Resource
2019 EN
David P. Bennett · Rachel Akeson · Yann Alibert
+41 more
The Kepler, K2 and TESS transit surveys are revolutionizing our understandingof planets orbiting close to their host stars and our understanding ofexoplanet systems in general, but there remains a gap in our understanding ofwide-orbit planets. This gap in our understanding must be filled if we are tounderstand planet formation and how it affects exoplanet habitability. Wesummarize current and planned exoplanet detection programs using a variety ofmethods: microlensing (including WFIRST), radial velocities, Gaia astrometry,and direct imaging. Finally, we discuss the prospects for joint analyses usingresults from multiple methods and obstacles that could hinder such analyses. We endorse the findings and recommendations published in the 2018 NationalAcademy report on Exoplanet Science Strategy. This white paper extends andcomplements the material presented therein.
Resource
2019 EN
Jennifer Yee · Rachel Akeson · Jay Anderson
+23 more
[abridged] WFIRST is uniquely capable of finding planets with masses as smallas Mars at separations comparable to Jupiter, i.e., beyond the current icelines of their stars. These planets fall between the close-in planets found byKepler and the wide separation gas giants seen by direct imaging and ice giantsinferred from ALMA observations. Furthermore, the smallest planets WFIRST candetect are smaller than the planets probed by RV and Gaia at comparableseparations. Interpreting planet populations to infer the underlying formationand evolutionary processes requires combining results from multiple detectionmethods to measure the full variation of planets as a function of planet size,orbital separation, and host star mass. Microlensing is the only way to findplanets from 0.5 to 5M_E at 1 to 5au. The case for a microlensing survey fromspace has not changed in the past 20 yrs: space allows wide-fielddiffraction-limited observations that resolve main-sequence stars in the bulge,which allows the detection and characterization of the smallest signalsincluding those from planets with masses at least as small as Mars. What haschanged is that ground-based (GB) microlensing is reaching its limits,underscoring the scientific necessity for a space-based survey. GB microlensinghas found a break in the mass-ratio distribution at about a Neptune, implyingthat these are the most common microlensing planet and that planets smallerthan this are rare. However, GB microlensing reaches its detection limits onlyslightly below the observed break. WFIRST will measure the shape of themass-ratio function below the break by finding numerous smaller planets: 500Neptunes, 600 gas giants, 200 Earths, and planets as small as 0.1M_E. Becauseit will also measure host masses and distances, WFIRST will also track thebehavior of the planet distribution as a function of separation and host starmass.
Resource
2019 EN
B. Scott Gaudi · Rachel Akeson · Jay Anderson
+21 more
The Wide Field Infrared Survey Telescope (WFIRST) will monitor $\sim 2$deg$^2$ toward the Galactic bulge in a wide ($\sim 1-2~\mu$m) W149 filter at15-minute cadence with exposure times of $\sim$50s for 6 seasons of 72 dayseach, for a total $\sim$41,000 exposures taken over $\sim$432 days, spread overthe 5-year prime mission. This will be one of the deepest exposures of the skyever taken, reaching a photon-noise photometric precision of 0.01 mag perexposure and collecting a total of $\sim 10^9$ photons over the course of thesurvey for a W149$_{\rm AB}\sim 21$ star. Of order $4 \times 10^7$ stars willbe monitored with W149$_{\rm AB}$<21, and 10$^8$ stars with W145$_{\rm AB}$<23.The WFIRST microlensing survey will detect $\sim$54,000 microlensing events, ofwhich roughly 1% ($\sim$500) will be due to isolated black holes, and $\sim$3%($\sim$1600) will be due to isolated neutron stars. It will be sensitive to(effectively) isolated compact objects with masses as low as the mass of Pluto,thereby enabling a measurement of the compact object mass function over 10orders of magnitude. Assuming photon-noise limited precision, it will detect$\sim 10^5$ transiting planets with sizes as small as $\sim 2~R_\oplus$,perform asteroseismology of $\sim 10^6$ giant stars, measure the proper motionsto $\sim 0.3\%$ and parallaxes to $\sim 10\%$ for the $\sim 6 \times 10^6$ diskand bulge stars in the survey area, and directly detect $\sim 5 \times 10^3$Trans-Neptunian objects (TNOs) with diameters down to $\sim 10$ km, as well asdetect $\sim 10^3$ occulations of stars by TNOs during the survey. All of thisscience will completely serendipitous, i.e., it will not require modificationsof the WFIRST optimal microlensing survey design. Allowing for some minordeviation from the optimal design, such as monitoring the Galactic center,would enable an even broader range of transformational science.
Resource
2019 EN
Ekta Rajput · N. K. Sahu
Controlled frames have been the subject of interest because of its ability toimprove the numerical efficiency of iterative algorithms for inverting theframe operator. In this paper, we introduce the notion of controlled $K$-framein Hilbert $C^{*}$-modules. We establish the equivalent condition forcontrolled $K$-frame. We investigate some operator theoretic characterizationsof controlled $K$-frames and controlled Bessel sequences. Moreover we establishthe relationship between the $K$-frames and controlled $K$-frames. We alsoinvestigate the invariance of a $C$-controlled $K$-frame under a suitable map$T$. At the end we prove a perturbation result for controlled$K$-frame.Controlled frames have been the subject of interest because of itsability to improve the numerical efficiency of iterative algorithms forinverting the frame operator. In this paper, we introduce the notion ofcontrolled $K$-frame in Hilbert $C^{*}$-modules. We establish the equivalentcondition for controlled $K$-frame. We investigate some operator theoreticcharacterizations of controlled $K$-frames and controlled Bessel sequences.Moreover we establish the relationship between the $K$-frames and controlled$K$-frames. We also investigate the invariance of a $C$-controlled $K$-frameunder a suitable map $T$. At the end we prove a perturbation result forcontrolled $K$-frame.
Resource
2019 EN
N. K. Sahu
To improve the numerical efficiency of iterative algorithms for inverting theframe operator, the controlled frame was introduced by Balazs et al.\cite{Balazs}, and has since been given more importance. In this paper, weintroduce the concept of controlled g-frames in Hilbert $C^{*}$-modules. Weestablish the equivalent condition for controlled $g$-frame using operatortheoretic approach. We investigate some operator theoretic characterizations ofcontrolled $g$-frames and controlled $g$-Bessel sequences. We also establishedthe relationship between $g$-frames and controlled $g$-frames in Hilbert$C^{*}$-modules. At the end we prove some perturbation results on controlled$g$-frames.
Resource
2019 EN
Brad Shutters · Jr. Timothy P. Hartke · Robert J. Sammelson
A limitation to molecular implementations of tile-based self-assembly systemsis the high rate of mismatch errors which has been observed to be between 1%and 10%. Controlling the physical conditions of the system to reduce thisintrinsic error rate $\epsilon$ prohibitively slows the growth rate of thesystem. This has motivated the development of techniques to redundantly encodeinformation in the tiles of a system in such a way that the rate of mismatcherrors in the final assembly is reduced even without a reduction in $\epsilon$.Winfree and Bekbolatov, and Chen and Goel, introduced such error-resilientsystems that reduce the mismatch error rate to $\epsilon^k$ by replacing eachtile in an error-prone system with a $k \times k$ block of tiles in theerror-resilient system, but this increases the number of tile types used by afactor of $k^2$, and the scale of the pattern produced by a factor of $k$.Reif, Sahu and Yin, and Sahu and Reif, introduced compact error-resilientsystems for the self-assembly of Boolean arrays that reduce the mismatch errorrate to $\epsilon^2$ without increasing the scale of the pattern produced. Inthis paper, we give a technique to design compact error-resilient systems forthe self-assembly of the recursively defined patterns introduced by Kautz andLathrop. We show that our compact error-resilient systems reduce the mismatcherror rate to $\epsilon^2$ by using the independent error model introduced bySahu and Reif. Surprisingly, our error-resilient systems use the same number oftile types as the error-prone system from which they are constructed.
Resource
2019 EN
Gaurav Sahu
Identifying emotion from speech is a non-trivial task pertaining to theambiguous definition of emotion itself. In this work, we adopt afeature-engineering based approach to tackle the task of speech emotionrecognition. Formalizing our problem as a multi-class classification problem,we compare the performance of two categories of models. For both, we extracteight hand-crafted features from the audio signal. In the first approach, theextracted features are used to train six traditional machine learningclassifiers, whereas the second approach is based on deep learning wherein abaseline feed-forward neural network and an LSTM-based classifier are trainedover the same features. In order to resolve ambiguity in communication, we alsoinclude features from the text domain. We report accuracy, f-score, precision,and recall for the different experiment settings we evaluated our models in.Overall, we show that lighter machine learning based models trained over a fewhand-crafted features are able to achieve performance comparable to the currentdeep learning based state-of-the-art method for emotion recognition.
Resource
2019 EN
Shashank Kotyan · Nishant Kumar · Pankaj Kumar Sahu
+1 more
Today, many of the aid systems deployed for visually impaired people aremostly made for a single purpose. Be it navigation, object detection, ordistance perceiving. Also, most of the deployed aid systems use indoornavigation which requires a pre-knowledge of the environment. These aid systemsoften fail to help visually impaired people in the unfamiliar scenario. In thispaper, we propose an aid system developed using object detection and depthperceivement to navigate a person without dashing into an object. The prototypedeveloped detects 90 different types of objects and compute their distancesfrom the user. We also, implemented a navigation feature to get input from theuser about the target destination and hence, navigate the impaired person tohis/her destination using Google Directions API. With this system, we built amulti-feature, high accuracy navigational aid system which can be deployed inthe wild and help the visually impaired people in their daily life bynavigating them effortlessly to their desired destination.
Resource
2019 EN
Shashank Kotyan · Nishant Kumar · Pankaj Kumar Sahu
+1 more
Today, many of the home automation systems deployed are mostly controlled byhumans. This control by humans restricts the automation of home appliances toan extent. Also, most of the deployed home automation systems use the Internetof Things technology to control the appliances. In this paper, we propose asystem developed using action recognition to fully automate the homeappliances. We recognize the three actions of a person (sitting, standing andlying) along with the recognition of an empty room. The accuracy of the systemwas 90% in the real-life test experiments. With this system, we remove thehuman intervention in home automation systems for controlling the homeappliances and at the same time we ensure the data privacy and reduce theenergy consumption by efficiently and optimally using home appliances.
Resource
2019 EN
Jianyu Wang · Anit Kumar Sahu · Zhouyi Yang
+2 more
This paper studies the problem of error-runtime trade-off, typicallyencountered in decentralized training based on stochastic gradient descent(SGD) using a given network. While a denser (sparser) network topology resultsin faster (slower) error convergence in terms of iterations, it incurs more(less) communication time/delay per iteration. In this paper, we proposeMATCHA, an algorithm that can achieve a win-win in this error-runtime trade-offfor any arbitrary network topology. The main idea of MATCHA is to parallelizeinter-node communication by decomposing the topology into matchings. Topreserve fast error convergence speed, it identifies and communicates morefrequently over critical links, and saves communication time by using otherlinks less frequently. Experiments on a suite of datasets and deep neuralnetworks validate the theoretical analyses and demonstrate that MATCHA takes upto $5\times$ less time than vanilla decentralized SGD to reach the sametraining loss.