A New Technique for Fitting Colour-Magnitude Diagrams Tim Naylor School of Physics, University of Exeter R. D. Jeffries Astrophysics Group, Keele University.

Slides:



Advertisements
Similar presentations
Shapelets Correlated with Surface Normals Produce Surfaces Peter Kovesi School of Computer Science & Software Engineering The University of Western Australia.
Advertisements

Empirical Isochrones and relative ages – is IC348 really that young? Abstract Abstract Interpreting observational studies of rotational evolution of stars.
Brief introduction on Logistic Regression
Welcome to PHYS 225a Lab Introduction, class rules, error analysis Julia Velkovska.
Sampling: Final and Initial Sample Size Determination
Bayesian Reasoning: Markov Chain Monte Carlo
SOLUTION 1: ECLIPSING BINARIES IN OPEN CLUSTERS The study of eclipsing binaries in open clusters allows strong constraints to be placed on theoretical.
Factor Analysis Purpose of Factor Analysis
Bayesian Analysis of X-ray Luminosity Functions A. Ptak (JHU) Abstract Often only a relatively small number of sources of a given class are detected in.
Evaluating Hypotheses
PROBABILITY AND SAMPLES: THE DISTRIBUTION OF SAMPLE MEANS.
Copyright © Cengage Learning. All rights reserved. 6 Point Estimation.
Chi Square Distribution (c2) and Least Squares Fitting
A Maximum Likelihood Method for Identifying the Components of Eclipsing Binary Stars Jonathan Devor and David Charbonneau Harvard-Smithsonian Center for.
Lehrstuhl für Informatik 2 Gabriella Kókai: Maschine Learning 1 Evaluating Hypotheses.
C82MCP Diploma Statistics School of Psychology University of Nottingham 1 Overview Central Limit Theorem The Normal Distribution The Standardised Normal.
Chapter 7 Probability and Samples: The Distribution of Sample Means
Determining the Size of
1 10. Joint Moments and Joint Characteristic Functions Following section 6, in this section we shall introduce various parameters to compactly represent.
Chapter 6 Random Error The Nature of Random Errors
Physics 114: Lecture 15 Probability Tests & Linear Fitting Dale E. Gary NJIT Physics Department.
1 Statistical Mechanics and Multi- Scale Simulation Methods ChBE Prof. C. Heath Turner Lecture 11 Some materials adapted from Prof. Keith E. Gubbins:
Review – Using Standard Deviation Here are eight test scores from a previous Stats 201 class: 35, 59, 70, 73, 75, 81, 84, 86. The mean and standard deviation.
Estimation of Statistical Parameters
Estimation in Sampling!? Chapter 7 – Statistical Problem Solving in Geography.
Lecture 12 Statistical Inference (Estimation) Point and Interval estimation By Aziza Munir.
1 Institute of Engineering Mechanics Leopold-Franzens University Innsbruck, Austria, EU H.J. Pradlwarter and G.I. Schuëller Confidence.
PHYSICS WORKSHOP Demystifying 9188/4 Yours truly T.V Madziva or
Module 1: Statistical Issues in Micro simulation Paul Sousa.
The Examination of Residuals. Examination of Residuals The fitting of models to data is done using an iterative approach. The first step is to fit a simple.
Orthographic Multiview Projection Multiview Projection.
CHAPTER 2 Statistical Inference, Exploratory Data Analysis and Data Science Process cse4/587-Sprint
Do accretion discs regulate the momentum of young stars? S.P. Littlefair 1,2, T. Naylor 2, R.D. Jeffries 3, B. Burningham 2, E. Saunders 2 1 Dept of Physics.
Can variability account for apparent age spreads in OB association colour-magnitude diagrams? Ben Burningham & Tim Naylor School of Physics, University.
A Photometric Study of Unstudied Open Clusters Berkeley 49 & 84 in the SDSS Jinhyuk Ryu and Myung Gyoon Lee Department of Physics & Astronomy, Seoul National.
LECTURE 3: ANALYSIS OF EXPERIMENTAL DATA
Worked Example Using R. > plot(y~x) >plot(epsilon1~x) This is a plot of residuals against the exploratory variable, x.
CVPR2013 Poster Detecting and Naming Actors in Movies using Generative Appearance Models.
1 8. One Function of Two Random Variables Given two random variables X and Y and a function g(x,y), we form a new random variable Z as Given the joint.
© 2008 McGraw-Hill Higher Education The Statistical Imagination Chapter 8. Parameter Estimation Using Confidence Intervals.
Lecture 18 Stellar populations. Stellar clusters Open clusters: contain stars loose structure Globular clusters: million stars centrally.
October 1, 2013Computer Vision Lecture 9: From Edges to Contours 1 Canny Edge Detector However, usually there will still be noise in the array E[i, j],
ECE 8443 – Pattern Recognition ECE 8527 – Introduction to Machine Learning and Pattern Recognition LECTURE 04: GAUSSIAN CLASSIFIERS Objectives: Whitening.
One Function of Two Random Variables
Copyright © Cengage Learning. All rights reserved. 9 Inferences Based on Two Samples.
Various Rupak Mahapatra (for Angela, Joel, Mike & Jeff) Timing Cuts.
CHAPTER 2.3 PROBABILITY DISTRIBUTIONS. 2.3 GAUSSIAN OR NORMAL ERROR DISTRIBUTION  The Gaussian distribution is an approximation to the binomial distribution.
CHAPTER – 1 UNCERTAINTIES IN MEASUREMENTS. 1.3 PARENT AND SAMPLE DISTRIBUTIONS  If we make a measurement x i in of a quantity x, we expect our observation.
Lecture 8: Measurement Errors 1. Objectives List some sources of measurement errors. Classify measurement errors into systematic and random errors. Study.
Age and distance for the unstudied open cluster Teutsch 7 In Sung Jang and Myung Gyoon Lee Department of Physics & Astronomy, Seoul National University.
ESTIMATION OF THE MEAN. 2 INTRO :: ESTIMATION Definition The assignment of plausible value(s) to a population parameter based on a value of a sample statistic.
Constraining population synthesis (and binary black hole inspiral rates) using binary neutron stars Richard O’Shaughnessy GWDAW
In Bayesian theory, a test statistics can be defined by taking the ratio of the Bayes factors for the two hypotheses: The ratio measures the probability.
G. Eigen, Paris, Introduction The SiPM response is non-linear and depends on operating voltage (V-V bd ) and temperature  SiPMs need monitoring.
CHAPTER- 3.1 ERROR ANALYSIS.  Now we shall further consider  how to estimate uncertainties in our measurements,  the sources of the uncertainties,
Richard Kass/F02P416 Lecture 6 1 Lecture 6 Chi Square Distribution (  2 ) and Least Squares Fitting Chi Square Distribution (  2 ) (See Taylor Ch 8,
The accuracy of averages We learned how to make inference from the sample to the population: Counting the percentages. Here we begin to learn how to make.
© 2010 Pearson Prentice Hall. All rights reserved Chapter Hypothesis Tests Regarding a Parameter 10.
Tom.h.wilson Department of Geology and Geography West Virginia University Morgantown, WV.
Yun, Hyuk Jin. Theory A.Nonuniformity Model where at location x, v is the measured signal, u is the true signal emitted by the tissue, is an unknown.
Virtual University of Pakistan
Probability plots.
Physics 114: Lecture 13 Probability Tests & Linear Fitting
Hypothesis Testing and Confidence Intervals (Part 1): Using the Standard Normal Lecture 8 Justin Kern October 10 and 12, 2017.
Latent Variables, Mixture Models and EM
Statistical Methods For Engineers
Geology Geomath Chapter 7 - Statistics tom.h.wilson
CHAPTER – 1.2 UNCERTAINTIES IN MEASUREMENTS.
These probabilities are the probabilities that individual values in a sample will fall in a 50 gram range, and thus represent the integral of individual.
CHAPTER – 1.2 UNCERTAINTIES IN MEASUREMENTS.
Presentation transcript:

A New Technique for Fitting Colour-Magnitude Diagrams Tim Naylor School of Physics, University of Exeter R. D. Jeffries Astrophysics Group, Keele University Abstract We present a new method for fitting colour magnitude diagrams of clusters and associations. The method allows the model to include binaries, and gives robust parameter uncertainties. WHAT’S THE PROBLEM? Despite the acquisition of many excellent colour-magnitude diagrams for young clusters and associations, and the calculation of good pre-main-sequence models, fitting these models to the data is still largely done “by eye”. There is a good reason for this, even though it is not clearly elucidated in the literature. Were the data drawn from a single star sequence the problem would become one of fitting an arbitrary line to data points with uncertainties in two dimensions. Even this is not a straightforward problem, but was largely solved by Flannery & Johnson (1982; ApJ ). However, this solution remains little used for the obvious reason that our data are not drawn from a single-star sequence, but from a population which contains a large fraction of binary stars. These binary stars lie above the (pre-)main- sequence, resulting in a two dimensional distribution of objects in the colour magnitude plane (see Figure 1). Faced with this, most galactic astronomers have taken the “by eye” approach, fitting the single-star model sequences to the lower envelope of the data. There have been more determined attempts to solve the problem in the extragalactic context, where the distributions are spread even further from single isochrones because star formation continues over large periods of time. Dolphin (2002, MNRAS , and references therein) bin the data in two dimensions, but in doing so blur out our hard-won photometric precision. This is especially serious in the case of clusters where the differences in the position of the isochrone with age are typically rather small. Tolstoy & Saha (1996; ApJ ) suggest making a two-dimensional simulation of the data, but only using a similar number of points to that in the original dataset. Thus some of the precision of the data is lost in the graininess of the model, and it is unclear how one could determine uncertainties in parameters. AN INTUITIVE SOLUTION Our solution to this problem can be envisaged in the following way. Figure 1 shows a grey scale model which includes binaries, where the intensity of the grey scale is the probability of finding a object at that colour and magnitude. Imagine moving the data points in Figure 1 over the grey scale, and collecting the values of the probability at the position of each data point. The product of all these values is clearly a goodness-of-fit statistic, and is maximised when the data of are placed correctly in colour and magnitude over the model. This method can be refined to include the (two dimensional) uncertainties of each data point (see below), at which point we call our statistic  2. It can be formally derived from maximum likelihood theory, and as such can be viewed as either a Bayesian or perfectly respectable Frequentist method. We have found that if the model is a single sequence with uncertainties in one dimension  2 is identical to  2, i.e.  2 is a special case of  2. One can derive uncertainties in the fitted parameters in a similar way to a  2 analysis, and we show in Figure 2 the  2 space for fitting the data of Figure 1. The expected correlation between distance modulus and age is clearly visible. FIGURE 1 The grey scale is the best fit model isochrone to the X-ray selected members of NGC2547 (circles). The data points have been dereddened and then shifted by the best fit distance modulus. FORMAL DEFINITION The formal definition of our statistic is given by where P i, the probability for a single data point at (c i,m i ) is given by the integral of the model  multiplied by P D the probability distribution due to the uncertainties for that data point. We can show this reduces to  2 if  is a line and P D a one-dimensional Gaussian. Then the product is only non-zero where the two intersect, and has a value proportional to the value of the Gaussian at that point. Thus the integral reduces to exp((c-c i ) 2 /  i 2 ), leading to the normal form for  2. FIGURE 2 A grey scale plot of  2 as a function of age and distance modulus for the model and data of Figure 1. The white contour is the 67% confidence level, whose structure is probably caused by the clipping procedure for data points lying outside the model. OTHER PARAMETERS There is a large range of possible parameters one could fit, but for NGC2547 we have been experimenting with binary fraction. Figure 3 shows the distribution of  2 from fitting the data and that expected from theoretical considerations. Clearly the data has too many points at high  2, which corresponds to too many data points in the region of low (but non-zero) probability in Figure 1. We have experimented in increasing the binary fraction, which increases the expected number of stars in this region of the CMD, which cures the problem, but does not significantly change the best-fit parameters for age and distance. FIGURE 3 The expected distribution of  2 (curve) compared with that obtained for the fit in Figure 1 (histogram). CONCLUSIONS  2 appears to be a very powerful technique for extracting robust parameters with uncertainties from colour magnitude diagrams. Although our own immediate interest is such datasets, it appears the method is very general, and should have many applications to sparse datasets, and datasets with uncertainties in two (or more) dimensions.