Planet WS Berlin 12/2003 Transit detecion on eclipsing binary systems Hans J. Deeg Instituto de Astrofísica de Canarias 1.Transit observations in eclipsing.

Slides:



Advertisements
Similar presentations
A Leading index for the Colombian economic activity Colombian economic activity Luis Fernando Melo Fabio Nieto Mario Ramos Luis Fernando Melo Fabio Nieto.
Advertisements

TEA Systems Corp. Confidential LithoWorks PEB On-Wafer analysis of two wafers August 8, 2003 Thermal analysis of two PEB plates: _peb_bake7.csv _peb_bake8.csv.
An Exemplar-Based Fitness Function with Dynamic Value Based on Sub-Problem Difficulty Steve Rowe June 2006 Srowe -at- cybernet -dot- com Please put "genetic"
March 24, 2004 Will H.264 Live Up to the Promise of MPEG-4 ? Vide / SURA March Marshall Eubanks Chief Technology Officer.
Chapter 5: Workforce. Chartbook 2003 Physician Workforce After dropping slightly in 1999, the number of active physicians per thousand population rose.
Sampling Research Questions
Stratospheric Measurements: Microwave Sounders I. Current Methods – MSU4/AMSU9 Diurnal Adjustment Merging II. Problems and Limitations III. Other AMSU.
13.1 Vis_2003 Data Visualization Lecture 13 Visualization of Very Large Datasets.
Object Recognition Using Locality-Sensitive Hashing of Shape Contexts Andrea Frome, Jitendra Malik Presented by Ilias Apostolopoulos.
Scotch Marine Boiler Design
$100 $200 $300 $400 $100 $200 $300 $400 $100 $200 $300 $400 $100 $200 $300 $400 $100 $200 $300 $400.
PLAnetary Transits and Oscillations of stars H. Rauer 1, C. Catala 2, D. Pollacco 3, S. Udry 4 and the PLATO Team 1: Institut für Planetenforschung, DLR.
Computational Complexity, Choosing Data Structures Svetlin Nakov Telerik Corporation
(SubLoc) Support vector machine approach for protein subcelluar localization prediction (SubLoc) Kim Hye Jin Intelligent Multimedia Lab
On Comparing Classifiers : Pitfalls to Avoid and Recommended Approach
Detection Chia-Hsin Cheng. Wireless Access Tech. Lab. CCU Wireless Access Tech. Lab. 2 Outlines Detection Theory Simple Binary Hypothesis Tests Bayes.
Chapter 10: Applications of Arrays and the class vector
1 Generating Network Topologies That Obey Power LawsPalmer/Steffan Carnegie Mellon Generating Network Topologies That Obey Power Laws Christopher R. Palmer.
Continued Psy 524 Ainsworth
Chapter 16 Goodness-of-Fit Tests and Contingency Tables
Object Recognition using Local Descriptors Javier Ruiz-del-Solar, and Patricio Loncomilla Center for Web Research Universidad de Chile.
Association Rule Mining
Basics & Motivation Technique & Observations Fits & Measurements Summary & Outlook Atmospheric Wave Workshop ESTEC 11/10/2011 Tobias Stangier I st Physics.
Created by Susan Neal $100 Fractions Addition Fractions Subtraction Fractions Multiplication Fractions Division General $200 $300 $400 $500 $100 $200.
A Partition Modelling Approach to Tomographic Problems Thomas Bodin & Malcolm Sambridge Research School of Earth Sciences, Australian National University.
Issues of Simultaneous Tests for Non-Inferiority and Superiority Tie-Hua Ng*, Ph. D. U.S. Food and Drug Administration Presented at MCP.
False positives in the Corot transiting planet search Goal: estimate the amount of ground-based observations necessary for the Corot ground-based follow-up.
Equal or Not. Equal or Not
Slippery Slope
Section 5: More Parallel Algorithms
How do we know when we know. Outline  What is Research  Measurement  Method Types  Statistical Reasoning  Issues in Human Factors.
Detecting changes in von Bertalanffy growth parameters (K, L ∞ ) in a hypothetical fish stock subject to size-selective fishing Otoliths (ear bones) from.
Nov 7, 2013 Lirong Xia Hypothesis testing and statistical decision theory.
Recent Results from NEOWISE Amy Mainzer JPL. WISE 4 imaging channels covering 3 – 25  m 40 cm telescope operating at
1 Wind Shear n Example of little or no vertical wind shear: 7 kts 6 kts 7 kts 6 kts There is very little change in the speed or direction of the wind with.
12-Apr-15 Analysis of Algorithms. 2 Time and space To analyze an algorithm means: developing a formula for predicting how fast an algorithm is, based.
Order Structure, Correspondence, and Shape Based Categories Presented by Piotr Dollar October 24, 2002 Stefan Carlsson.
Principal Component Analysis Based on L1-Norm Maximization Nojun Kwak IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008.
Upper-limit on Sco X-1 S2 preliminary results C Messenger, V Re and A Vecchio on behalf of PULG LSC General Meeting LHO, 10 th – 13 th November 2003.
P. Bordé 1, B. Samuel 1, A. Léger 1, M. Ollivier 1 & D. Rouan 2 1 IAS, Université Paris-Sud 11 2 LESIA, Observatoire de Paris First International Corot.
Photometric follow-up of transiting planet candidates Marton Hidas UNSW.
Oversampling mode Roi Alonso, Pierre-Yves Chabaud Christian Surace, Raphael Cautain, P. Barge.
INTRINSIC PLANETARY FREQUENCIES BASED ON KEPLER OBSERVATIONS William Borucki, NASA Ames Kepler Mission Objectives; Determine the Frequency of Earth-size.
Detection of Terrestrial Extra-Solar Planets via Gravitational Microlensing David Bennett University of Notre Dame.
Properties of binary systems found in the CoRoT exoplanet search J.M. Almenara, H. Deeg, CoRoT Detection WG, CoRoT Ground-Based Follow-Up WG, Planets in.
Two and a half problems in homogenization of climate series concluding remarks to Daily Stew Ralf Lindau.
October 8, 2013Computer Vision Lecture 11: The Hough Transform 1 Fitting Curve Models to Edges Most contours can be well described by combining several.
What stellar properties can be learnt from planetary transits Adriana Válio Roque da Silva CRAAM/Mackenzie.
Adriana V. R. Silva CRAAM/Mackenzie COROT /11/2005.
Hough Transform on the GPU Presented by Javier Garcia November 29, 2010.
Search for planetary candidates within the OGLE stars Adriana V. R. Silva & Patrícia C. Cruz CRAAM/Mackenzie COROT /11/2005.
The experience of BEST Heike Rauer and the BEST Team Institut für Planetenforschung Deutsches Zentrum für Luft- und Raumfahrt e.V. (DLR) and Zentrum für.
Sparse Signals Reconstruction Via Adaptive Iterative Greedy Algorithm Ahmed Aziz, Ahmed Salim, Walid Osamy Presenter : 張庭豪 International Journal of Computer.
Dec 16, 2005GWDAW-10, Brownsville Population Study of Gamma Ray Bursts S. D. Mohanty The University of Texas at Brownsville.
October 16, 2014Computer Vision Lecture 12: Image Segmentation II 1 Hough Transform The Hough transform is a very general technique for feature detection.
CoRoT Week 10, Nice - 06/06/2006 OverSampling Mode C. Quentin, R. Cautain, C. Surace, R. Savalle, J-C. Meunier P.Barge, R. Alonso, M. Deleuil, C. Moutou.
Wonjun Kim and Changick Kim, Member, IEEE
Digital Image Processing Lecture 17: Segmentation: Canny Edge Detector & Hough Transform Prof. Charlene Tsai.
Machine Vision Edge Detection Techniques ENT 273 Lecture 6 Hema C.R.
The joint influence of break and noise variance on break detection Ralf Lindau & Victor Venema University of Bonn Germany.
Comparing Groups April 6-7, 2017 CS 160 – Section 10.
Circumbinary Planet Detection with PLATO
IAU253 Transiting Planets: May
Observation of Diffractively Produced W- and Z-Bosons
Fitting Curve Models to Edges
Pre-Cursor Data Needed for JWST Transit and Eclipse Observations
Goal: To learn about the Kepler Mission and the Transit Method
UVIS Calibration Update
Feature extraction and alignment for LC/MS data
Observation of Diffractively Produced W- and Z-Bosons
Presentation transcript:

Planet WS Berlin 12/2003 Transit detecion on eclipsing binary systems Hans J. Deeg Instituto de Astrofísica de Canarias 1.Transit observations in eclipsing binary systems, why and how 2.Transit detection algorithms for EB's 3.Considerations for computing requirements

Planet WS Berlin 12/2003 Why are EB transits of interest to be observed The probability to detect planets around EB's with transits is higher: Since the EB plane is aligned with the observer, the planetary one is it likely as well ->higher probability for correct alignement Non-aligned planetary planes would precess around EB rotation axis, and cause periodically transits as well (Schneider, 1994) Detection of circumbinary planets would be a very significant extension to the variety of known planets.

Planet WS Berlin 12/2003 EB Transit lightcurves 1 Time and duration of transits depends on phase of EB during transit.

Planet WS Berlin 12/2003 EB transit lightcurves 2: after removal of binary eclipse 'Normal' case: planet transits one component complicated shapes: planetary transit during binary eclipse A detection algorithm needs to model these curves by considering all possible periods AND phases of EB at planet crossing P=9d P=36d For longer periodic planets v pl << v EB : trains of several irregularly spaced transits per EB crossing v p l < v EB v pl > v EB

Planet WS Berlin 12/2003 EB Transit detection algorithm (TDA) development Jenkins,Doyle & Cullers 1996: propose a matching filter TDA.The detection statistics C is obtained from a scalar multiplication of the vectors representing model-lc and observed data. Jenkins et al. present this TDA in the context of the transit detection of EBs TEP project ( ; Deeg et al 1998,Doyle et al 2000.) detection statistics is obtained from a comparison of hypotheses transit-present vs transit-absent.

Planet WS Berlin 12/2003 EB transit detecion algorithm, general There is nothing special about them, except for the first step: i) removal of binary eclipses from lightcurve. Use fit obtained from several good observations of eclipses. ii) model signal for Planet(E i,P i,{R pl,ecc}) E: Epoch [E 0,..,E 0 +P] (or phase [0..2  ]) iii) compare model to data -> detection statistical value C(E i,P i ) within search space [E 0,..,E 0 +P], and [Pmin..Pmax], (R pl, ecc), repeat steps ii) and iii) for all values of E, P that generate significantly different model signals

Planet WS Berlin 12/2003 EB transit detecion vs single star TD Difference of EB algorithms is in the generation of model transits and in finding an efficient way to compare them to the data. -> EB transit detection is a much larger computational problem than single star TD on the order of the number of epoch steps needed For single star transits: Transits are strictly periodic. Transit shape, duration is invariant against Epoch (slowly varying if eccentric) EB transits: Transits are semi-periodic and shapes/durations depend strongly on planet's epoch

Planet WS Berlin 12/2003 The TEP project , Doyle, Deeg, Schneider, Jenkins et al. Observed CM Dra for 1000hrs (Deeg et al 1998, Doyle et al 2000) TDA: searched data from for P=7-60 days over all phases of CM Dra ~ 2 month computing time with Sun Ultra5 (300MHz) a list of planet candidates was published. Follow-up observations at predicted transit times did rule out all of them

Planet WS Berlin 12/2003 Algorithm used in TEP Adapted for analysis of ground based data with unknown extinction slope: Compares data to two models: -one with transits (E,P) and extinction fits to all nights (H1) -one without transit, only extinction fit to all nights (H0) Detection stat. C is difference between residuals: C(E,P) =  |resid H0 | – |resid H1 | This algorithm is NOT specific for eclipsing binaries – but for problematic ground-based data! Doyle et al. 2000

Planet WS Berlin 12/2003 A transit candidate with P ≈ 11d Doyle et al. 2000

Planet WS Berlin 12/2003 Computational Requirements: TD stepsize Minimum TD step sizes to not miss a transit (Jenkins et al. 1996): If a planet (E,P) is tested by a detection test with (P±  P,E) or (P,E±  E), its detection statistics is reduced 'signficantly' (to 1/2 in case of correlation). Planet (E,P) Model (E,P) c=1 Model (E+  E,P),  E=t d /2 tdtd P t obs c=1/2 Model (E,P+  P),  P=P t d /t obs c=1/2 detc. stat.

Planet WS Berlin 12/2003 Computational Requirements 1 Number of epoch and period steps in TDA: t obs : duration of observations t d : duration of transit (  fundamental mesh size) P : period of planet in Epoch:  E=t d /2 -> N E = P /  E = 2 P /t d in Period:  P=P t d /t obs -> N P = (t obs /t d ) ln(P max /P min ) total: N step ≈ 2 t obs t d - 2 (P max - P min ) Ignored here is a weak dependency of t d ~ P 1/3

Planet WS Berlin 12/2003 Computational Requirements 2 Total Number of operations of TDA: N op = k N step N pts, = k 2 t obs t d - 2 (P max - P min ) N pts where k is a factor depending on the TDA used (k might be < 1), N pts is the number of data pts. For equidistant points: N pts = t obs /t inc and then: N op = k 2 t d -2 (P max -P min ) t obs 2 t inc -1 (1) Examples: (k =1, P min =7d, P max =60d, t d =0.0139d = 20min) TEP: N op ≈ (t obs = 1526d, N pts =24 874) COROT: N op ≈ (t obs =150d, N pts =13 500, t inc =16min) Kepler/Eddi: N op ≈ ( t obs =1100d, N pts = , t inc =10min)

Planet WS Berlin 12/2003 Slicing the lightcurve: saving computing steps Principle: running a TDA on shorter sections of a lightcurve, and co-adding detection statistics from individual sections. If a lightcurve of N pts points is divided into n s sections of N pts,s points, of length t obs,s, the computing requirement N op,s for each section is: N op,s = k 2 t d -2 (P max -P min ) t obs,s N pts,s. The total number of operations is then given by: N op = n s N op,s = k 2 t d -2 (P max -P min ) t obs,s N pts (2) i.e. N op is reduced by t obs /t obs,s against (1) This scheme was used by TEP project, dividing the observations of t obs =1529d into 5 sections corresponding to the yearly ( ) observing seasons, using t obs = 200d for each of them (mid-march to october). Savings: 1526/200 ≈ 7.6 Only this made the computation feasable with a CPU time of 2 months

Planet WS Berlin 12/2003 Slicing: Adding of detection statistics 1994 (186h) 1995 (185h) 1996 (246h) 1997 (250h)  P E maps C(E,P) of small region around a candidate with P ≈ 36d, using t d = 20min, t obs =200d  epoch E is an offset {0..P} from reference Epoch JD 0 = (9 Oct 1995) Bright: high C(E,P) 36,1d 36,7d JD 0 +0JD 0 +3 dominant lines are caused by individual features in lightcurve coadded C(E,P) arrays: Once maxima in C(E i,T i ) map are found, TDA needs to be rerun with a finer mesh (t d,fine ≈ 1/4 t d ) in small area in E,P around all maximum to find best value of C in (E,P) space. In TEP, this fine-mesh search was performed around the 5000 highest maxima -> final list of planet candidates, subject to follow-up observations

Planet WS Berlin 12/2003 Slicing for equidistant data starting from (2) : N op = k 2 t d -2 (P max -P min ) t obs,s N pts with N pts = t obs /t inc the total number of computations is: N op = k 2 t d -2 (P max -P min ) t obs,s t obs t inc -1 (3) now, N op  t obs, instead of N op  t obs 2 without 'slicing'. Savings from slicing is  n s.. Drawbacks: added complexity. Slicing allows running the TDA on shorter sections of data, with the possiblitity to add matrizes C of later data at any time (useful for missions with long t obs )

Planet WS Berlin 12/2003 Analysis of COROT EB data vs TEP -data span only 150 days, not 5 years as in TEP -homogenous noise characteristics in data set -no atmosphere with ‘unkown’ extinction -computers became faster With O(100) EB's to analyze in COROT fields, no serious computing obstacles are expected. Selection of best TDA-kernel (matched filter, BLS,..) is TBD and requires lightcurves with realistic noise simulations.

Planet WS Berlin 12/2003 Summary 1. algorithms to analyze EB lightcurves: fundamentally similar to single stars; need to model more complex transits; more 'tricky' to obtain efficient transit detection, as there are fewer invariants in transit lightcurves. 2. Slicing the data into smaller sections improves computational efficiency of TDA, adding complexity. Also, allows efficient adding of additional data. ¿How useful may this method be to single star TD? - TBD 3. COROT EB analysis is not expected to cause serious difficulties. Selection of best TDA is TBD

Planet WS Berlin 12/2003 References Deeg, H.J., et al.1998, A&A 338, 479 Doyle, L.R., et al. 2000, ApJ, 535, 338 Jenkins, J. M.; Doyle, L. R.; Cullers, D. K. 1996, Icarus 119, 244 Schneider, J. 1994, Planet Space Sci. 42, 539