) and the Subaru Telescope. Estimation of shear employs ensemble averaging and responsivity calibration strategies similar to those discussed in publications from the Kavli Institute for Cosmology, the Institute of Theoretical Astrophysics (ITA), and the Pontificia Universidad Católica de Chile.

Applications

lensfit has been applied to cosmic shear measurement in CFHTLenS, KiDS, and DES science products feeding cosmological parameter estimation efforts by collaborations including Planck, WMAP teams, and the Baryon Oscillation Spectroscopic Survey (BOSS). Data products processed with lensfit have contributed to studies on dark energy constraints pursued by the Dark Energy Task Force, and cross-correlation analyses with galaxy clustering from the Sloan Digital Sky Survey III team and the Two Micron All Sky Survey (2MASS). Lensfit-derived catalogs have supported weak-lensing mass reconstructions of galaxy clusters observed by the South Pole Telescope collaboration, the Atacama Cosmology Telescope project, and investigations by the European Southern Observatory surveys. The method has been integrated in multi-probe analyses alongside supernova samples curated by the Supernova Cosmology Project and the SNLS team.

Validation and Performance

Performance validation of lensfit used image simulation campaigns and community-led challenges such as the Shear Testing Programme (STEP), the GRavitational lEnsing Accuracy Testing (GREAT) challenges, and dedicated simulation efforts by the Cosmology Large Angular Scale Surveyor (CLASS) community. Comparisons were made to alternative shear estimators developed by teams around the Astrophysical Institute Potsdam, the National Optical Astronomy Observatory (NOAO), and the Max Planck Institute for Astrophysics. Metrics included multiplicative and additive bias estimates, tested on synthetic datasets modeled after observations from the Canada–France–Hawaii Telescope, Subaru, and the Cerro Tololo Inter-American Observatory (CTIO). lensfit demonstrated competitive bias control when combined with calibration strategies advanced by groups at the Jet Propulsion Laboratory (JPL) and the Lawrence Berkeley National Laboratory (LBNL).

History and Development

lensfit originated in the late 2000s from collaborations involving researchers at University of British Columbia, University of Edinburgh, and University of Oxford, motivated by the needs of CFHTLenS and preceded by methods trialed in the STEP collaborations. Subsequent development incorporated feedback from KiDS teams at Leiden Observatory, DES consortium members at Fermilab and SLAC National Accelerator Laboratory, and the Euclid Consortium planning groups. The software evolved through iterations influenced by results from GREAT08, GREAT10, and GREAT3 community exercises organized by groups including University College London and the University of Pennsylvania. Methodological refinements drew on statistical practices in astrophysics departments at Princeton University, Columbia University, and the University of Chicago.

Software and Implementation

Implementations of lensfit have been written in compiled languages and scripting interfaces compatible with data reduction environments used by teams at the European Southern Observatory, the Space Telescope Science Institute, and the National Radio Astronomy Observatory (NRAO). The package interfaces with image processing tools such as SExtractor used by the Max Planck Society teams, PSFEx employed by CFHTLenS and KiDS pipelines, and Astropy libraries developed by the Astropy Project community. Deployment in survey pipelines has been coordinated with data management centers like the Cambridge Astronomy Survey Unit (CASU), the NOAO Science Data Management, and the Canadian Astronomy Data Centre (CADC). Integration with cosmological inference frameworks used by Planck Collaboration and the DES Science Collaboration has been achieved through catalog formats compatible with TOPCAT and CDS services at the Strasbourg Astronomical Data Center.

Limitations and Future Directions

Known limitations include sensitivity to model bias for complex morphologies observed with instruments such as HST and JWST, and challenges in PSF interpolation under varying atmospheric and optical conditions encountered at Mauna Kea and Cerro Paranal observatories. Ongoing directions involve hybrid machine-learning enhancements researched at DeepMind collaborations with astronomical groups, efforts to unify model-fitting with moment-based methods advocated by groups at the University of Michigan and Carnegie Observatories, and adaptations for next-generation surveys by the Rubin Observatory LSST Science Collaboration and the Euclid Consortium. Future validation will likely involve cross-comparisons with methods from the HSC team, joint analyses with spectroscopic programs like SDSS-IV (eBOSS) and DESI, and participation in forthcoming community challenges organized by the GREAT series and international survey working groups.

Category:Gravitational lensing