資源描述:
《Time Series Analysis時(shí)間系列分析》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、TimeSeriesAnalysisOutline1Timeseriesinastronomy2Frequencydomainmethods3Timedomainmethods4ReferencesTimeseriesinastronomyPeriodicphenomena:binaryorbits(stars,extrasolarplanets);stellarrotation(radiopulsars);pulsation(helioseismology,Cepheids)Stochasticphenomena:accretion(C
2、Vs,X-raybinaries,Seyfertgals,quasars);scintillation(interplanetary&interstellarmedia);jetvariations(blazars)Explosivephenomena:thermonuclear(novae,X-raybursts),magneticreconnection(solar/stellar?ares),stardeath(supernovae,gamma-raybursts)Di?cultiesinastronomicaltimeseries
3、Gappeddatastreams:Diurnal&monthlycycles;satelliteorbitalcycles;telescopeallocationsHeteroscedasticmeasurementerrors:Signal-to-noiseratiodi?ersfrompointtopointPoissonprocesses:Individualphoton/particleeventsinhigh-energyastronomyImportantFourierFunctionsDiscreteFourierTran
4、sformXn?1/2d(ωj)=nxtexp(?2πitωj)t=1XnXn?1/2?1/2d(ωj)=nxtcos(2πiωjt)?inxtsin(2πiωjt)t=1t=1Classical(Schuster)Periodogram2I(ωj)=
5、d(ωj)
6、SpectralDensityhX=∞f(ω)=exp(?2πiωh)γ(h)h=?∞Fourieranalysisrevealsnothingoftheevolutionintime,butratherrevealsthevarianceofthesignalatdi?ere
7、ntfrequencies.Itcanbeprovedthattheclassicalperiodogramisanestimatorofthespectraldensity,theFouriertransformoftheautocovariancefunction.Formally,theprobabilityofaperiodicsignalinGaussiannoiseisP∝ed(ωj)/σ2.GingaobservationsofX-raybinaryGX5-1GX5-1isabinarystarsystemwithgasfr
8、omanormalcompanionaccretingontoaneutronstar.HighlyvariableX-raysareproducedintheinneraccretiondisk.XRBtimeseriesoftenshow‘rednoise’and‘quasi-periodicoscillations’,probablyfrominhomogeneitiesinthedisk.Weplotbelowthe?rst5000of65,536countratesfromGingasatelliteobservationsdu
9、ringthe1980s.gx=scan(”?/Desktop/CASt/SumSch/TSA/GX.dat”)t=1:5000plot(t,gx[1:5000],pch=20)FastFourierTransformoftheGX5-1timeseriesrevealsthe‘rednoise’(highspectralamplitudeatsmallfrequencies),theQPO(broadenedspectralpeakaround0.35),andwhitenoise.f=0:32768/65536I=(4/65536)*
10、abs(?t(gx)/sqrt(65536))?2plot(f[2:60000],I[2:60000],type=”l”,xlab=”Frequency”)Limitationsofthesp