Minsu Park
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Minsu Park has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
Bhuvnesh Jain
Center for Particle Cosmology, Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA
3 shared publications- Marco Raveri
Department of Physics and INFN, University of Genova, Via Dodecaneso 33, 16146
3 shared publications - Christina D. Kreisch
Department of Astrophysical Sciences, Princeton University, Princeton, New Jersey 08544, USA
2 shared publications Jo Dunkley
Joseph Henry Laboratories of Physics, Jadwin Hall, Princeton University, Princeton, New Jersey 08544, USA
2 shared publications- Sunao Sugiyama
University of Pennsylvania
2 shared publications Adam D. Hincks
David A. Dunlap Department of Astronomy and Astrophysics, University of Toronto, 50 St. George St., Toronto, Ontario M5S 3H4, Canada
1 shared publication
Latest Publications
- 2 December 2025
- Published by American Physical Society (APS) in Physical Review D
- Vol. 112 (12)
Data compression with noise suppression for inference under noisy covariance
- 2 December 2025
- Published by American Physical Society (APS) in Physical Review D
- Vol. 112 (12)
Data compression with noise suppression for inference under noisy covariance
- Open Access
Data Compression with Noise Suppression for Inference under Noisy Covariance
In many fields including cosmology, statistical inference often relies on Gaussian likelihoods whose covariance matrices are estimated from a finite number of simulations. This finite-sample estimation introduces noise into the covariance, which propagates to parameter estimates, a phenomenon known as the Dodelson-Schneider (DS) effect, leading to inflated uncertainties. While the Massively Optimized Parameter Estimation and Data compression (MOPED) algorithm offers lossless Fisher information-preserving compression, it does not mitigate the DS effect when the compression matrix itself is derived from noisy covariances. In this paper, we propose a modified compression scheme, powered MOPED ($p$-MOPED), which suppresses noise propagation by balancing information retention and covariance estimate noise reduction through a tunable power-law transformation of the sample correlation matrix. We test $p$-MOPED against standard and diagonal MOPED on toy models and on cosmological data from the Subaru Hyper Suprime-Cam Year 3 weak lensing survey. Our results demonstrate that $p$-MOPED consistently outperforms other approaches, especially in regimes with limited simulations, offering a robust compression strategy for high-dimensional data analyses under practical constraints.
Open AccessData Compression with Noise Suppression for Inference under Noisy Covariance
In many fields including cosmology, statistical inference often relies on Gaussian likelihoods whose covariance matrices are estimated from a finite number of simulations. This finite-sample estimation introduces noise into the covariance, which propagates to parameter estimates, a phenomenon known as the Dodelson-Schneider (DS) effect, leading to inflated uncertainties. While the Massively Optimized Parameter Estimation and Data compression (MOPED) algorithm offers lossless Fisher information-preserving compression, it does not mitigate the DS effect when the compression matrix itself is derived from noisy covariances. In this paper, we propose a modified compression scheme, powered MOPED ($p$-MOPED), which suppresses noise propagation by balancing information retention and covariance estimate noise reduction through a tunable power-law transformation of the sample correlation matrix. We test $p$-MOPED against standard and diagonal MOPED on toy models and on cosmological data from the Subaru Hyper Suprime-Cam Year 3 weak lensing survey. Our results demonstrate that $p$-MOPED consistently outperforms other approaches, especially in regimes with limited simulations, offering a robust compression strategy for high-dimensional data analyses under practical constraints.
- 10 March 2025
- Published by American Physical Society (APS) in Physical Review D
- Vol. 111 (6)
Dimensionality reduction techniques for statistical inference in cosmology
- 10 March 2025
- Published by American Physical Society (APS) in Physical Review D
- Vol. 111 (6)
Dimensionality reduction techniques for statistical inference in cosmology
- 1 February 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (4) , 043501
Atacama Cosmology Telescope: The persistence of neutrino self-interaction in cosmological measurements
We use data from the Atacama Cosmology Telescope (ACT) DR4 to search for the presence of neutrino self-interaction in the cosmic microwave background. Consistent with prior works, the posterior distributions we find are bimodal, with one mode consistent with and one where neutrinos strongly self-interact. By combining ACT data with large-scale information from WMAP, we find that a delayed onset of neutrino free streaming caused by significantly strong neutrino self-interaction is compatible with these data at the level. As seen in the past, the preference shifts to with the inclusion of Planck data. We determine that the preference for strong neutrino self-interaction is largely driven by angular scales corresponding to in the ACT E-mode polarization data. This region is expected to be key to discriminate between neutrino self-interacting modes and will soon be probed with more sensitive data.- 1 February 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (4) , 043501
Atacama Cosmology Telescope: The persistence of neutrino self-interaction in cosmological measurements
We use data from the Atacama Cosmology Telescope (ACT) DR4 to search for the presence of neutrino self-interaction in the cosmic microwave background. Consistent with prior works, the posterior distributions we find are bimodal, with one mode consistent with and one where neutrinos strongly self-interact. By combining ACT data with large-scale information from WMAP, we find that a delayed onset of neutrino free streaming caused by significantly strong neutrino self-interaction is compatible with these data at the level. As seen in the past, the preference shifts to with the inclusion of Planck data. We determine that the preference for strong neutrino self-interaction is largely driven by angular scales corresponding to in the ACT E-mode polarization data. This region is expected to be key to discriminate between neutrino self-interacting modes and will soon be probed with more sensitive data.- Open Access
- 20 March 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 107 (6) , 063530
Beyond dark energy Fisher forecasts: How the Dark Energy Spectroscopic Instrument will constrain LCDM and quintessence models
We baseline with current cosmological observations to forecast the power of the Dark Energy Spectroscopic Instrument (DESI) in two ways: (1) the gain in constraining power of parameter combinations in the standard model, and (2) the reconstruction of quintessence models of dark energy. For the former task we use a recently developed formalism to extract the leading parameter combinations constrained by different combinations of cosmological survey data. For the latter, we perform a nonparametric reconstruction of quintessence using the effective field theory of dark energy. Using mock DESI observations of the Hubble parameter, angular diameter distance, and growth rate, we find that DESI will provide significant improvements over current datasets on and quintessence constraints. Including DESI mocks in our analysis improves constraints on , , and by a factor of two, where the improvement results almost entirely from the angular diameter distance and growth of structure measurements. Our quintessence reconstruction suggests that DESI will considerably improve constraints on a range of quintessence properties, such as the reconstructed potential, scalar field excursion, and the dark energy equation of state. The angular diameter distance measurements are particularly constraining in the presence of a non- signal in which the potential cannot be accounted for by shifts in and .- 20 March 2023
- Published by American Physical Society (APS) in Physical Review D
- Vol. 107 (6) , 063530
Open AccessBeyond dark energy Fisher forecasts: How the Dark Energy Spectroscopic Instrument will constrain LCDM and quintessence models
We baseline with current cosmological observations to forecast the power of the Dark Energy Spectroscopic Instrument (DESI) in two ways: (1) the gain in constraining power of parameter combinations in the standard model, and (2) the reconstruction of quintessence models of dark energy. For the former task we use a recently developed formalism to extract the leading parameter combinations constrained by different combinations of cosmological survey data. For the latter, we perform a nonparametric reconstruction of quintessence using the effective field theory of dark energy. Using mock DESI observations of the Hubble parameter, angular diameter distance, and growth rate, we find that DESI will provide significant improvements over current datasets on and quintessence constraints. Including DESI mocks in our analysis improves constraints on , , and by a factor of two, where the improvement results almost entirely from the angular diameter distance and growth of structure measurements. Our quintessence reconstruction suggests that DESI will considerably improve constraints on a range of quintessence properties, such as the reconstructed potential, scalar field excursion, and the dark energy equation of state. The angular diameter distance measurements are particularly constraining in the presence of a non- signal in which the potential cannot be accounted for by shifts in and . - 25 March 2022
- Published by American Physical Society (APS) in Physical Review D
- Vol. 105 (6) , 063529
What does a cosmological experiment really measure? Covariant posterior decomposition with normalizing flows
We present methods to rigorously extract parameter combinations that are constrained by data from posterior distributions. The standard approach uses linear methods that apply to Gaussian distributions. We show the limitations of the linear methods for current surveys and develop nonlinear methods that can be used with non-Gaussian distributions and are independent of the parameter basis. These are made possible by the use of machine-learning models, normalizing flows, to learn posterior distributions from their samples. These models allow us to obtain the local covariance of the posterior at all positions in parameter space and use its inverse, the Fisher matrix, as a local metric over parameter space. The posterior distribution can then be nonlinearly decomposed into the leading constrained parameter combinations via parallel transport in the metric space. We test our methods on two non-Gaussian, benchmark examples and then apply them to the parameter posteriors of the Dark Energy Survey and Planck CMB lensing. We illustrate how our method automatically learns the survey-specific, best constrained effective amplitude parameter for cosmic shear alone, cosmic shear and galaxy clustering, and CMB lensing. We also identify constrained parameter combinations in the full parameter space, and as an application we estimate the Hubble constant from large-structure data alone.- 25 March 2022
- Published by American Physical Society (APS) in Physical Review D
- Vol. 105 (6) , 063529
What does a cosmological experiment really measure? Covariant posterior decomposition with normalizing flows
We present methods to rigorously extract parameter combinations that are constrained by data from posterior distributions. The standard approach uses linear methods that apply to Gaussian distributions. We show the limitations of the linear methods for current surveys and develop nonlinear methods that can be used with non-Gaussian distributions and are independent of the parameter basis. These are made possible by the use of machine-learning models, normalizing flows, to learn posterior distributions from their samples. These models allow us to obtain the local covariance of the posterior at all positions in parameter space and use its inverse, the Fisher matrix, as a local metric over parameter space. The posterior distribution can then be nonlinearly decomposed into the leading constrained parameter combinations via parallel transport in the metric space. We test our methods on two non-Gaussian, benchmark examples and then apply them to the parameter posteriors of the Dark Energy Survey and Planck CMB lensing. We illustrate how our method automatically learns the survey-specific, best constrained effective amplitude parameter for cosmic shear alone, cosmic shear and galaxy clustering, and CMB lensing. We also identify constrained parameter combinations in the full parameter space, and as an application we estimate the Hubble constant from large-structure data alone.- 21 May 2021
- Published by American Physical Society (APS) in Physical Review D
- Vol. 103 (10) , 103530
Reconstructing quintessence
We present an effective-field-theory-based reconstruction of quintessence models of dark energy directly from cosmological data. We show that current cosmological data possess enough constraining power to test several quintessence model properties for redshifts with no assumptions about the behavior of the scalar field potential. We use measurements of the cosmic microwave background, supernova distances, and the clustering and lensing of galaxies to constrain the evolution of the dark energy equation of state, swampland conjectures, the shape of the scalar field reconstructed potential, and the structure of its phase space. The standard cosmological model still remains favored by data and, within quintessence models, deviations from its expansion history are bounded to be below the 10% level at 95% confidence at any redshift below .- 21 May 2021
- Published by American Physical Society (APS) in Physical Review D
- Vol. 103 (10) , 103530
Reconstructing quintessence
We present an effective-field-theory-based reconstruction of quintessence models of dark energy directly from cosmological data. We show that current cosmological data possess enough constraining power to test several quintessence model properties for redshifts with no assumptions about the behavior of the scalar field potential. We use measurements of the cosmic microwave background, supernova distances, and the clustering and lensing of galaxies to constrain the evolution of the dark energy equation of state, swampland conjectures, the shape of the scalar field reconstructed potential, and the structure of its phase space. The standard cosmological model still remains favored by data and, within quintessence models, deviations from its expansion history are bounded to be below the 10% level at 95% confidence at any redshift below .- 20 September 2019
- Published by American Physical Society (APS) in Physical Review D
- Vol. 100 (6) , 063524
or self-interacting neutrinos: How CMB data can tell the two models apartOf the many proposed extensions to the paradigm, a model in which neutrinos self-interact until close to the epoch of matter-radiation equality has been shown to provide a good fit to current cosmic microwave background (CMB) data, while at the same time alleviating tensions with late-time measurements of the expansion rate and matter fluctuation amplitude. Interestingly, CMB fits to this model either pick out a specific large value of the neutrino interaction strength, or are consistent with the extremely weak neutrino interaction found in , resulting in a bimodal posterior distribution for the neutrino self-interaction cross section. In this paper, we explore why current cosmological data select this particular large neutrino self-interaction strength, and by consequence, disfavor intermediate values of the self-interaction cross section. We show how it is the CMB temperature anisotropies, most recently measured by the Planck satellite, that produce this bimodality. We also establish that smaller scale temperature data, and improved polarization data measuring the temperature-polarization cross-correlation, will best constrain the neutrino self-interaction strength. We forecast that the upcoming Simons Observatory should be capable of distinguishing between the models.- 20 September 2019
- Published by American Physical Society (APS) in Physical Review D
- Vol. 100 (6) , 063524
or self-interacting neutrinos: How CMB data can tell the two models apartOf the many proposed extensions to the paradigm, a model in which neutrinos self-interact until close to the epoch of matter-radiation equality has been shown to provide a good fit to current cosmic microwave background (CMB) data, while at the same time alleviating tensions with late-time measurements of the expansion rate and matter fluctuation amplitude. Interestingly, CMB fits to this model either pick out a specific large value of the neutrino interaction strength, or are consistent with the extremely weak neutrino interaction found in , resulting in a bimodal posterior distribution for the neutrino self-interaction cross section. In this paper, we explore why current cosmological data select this particular large neutrino self-interaction strength, and by consequence, disfavor intermediate values of the self-interaction cross section. We show how it is the CMB temperature anisotropies, most recently measured by the Planck satellite, that produce this bimodality. We also establish that smaller scale temperature data, and improved polarization data measuring the temperature-polarization cross-correlation, will best constrain the neutrino self-interaction strength. We forecast that the upcoming Simons Observatory should be capable of distinguishing between the models.