서브메뉴
검색
Optimizing Parameter Inference and Foreground Removal Techniques for Cosmic Microwave Background Data Analysis
Optimizing Parameter Inference and Foreground Removal Techniques for Cosmic Microwave Background Data Analysis
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202104642
- ISBN
- 9798286499793
- DDC
- 530
- 서명/저자
- Optimizing Parameter Inference and Foreground Removal Techniques for Cosmic Microwave Background Data Analysis
- 발행사항
- [Sl] : Columbia University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 524 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-01, Section: B.
- 주기사항
- Advisor: Hill, James C.
- 학위논문주기
- Thesis (Ph.D.)--Columbia University, 2025.
- 초록/해제
- 요약The cosmic microwave background (CMB) is the earliest observable light in the Universe, providing insight into the early Universe, how it evolved, and how it continues to evolve. As CMB photons travel to us today, they interact with matter in galaxies, leading to secondary anisotropies. These secondary anisotropies and other microwave foregrounds provide useful astrophysical information. This work covers a variety of topics in CMB data analysis, with introductory information in Chapter 1.Maps of the microwave sky are often compressed into angular power spectra. Chapter 2 explores the problem of computing angular power spectra from maps that are masked. In particular, it focuses on cases where the masks applied may be correlated with the maps themselves, such as point source masks that are often used. The work develops a new formalism for computing these spectra, correcting a result that has been widely used in CMB data analyses for the past two decades. Applying the result to simulations confirms that the new formalism recovers expected angular power spectra to machine precision, whereas the original approach could yield biases of up to 10% in the measured power spectra.Once computed, angular power spectra are used in cosmological parameter inference pipelines. Usually, the power spectra are measured across various frequency channels, and templates must be fit to extract both the CMB and astrophysical foregrounds. However, these multifrequency power spectra only take into account Gaussian information in the fields. While the primary CMB is a Gaussian random field, astrophysical foregrounds can be highly non-Gaussian. Thus, information is lost in the power spectrum compression. Chapters 3 and 4 explore how to use the power spectra of needlet internal linear combination (NILC) component-separated maps to do parameter inference in a way that accounts for non-Gaussian information. Chapter 3 builds on the formalism from Chapter 2 to develop analytic expressions for these spectra and validates them on simulations. Chapter 4 investigates using simulation-based likelihood-free inference on NILC power spectra, showing that it can reduce parameter posterior areas by 60% when large non-Gaussian foregrounds are present. This result has implications for primordial \uD835\uDC35-mode detection, where the large non-Gaussian dust foreground is significantly larger than current constraints on the tensor-to-scalar ratio r.Chapters 5, 6, and 7 focus on microwave foregrounds and foreground removal strategies. Chapter 5 presents the foreground model of the Atacama Cosmology Telescope (ACT) Collaboration Data Release 6 (DR6). The work validates the foreground model by assessing its consistency with existing data, investigating foreground model variations and extensions, and validating the full analysis pipeline on realistic non-Gaussian simulations, with correlations among the different foregrounds. Chapter 6 presents new foreground cleaning methods, specifically by using large-scale structure tracers to clean the cosmic infrared background (CIB) and thermal Sunyaev-Zel'dovich (tSZ) effect for enhancing measurements of the CMB blackbody temperature power spectrum, forecasting signal-to-noise improvements of up to 50% using future galaxy surveys. Chapter 7 presents an algorithm for CIB cleaning for the purpose of tSZ cross-correlations, achieving a factor of 1.6 improvement in the signal-to-noise ratio on simulations, as compared with standard approaches.Chapters 8 and 9 present constraints on beyond ΛCDM cosmological models. Chapter 8 uses neural network emulators emulators of Boltzmann codes to constrain the early dark energy (EDE) model using Baryon Acoustic Oscillation (BAO) data from the Dark Energy Spectroscopic Instrument (DESI). The emulators achieve 100x speedups in parameter inference pipelines. The work finds that there is no preference for the EDE model with DESI Year 1 (Y1) data, with the maximum fractional contribution of EDE to the cosmic energy budget being \uD835\uDC53EDE 0.091 (95% CL) using Planck CMB, CMB lensing, and DESI BAO. Finally, Chapter 9 is adapted from a larger work by the ACT Collaboration, using the DR6 data to constrain several extended cosmological models. The work found no preference for any beyond ΛCDM model. Chapter 9 specifically focuses on ACT DR6 constraints on two models: EDE and a modified gravity model. Using the DR6 data along with additional datasets including DESI BAO and CMB lensing data, the work places tight constraints on both models, and finds no statistically significant preference for them over ΛCDM. For the EDE model, \uD835\uDC53EDE 0.12 (95% CL), and the Hubble constant is \uD835\uDC3B0 = 69.9 +0.8 −1.5 km/s/Mpc (68% CL). For the modified gravity model, the growth index is \uD835\uDEFE = 0.663 ± 0.052 (68% CL), and the amplitude of density fluctuations is \uD835\uDC468 = 0.799 ± 0.012 (68% CL).
- 일반주제명
- Physics
- 일반주제명
- Astrophysics
- 일반주제명
- Remote sensing
- 일반주제명
- Applied physics
- 기타저자
- Columbia University Physics
- 기본자료저록
- Dissertations Abstracts International. 87-01B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017358314
■00520260202104642
■006m o d
■007cr#unu||||||||
■020 ▼a9798286499793
■035 ▼a(MiAaPQ)AAI32114232
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a530
■1001 ▼aSurrao, Kristen Marie.
■24510▼aOptimizing Parameter Inference and Foreground Removal Techniques for Cosmic Microwave Background Data Analysis
■260 ▼a[Sl]▼bColumbia University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a524 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-01, Section: B.
■500 ▼aAdvisor: Hill, James C.
■5021 ▼aThesis (Ph.D.)--Columbia University, 2025.
■520 ▼aThe cosmic microwave background (CMB) is the earliest observable light in the Universe, providing insight into the early Universe, how it evolved, and how it continues to evolve. As CMB photons travel to us today, they interact with matter in galaxies, leading to secondary anisotropies. These secondary anisotropies and other microwave foregrounds provide useful astrophysical information. This work covers a variety of topics in CMB data analysis, with introductory information in Chapter 1.Maps of the microwave sky are often compressed into angular power spectra. Chapter 2 explores the problem of computing angular power spectra from maps that are masked. In particular, it focuses on cases where the masks applied may be correlated with the maps themselves, such as point source masks that are often used. The work develops a new formalism for computing these spectra, correcting a result that has been widely used in CMB data analyses for the past two decades. Applying the result to simulations confirms that the new formalism recovers expected angular power spectra to machine precision, whereas the original approach could yield biases of up to 10% in the measured power spectra.Once computed, angular power spectra are used in cosmological parameter inference pipelines. Usually, the power spectra are measured across various frequency channels, and templates must be fit to extract both the CMB and astrophysical foregrounds. However, these multifrequency power spectra only take into account Gaussian information in the fields. While the primary CMB is a Gaussian random field, astrophysical foregrounds can be highly non-Gaussian. Thus, information is lost in the power spectrum compression. Chapters 3 and 4 explore how to use the power spectra of needlet internal linear combination (NILC) component-separated maps to do parameter inference in a way that accounts for non-Gaussian information. Chapter 3 builds on the formalism from Chapter 2 to develop analytic expressions for these spectra and validates them on simulations. Chapter 4 investigates using simulation-based likelihood-free inference on NILC power spectra, showing that it can reduce parameter posterior areas by 60% when large non-Gaussian foregrounds are present. This result has implications for primordial \uD835\uDC35-mode detection, where the large non-Gaussian dust foreground is significantly larger than current constraints on the tensor-to-scalar ratio r.Chapters 5, 6, and 7 focus on microwave foregrounds and foreground removal strategies. Chapter 5 presents the foreground model of the Atacama Cosmology Telescope (ACT) Collaboration Data Release 6 (DR6). The work validates the foreground model by assessing its consistency with existing data, investigating foreground model variations and extensions, and validating the full analysis pipeline on realistic non-Gaussian simulations, with correlations among the different foregrounds. Chapter 6 presents new foreground cleaning methods, specifically by using large-scale structure tracers to clean the cosmic infrared background (CIB) and thermal Sunyaev-Zel'dovich (tSZ) effect for enhancing measurements of the CMB blackbody temperature power spectrum, forecasting signal-to-noise improvements of up to 50% using future galaxy surveys. Chapter 7 presents an algorithm for CIB cleaning for the purpose of tSZ cross-correlations, achieving a factor of 1.6 improvement in the signal-to-noise ratio on simulations, as compared with standard approaches.Chapters 8 and 9 present constraints on beyond ΛCDM cosmological models. Chapter 8 uses neural network emulators emulators of Boltzmann codes to constrain the early dark energy (EDE) model using Baryon Acoustic Oscillation (BAO) data from the Dark Energy Spectroscopic Instrument (DESI). The emulators achieve 100x speedups in parameter inference pipelines. The work finds that there is no preference for the EDE model with DESI Year 1 (Y1) data, with the maximum fractional contribution of EDE to the cosmic energy budget being \uD835\uDC53EDE 0.091 (95% CL) using Planck CMB, CMB lensing, and DESI BAO. Finally, Chapter 9 is adapted from a larger work by the ACT Collaboration, using the DR6 data to constrain several extended cosmological models. The work found no preference for any beyond ΛCDM model. Chapter 9 specifically focuses on ACT DR6 constraints on two models: EDE and a modified gravity model. Using the DR6 data along with additional datasets including DESI BAO and CMB lensing data, the work places tight constraints on both models, and finds no statistically significant preference for them over ΛCDM. For the EDE model, \uD835\uDC53EDE 0.12 (95% CL), and the Hubble constant is \uD835\uDC3B0 = 69.9 +0.8 −1.5 km/s/Mpc (68% CL). For the modified gravity model, the growth index is \uD835\uDEFE = 0.663 ± 0.052 (68% CL), and the amplitude of density fluctuations is \uD835\uDC468 = 0.799 ± 0.012 (68% CL).
■590 ▼aSchool code: 0054.
■650 4▼aPhysics
■650 4▼aAstrophysics
■650 4▼aRemote sensing
■650 4▼aApplied physics
■653 ▼aCosmic microwave background
■653 ▼aForeground removal
■653 ▼aNon-Gaussian foregrounds
■690 ▼a0605
■690 ▼a0596
■690 ▼a0799
■690 ▼a0215
■71020▼aColumbia University▼bPhysics.
■7730 ▼tDissertations Abstracts International▼g87-01B.
■790 ▼a0054
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17358314▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


