서브메뉴
검색
Methods for Weak Lensing Systematics in the Era of the Vera C. Rubin Observatory
Methods for Weak Lensing Systematics in the Era of the Vera C. Rubin Observatory
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105232
- ISBN
- 9798291567371
- DDC
- 310
- 서명/저자
- Methods for Weak Lensing Systematics in the Era of the Vera C. Rubin Observatory
- 발행사항
- [Sl] : University of Michigan, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 157 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-03, Section: B.
- 주기사항
- Advisor: Avestruz, Camille.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2025.
- 초록/해제
- 요약One of the biggest mysteries in Cosmology today relates to the nature of dark energy. Dark energy drives the observed cosmic expansion of the universe by counteracting the effects of gravity. The existence of dark energy could imply an exotic new type of substance, or the breakdown of General Relativity at cosmological scales. Any of these options would have significant implications for fundamental physics. Modern dark energy surveys, such as the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST), are designed to improve our understanding of dark energy by precisely measuring its properties.A powerful probe of dark energy is weak lensing, which refers to the deflection of light rays from distant astronomical sources due to matter along their path to us. Weak lensing induces a coherent alignment of observed galaxy shapes which directly depends on the integrated matter density along the line of sight, so-called cosmic shear. Measurements of cosmic shear can be used to understand the distribution and evolution of matter in our Universe, which in turn allow us to constrain dark energy properties. LSST will dramatically increase the statistical power of weak lensing surveys by observing tens of billions of galaxies during its 10 year mission. However, their power will be limited by a wide range of theoretical and observational systematics.This thesis develops software tools and algorithms to explore and mitigate weak lensing related systematics the LSST survey will face. One such systematic is blending, which refers to the visual overlap of light sources in astronomical images. Given the increased depth and number density of sources in LSST, we expect a large fraction of observable sources will be blended. Blending can create several types of biases in weak lensing measurements, and some of these remain unaccounted for by contemporary algorithms.First, we present a software package, the BlendingToolKit, with the goal of providing a framework to study blending-related measurement biases in a controlled way. It provides customizable simulations of galaxy blends; a framework to standardize the input and output of deblenders; and a library of relevant metrics related to detection, morphology reconstruction, and flux recovery of these sources. This package is actively being used in the LSST Dark Energy Science Collaboration to aid the development of new deblending methods.The next portion of the thesis is devoted to a probabilistic algorithm to mitigate blending-related biases: BLISS. This novel method uses simulation based inference (SBI) to produce a probabilistic catalog of blended galaxies which captures corresponding measurement uncertainty. We test our approach in LSST-like image simulations and demonstrate a significant improvement in recovering the flux of highly blended sources.Finally, we present a modern implementation of the hierarchical Bayesian shear inference framework in Schneider et al. 2015. Our implementation leverages GPUs and gradient-based samplers to improve upon its runtime by an order of magnitude. We rigorously test our algorithm on a set of 300k isolated parametric galaxies with realistic pixel and shape noise levels. We find that resulting shear measurements present no significant level of pixel noise bias, and that these meet the LSST requirement.Together, these three studies represent a significant contribution towards leveraging modern statistical and hardware tools to reduce the impact of image-level systematics in weak lensing measurements for dark energy surveys.
- 일반주제명
- Statistics
- 일반주제명
- Physics
- 일반주제명
- Astronomy
- 일반주제명
- Astrophysics
- 키워드
- Cosmology
- 키워드
- Weak lensing
- 키워드
- Cosmic shear
- 키워드
- Galaxy blending
- 키워드
- Machine learning
- 기타저자
- University of Michigan Physics
- 기본자료저록
- Dissertations Abstracts International. 87-03B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017359890
■00520260202105232
■006m o d
■007cr#unu||||||||
■020 ▼a9798291567371
■035 ▼a(MiAaPQ)AAI32271904
■035 ▼a(MiAaPQ)umichrackham006324
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a310
■1001 ▼aMendoza Serrano, Ismael Salvador.
■24510▼aMethods for Weak Lensing Systematics in the Era of the Vera C. Rubin Observatory
■260 ▼a[Sl]▼bUniversity of Michigan▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a157 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-03, Section: B.
■500 ▼aAdvisor: Avestruz, Camille.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2025.
■520 ▼aOne of the biggest mysteries in Cosmology today relates to the nature of dark energy. Dark energy drives the observed cosmic expansion of the universe by counteracting the effects of gravity. The existence of dark energy could imply an exotic new type of substance, or the breakdown of General Relativity at cosmological scales. Any of these options would have significant implications for fundamental physics. Modern dark energy surveys, such as the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST), are designed to improve our understanding of dark energy by precisely measuring its properties.A powerful probe of dark energy is weak lensing, which refers to the deflection of light rays from distant astronomical sources due to matter along their path to us. Weak lensing induces a coherent alignment of observed galaxy shapes which directly depends on the integrated matter density along the line of sight, so-called cosmic shear. Measurements of cosmic shear can be used to understand the distribution and evolution of matter in our Universe, which in turn allow us to constrain dark energy properties. LSST will dramatically increase the statistical power of weak lensing surveys by observing tens of billions of galaxies during its 10 year mission. However, their power will be limited by a wide range of theoretical and observational systematics.This thesis develops software tools and algorithms to explore and mitigate weak lensing related systematics the LSST survey will face. One such systematic is blending, which refers to the visual overlap of light sources in astronomical images. Given the increased depth and number density of sources in LSST, we expect a large fraction of observable sources will be blended. Blending can create several types of biases in weak lensing measurements, and some of these remain unaccounted for by contemporary algorithms.First, we present a software package, the BlendingToolKit, with the goal of providing a framework to study blending-related measurement biases in a controlled way. It provides customizable simulations of galaxy blends; a framework to standardize the input and output of deblenders; and a library of relevant metrics related to detection, morphology reconstruction, and flux recovery of these sources. This package is actively being used in the LSST Dark Energy Science Collaboration to aid the development of new deblending methods.The next portion of the thesis is devoted to a probabilistic algorithm to mitigate blending-related biases: BLISS. This novel method uses simulation based inference (SBI) to produce a probabilistic catalog of blended galaxies which captures corresponding measurement uncertainty. We test our approach in LSST-like image simulations and demonstrate a significant improvement in recovering the flux of highly blended sources.Finally, we present a modern implementation of the hierarchical Bayesian shear inference framework in Schneider et al. 2015. Our implementation leverages GPUs and gradient-based samplers to improve upon its runtime by an order of magnitude. We rigorously test our algorithm on a set of 300k isolated parametric galaxies with realistic pixel and shape noise levels. We find that resulting shear measurements present no significant level of pixel noise bias, and that these meet the LSST requirement.Together, these three studies represent a significant contribution towards leveraging modern statistical and hardware tools to reduce the impact of image-level systematics in weak lensing measurements for dark energy surveys.
■590 ▼aSchool code: 0127.
■650 4▼aStatistics
■650 4▼aPhysics
■650 4▼aAstronomy
■650 4▼aAstrophysics
■653 ▼aCosmology
■653 ▼aWeak lensing
■653 ▼aCosmic shear
■653 ▼aGalaxy blending
■653 ▼aBayesian statistics
■653 ▼aMachine learning
■690 ▼a0606
■690 ▼a0605
■690 ▼a0463
■690 ▼a0596
■71020▼aUniversity of Michigan▼bPhysics.
■7730 ▼tDissertations Abstracts International▼g87-03B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359890▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


