Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userCosmicRIM : Reconstructing Early Universe by Combining Differentiable Simulations with Recurrent Inference Machines
73
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
Reconstructing the Gaussian initial conditions at the beginning of the Universe from the survey data in a forward modeling framework is a major challenge in cosmology. This requires solving a high dimensional inverse problem with an expensive, non-linear forward model: a cosmological N-body simulation. While intractable until recently, we propose to solve this inference problem using an automatically differentiable N-body solver, combined with a recurrent networks to learn the inference scheme and obtain the maximum-a-posteriori (MAP) estimate of the initial conditions of the Universe. We demonstrate using realistic cosmological observables that learnt inference is 40 times faster than traditional algorithms such as ADAM and LBFGS, which require specialized annealing schemes, and obtains solution of higher quality.
rate research
Read More
We test the mutual consistency between the baryon acoustic oscillation measurements from the eBOSS SDSS final release, as well as the Pantheon supernova compilation in a model independent fashion using Gaussian process regression. We also test their joint consistency with the $Lambda$CDM model, also in a model independent fashion. We also use Gaussian process regression to reconstruct the expansion history that is preferred by these two datasets. While this methodology finds no significant preference for model flexibility beyond $Lambda$CDM, we are able to generate a number of reconstructed expansion histories that fit the data better than the best-fit $Lambda$CDM model. These example expansion histories may point the way towards modifications to $Lambda$CDM. We also constrain the parameters $Omega_k$ and $H_0r_d$ both with $Lambda$CDM and with Gaussian process regression. We find that $H_0r_d =10030 pm 130$ km/s and $Omega_k = 0.05 pm 0.10$ for $Lambda$CDM and that $H_0r_d = 10040 pm 140$ km/s and $Omega_k = 0.02 pm 0.20$ for the Gaussian process case.
We perform direct numerical simulations of magnetohydrodynamic turbulence in the early universe and numerically compute the resulting stochastic background of gravitational waves and relic magnetic fields. These simulations do not make the simplifying assumptions of earlier analytic work. If the turbulence is assumed to have an energy-carrying scale that is about a hundredth of the Hubble radius at the time of generation, as expected in a first-order phase transition, the peak of gravitational wave power will be in the mHz frequency range for a signal produced at the electroweak scale. The efficiency of gravitational wave (GW) production varies significantly with how the turbulence is driven. Detectability of turbulence at the electroweak scale by the planned Laser Interferometer Space Antenna (LISA) requires anywhere from 0.1% to 10% of the thermal plasma energy density to be in plasma motions or magnetic fields, depending on the model of the driving process. Our results predict a new universal form below the spectral peak frequency that is shallower than previously thought. This implies larger values of the GW energy spectra in the low-frequency range. This extends the range where turbulence is detectable with LISA to lower frequencies, corresponding to higher energy scales than the assumed energy-carrying scale.
Objective: To allow efficient learning using the Recurrent Inference Machine (RIM) for image reconstruction whereas not being strictly dependent on the training data distribution so that unseen modalities and pathologies are still accurately recovered. Methods: Theoretically, the RIM learns to solve the inverse problem of accelerated-MRI reconstruction whereas being robust to variable imaging conditions. The efficiency and generalization capabilities with different training datasets were studied, as well as recurrent network units with decreasing complexity: the Gated Recurrent Unit (GRU), the Minimal Gated Unit (MGU), and the Independently Recurrent Neural Network (IndRNN), to reduce inference times. Validation was performed against Compressed Sensing (CS) and further assessed based on data unseen during training. A pathology study was conducted by reconstructing simulated white matter lesions and prospectively undersampled data of a Multiple Sclerosis patient. Results: Training on a single modality of 3T $T_1$-weighted brain data appeared sufficient to also reconstruct 7T $T_{2}^*$-weighted brain and 3T $T_2$-weighted knee data. The IndRNN is an efficient recurrent unit, reducing inference time by 68% compared to CS, whereas maintaining performance. The RIM was able to reconstruct lesions unseen during training more accurately than CS when trained on $T_2$-weighted knee data. Training on $T_1$-weighted brain data and on combined data slightly enhanced the signal compared to CS. Conclusion: The RIM is efficient when decreasing its complexity, which reduces the inference time, whereas still being able to reconstruct data and pathology that was unseen during training.
The Integrated Sachs-Wolfe (ISW) effect is a large-angle modulation of the cosmic microwave background (CMB), generated when CMB photons traverse evolving potential wells associated with large scale structure (LSS). Recent efforts have been made to reconstruct maps of the ISW signal using information from surveys of galaxies and other LSS tracers, but investigation into how survey systematics affect their reliability has so far been limited. Using simulated ISW and LSS maps, we study the impact of galaxy survey properties and systematic errors on the accuracy of reconstructed ISW signal. We find that systematics that affect the observed distribution of galaxies along the line of sight, such as photo-z and bias-evolution related errors, have a relatively minor impact on reconstruction quality. In contrast, however, we find that direction-dependent calibration errors can be very harmful. Specifically, we find that in order to avoid significant degradation of our reconstruction quality statistics, direction-dependent number density fluctuations due to systematics must be controlled so that their variance is smaller than $10^{-6}$ (which corresponds to a 0.1% calibration). Additionally, we explore the implications of our results for attempts to use reconstructed ISW maps to shed light on the origin of large-angle CMB alignments. We find that there is only a weak correlation between the true and reconstructed angular momentum dispersion, which quantifies alignment, even for reconstructed ISW maps which are fairly accurate overall.
With the Swift detection of GRB090423 at z = 8.2, it was confirmed that GRBs are now detectable at (significantly) larger redshifts than AGN, and so can indeed be used as probes of the Early Universe. The proposed Energetic X-ray Imaging Survey Telescope (EXIST) mission has been designed to detect and promptly measure redshifts and both soft X-ray (0.1 - 10 keV) and simultaneous nUV-nIR (0.3 - 2.3microns) imaging and spectra for GRBs out to redshifts z ~18, which encompasses (or even exceeds) current estimates for Pop III stars that are expected to be massive and possibly GRB sources. Scaling from Swift for the ~10X greater sensitivity of EXIST, more than 100 GRBs at z >=8 may be detected and would provide direct constraints on the formation and evolution of the first stars and galaxies. For GRBs at redshifts z >= 8, with Lyman breaks at greater than 1.12microns, spectra at resolution R = 30 or R = 3000 for afterglows with AB magnitudes brighter than 24 or 20 (respectively) within ~3000sec of trigger will directly probe the Epoch of Reionization, formation of galaxies, and cosmic star formation rate. The proposed EXIST mission can probe these questions, and many others, given its unparalleled combination of sensitivity and spatial-spectral-temporal coverage and resolution. Here we provide an overview of the key science objectives for GRBs as probes of the early Universe and of extreme physics, and the mission plan and technical readiness to bring this to EXIST.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
