arXiv subjects

Sayantan Auddy

Publications and source records attributed to Sayantan Auddy.

34 records · Page 2Linked to original sources

arXiv2021

DPNNet-2.0 Part I: Finding hidden planets from simulated images of protoplanetary disk gaps

The observed sub-structures, like annular gaps, in dust emissions from protoplanetary disk, are often interpreted as signatures of embedded planets. Fitting a model of planetary gaps to these observed features using customized simulations or empirical relations can reveal the characteristics of the hidden planets. However, customized fitting is often impractical owing to the increasing sample size and the complexity of disk-planet interaction. In this paper we introduce the architecture of DPNNet-2.0, second in the series after DPNNet \citep{aud20}, designed using a Convolutional Neural Network ( CNN, here specifically ResNet50) for predicting exoplanet masses directly from simulated images of protoplanetary disks hosting a single planet. DPNNet-2.0 additionally consists of a multi-input framework that uses both a CNN and multi-layer perceptron (a class of artificial neural network) for processing image and disk parameters simultaneously. This enables DPNNet-2.0 to be trained using images directly, with the added option of considering disk parameters (disk viscosities, disk temperatures, disk surface density profiles, dust abundances, and particle Stokes numbers) generated from disk-planet hydrodynamic simulations as inputs. This work provides the required framework and is the first step towards the use of computer vision (implementing CNN) to directly extract mass of an exoplanet from planetary gaps observed in dust-surface density maps by telescopes such as the Atacama Large (sub-)Millimeter Array.

astro-ph.EP↗

arXiv2020

How do Offline Measures for Exploration in Reinforcement Learning behave?

Sufficient exploration is paramount for the success of a reinforcement learning agent. Yet, exploration is rarely assessed in an algorithm-independent way. We compare the behavior of three data-based, offline exploration metrics described in the literature on intuitive simple distributions and highlight problems to be aware of when using them. We propose a fourth metric,uniform relative entropy, and implement it using either a k-nearest-neighbor or a nearest-neighbor-ratio estimator, highlighting that the implementation choices have a profound impact on these measures.

cs.LG↗

arXiv2020

A Machine Learning model to infer planet masses from gaps observed in protoplanetary disks

Observations of bright protoplanetary disks often show annular gaps in their dust emission. One interpretation of these gaps is disk-planet interaction. If so, fitting models of planetary gaps to observed protoplanetary disk gaps can reveal the presence of hidden planets. However, future surveys are expected to produce an ever-increasing number of protoplanetary disks with gaps. In this case, performing a customized fitting for each target becomes impractical owing to the complexity of disk-planet interaction. To this end, we introduce DPNNet (Disk Planet Neural Network), an efficient model of planetary gaps by exploiting the power of machine learning. We train a deep neural network with a large number of dusty disk-planet hydrodynamic simulations across a range of planet masses, disk temperatures, disk viscosities, disk surface density profiles, particle Stokes numbers, and dust abundances. The network can then be deployed to extract the planet mass for a given gap morphology. In this work, first in a series, we focus on the basic concepts of our machine learning framework. We demonstrate its utility by applying it to the dust gaps observed in the protoplanetary disk around HL Tau at $10$ au, $30$ au, and $80$ au. Our network predict planet masses of $80 \, M_{\rm Earth}$, $63 \, M_{\rm Earth}$, and $70 \, M_{\rm Earth}$, respectively, which are comparable to other studies based on specialized simulations. We discuss the key advantages of our DPNNet in its flexibility to incorporate new physics, any number of parameters and predictions, and its potential to ultimately replace hydrodynamical simulations for disk observers and modelers.

astro-ph.EP↗

arXiv2020

Magnetic Field Structure in Spheroidal Star-Forming Clouds. II. Estimating Field Structure from Observed Maps

This paper presents models to estimate the structure of density and magnetic field strength in spheroidal condensations, from maps of their column density and their polarization of magnetically aligned dust grains. The density model is obtained by fitting a column density map with an embedded p = 2 Plummer spheroid of any aspect ratio and inclination. The magnetic properties are based on the density model, on the Davis-Chandrasekhar-Fermi (DCF) model of Alfvénic fluctuations, and on the Spheroid Flux Freezing (SFF) model of mass and flux conservation in Paper I. The field strength model has the resolution of the column density map, which is finer than the resolution of the DCF estimate of field strength. The models are applied to ALMA observations of the envelope of the protostar BHR71 IRS1. Column density fits give the density model, from (2.0 +- 0.4) x 10^5 cm^-3 to (7 +- 1) x 10^7 cm^-3 . The density model predicts the field directions map, which fits the polarization map best within 1100 au, with standard deviation of angle differences 17°. In this region the DCF mean field strength is 0.7 +- 0.2 mG and the envelope mass is supercritical, with ratio of mass to magnetic critical mass 1.5 +- 0.4. The SFF field strength profile scales with the DCF field strength, from 60 x 10μG to 4+-1 mG. The spatial resolution of the SFF field strength estimate is finer than the DCF resolution by a factor ~7, and the peak SFF field strength exceeds the DCF field strength by a factor ~5.

astro-ph.GA↗

arXiv2019

Hierarchical Control for Bipedal Locomotion using Central Pattern Generators and Neural Networks

The complexity of bipedal locomotion may be attributed to the difficulty in synchronizing joint movements while at the same time achieving high-level objectives such as walking in a particular direction. Artificial central pattern generators (CPGs) can produce synchronized joint movements and have been used in the past for bipedal locomotion. However, most existing CPG-based approaches do not address the problem of high-level control explicitly. We propose a novel hierarchical control mechanism for bipedal locomotion where an optimized CPG network is used for joint control and a neural network acts as a high-level controller for modulating the CPG network. By separating motion generation from motion modulation, the high-level controller does not need to control individual joints directly but instead can develop to achieve a higher goal using a low-dimensional control signal. The feasibility of the hierarchical controller is demonstrated through simulation experiments using the Neuro-Inspired Companion (NICO) robot. Experimental results demonstrate the controller's ability to function even without the availability of an exact robot model.

cs.NE↗

arXiv2019

The Transition from a Lognormal to a Power-Law Column Density Distribution in Molecular Clouds: An Imprint of the Initial Magnetic Field and Turbulence

We introduce a theory for the development of a transitional column density $Σ_{\rm TP}$ between the lognormal and the power-law forms of the probability distribution function (PDF) in a molecular cloud. Our turbulent magnetohydrodynamic simulations show that the value of $Σ_{\rm TP}$ increases as the strength of both the initial magnetic field and turbulence increases. We develop an analytic expression for $Σ_{\rm TP}$ based on the interplay of turbulence, a (strong) magnetic field, and gravity. The transition value $Σ_{\rm TP}$ scales with $\mathcal{M}^2_{\rm 0}$, the square of the initial sonic Mach number, and $β_{0}$, the initial ratio of gas pressure to magnetic pressure. We fit the variation of $Σ_{\rm TP}$ among different model clouds as a function of $\mathcal{M}^2_{\rm 0} β_{0}$, or equivalently the square of the initial Alfvénic Mach number $\mathcal{M}^2_{\rm A0}$. This implies that the transition value $Σ_{\rm TP}$ is an imprint of cloud initial conditions and is set by turbulent compression of a magnetic cloud. Physically, the value of $Σ_{\rm TP}$ denotes the boundary above which the mass-to-flux ratio becomes supercritical and gravity drives the evolution.

astro-ph.SR↗

arXiv2019

Magnetic Field Structure of Dense Cores using Spectroscopic Methods

We develop a new ''core field structure'' (CFS) model to predict the magnetic field strength and magnetic field fluctuation profile of dense cores using gas kinematics. We use spatially resolved observations of the nonthermal velocity dispersion from the Green Bank Ammonia survey along with column density maps from SCUBA-2 to estimate the magnetic field strength across seven dense cores located in the L1688 region of Ophiuchus. The CFS model predicts the profile of the relative field fluctuation, which is related to the observable dispersion in direction of the polarization vectors. Within the context of our model we find that all the cores have a transcritical mass-to-flux ratio.

astro-ph.GA↗

arXiv2018

Magnetic Field Structure in Spheroidal Star-Forming Clouds

A model of magnetic field structure is presented to help test the prevalence of flux freezing in star-forming clouds of various shapes, orientations, and degrees of central concentration, and to estimate their magnetic field strength. The model is based on weak-field flux freezing in centrally condensed Plummer spheres and spheroids of oblate and prolate shape. For a spheroid of given density contrast, aspect ratio, and inclination, the model estimates the local field strength and direction, and the global field pattern of hourglass shape. Comparisons with a polarization simulation indicate typical angle agreement within 1 - 10 degrees. Scalable analytic expressions are given to match observed polarization patterns, and to provide inputs to radiative transfer codes for more accurate predictions. The model may apply to polarization observations of dense cores, elongated filamentary clouds, and magnetized circumstellar disks.

astro-ph.GA↗

arXiv2017

From Molecular Clouds to the IMF: Spatial and Temporal Effects

We review star formation in molecular clouds and describe why magnetic fields may be important and how they can influence filamentary structure and the column density probability distribution function (PDF). We also comment on the origin of the stellar and substellar initial mass function (IMF), which may require explanations beyond a simple Jeans length argument in turbulent molecular clouds. A mathematical model of the modified lognormal power-law (MLP) distribution function provides a framework within which to connect accretion processes with the IMF.

astro-ph.GA↗

arXiv2017

The Effect of Magnetic Fields and Ambipolar Diffusion on the Column Density Probability Distribution Function in Molecular Clouds

Simulations generally show that non-self-gravitating clouds have a lognormal column density ($Σ$) probability distribution function (PDF), while self-gravitating clouds with active star formation develop a distinct power-law tail at high column density. Although the growth of the power law can be attributed to gravitational contraction leading to the formation of condensed cores, it is often debated if an observed lognormal shape is a direct consequence of supersonic turbulence alone, or even if it is really observed in molecular clouds. In this paper we run three-dimensional magnetohydrodynamic simulations including ambipolar diffusion with different initial conditions to see the effect of strong magnetic fields and nonlinear initial velocity perturbations on the evolution of the column density PDFs. Our simulations show that column density PDFs of clouds with supercritical mass-to-flux ratio, with either linear perturbations or nonlinear turbulence, quickly develop a power-law tail such that $dN/d \log Σ\propto Σ^{-α}$ with index $α\simeq 2$. Interestingly, clouds with subcritical mass-to-flux ratio also proceed directly to a power-law PDF, but with a much steeper index $α\simeq 4$. This is a result of gravitationally-driven ambipolar diffusion. However, for nonlinear perturbations with a turbulent spectrum ($v_{k}^{2} \propto k^{-4}$), the column density PDFs of subcritical clouds do retain a lognormal shape for a major part of the cloud evolution, and only develop a distinct power-law tail with index $α\simeq 2$ at greater column density when supercritical pockets are formed.

astro-ph.GA↗

arXiv2017

Magnetic Ribbons: A Minimum Hypothesis Model for Filaments

We develop a magnetic ribbon model for molecular cloud filaments. These result from turbulent compression in a molecular cloud in which the background magnetic field sets a preferred direction. We use our model to calculate a synthetic observed relation between apparent width in projection versus observed column density. The relationship is relatively flat, in rough agreement with the observations, and unlike the simple expectation based on a Jeans length argument.

astro-ph.GA↗

arXiv2017

The Anomalous Magnetic Moment of a Photon Propagating in a Magnetic Field

We analyze the spectrum of the Hamiltonian of a photon propagating in a strong magnetic field $B\sim B_{\rm{cr}}$, where $B_{\rm cr}= \frac{m^2}{e} \simeq 4.4 \times 10^{13}$ Gauss is the Schwinger critical field . We show that the expected value of the Hamiltonian of a quantized photon for a perpendicular mode is a concave function of the magnetic field $B$. We show by a partially analytic and numerical method that the anomalous magnetic moment of a photon in the one loop approximation is a non - decreasing function of the magnetic field $B$ in the range $0\leq B \leq 30 \, B_{\rm cr}$ We provide a numerical representation of the expression for the anomalous magnetic moment in terms of special functions. We find that the anomalous magnetic moment $μ_γ$ of a photon for $B=30\, B_{\rm cr }$ is $8/3$ of the anomalous magnetic moment of a photon for $B = 1/2 ~ B_{\rm cr}$.

quant-ph↗

arXiv2016

A Magnetic Ribbon Model for Star-Forming Filaments

We develop a magnetic ribbon model for molecular cloud filaments. These result from turbulent compression in a molecular cloud in which the background magnetic field sets a preferred direction. We argue that this is a natural model for filaments and is based on the interplay between turbulence, strong magnetic fields, and gravitationally-driven ambipolar diffusion, rather than pure gravity and thermal pressure. An analytic model for the formation of magnetic ribbons that is based on numerical simulations is used to derive a lateral width of a magnetic ribbon. This differs from the thickness along the magnetic field direction, which is essentially the Jeans scale. We use our model to calculate a synthetic observed relation between apparent width in projection versus observed column density. The relationship is relatively flat, similar to observations, and unlike the simple expectation based on a Jeans length argument.

astro-ph.GA↗

arXiv2016

Analytic Models of Brown Dwarfs and The Substellar Mass Limit

We present the current status of the analytic theory of brown dwarf evolution and the lower mass limit of the hydrogen burning main sequence stars. In the spirit of a simplified analytic theory we also introduce some modifications to the existing models. We give an exact expression for the pressure of an ideal non-relativistic Fermi gas at a finite temperature, therefore allowing for non-zero values of the degeneracy parameter ($ψ= \frac{kT}{μ_{F}}$, where $μ_{F}$ is the Fermi energy). We review the derivation of surface luminosity using an entropy matching condition and the first-order phase transition between the molecular hydrogen in the outer envelope and the partially-ionized hydrogen in the inner region. We also discuss the results of modern simulations of the plasma phase transition, which illustrate the uncertainties in determining its critical temperature. Based on the existing models and with some simple modification we find the maximum mass for a brown dwarf to be in the range $0.064M_\odot-0.087M_\odot$. An analytic formula for the luminosity evolution allows us to estimate the time period of the non-steady state (i.e., non-main sequence) nuclear burning for substellar objects. Standard models also predict that stars that are just above the substellar mass limit can reach an extremely low luminosity main sequence after at least a few million years of evolution, and sometimes much longer. We estimate that $\simeq 11 \%$ of stars take longer than $10^7$ yr to reach the main-sequence, and $\simeq 5 \%$ of stars take longer than $10^8$ yr.

astro-ph.SR↗

arXiv2015

Gravitational Wave Background in the Quasi-Steady State Cosmology

This paper calculates the expected gravitational wave background (GWB) in the quasi-steady state cosmology (QSSC). The principal sources of gravitational waves in the QSSC are the minicreation events (MCE). With suitable assumptions the GWB can be computed both numerically and with analytical methods. It is argued that the GWB in QSSC differs from that predicted for the standard cosmology and a future technology of detectors will be able to decide between the two predictions. We also derive a formula for the flux density of a typical extragalactic source of gravitational waves.

astro-ph.CO↗

arXiv2015

The MLP Distribution: A Modified Lognormal Power-Law Model for the Stellar Initial Mass Function

This work explores the mathematical properties of a distribution introduced by Basu & Jones (2004), and applies it to model the stellar initial mass function (IMF). The distribution arises simply from an initial lognormal distribution, requiring that each object in it subsequently undergoes exponential growth but with an exponential distribution of growth lifetimes. This leads to a modified lognormal with a power-law tail (MLP) distribution, which can in fact be applied to a wide range of fields where distributions are observed to have a lognormal-like body and a power-law tail. We derive important properties of the MLP distribution, like the cumulative distribution, the mean, variance, arbitrary raw moments, and a random number generator. These analytic properties of the distribution can be used to facilitate application to modeling the IMF. We demonstrate how the MLP function provides an excellent fit to the IMF compiled by Chabrier (2005) and how this fit can be used to quickly identify quantities like the mean, median, and mode, as well as number and mass fractions in different mass intervals.

astro-ph.SR↗