arXiv subjects

Richard Galvez

Publications and source records attributed to Richard Galvez.

16 recordsLinked to original sources

arXiv2024

A Foundation Model for the Solar Dynamics Observatory

SDO-FM is a foundation model using data from NASA's Solar Dynamics Observatory (SDO) spacecraft; integrating three separate instruments to encapsulate the Sun's complex physical interactions into a multi-modal embedding space. This model can be used to streamline scientific investigations involving SDO by making the enormous datasets more computationally accessible for heliophysics research and enable investigations that require instrument fusion. We discuss four key components: an ingestion pipeline to create machine learning ready datasets, the model architecture and training approach, resultant embeddings and fine-tunable models, and finally downstream fine-tuned applications. A key component of this effort has been to include subject matter specialists at each stage of development; reviewing the scientific value and providing guidance for model architecture, dataset, and training paradigm decisions. This paper marks release of our pretrained models and embedding datasets, available to the community on Hugging Face and sdofm.org.

astro-ph.SR↗

arXiv2024

Virtual EVE: a Deep Learning Model for Solar Irradiance Prediction

Understanding space weather is vital for the protection of our terrestrial and space infrastructure. In order to predict space weather accurately, large amounts of data are required, particularly in the extreme ultraviolet (EUV) spectrum. An exquisite source of information for such data is provided by the Solar Dynamic Observatory (SDO), which has been gathering solar measurements for the past 13 years. However, after a malfunction in 2014 affecting the onboard Multiple EUV Grating Spectrograph A (MEGS-A) instrument, the scientific output in terms of EUV measurements has been significantly degraded. Building upon existing research, we propose to utilize deep learning for the virtualization of the defective instrument. Our architecture features a linear component and a convolutional neural network (CNN) -- with EfficientNet as a backbone. The architecture utilizes as input grayscale images of the Sun at multiple frequencies -- provided by the Atmospheric Imaging Assembly (AIA) -- as well as solar magnetograms produced by the Helioseismic and Magnetic Imager (HMI). Our findings highlight how AIA data are all that is needed for accurate predictions of solar irradiance. Additionally, our model constitutes an improvement with respect to the state-of-the-art in the field, further promoting the idea of deep learning as a viable option for the virtualization of scientific instruments.

astro-ph.SR↗

arXiv2019

Modeling the Gaia Color-Magnitude Diagram with Bayesian Neural Flows to Constrain Distance Estimates

We demonstrate an algorithm for learning a flexible color-magnitude diagram from noisy parallax and photometry measurements using a normalizing flow, a deep neural network capable of learning an arbitrary multi-dimensional probability distribution. We present a catalog of 640M photometric distance posteriors to nearby stars derived from this data-driven model using Gaia DR2 photometry and parallaxes. Dust estimation and dereddening is done iteratively inside the model and without prior distance information, using the Bayestar map. The signal-to-noise (precision) of distance measurements improves on average by more than 48% over the raw Gaia data, and we also demonstrate how the accuracy of distances have improved over other models, especially in the noisy-parallax regime. Applications are discussed, including significantly improved Milky Way disk separation and substructure detection. We conclude with a discussion of future work, which exploits the normalizing flow architecture to allow us to exactly marginalize over missing photometry, enabling the inclusion of many surveys without losing coverage.

astro-ph.IM↗

arXiv2019

A Recommendation Algorithm to Predict Giant Exoplanet Host Stars Using Stellar Elemental Abundances

The presence of certain elements within a star, and by extension its planet, strongly impacts the formation and evolution of the planetary system. The positive correlation between a host star's iron-content and the presence of an orbiting giant exoplanet has been confirmed; however, the importance of other elements in predicting giant planet occurrence is less certain despite their central role in shaping internal planetary structure. We designed and applied a machine learning algorithm to the Hypatia Catalog (Hinkel et a. 2014) to analyze the stellar abundance patterns of known host stars to determine those elements important in identifying potential giant exoplanet host stars. We analyzed a variety of different elements ensembles, namely volatiles, lithophiles, siderophiles, and Fe. We show that the relative abundances of oxygen, carbon, and sodium, in addition to iron, are influential indicators of the presence of a giant planet. We demonstrate the predictive power of our algorithm by analyzing stars with known giant planets and found that they had median 75% prediction score. We present a list of ~350 stars with no currently discovered planets that have a $\geq$90% prediction probability likelihood of hosting a giant exoplanet. We investigated archival HARPS data and found significant trends that HIP62345, HIP71803, and HIP10278 host long-period giant planet companions with estimated minimum $M_p\sin(i)$ values of 3.7, 6.8, and 8.5 M$_{J}$, respectively. We anticipate that our findings will revolutionize future target selection, the role that elements play in giant planet formation, and the determination of giant planet interior structure models.

astro-ph.EP↗

arXiv2019

A Machine Learning Dataset Prepared From the NASA Solar Dynamics Observatory Mission

In this paper we present a curated dataset from the NASA Solar Dynamics Observatory (SDO) mission in a format suitable for machine learning research. Beginning from level 1 scientific products we have processed various instrumental corrections, downsampled to manageable spatial and temporal resolutions, and synchronized observations spatially and temporally. We illustrate the use of this dataset with two example applications: forecasting future EVE irradiance from present EVE irradiance and translating HMI observations into AIA observations. For each application we provide metrics and baselines for future model comparison. We anticipate this curated dataset will facilitate machine learning research in heliophysics and the physical sciences generally, increasing the scientific return of the SDO mission. This work is a direct result of the 2018 NASA Frontier Development Laboratory Program. Please see the appendix for access to the dataset.

astro-ph.SR↗

arXiv2017

Cosmology with Independently Varying Neutrino Temperature and Number

We consider Big Bang nucleosynthesis and the cosmic microwave background when both the neutrino temperature and neutrino number are allowed to vary from their standard values. The neutrino temperature is assumed to differ from its standard model value by a fixed factor from Big Bang nucleosynthesis up to the present. In this scenario, the effective number of relativistic degrees of freedom, $N_{\rm eff}^{\rm CMB}$, derived from observations of the cosmic microwave background is not equal to the true number of neutrinos, $N_ν$. We determine the element abundances predicted by Big Bang nucleosynthesis as a function of the neutrino number and temperature, converting the latter to the equivalent value of $N_{\rm eff}^{\rm CMB}$. We find that a value of $N_{\rm eff}^{\rm CMB} \approx 3$ can be made consistent with $N_ν= 4$ with a decrease in the neutrino temperature of $\sim 5\%$, while $N_ν= 5$ is excluded for any value of $N_{\rm eff}^{\rm CMB}$. No observationally-allowed values for $N_{\rm eff}^{\rm CMB}$ and $N_ν$ can solve the lithium problem.

astro-ph.CO↗

arXiv2016

Kahler Moduli Inflation in Type IIB Compactifications: A random tumble through the Calabi-Yau landscape

In this paper we present an initial exploration of the Calabi-Yau landscape in the context of Kahler moduli inflation. We review how the slow-roll requirement on the scalar potential translates to a geometric constraint on the Kahler geometry of the vacuum. This constraint leads to a hard bound on the moduli space geometry and we consider the effects of this constraint on the string landscape that arises in type IIB string compactifications on an O3/O7 orientifold. Most notably we find that the inflationary constraint is independent of the moduli space dimension and only 6.57% of geometries inspected support high-scale Kahler moduli inflation.

hep-th↗

arXiv2013

Supersymmetry, Nonthermal Dark Matter and Precision Cosmology

Within the Minimal Supersymmetric Standard Model (MSSM), LHC bounds suggest that scalar superpartner masses are far above the electroweak scale. Given a high superpartner mass, nonthermal dark matter is a viable alternative to WIMP dark matter generated via freezeout. In the presence of moduli fields nonthermal dark matter production is associated with a long matter dominated phase, modifying the spectral index and primordial tensor amplitude relative to those in a thermalized primordial universe. Nonthermal dark matter can have a higher self-interaction cross-section than its thermal counterpart, enhancing astrophysical bounds on its annihilation signals. We constrain the contributions to the neutralino mass from the bino, wino and higgsino using existing astrophysical bounds and direct detection experiments for models with nonthermal neutralino dark matter. Using these constraints we quantify the expected change to inflationary observables resulting from the nonthermal phase.

hep-ph↗

arXiv2012

Phase Structure of Lattice N=4 Super Yang-Mills

We make a first study of the phase diagram of four-dimensional N=4 super Yang-Mills theory regulated on a space-time lattice. The lattice formulation we employ is both gauge invariant and retains at all lattice spacings one exactly preserved supersymmetry charge. Our numerical results are consistent with the existence of a single deconfined phase at all observed values of the bare coupling.

hep-lat↗

arXiv2012

Non-abelian gauged NJL models on the lattice

We use Monte Carlo simulation to probe the phase structure of a SU(2) gauge theory containing $N_f$ Dirac fermion flavors transforming in the fundamental representation of the group and interacting through an additional four fermion term. Pairs of physical flavors are implemented using the two tastes present in a reduced staggered fermion formulation of the theory. The resultant lattice theory is invariant under a set of shift symmetries which correspond to a discrete subgroup of the continuum chiral-flavor symmetry. The pseudoreal character of the representation guarantees that the theory has no sign problem. For the case of $N_f=4$ we observe a crossover in the behavior of the chiral condensate for strong four fermi coupling associated with the generation of a dynamical mass for the fermions. At weak gauge coupling this crossover is consistent with the usual continuous phase transition seen in the pure (ungauged) NJL model. However, if the gauge coupling is strong enough to cause confinement we observe a much more rapid crossover in the chiral condensate consistent with a first order phase transition

hep-lat↗

arXiv2012

Investigating the sign problem for two-dimensional $\mathcal{N}=(2,2)$ and $\mathcal{N}=(8,8)$ lattice super Yang--Mills theories

Recently there has been some controversy in the literature concerning the existence of a fermion sign problem in the $\mathcal{N}=(2,2)$ supersymmetric Yang--Mills (SYM) theories on the lattice. In this work, we address this issue by conducting Monte Carlo simulations not only for $\mathcal{N}=(2,2)$ but also for $\mathcal{N}=(8,8)$ SYM in two dimensions for the U(N) theories with N=2, using the new ideas derived from topological twisting followed by geometric discretization. Our results from simulations provide the evidence that these theories do {\it not} suffer from a sign problem as the continuum limit is approached. These results thus boost confidence that these new lattice formulations can be used successfully to explore the nonperturbative aspects of the four-dimensional $\mathcal{N}=4$ SYM theory.

hep-lat↗

arXiv2011

Supersymmetric gauge theories on the lattice: Pfaffian phases and the Neuberger 0/0 problem

Recently a class of supersymmetric gauge theories have been successfully implemented on the lattice. However, there has been an ongoing debate on whether lattice versions of some of these theories suffer from a sign problem, with independent simulations for the ${\cal N} = (2, 2)$ supersymmetric Yang-Mills theories in two dimensions yielding seemingly contradictory results. Here, we address this issue from an interesting theoretical point of view. We conjecture that the sign problem observed in some of the simulations is related to the so called Neuberger 0/0 problem, which arises in ordinary non-supersymmetric lattice gauge theories, and prevents the realization of Becchi-Rouet-Stora-Tyutin symmetry on the lattice. After discussing why we expect a sign problem in certain classes of supersymmetric lattice gauge theories far from the continuum limit, we argue that these problems can be evaded by use of a non-compact parametrization of the gauge link fields.

hep-lat↗

arXiv2011

On the sign problem in 2D lattice super Yang--Mills

In recent years a new class of supersymmetric lattice theories have been proposed which retain one or more exact supersymmetries for non-zero lattice spacing. Recently there has been some controversy in the literature concerning whether these theories suffer from a sign problem. In this paper we address this issue by conducting simulations of the N=(2, 2) and N=(8, 8) supersymmetric Yang--Mills theories in two dimensions for the U(N) theories with N=2,3,4, using the new twisted lattice formulations. Our results provide evidence that these theories do not suffer from a sign problem in the continuum limit. These results thus boost confidence that the new lattice formulations can be used successfully to explore non-perturbative aspects of four-dimensional N=4 supersymmetric Yang--Mills theory.

hep-lat↗

arXiv2011

Accelerating the solution of families of shifted linear systems with CUDA

We describe the GPU implementation of shifted or multimass iterative solvers for sparse linear systems of the sort encountered in lattice gauge theory. We provide a generic tool that can be used by those without GPU programming experience to accelerate the simulation of a wide array of theories. We stress genericity, which is important to allow the simulation of candidate theories for new physics at LHC, and for the study of various supersymmetric theories. We find significant speed ups, which we conservatively bound below at at least twelve times, that promise to put a variety of research questions within practical reach.

hep-lat↗

arXiv2010

Realization of Center Symmetry in Two Adjoint Flavor Large-N Yang-Mills

We report on the results of numerical simulations of $SU(N)$ lattice Yang Mills with two flavors of (light) Wilson fermion in the adjoint representation. We analytically and numerically address the question of center symmetry realization on lattices with $Γ$ sites in each direction in the large-$N$ limit. We show, by a weak coupling calculation that, for massless fermions, center symmetry realization is independent of $Γ$, and is unbroken. Then, we extend our result by conducting simulations at non zero mass and finite gauge coupling. Our results indicate that center symmetry is intact for a range of fermion mass in the vicinity of the critical line on lattices of volume $2^4$. This observation makes it possible to compute infinite volume physical observables using small volume simulations in the limit $N\to\infty$, with possible applications to the determination of the conformal window in gauge theories with adjoint fermions.

hep-lat↗

arXiv2009

Large N QCD in two dimensions with a baryonic chemical potential

We consider large N gauge theory on a two dimensional lattice in the presence of a baryonic chemical potential. We work with one copy of naive fermion and argue that reduction holds even in the presence of a chemical potential. Analytical arguments supported by numerical studies show that there is no phase transition as a function of the baryonic chemical potential.

hep-lat↗