Analysis of Fermi GRB T 90 distribution
|
|
- Neal Richards
- 5 years ago
- Views:
Transcription
1 Analysis of Fermi GRB T 90 distribution Mariusz Tarnopolski Astronomical Observatory Jagiellonian University 23 July 2015 Cosmology School, Kielce Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
2 Presentation plan 1 Introduction and overview Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
3 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
4 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs χ 2 tting Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
5 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs χ 2 tting Maximum log-likelihood tting Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
6 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs χ 2 tting Maximum log-likelihood tting 3 Hurst Exponents (HEs) & Machine Learning (ML) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
7 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs χ 2 tting Maximum log-likelihood tting 3 Hurst Exponents (HEs) & Machine Learning (ML) 4 On the limit between short and long GRBs Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
8 Presentation plan 1 Introduction and overview 2 T 90 distributions of Fermi GRBs χ 2 tting Maximum log-likelihood tting 3 Hurst Exponents (HEs) & Machine Learning (ML) 4 On the limit between short and long GRBs 5 Conclusions Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
9 Satellites Introduction and overview CGRO/BATSE Swift/BAT RHESSI BeppoSAX/GRBM Fermi/GBM HETE-2/FREGATE INTEGRAL/SPI-ACS SUZAKU Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
10 Introduction and overview KONUS (Mazets et al. 1981) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
11 BATSE 1B Introduction and overview Kouveliotou et al. (1993) tted a quadratic function between the two peaks of 222 GRBs and determined its minimum to be at (1.2 ± 0.4) s, which rounded of to the next integer bin edge, is 2.0 s. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
12 Introduction and overview BATSE 3B (Horváth 1998) Counts logt 90 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
13 Introduction and overview BATSE 4B (current) (Horváth 2002) 0.7 PDF MLE chi2 Horvath logt 90 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
14 Introduction and overview Swift (Horváth et al. 2008) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
15 Introduction and overview Swift (Huja et al. 2009) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
16 Introduction and overview Swift (Huja & ípa 2009) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
17 Introduction and overview RHESSI ( ípa et al. 2009) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
18 Introduction and overview BeppoSAX (Horváth 2009) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
19 T90 distributions of Fermi GRBs χ 2 tting A mixture of Gaussians: f k = k A i N i (µ i, σ 2) i i=1 f k = k i=1 A 2πσi i exp ( (x µ i ) 2 2σ 2 i ) is tted to a histogram of log (T 90 ). A signicance level of α = 0.05 is adopted; 25 binnings are applied, dened by the bin widths w from 0.30 to 0.06 with a step of The corresponding number of bins range from 15 to 69. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
20 T90 distributions of Fermi GRBs χ 2 tting w A mixture of Gaussians: f k = k A i N i (µ i, σ 2) i i=1 f k = k i=1 A 2πσi i exp ( (x µ i ) 2 2σ 2 i is tted to a histogram of log (T 90 ). A signicance level of α = 0.05 is adopted; 25 binnings are applied, dened by the bin widths w from 0.30 to 0.06 with a step of The corresponding number of bins range from 15 to 69. ) w w w w Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41 logt 90
21 T90 distributions of Fermi GRBs χ 2 tting Table 1: Parameters of a two-gaussian t w i µ i σ i A i χ 2 p-val Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
22 T90 distributions of Fermi GRBs χ 2 tting 300 w w w w w w w w w w logt 90 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41 logt 90
23 T90 distributions of Fermi GRBs χ 2 tting Table 2: Parameters of a three-gaussian t w i µ i σ i A i χ 2 p-val Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
24 T90 distributions of Fermi GRBs χ 2 tting χ 2 = χ 2 1 χ 2 2. = χ 2 ( ν) Table 3: Improvements of a three-gaussian over a two-gaussian t w χ 2 p-value Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
25 T90 distributions of Fermi GRBs χ 2 tting 2.5 BATSE 3B BATSE 4B Swift Swift Swfit RHESSI X Fermi logt Horváth 1998 Horváth 2002 Huja et al ípa et al Horváth 2009 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
26 Results 1 T90 distributions of Fermi GRBs χ 2 tting T 90 distribution of Fermi GRBs is bimodal no evidence for a (phenomenological) third (intermediate) class Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
27 T90 distributions of Fermi GRBs Maximum log-likelihood tting It feels like a waste of data to bin 1600 events into a few dozens of bins. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
28 T90 distributions of Fermi GRBs Maximum log-likelihood tting It feels like a waste of data to bin 1600 events into a few dozens of bins. Having a distribution with a PDF given by f = f (x; θ) (possibly a mixture), where θ = {θ i } p is a set of parameters, the log-likelihood function is i=1 dened as L = N i=1 log f (x i ; θ), where {x i } N i=1 are the datapoints from the sample to which a distribution is tted. The tting is performed by searching a set of parameters θ for which the log-likelihood L is maximized. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
29 T90 distributions of Fermi GRBs Maximum log-likelihood tting For nested as well as non-nested models, the Akaike information criterion (AIC ) may be applied. The AIC is dened as AIC = 2p 2L p. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
30 T90 distributions of Fermi GRBs Maximum log-likelihood tting For nested as well as non-nested models, the Akaike information criterion (AIC ) may be applied. The AIC is dened as AIC = 2p 2L p. A preferred model is the one that minimizes AIC. The formulation of AIC penalizes the use of an overly excessive number of parameters, hence discourages overtting. Among candidate models with AIC i, let AIC min denote the smallest. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
31 T90 distributions of Fermi GRBs Maximum log-likelihood tting For nested as well as non-nested models, the Akaike information criterion (AIC ) may be applied. The AIC is dened as AIC = 2p 2L p. A preferred model is the one that minimizes AIC. The formulation of AIC penalizes the use of an overly excessive number of parameters, hence discourages overtting. Among candidate models with AIC i, let AIC min denote the smallest. Then, Pr i = exp ( i 2 ), where i = AIC min AIC i, can be interpreted as the relative (compared to AIC min ) probability that the i-th model minimizes the AIC. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
32 T90 distributions of Fermi GRBs Maximum log-likelihood tting A mixture of k standard normal (Gaussian) N (µ, σ 2 ) distributions: f (N ) k (x) = k i=1 A i ϕ ( x µ i σ i ) = k i=1 A i 2πσi exp ( (x µ i) 2 2σ 2 i ) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
33 T90 distributions of Fermi GRBs Maximum log-likelihood tting A mixture of k standard normal (Gaussian) N (µ, σ 2 ) distributions: f (N ) k (x) = k i=1 A i ϕ ( x µ i σ i ) = k i=1 A mixture of k skew normal (SN) distributions: f (SN) k (x) = k i=1 A i 2πσi exp ( (x µ i) 2 2A i ϕ ( x µ i x µ i ) Φ (α i ) σ i σ i 2σ 2 i ) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
34 T90 distributions of Fermi GRBs Maximum log-likelihood tting A mixture of k standard normal (Gaussian) N (µ, σ 2 ) distributions: f (N ) k (x) = k i=1 A i ϕ ( x µ i σ i ) = k i=1 A mixture of k skew normal (SN) distributions: f (SN) k (x) = k i=1 A i 2πσi exp ( (x µ i) 2 2A i ϕ ( x µ i x µ i ) Φ (α i ) σ i σ i A mixture of k sinh-arcsinh (SAS) distributions: f (SAS) (x) = k A i k σ [1 i + ( x µ i σ i ) ] βi cosh [β i sinh 1 ( x µ i σ i ) δ i ] i=1 F (SAS) (x) = k exp [ 1 k 2 sinh [β i sinh 1 ( x µ i σ i ) δ i ] 2 ] i=1 2σ 2 i ) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
35 T90 distributions of Fermi GRBs Maximum log-likelihood tting A mixture of k standard normal (Gaussian) N (µ, σ 2 ) distributions: f (N ) k (x) = k i=1 A i ϕ ( x µ i σ i ) = k i=1 A mixture of k skew normal (SN) distributions: f (SN) k (x) = k i=1 A i 2πσi exp ( (x µ i) 2 2A i ϕ ( x µ i x µ i ) Φ (α i ) σ i σ i A mixture of k sinh-arcsinh (SAS) distributions: f (SAS) (x) = k A i k σ [1 i + ( x µ i σ i ) ] βi cosh [β i sinh 1 ( x µ i σ i ) δ i ] i=1 F (SAS) (x) = k exp [ 1 k 2 sinh [β i sinh 1 ( x µ i σ i ) δ i ] 2 ] i=1 A mixture of k alpha-skew-normal (ASN) distributions: f (ASN) k (x) = k i=1 x µ (1 α i i A i 2 + α 2 i σ i ) σ 2 i ϕ ( x µ i σ i ) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41 )
36 T90 distributions of Fermi GRBs Maximum log-likelihood tting AIC number of components AIC vs. number of components in a mixture of standard normal distributions. The minimal value corresponds to a three-gaussian. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
37 T90 distributions of Fermi GRBs Maximum log-likelihood tting PDF a b c d e f g h Distributions tted to log T 90 data. Color dashed curves are the components of the (black solid) mixture distribution. The panels show a mixture of (a) two standard Gaussians, (b) three standard Gaussians, (c) two skew-normal, (d) three skew-normal, (e) two sinh-arcsinh, (f) three sinh-arcsinh, (g) one alpha-skew-normal, and (h) two alpha-skew-normal distributions. logt 90 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
38 T90 distributions of Fermi GRBs Maximum log-likelihood tting 2 G 3 G 2 SN 3 SN 2 SAS 3 SAS 1 ASN 2 ASN AIC Pr AIC a b c d e f g h Pr AIC and relative probability (Pr ) for the models examined. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
39 Results 2 T90 distributions of Fermi GRBs Maximum log-likelihood tting Log-likelihood method supported the non-existence of a third (intermediate) component in the T 90 distribution of Fermi. A two-component mixture of skewed distributions (2-SN and 2-SAS) describes the data better than a three-gaussian. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
40 Hurst Exponents (HEs) & Machine Learning (ML) Methods HE denition HE is a measure of persistency/long-term memory/self-similarity of a process. Two ways of dening: 1 a process Y (t) (non-stationary) is self-similar with self-similarity parameter H, if Y (λt). = λ H Y (t) 2 a process X (t) (stationary) is self-similar if α (0, 2): lim ρ(t) τ α, τ α = 2 2H Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
41 Hurst Exponents (HEs) & Machine Learning (ML) Methods HE properties 0 < H 1 H = 0.5 for a random walk (Brownion motion) H < 0.5 for anti-persistent (anti-correlated, short memory) process H > 0.5 for persistent (correlated, long memory) process H = 1 for periodic time series fractal dimension D = 2 H Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
42 Hurst Exponents (HEs) & Machine Learning (ML) 8 6 Counts HE Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
43 Hurst Exponents (HEs) & Machine Learning (ML) Methods MVTS Light curves are binned in to narrow time bins. Optimum bin-width at which the non-statistical variability in the light curve becomes signicant. Prompt duration emission and equal duration of background region. Ratio of the variances of the GRB and the background divided by the bin-width as a function of bin-width. For binnings beyond the minimum the signal is indistinguishable from Poissonian uctuations. Left from minimum, signicant variability in the light curve may vanish (coarse binning). Optimum bin-width is at the minimum. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
44 Hurst Exponents (HEs) & Machine Learning (ML) Methods SVM Not probabilistic, but methods exist (probability calibration, e.g. distance to the hyperplane) to make it probabilistic. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
45 Hurst Exponents (HEs) & Machine Learning (ML) Methods Monte Carlo & SVM ( 22 ) = 231 subsamples from short GRBs; for each, 435 subsamples of (out of 46) long GRBs training set. Remaining validation set realisations. Success ratio r : r short and r long ; r tot = 2r short+4r long. 6 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
46 Hurst Exponents (HEs) & Machine Learning (ML) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
47 Hurst Exponents (HEs) & Machine Learning (ML) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
48 Hurst Exponents (HEs) & Machine Learning (ML) Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
49 Hurst Exponents (HEs) & Machine Learning (ML) Results 3 1 H and MVTS alone give unsatisfactory classications 2 T 90 works as expected 3 (H, log MVTS) unsatisfactory 4 (H, log T 90 ) better than H and log T 90 alone 5 (log MVTS, log T 90 ) worse, better 6 complementing (log MVTS, log T 90 ) with HEs accuracy increased by 7%; comparable to T 90 alone. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
50 On the limit between short and long GRBs As in (Kouveliotou et al. 1993), the limitting T 90 value is found as a local minimum. ML t of a two-gaussian instead of a parabola. Datasets: BATSE 1B (for comparison; 226 events), BATSE current, Swift, BeppoSAX and Fermi ( events). Parameter errors: parametric bootstrap. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
51 On the limit between short and long GRBs PDF 0.5 (a) 0.6 (b) (c) (d) (e) logt 90 Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
52 On the limit between short and long GRBs Table 4: Parameters of the ts. Errors are estimated using the bootstrap method. Label Dataset N i µ i δµ i σ i δσ i A i δa i min. δmin. (a) BATSE 1B (b) BATSE current (c) Swift (d) BeppoSAX (e) Fermi Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
53 On the limit between short and long GRBs Results 4 Datasets from Swift and BeepoSAX are unimodal, hence no new limit may be inferred. BATSE 1B and Fermi are consistent with the conventional 2 s value. A limit of 3.38 ± 0.27 s was obtained for BATSE current. This leads to diminishing the fraction of long GRBs in the sampe by 4%. T 90 is an ambiguous GRB type indicator. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
54 Conclusions Conclusions Both χ 2 and ML tting lead to a bimodal T 90 distribution This is a hint that there are only two GRB classes Two types of two-component skewed distributions are a better t than a three-gaussian It is unlikely that the third, intermediate-duration, GRB class is a real physical phenomenon It was suggested (Zitouni 2015) that an assymetric T 90 distribution may be due to an assymetric distribution of envelope masses of the progenitors Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
55 Conclusions Conclusions Both χ 2 and ML tting lead to a bimodal T 90 distribution This is a hint that there are only two GRB classes Two types of two-component skewed distributions are a better t than a three-gaussian It is unlikely that the third, intermediate-duration, GRB class is a real physical phenomenon It was suggested (Zitouni 2015) that an assymetric T 90 distribution may be due to an assymetric distribution of envelope masses of the progenitors HE might serve as a GRB class indicator including it in the SVM scheme increased accuracy by 7% Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
56 Conclusions Conclusions Both χ 2 and ML tting lead to a bimodal T 90 distribution This is a hint that there are only two GRB classes Two types of two-component skewed distributions are a better t than a three-gaussian It is unlikely that the third, intermediate-duration, GRB class is a real physical phenomenon It was suggested (Zitouni 2015) that an assymetric T 90 distribution may be due to an assymetric distribution of envelope masses of the progenitors HE might serve as a GRB class indicator including it in the SVM scheme increased accuracy by 7% A division between short and long GRBs at T 90 of 3.38 s is more appropriate for the BATSE current dataset than the the conventional value of 2 s. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
57 References [1] Tarnopolski M., 2015, A&A, in press (arxiv: ) [2] Tarnopolski M., 2015 (arxiv: ) [3] Tarnopolski M., 2015 (arxiv: ) [4] Tarnopolski M., 2015 (arxiv: ) [5] Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
58 References [1] Tarnopolski M., 2015, A&A, in press (arxiv: ) [2] Tarnopolski M., 2015 (arxiv: ) [3] Tarnopolski M., 2015 (arxiv: ) [4] Tarnopolski M., 2015 (arxiv: ) [5] Thank you for your attention. Mariusz Tarnopolski (AO JU) Fermi GRBs 23 July / 41
Rest-frame properties of gamma-ray bursts observed by the Fermi Gamma-Ray Burst Monitor
Rest-frame properties of gamma-ray bursts observed by the Fermi Gamma-Ray Burst Monitor on behalf of the Fermi/GBM collaboration Max Planck Institute for extraterrestrial Physics, Giessenbachstr. 1., 85748
More informationParametric Models. Dr. Shuang LIANG. School of Software Engineering TongJi University Fall, 2012
Parametric Models Dr. Shuang LIANG School of Software Engineering TongJi University Fall, 2012 Today s Topics Maximum Likelihood Estimation Bayesian Density Estimation Today s Topics Maximum Likelihood
More informationIntroduction to Machine Learning Midterm, Tues April 8
Introduction to Machine Learning 10-701 Midterm, Tues April 8 [1 point] Name: Andrew ID: Instructions: You are allowed a (two-sided) sheet of notes. Exam ends at 2:45pm Take a deep breath and don t spend
More informationLinear Models for Classification
Linear Models for Classification Oliver Schulte - CMPT 726 Bishop PRML Ch. 4 Classification: Hand-written Digit Recognition CHINE INTELLIGENCE, VOL. 24, NO. 24, APRIL 2002 x i = t i = (0, 0, 0, 1, 0, 0,
More informationExpected and unexpected gamma-ray emission from GRBs in light of AGILE and Fermi. Marco Tavani (INAF & University of Rome Tor Vergata)
Expected and unexpected gamma-ray emission from GRBs in light of AGILE and Fermi Marco Tavani (INAF & University of Rome Tor Vergata) The EGRET heritage; Outline Description of the AGILE detectors for
More informationTEMPORAL DECOMPOSITION STUDIES OF GRB LIGHTCURVES arxiv: v2 [astro-ph.he] 18 Feb 2013 Narayana P. Bhat 1
Title : will be set by the publisher Editors : will be set by the publisher EAS Publications Series, Vol.?, 2018 TEMPORAL DECOMPOSITION STUDIES OF GRB LIGHTCURVES arxiv:1301.4180v2 [astro-ph.he] 18 Feb
More informationBIO5312 Biostatistics Lecture 13: Maximum Likelihood Estimation
BIO5312 Biostatistics Lecture 13: Maximum Likelihood Estimation Yujin Chung November 29th, 2016 Fall 2016 Yujin Chung Lec13: MLE Fall 2016 1/24 Previous Parametric tests Mean comparisons (normality assumption)
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.utstat.utoronto.ca/~rsalakhu/ Sidney Smith Hall, Room 6002 Lecture 3 Linear
More informationLearning with Ensembles: How. over-tting can be useful. Anders Krogh Copenhagen, Denmark. Abstract
Published in: Advances in Neural Information Processing Systems 8, D S Touretzky, M C Mozer, and M E Hasselmo (eds.), MIT Press, Cambridge, MA, pages 190-196, 1996. Learning with Ensembles: How over-tting
More informationReview. DS GA 1002 Statistical and Mathematical Models. Carlos Fernandez-Granda
Review DS GA 1002 Statistical and Mathematical Models http://www.cims.nyu.edu/~cfgranda/pages/dsga1002_fall16 Carlos Fernandez-Granda Probability and statistics Probability: Framework for dealing with
More informationCurve Fitting Re-visited, Bishop1.2.5
Curve Fitting Re-visited, Bishop1.2.5 Maximum Likelihood Bishop 1.2.5 Model Likelihood differentiation p(t x, w, β) = Maximum Likelihood N N ( t n y(x n, w), β 1). (1.61) n=1 As we did in the case of the
More informationWidths. Center Fluctuations. Centers. Centers. Widths
Radial Basis Functions: a Bayesian treatment David Barber Bernhard Schottky Neural Computing Research Group Department of Applied Mathematics and Computer Science Aston University, Birmingham B4 7ET, U.K.
More informationUNIVERSITY of PENNSYLVANIA CIS 520: Machine Learning Final, Fall 2013
UNIVERSITY of PENNSYLVANIA CIS 520: Machine Learning Final, Fall 2013 Exam policy: This exam allows two one-page, two-sided cheat sheets; No other materials. Time: 2 hours. Be sure to write your name and
More informationChapter 9. Non-Parametric Density Function Estimation
9-1 Density Estimation Version 1.1 Chapter 9 Non-Parametric Density Function Estimation 9.1. Introduction We have discussed several estimation techniques: method of moments, maximum likelihood, and least
More informationSTA 4273H: Sta-s-cal Machine Learning
STA 4273H: Sta-s-cal Machine Learning Russ Salakhutdinov Department of Computer Science! Department of Statistical Sciences! rsalakhu@cs.toronto.edu! h0p://www.cs.utoronto.ca/~rsalakhu/ Lecture 2 In our
More informationThe Poisson transform for unnormalised statistical models. Nicolas Chopin (ENSAE) joint work with Simon Barthelmé (CNRS, Gipsa-LAB)
The Poisson transform for unnormalised statistical models Nicolas Chopin (ENSAE) joint work with Simon Barthelmé (CNRS, Gipsa-LAB) Part I Unnormalised statistical models Unnormalised statistical models
More informationDensity Estimation. Seungjin Choi
Density Estimation Seungjin Choi Department of Computer Science and Engineering Pohang University of Science and Technology 77 Cheongam-ro, Nam-gu, Pohang 37673, Korea seungjin@postech.ac.kr http://mlg.postech.ac.kr/
More informationBayesian Learning (II)
Universität Potsdam Institut für Informatik Lehrstuhl Maschinelles Lernen Bayesian Learning (II) Niels Landwehr Overview Probabilities, expected values, variance Basic concepts of Bayesian learning MAP
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.utstat.utoronto.ca/~rsalakhu/ Sidney Smith Hall, Room 6002 Lecture 7 Approximate
More informationarxiv:astro-ph/ v2 17 May 1998
A 3 rd CLASS OF GAMMA RAY BURSTS? I. Horváth Department of Astronomy & Astrophysics, Pennsylvania State University, arxiv:astro-ph/9803077v2 17 May 1998 525 Davey Laboratory, University Park, PA 16802,
More informationChapter 9. Non-Parametric Density Function Estimation
9-1 Density Estimation Version 1.2 Chapter 9 Non-Parametric Density Function Estimation 9.1. Introduction We have discussed several estimation techniques: method of moments, maximum likelihood, and least
More informationData modelling Parameter estimation
ASTR509-10 Data modelling Parameter estimation Pierre-Simon, Marquis de Laplace 1749-1827 Under Napoleon, Laplace was a member, then chancellor, of the Senate, and received the Legion of Honour in 1805.
More informationProbabilistic modeling. The slides are closely adapted from Subhransu Maji s slides
Probabilistic modeling The slides are closely adapted from Subhransu Maji s slides Overview So far the models and algorithms you have learned about are relatively disconnected Probabilistic modeling framework
More informationDensity Estimation: ML, MAP, Bayesian estimation
Density Estimation: ML, MAP, Bayesian estimation CE-725: Statistical Pattern Recognition Sharif University of Technology Spring 2013 Soleymani Outline Introduction Maximum-Likelihood Estimation Maximum
More informationMachine Learning. Probabilistic KNN.
Machine Learning. Mark Girolami girolami@dcs.gla.ac.uk Department of Computing Science University of Glasgow June 21, 2007 p. 1/3 KNN is a remarkably simple algorithm with proven error-rates June 21, 2007
More informationSupport Vector Machines
Support Vector Machines Stephan Dreiseitl University of Applied Sciences Upper Austria at Hagenberg Harvard-MIT Division of Health Sciences and Technology HST.951J: Medical Decision Support Overview Motivation
More informationCourse 495: Advanced Statistical Machine Learning/Pattern Recognition
Course 495: Advanced Statistical Machine Learning/Pattern Recognition Goal (Lecture): To present Probabilistic Principal Component Analysis (PPCA) using both Maximum Likelihood (ML) and Expectation Maximization
More informationPrompt Emission Properties of Swift GRBs
Prompt Emission Properties of GRBs T. Sakamoto (Aoyama Gakuin University) Outline Collaboration with KW team BAT 3 rd GRB catalog Duration and hardness Global properties of BAT GRBs Pre-/Post-GRB emission
More informationLectures in AstroStatistics: Topics in Machine Learning for Astronomers
Lectures in AstroStatistics: Topics in Machine Learning for Astronomers Jessi Cisewski Yale University American Astronomical Society Meeting Wednesday, January 6, 2016 1 Statistical Learning - learning
More informationBayesian Learning. HT2015: SC4 Statistical Data Mining and Machine Learning. Maximum Likelihood Principle. The Bayesian Learning Framework
HT5: SC4 Statistical Data Mining and Machine Learning Dino Sejdinovic Department of Statistics Oxford http://www.stats.ox.ac.uk/~sejdinov/sdmml.html Maximum Likelihood Principle A generative model for
More informationLinear Regression and Its Applications
Linear Regression and Its Applications Predrag Radivojac October 13, 2014 Given a data set D = {(x i, y i )} n the objective is to learn the relationship between features and the target. We usually start
More informationMISCELLANEOUS TOPICS RELATED TO LIKELIHOOD. Copyright c 2012 (Iowa State University) Statistics / 30
MISCELLANEOUS TOPICS RELATED TO LIKELIHOOD Copyright c 2012 (Iowa State University) Statistics 511 1 / 30 INFORMATION CRITERIA Akaike s Information criterion is given by AIC = 2l(ˆθ) + 2k, where l(ˆθ)
More informationHandling uncertainties in background shapes: the discrete profiling method
Journal of Instrumentation OPEN ACCESS Handling uncertainties in background shapes: the discrete profiling method To cite this article: P.D. Dauncey et al 5 JINST P45 View the article online for updates
More informationTerminology Suppose we have N observations {x(n)} N 1. Estimators as Random Variables. {x(n)} N 1
Estimation Theory Overview Properties Bias, Variance, and Mean Square Error Cramér-Rao lower bound Maximum likelihood Consistency Confidence intervals Properties of the mean estimator Properties of the
More informationStatistical Data Analysis Stat 3: p-values, parameter estimation
Statistical Data Analysis Stat 3: p-values, parameter estimation London Postgraduate Lectures on Particle Physics; University of London MSci course PH4515 Glen Cowan Physics Department Royal Holloway,
More informationTemporal context calibrates interval timing
Temporal context calibrates interval timing, Mehrdad Jazayeri & Michael N. Shadlen Helen Hay Whitney Foundation HHMI, NPRC, Department of Physiology and Biophysics, University of Washington, Seattle, Washington
More informationCosmology with Gamma-Ray. Bursts. Lorenzo Amati. May 29, 2012
Cosmology with Gamma-Ray Bursts Lorenzo Amati Italian National Institute for Astrophysics (INAF), Bologna May 29, 2012 Outline Gamma-Ray Bursts: the brightest cosmological sources The Ep,i - intensity
More informationAUTOMATIC CONTROL COMMUNICATION SYSTEMS LINKÖPINGS UNIVERSITET. Questions AUTOMATIC CONTROL COMMUNICATION SYSTEMS LINKÖPINGS UNIVERSITET
The Problem Identification of Linear and onlinear Dynamical Systems Theme : Curve Fitting Division of Automatic Control Linköping University Sweden Data from Gripen Questions How do the control surface
More informationBias-Variance Tradeoff
What s learning, revisited Overfitting Generative versus Discriminative Logistic Regression Machine Learning 10701/15781 Carlos Guestrin Carnegie Mellon University September 19 th, 2007 Bias-Variance Tradeoff
More informationThe Bayes classifier
The Bayes classifier Consider where is a random vector in is a random variable (depending on ) Let be a classifier with probability of error/risk given by The Bayes classifier (denoted ) is the optimal
More informationData Analysis I. Dr Martin Hendry, Dept of Physics and Astronomy University of Glasgow, UK. 10 lectures, beginning October 2006
Astronomical p( y x, I) p( x, I) p ( x y, I) = p( y, I) Data Analysis I Dr Martin Hendry, Dept of Physics and Astronomy University of Glasgow, UK 10 lectures, beginning October 2006 4. Monte Carlo Methods
More informationCSCI-567: Machine Learning (Spring 2019)
CSCI-567: Machine Learning (Spring 2019) Prof. Victor Adamchik U of Southern California Mar. 19, 2019 March 19, 2019 1 / 43 Administration March 19, 2019 2 / 43 Administration TA3 is due this week March
More informationParametric Modelling of Over-dispersed Count Data. Part III / MMath (Applied Statistics) 1
Parametric Modelling of Over-dispersed Count Data Part III / MMath (Applied Statistics) 1 Introduction Poisson regression is the de facto approach for handling count data What happens then when Poisson
More informationReport on lab session, Monday 20 February, 5-6pm
Report on la session, Monday 0 Feruary, 5-6pm Read the data as follows. The variale of interest is the adjusted closing price of Coca Cola in the 7th column of the data file KO.csv. dat=read.csv("ko.csv",header=true)
More informationF & B Approaches to a simple model
A6523 Signal Modeling, Statistical Inference and Data Mining in Astrophysics Spring 215 http://www.astro.cornell.edu/~cordes/a6523 Lecture 11 Applications: Model comparison Challenges in large-scale surveys
More informationThe Bootstrap: Theory and Applications. Biing-Shen Kuo National Chengchi University
The Bootstrap: Theory and Applications Biing-Shen Kuo National Chengchi University Motivation: Poor Asymptotic Approximation Most of statistical inference relies on asymptotic theory. Motivation: Poor
More informationMachine Learning, Midterm Exam: Spring 2009 SOLUTION
10-601 Machine Learning, Midterm Exam: Spring 2009 SOLUTION March 4, 2009 Please put your name at the top of the table below. If you need more room to work out your answer to a question, use the back of
More informationMachine Learning. Gaussian Mixture Models. Zhiyao Duan & Bryan Pardo, Machine Learning: EECS 349 Fall
Machine Learning Gaussian Mixture Models Zhiyao Duan & Bryan Pardo, Machine Learning: EECS 349 Fall 2012 1 The Generative Model POV We think of the data as being generated from some process. We assume
More informationCMU-Q Lecture 24:
CMU-Q 15-381 Lecture 24: Supervised Learning 2 Teacher: Gianni A. Di Caro SUPERVISED LEARNING Hypotheses space Hypothesis function Labeled Given Errors Performance criteria Given a collection of input
More informationLecture : Probabilistic Machine Learning
Lecture : Probabilistic Machine Learning Riashat Islam Reasoning and Learning Lab McGill University September 11, 2018 ML : Many Methods with Many Links Modelling Views of Machine Learning Machine Learning
More informationMathematical Tools for Neuroscience (NEU 314) Princeton University, Spring 2016 Jonathan Pillow. Homework 8: Logistic Regression & Information Theory
Mathematical Tools for Neuroscience (NEU 34) Princeton University, Spring 206 Jonathan Pillow Homework 8: Logistic Regression & Information Theory Due: Tuesday, April 26, 9:59am Optimization Toolbox One
More informationRecent Advances in Bayesian Inference Techniques
Recent Advances in Bayesian Inference Techniques Christopher M. Bishop Microsoft Research, Cambridge, U.K. research.microsoft.com/~cmbishop SIAM Conference on Data Mining, April 2004 Abstract Bayesian
More informationHypothesis testing:power, test statistic CMS:
Hypothesis testing:power, test statistic The more sensitive the test, the better it can discriminate between the null and the alternative hypothesis, quantitatively, maximal power In order to achieve this
More informationPattern Recognition and Machine Learning. Bishop Chapter 2: Probability Distributions
Pattern Recognition and Machine Learning Chapter 2: Probability Distributions Cécile Amblard Alex Kläser Jakob Verbeek October 11, 27 Probability Distributions: General Density Estimation: given a finite
More informationNon-parametric Methods
Non-parametric Methods Machine Learning Torsten Möller Möller/Mori 1 Reading Chapter 2 of Pattern Recognition and Machine Learning by Bishop (with an emphasis on section 2.5) Möller/Mori 2 Outline Last
More informationReport on lab session, Monday 27 February, 5-6pm
Report on la session, Monday 7 Feruary, 5-6pm Read the data as follows. The variale of interest is the adjusted closing price of Coca Cola in the 7th column of the data file KO.csv. dat=read.csv("ko.csv",header=true)
More informationUncertainty Quantification for Machine Learning and Statistical Models
Uncertainty Quantification for Machine Learning and Statistical Models David J. Stracuzzi Joint work with: Max Chen, Michael Darling, Stephen Dauphin, Matt Peterson, and Chris Young Sandia National Laboratories
More informationWarm up: risk prediction with logistic regression
Warm up: risk prediction with logistic regression Boss gives you a bunch of data on loans defaulting or not: {(x i,y i )} n i= x i 2 R d, y i 2 {, } You model the data as: P (Y = y x, w) = + exp( yw T
More informationSTAT 535 Lecture 5 November, 2018 Brief overview of Model Selection and Regularization c Marina Meilă
STAT 535 Lecture 5 November, 2018 Brief overview of Model Selection and Regularization c Marina Meilă mmp@stat.washington.edu Reading: Murphy: BIC, AIC 8.4.2 (pp 255), SRM 6.5 (pp 204) Hastie, Tibshirani
More informationIntroduction to Machine Learning
Introduction to Machine Learning Linear Regression Varun Chandola Computer Science & Engineering State University of New York at Buffalo Buffalo, NY, USA chandola@buffalo.edu Chandola@UB CSE 474/574 1
More informationPrinceton University Observatory preprint POP-567. Submitted to Astrophysical Journal Letters, 2 June 1994
Princeton University Observatory preprint POP-567 Submitted to Astrophysical Journal Letters, 2 June 1994 On the nature of gamma-ray burst time dilations Ralph A.M.J. Wijers 1 and Bohdan Paczynski 2 Princeton
More informationMeasuring cosmological parameters with GRBs
Measuring cosmological parameters with GRBs Lorenzo Amati Italian National Institute for Astrophysics (INAF), Bologna with contributions by: M. Della Valle, S. Dichiara, F. Frontera, C. Guidorzi Why look
More informationMaximum Likelihood Estimation. only training data is available to design a classifier
Introduction to Pattern Recognition [ Part 5 ] Mahdi Vasighi Introduction Bayesian Decision Theory shows that we could design an optimal classifier if we knew: P( i ) : priors p(x i ) : class-conditional
More informationAn Information Criteria for Order-restricted Inference
An Information Criteria for Order-restricted Inference Nan Lin a, Tianqing Liu 2,b, and Baoxue Zhang,2,b a Department of Mathematics, Washington University in Saint Louis, Saint Louis, MO 633, U.S.A. b
More informationarxiv:astro-ph/ v1 29 May 2000
Effects of Luminosity Functions Induced by Relativistic Beaming on Statistics of Cosmological Gamma-Ray Bursts Chunglee Kim, Heon-Young Chang, and Insu Yi arxiv:astro-ph/5556v 29 May 2 Korea Institute
More informationBayesian Machine Learning
Bayesian Machine Learning Andrew Gordon Wilson ORIE 6741 Lecture 2: Bayesian Basics https://people.orie.cornell.edu/andrew/orie6741 Cornell University August 25, 2016 1 / 17 Canonical Machine Learning
More informationA NEW INFORMATION THEORETIC APPROACH TO ORDER ESTIMATION PROBLEM. Massachusetts Institute of Technology, Cambridge, MA 02139, U.S.A.
A EW IFORMATIO THEORETIC APPROACH TO ORDER ESTIMATIO PROBLEM Soosan Beheshti Munther A. Dahleh Massachusetts Institute of Technology, Cambridge, MA 0239, U.S.A. Abstract: We introduce a new method of model
More informationBayesian Methods for Machine Learning
Bayesian Methods for Machine Learning CS 584: Big Data Analytics Material adapted from Radford Neal s tutorial (http://ftp.cs.utoronto.ca/pub/radford/bayes-tut.pdf), Zoubin Ghahramni (http://hunch.net/~coms-4771/zoubin_ghahramani_bayesian_learning.pdf),
More informationPrecise Interplanetary Network Localization of a New Soft. Gamma Repeater, SGR
Precise Interplanetary Network Localization of a New Soft Gamma Repeater, SGR1627-41 K. Hurley University of California, Berkeley, Space Sciences Laboratory, Berkeley, CA 94720-7450 C. Kouveliotou Universities
More informationPATTERN RECOGNITION AND MACHINE LEARNING
PATTERN RECOGNITION AND MACHINE LEARNING Chapter 1. Introduction Shuai Huang April 21, 2014 Outline 1 What is Machine Learning? 2 Curve Fitting 3 Probability Theory 4 Model Selection 5 The curse of dimensionality
More informationContents Lecture 4. Lecture 4 Linear Discriminant Analysis. Summary of Lecture 3 (II/II) Summary of Lecture 3 (I/II)
Contents Lecture Lecture Linear Discriminant Analysis Fredrik Lindsten Division of Systems and Control Department of Information Technology Uppsala University Email: fredriklindsten@ituuse Summary of lecture
More informationIntroduction to Error Analysis
Introduction to Error Analysis Part 1: the Basics Andrei Gritsan based on lectures by Petar Maksimović February 1, 2010 Overview Definitions Reporting results and rounding Accuracy vs precision systematic
More informationNews from the Niels Bohr International Academy
News from the Niels Bohr International Academy What is a gamma-ray burst? The development of our understanding of the phenomenon illustrated by important events (including very recent results on gravitational
More informationProbability distribution functions
Probability distribution functions Everything there is to know about random variables Imperial Data Analysis Workshop (2018) Elena Sellentin Sterrewacht Universiteit Leiden, NL Ln(a) Sellentin Universiteit
More informationPattern Recognition. Parameter Estimation of Probability Density Functions
Pattern Recognition Parameter Estimation of Probability Density Functions Classification Problem (Review) The classification problem is to assign an arbitrary feature vector x F to one of c classes. The
More informationStatistical Learning Reading Assignments
Statistical Learning Reading Assignments S. Gong et al. Dynamic Vision: From Images to Face Recognition, Imperial College Press, 2001 (Chapt. 3, hard copy). T. Evgeniou, M. Pontil, and T. Poggio, "Statistical
More informationPATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 2: PROBABILITY DISTRIBUTIONS
PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 2: PROBABILITY DISTRIBUTIONS Parametric Distributions Basic building blocks: Need to determine given Representation: or? Recall Curve Fitting Binary Variables
More informationExpectation Propagation for Approximate Bayesian Inference
Expectation Propagation for Approximate Bayesian Inference José Miguel Hernández Lobato Universidad Autónoma de Madrid, Computer Science Department February 5, 2007 1/ 24 Bayesian Inference Inference Given
More informationLikelihood-Based Methods
Likelihood-Based Methods Handbook of Spatial Statistics, Chapter 4 Susheela Singh September 22, 2016 OVERVIEW INTRODUCTION MAXIMUM LIKELIHOOD ESTIMATION (ML) RESTRICTED MAXIMUM LIKELIHOOD ESTIMATION (REML)
More informationIntroduction. Chapter 1
Chapter 1 Introduction In this book we will be concerned with supervised learning, which is the problem of learning input-output mappings from empirical data (the training dataset). Depending on the characteristics
More information9. Model Selection. statistical models. overview of model selection. information criteria. goodness-of-fit measures
FE661 - Statistical Methods for Financial Engineering 9. Model Selection Jitkomut Songsiri statistical models overview of model selection information criteria goodness-of-fit measures 9-1 Statistical models
More informationRecall the Basics of Hypothesis Testing
Recall the Basics of Hypothesis Testing The level of significance α, (size of test) is defined as the probability of X falling in w (rejecting H 0 ) when H 0 is true: P(X w H 0 ) = α. H 0 TRUE H 1 TRUE
More informationVariability within multi-component systems. Bayesian inference in probabilistic risk assessment The current state of the art
PhD seminar series Probabilistics in Engineering : g Bayesian networks and Bayesian hierarchical analysis in engeering g Conducted by Prof. Dr. Maes, Prof. Dr. Faber and Dr. Nishijima Variability within
More informationPhysics 403. Segev BenZvi. Classical Hypothesis Testing: The Likelihood Ratio Test. Department of Physics and Astronomy University of Rochester
Physics 403 Classical Hypothesis Testing: The Likelihood Ratio Test Segev BenZvi Department of Physics and Astronomy University of Rochester Table of Contents 1 Bayesian Hypothesis Testing Posterior Odds
More informationy Xw 2 2 y Xw λ w 2 2
CS 189 Introduction to Machine Learning Spring 2018 Note 4 1 MLE and MAP for Regression (Part I) So far, we ve explored two approaches of the regression framework, Ordinary Least Squares and Ridge Regression:
More informationMachine Learning Srihari. Probability Theory. Sargur N. Srihari
Probability Theory Sargur N. Srihari srihari@cedar.buffalo.edu 1 Probability Theory with Several Variables Key concept is dealing with uncertainty Due to noise and finite data sets Framework for quantification
More informationLTI Systems, Additive Noise, and Order Estimation
LTI Systems, Additive oise, and Order Estimation Soosan Beheshti, Munther A. Dahleh Laboratory for Information and Decision Systems Department of Electrical Engineering and Computer Science Massachusetts
More informationIntroduction to Machine Learning. Maximum Likelihood and Bayesian Inference. Lecturers: Eran Halperin, Yishay Mansour, Lior Wolf
1 Introduction to Machine Learning Maximum Likelihood and Bayesian Inference Lecturers: Eran Halperin, Yishay Mansour, Lior Wolf 2013-14 We know that X ~ B(n,p), but we do not know p. We get a random sample
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Computer Science! Department of Statistical Sciences! rsalakhu@cs.toronto.edu! h0p://www.cs.utoronto.ca/~rsalakhu/ Lecture 7 Approximate
More informationLECTURE NOTE #3 PROF. ALAN YUILLE
LECTURE NOTE #3 PROF. ALAN YUILLE 1. Three Topics (1) Precision and Recall Curves. Receiver Operating Characteristic Curves (ROC). What to do if we do not fix the loss function? (2) The Curse of Dimensionality.
More informationExpectation Maximization Algorithm
Expectation Maximization Algorithm Vibhav Gogate The University of Texas at Dallas Slides adapted from Carlos Guestrin, Dan Klein, Luke Zettlemoyer and Dan Weld The Evils of Hard Assignments? Clusters
More informationLECTURE NOTE #8 PROF. ALAN YUILLE. Can we find a linear classifier that separates the position and negative examples?
LECTURE NOTE #8 PROF. ALAN YUILLE 1. Linear Classifiers and Perceptrons A dataset contains N samples: { (x µ, y µ ) : µ = 1 to N }, y µ {±1} Can we find a linear classifier that separates the position
More informationGaussian Mixture Models
Gaussian Mixture Models Pradeep Ravikumar Co-instructor: Manuela Veloso Machine Learning 10-701 Some slides courtesy of Eric Xing, Carlos Guestrin (One) bad case for K- means Clusters may overlap Some
More informationECE521 lecture 4: 19 January Optimization, MLE, regularization
ECE521 lecture 4: 19 January 2017 Optimization, MLE, regularization First four lectures Lectures 1 and 2: Intro to ML Probability review Types of loss functions and algorithms Lecture 3: KNN Convexity
More informationQuantifying Stochastic Model Errors via Robust Optimization
Quantifying Stochastic Model Errors via Robust Optimization IPAM Workshop on Uncertainty Quantification for Multiscale Stochastic Systems and Applications Jan 19, 2016 Henry Lam Industrial & Operations
More informationInvestigation of Possible Biases in Tau Neutrino Mass Limits
Investigation of Possible Biases in Tau Neutrino Mass Limits Kyle Armour Departments of Physics and Mathematics, University of California, San Diego, La Jolla, CA 92093 (Dated: August 8, 2003) We study
More informationBAYESIAN DECISION THEORY
Last updated: September 17, 2012 BAYESIAN DECISION THEORY Problems 2 The following problems from the textbook are relevant: 2.1 2.9, 2.11, 2.17 For this week, please at least solve Problem 2.3. We will
More information6 Reweighting. 6.1 Reweighting in Monte Carlo simulations
6 Reweighting The configurations generated in a Monte Carlo simulation contain a huge amount of information, from which we usually distill a couple of numbers. It would be a shame to waste all that information.
More informationHomework 1 Solutions Probability, Maximum Likelihood Estimation (MLE), Bayes Rule, knn
Homework 1 Solutions Probability, Maximum Likelihood Estimation (MLE), Bayes Rule, knn CMU 10-701: Machine Learning (Fall 2016) https://piazza.com/class/is95mzbrvpn63d OUT: September 13th DUE: September
More informationPerformance Comparison of K-Means and Expectation Maximization with Gaussian Mixture Models for Clustering EE6540 Final Project
Performance Comparison of K-Means and Expectation Maximization with Gaussian Mixture Models for Clustering EE6540 Final Project Devin Cornell & Sushruth Sastry May 2015 1 Abstract In this article, we explore
More information