Stationary Stochastic Processes Table of Formulas, 2016

Σχετικά έγγραφα
Stationary Stochastic Processes Table of Formulas, 2017

Probability and Random Processes (Part II)

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 4, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 2

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 4, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 4

Statistics 104: Quantitative Methods for Economics Formula and Theorem Review

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 4, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 3

Introduction to the ML Estimation of ARMA processes

Solution Series 9. i=1 x i and i=1 x i.

Aquinas College. Edexcel Mathematical formulae and statistics tables DO NOT WRITE ON THIS BOOKLET

HW 3 Solutions 1. a) I use the auto.arima R function to search over models using AIC and decide on an ARMA(3,1)

Chapter 5, 6 Multiple Random Variables ENCS Probability and Stochastic Processes

Solutions to Exercise Sheet 5

6.3 Forecasting ARMA processes

HMY 799 1: Αναγνώριση Συστημάτων

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 3, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 3

5.4 The Poisson Distribution.

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 3, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 5

Homework 8 Model Solution Section

HOMEWORK#1. t E(x) = 1 λ = (b) Find the median lifetime of a randomly selected light bulb. Answer:

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 3, Νέα Κτίρια ΣΗΜΜΥ Θεωρία Πιθανοτήτων & Στοχαστικές Ανελίξεις - 1

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

Main source: "Discrete-time systems and computer control" by Α. ΣΚΟΔΡΑΣ ΨΗΦΙΑΚΟΣ ΕΛΕΓΧΟΣ ΔΙΑΛΕΞΗ 4 ΔΙΑΦΑΝΕΙΑ 1

Linear System Response to Random Inputs. M. Sami Fadali Professor of Electrical Engineering University of Nevada

Durbin-Levinson recursive method

Introduction to Time Series Analysis. Lecture 16.

List MF20. List of Formulae and Statistical Tables. Cambridge Pre-U Mathematics (9794) and Further Mathematics (9795)

Estimation for ARMA Processes with Stable Noise. Matt Calder & Richard A. Davis Colorado State University

Biostatistics for Health Sciences Review Sheet

Asymptotic distribution of MLE

Second Order Partial Differential Equations

CT Correlation (2B) Young Won Lim 8/15/14

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

Limit theorems under sublinear expectations and probabilities

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

Fundamentals of Signals, Systems and Filtering

P AND P. P : actual probability. P : risk neutral probability. Realtionship: mutual absolute continuity P P. For example:

FORMULAS FOR STATISTICS 1

SPECIAL FUNCTIONS and POLYNOMIALS

HMY 799 1: Αναγνώριση Συστημάτων

4.6 Autoregressive Moving Average Model ARMA(1,1)

Nachrichtentechnik I WS 2005/2006

APPENDICES APPENDIX A. STATISTICAL TABLES AND CHARTS 651 APPENDIX B. BIBLIOGRAPHY 677 APPENDIX C. ANSWERS TO SELECTED EXERCISES 679

Outline Analog Communications. Lecture 05 Angle Modulation. Instantaneous Frequency and Frequency Deviation. Angle Modulation. Pierluigi SALVO ROSSI

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

w o = R 1 p. (1) R = p =. = 1

2. THEORY OF EQUATIONS. PREVIOUS EAMCET Bits.

Bayesian statistics. DS GA 1002 Probability and Statistics for Data Science.

D Alembert s Solution to the Wave Equation

MOTORCAR INSURANCE I

Module 5. February 14, h 0min

A Bonus-Malus System as a Markov Set-Chain. Małgorzata Niemiec Warsaw School of Economics Institute of Econometrics

Ηλεκτρονικοί Υπολογιστές IV


Lecture 7: Overdispersion in Poisson regression

6. MAXIMUM LIKELIHOOD ESTIMATION

ST5224: Advanced Statistical Theory II

Fundamentals of Probability: A First Course. Anirban DasGupta

DERIVATION OF MILES EQUATION FOR AN APPLIED FORCE Revision C

Ψηφιακές Επικοινωνίες

Διάλεξη 6. Fourier Ανάλυση Σημάτων. (Επανάληψη Κεφ Κεφ. 10.3, ) Ανάλυση σημάτων. Τι πρέπει να προσέξουμε

Web-based supplementary materials for Bayesian Quantile Regression for Ordinal Longitudinal Data

CHAPTER 101 FOURIER SERIES FOR PERIODIC FUNCTIONS OF PERIOD

Probability theory STATISTICAL METHODS FOR SAFETY ANALYSIS FMS065 TABLE OF FORMULÆ (2016) Basic probability theory. One-dimensional random variables

ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 3, Νέα Κτίρια ΣΗΜΜΥ Ανάλυση Επικοινωνιακών Σημάτων κατά Fourier


Section 8.3 Trigonometric Equations

Gaussian related distributions

The circle theorem and related theorems for Gauss-type quadrature rules

Additional Results for the Pareto/NBD Model

Math221: HW# 1 solutions

A Two-Sided Laplace Inversion Algorithm with Computable Error Bounds and Its Applications in Financial Engineering

Lecture 21: Properties and robustness of LSE

Stationary ARMA Processes

HMY 429: Εισαγωγή στην Επεξεργασία Ψηφιακών. Χρόνου (Ι)

ON NEGATIVE MOMENTS OF CERTAIN DISCRETE DISTRIBUTIONS


ΣΤΟΧΑΣΤΙΚΑ ΣΥΣΤΗΜΑΤΑ & ΕΠΙΚΟΙΝΩΝΙΕΣ 1o Τμήμα (Α - Κ): Αμφιθέατρο 4, Νέα Κτίρια ΣΗΜΜΥ Ανάλυση Επικοινωνιακών Σημάτων κατά Fourier

Using Long-Run Consumption-Return Correlations to Test Asset Pricing Models : Internet Appendix

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

Fourier Analysis of Waves

Uniform Convergence of Fourier Series Michael Taylor

5. Choice under Uncertainty

Stationary ARMA Processes

Jesse Maassen and Mark Lundstrom Purdue University November 25, 2013

HMY 799 1: Αναγνώριση Συστημάτων

Ανάλυση Θορύβου Σε Γραμμικά Κυκλώματα

Exercises to Statistics of Material Fatigue No. 5

Lecture 34 Bootstrap confidence intervals

2. ARMA 1. 1 This part is based on H and BD.

EE512: Error Control Coding

Srednicki Chapter 55

ESTIMATION OF SYSTEM RELIABILITY IN A TWO COMPONENT STRESS-STRENGTH MODELS DAVID D. HANAGAL

ΗΜΥ 220: ΣΗΜΑΤΑ ΚΑΙ ΣΥΣΤΗΜΑΤΑ Ι Ακαδημαϊκό έτος Εαρινό Εξάμηνο Κατ οίκον εργασία αρ. 2

Anti-Final CS/SE 3341 SOLUTIONS

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

1. (a) (5 points) Find the unit tangent and unit normal vectors T and N to the curve. r(t) = 3cost, 4t, 3sint

ECE598: Information-theoretic methods in high-dimensional statistics Spring 2016

Notations. Primary definition. Specific values. General characteristics. Series representations. Traditional name. Traditional notation

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

Other Test Constructions: Likelihood Ratio & Bayes Tests

Transcript:

Stationary Stochastic Processes, 06 Stationary Stochastic Processes Table of Formulas, 06 Basics of probability theory The following is valid for probabilities: P(Ω), where Ω is all possible outcomes 0 P(A), where A is some event P(A c ) P(A), where A c is the complement of A P(A B) P(A) + P(B), if the events A and B are mutually exclusive The addition law of probability: P(A B) P(A) + P(B) P(A B) The conditional probability: P(B A) P(A B) P(A) A and B are independent P(A B) P(A) P(B) Stochastic variables p X (k) k x Distribution functions: F X (x 0 ) P(X x 0 ) 0 x0 f X (x) dx k p X (k) k Expected value: E[X] m X x f X (x) dx (X discrete) (X continuous) (X discrete) (X continuous) (k m X ) p X (k) (X discrete) Variance: V[X] E[X ] m X k (x m X ) f X (x) dx (X continuous) Rules for expected value and variance (a and b constants): E[aX + b] ae[x] + b V[aX] a V[X] V[X + b] V[X] E[X + Y] E[X] + E[Y] V[X + Y] V[X] + V[Y] + C[X, Y]

Stationary Stochastic Processes, 06 Covariance: C[X, Y] E[(X m X )(Y m Y )] E[XY] m X m Y Correlation coefficient: ρ[x, Y] C[X, Y] V[X] V[Y] Taylor series expansions ("Gauss approximations"): E[g(X,..., X n )] g(e[x ],..., E[X n ]) V[g(X,..., X n )] c i V[X i ] + c i c j C[X i, X j ] i<j where c i g (x,..., x n ) x i xk E[X k ], k i The two-dimensional probability density function of a jointly Gaussian random variable, (X, Y), with E[X] E[Y] 0, V[X] V[Y] and C[X, Y] ρ is { } f X,Y (x, y) π ρ exp ( ρ ) (x ρxy + y ) Stationary stochastic processes Estimation of expected value: ˆm n n t X t V [ ˆm n ] n n nv [ ˆm n ] τ τ n+ r X (τ) (n τ )r X (τ) If ˆm n N(m, V[ ˆm n ]), the confidence interval for m is for large n I m : { ˆm n λ α/ V[ ˆmn ], ˆm n + λ α/ V[ ˆmn ]} with confidence level α. For confidence level 0.95, α 0.05 and λ α/ λ 0.05.96. Estimation of covariance function: ˆr n (τ) n τ (X t m X )(X t+τ m X ) for τ 0 n t

Stationary Stochastic Processes, 06 3 where m X is replaced by ˆm n if m X is unknown. If X t is Gaussian, with mean m X and covariance function r X (τ), such that τ0 r X(τ) <, then for t s + τ, nc [ˆr n (s),ˆr n (t)] u The Poisson process and the Wiener process {r X (u)r X (u + τ) + r X (u s)r X (u + t)} when n A simply increasing process {X(t), t 0} is a homogeneous Poisson process, if X(0) 0 and X(t) has stationary, independent increments. If the intensity is λ, E[X(t)] λt V[X(t)] λt r X (s, t) λ min(s, t) The interarrival times are independent and exponentially distributed with mean value /λ. A Gaussian process {X(t), t 0} is a Wiener process, if X(0) 0, and X(t) has independent increments, where X(t) X(t + h) N(0, σ h), E[X(t)] 0 V[X(t)] σ t r X (s, t) σ min(s, t) Spectral representations Relations between covariance function r X (τ) and spectral density R X (f): Continuous time Discrete time r X (τ) R X(f)e iπfτ df r X (τ) / / R X(f)e iπfτ df R X (f) r X(τ)e iπfτ dτ R X (f) τ r X(τ)e iπfτ Folding (aliasing): Let {Z t, t 0, ±d, ±d,... } be the continuous time process Y(t) sampled with time interval d and sampling frequency f s /d: R Z (f) R Y (f + kf s ) for f s / < f f s / k

4 Stationary Stochastic Processes, 06 Sum of harmonic components with random phase and amplitude: X(t) A 0 + A k cos(πf k t + φ k ) where φ k Rect(0, π), A k, k 0,..., n, are independent and E[A 0 ] 0. Covariance function: k r X (τ) σ 0 + where σ 0 E [A 0] and σ k E [A k ] /. Spectral density: σ k cos πf k τ k R X (f) k n b k δ fk (f), where b 0 σ 0 E [A 0], and b k σ k / E [A k ] /4. Linear filters - general theory Impulse response h(u): Y(t) h(u)x(t u) du u h(u)x(t u) (continuous time) (discrete time) Relation between covariance functions: h(u)h(v) r X(τ + u v) du dv r Y (τ) v h(u)h(v) r X(τ + u v) u (continuous time) (discrete time) Relation between spectral densities: R Y (f) H(f) R X (f) where H(f) is the frequency function corresponding to the impulse response h(n). Differentiation: X (t) exists (in quadratic mean) if r X (t) exists. This is equivalent

Stationary Stochastic Processes, 06 5 to (πf) R(f)df <. If X (t) exists, the following relations hold: r X (τ) r X(τ) R X (f) (πf) R X (f) V [X (t)] (πf) R X (f) df r X,X (τ) r X(τ) r X (j),x (k)(τ) ( )j r (j+k) X (τ) Integration: [ E ] g(s)x(s) ds g(s)e[x(s)] ds [ C g(s)x(s) ds, ] h(t)y(t) dt g(s)h(t) C[X(s), Y(t)] ds dt Cross-covariance and cross-spectrum: r X,Y (τ) C[X(t), Y(t + τ)] e iπfτ R X,Y (f) df R X,Y (f) H(f)R X (f) A X,Y (f)e iφ X,Y(f) where A X,Y (f) is the amplitude spectrum and Φ X,Y (f) the phase spectrum. The squared coherence spectrum is κ X,Y(f) A X,Y (f) R X (f)r Y (f) AR- MA- and ARMA-models White noise in discrete time: {e t, t 0, ±,...}, E[e t ] 0 and V[e t ] σ : R e (f) σ for / f / AR(p)-process: (a 0 ) X t + a X t + a X t +... + a p X t p e t

6 Stationary Stochastic Processes, 06 Yule-Walker equations for covariance function: r X (k) + a r X (k ) +... + a p r X (k p) Spectral density: { σ for k 0 0 for k,,... MA(q)-process: (b 0 ) R X (f) σ p k0 a ke iπfk Covariance function: Spectral density: X t e t + b e t + b e t +... + b q e t q r X (τ) Matched filter and Wiener filter Matched filter: with white noise: with colored noise: s(t u) c { σ j kτ b jb k for τ q 0 for τ > q q R X (f) σ b k e iπfk k0 h(u) c s(t u) SNR N 0 s(t u) du SNR c h(v)r N (u v) dv h(u)h(v)r N (u v) du dv Wiener filter: H(f) SNR R S (f) R S (f) + R N (f) RS (f) df RS (f)r N (f) R S (f)+r N (f) df

Stationary Stochastic Processes, 06 7 Spectral estimation Periodogram of the sequence {x(t), t 0,,,... n }, where X (f) n t0 x(t)e iπft. ] E [ˆRx (f) ˆR x (f) X (f) n τ / / k n (τ)r X (τ)e iπfτ K n (f u)r X (u)du where k n (τ) τ for n + τ n and K n n(f) n τ n+ k n(τ)e iπfτ. ] { R V [ˆRx (f) X (f) for 0 < f < / R X (f) for f 0, ±/ The distribution of the periodogram estimate is ˆR x (f) R X (f) χ () for 0 < f < / Modified periodogram ˆR w (f) n x(t)w(t)e iπft n t0 / X (ν)w(f ν)dν n Lag-windowing Averaging of spectrum ˆR lw (f) / τ / / ˆR av (f) K k Ln (τ)ˆr x (τ)e iπfτ K Ln (f ν)ˆr x (ν)dν K ˆR x,j (f) where K different spectrum estimates, ˆR x,j (f), j... K, are used. The distribution is ˆR av (f) R X (f) χ (K) for 0 < f < / K j

8 Stationary Stochastic Processes, 06 Fourier transforms g(τ) (α > 0) G(f) e iπfτ g(τ) dτ e α τ α α +(πf) α +τ π α e πα f τ e α τ (α (πf) ) (α +(πf) ) τ k e α τ k! {(α + iπf) k+ + (α iπf) k+ } (α +(πf) ) k+ e ατ π/α exp( (πf) 4α ) e α τ cos(πf 0 τ) e α τ sin(πf 0 τ) { α if τ 0 sin(πατ) πτ if τ 0 { α τ if τ α 0 if τ > α g(τ)h(τ) g(τ) h(τ) g(t)h(τ t)dt g (τ) α α α +(πf 0 + πf) α +(πf 0 +πf) πf 0 πf α +(πf 0 + πf 0+πf πf) α +(πf 0 +πf) { / if f α 0 if f > α { α ( ( if f 0 α (πf) cos πf )) α if f 0 G(f) H(f) G(ν)H(f ν)dν G(f)H(f) iπf G(f) g(ατ) α G( f α ) α g( τ α ) G(αf) g(τ τ 0 ) G(f)e iπfτ 0 g(τ)e iπf 0τ G(f f 0 )

Stationary Stochastic Processes, 06 9 Gaussian distribution table F(x) Φ(x) x 0.00 0.0 0.0 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.0 0.50000 0.50399 0.50798 0.597 0.5595 0.5994 0.539 0.5790 0.5388 0.53586 0. 0.53983 0.54380 0.54776 0.557 0.55567 0.5596 0.56356 0.56749 0.574 0.57535 0. 0.5796 0.5837 0.58706 0.59095 0.59483 0.5987 0.6057 0.6064 0.606 0.6409 0.3 0.679 0.67 0.655 0.6930 0.63307 0.63683 0.64058 0.6443 0.64803 0.6573 0.4 0.6554 0.6590 0.6676 0.66640 0.67003 0.67364 0.6774 0.6808 0.68439 0.68793 0.5 0.6946 0.69497 0.69847 0.7094 0.70540 0.70884 0.76 0.7566 0.7904 0.740 0.6 0.7575 0.7907 0.7337 0.73565 0.7389 0.745 0.74537 0.74857 0.7575 0.75490 0.7 0.75804 0.765 0.7644 0.76730 0.77035 0.77337 0.77637 0.77935 0.7830 0.7854 0.8 0.7884 0.7903 0.79389 0.79673 0.79955 0.8034 0.805 0.80785 0.8057 0.837 0.9 0.8594 0.8859 0.8 0.838 0.8639 0.8894 0.8347 0.83398 0.83646 0.8389.0 0.8434 0.84375 0.8464 0.84849 0.85083 0.8534 0.85543 0.85769 0.85993 0.864. 0.86433 0.86650 0.86864 0.87076 0.8786 0.87493 0.87698 0.87900 0.8800 0.8898. 0.88493 0.88686 0.88877 0.89065 0.895 0.89435 0.8967 0.89796 0.89973 0.9047.3 0.9030 0.90490 0.90658 0.9084 0.90988 0.949 0.9309 0.9466 0.96 0.9774.4 0.994 0.9073 0.90 0.9364 0.9507 0.9647 0.9785 0.99 0.93056 0.9389.5 0.9339 0.93448 0.93574 0.93699 0.938 0.93943 0.9406 0.9479 0.9495 0.94408.6 0.9450 0.94630 0.94738 0.94845 0.94950 0.95053 0.9554 0.9554 0.9535 0.95449.7 0.95543 0.95637 0.9578 0.9588 0.95907 0.95994 0.96080 0.9664 0.9646 0.9637.8 0.96407 0.96485 0.9656 0.96638 0.967 0.96784 0.96856 0.9696 0.96995 0.9706.9 0.978 0.9793 0.9757 0.9730 0.9738 0.9744 0.97500 0.97558 0.9765 0.97670.0 0.9775 0.97778 0.9783 0.9788 0.9793 0.9798 0.98030 0.98077 0.984 0.9869. 0.984 0.9857 0.98300 0.9834 0.9838 0.984 0.9846 0.98500 0.98537 0.98574. 0.9860 0.98645 0.98679 0.9873 0.98745 0.98778 0.98809 0.98840 0.98870 0.98899.3 0.9898 0.98956 0.98983 0.9900 0.99036 0.9906 0.99086 0.99 0.9934 0.9958.4 0.9980 0.990 0.994 0.9945 0.9966 0.9986 0.99305 0.9934 0.99343 0.9936.5 0.99379 0.99396 0.9943 0.99430 0.99446 0.9946 0.99477 0.9949 0.99506 0.9950.6 0.99534 0.99547 0.99560 0.99573 0.99585 0.99598 0.99609 0.996 0.9963 0.99643.7 0.99653 0.99664 0.99674 0.99683 0.99693 0.9970 0.997 0.9970 0.9978 0.99736.8 0.99744 0.9975 0.99760 0.99767 0.99774 0.9978 0.99788 0.99795 0.9980 0.99807.9 0.9983 0.9989 0.9985 0.9983 0.99836 0.9984 0.99846 0.9985 0.99856 0.9986 3.0 0.99865 0.99869 0.99874 0.99878 0.9988 0.99886 0.99889 0.99893 0.99896 0.99900 3. 0.99903 0.99906 0.9990 0.9993 0.9996 0.9998 0.999 0.9994 0.9996 0.9999 3. 0.9993 0.99934 0.99936 0.99938 0.99940 0.9994 0.99944 0.99946 0.99948 0.99950 3.3 0.9995 0.99953 0.99955 0.99957 0.99958 0.99960 0.9996 0.9996 0.99964 0.99965 3.4 0.99966 0.99968 0.99969 0.99970 0.9997 0.9997 0.99973 0.99974 0.99975 0.99976 3.5 0.99977 0.99978 0.99978 0.99979 0.99980 0.9998 0.9998 0.9998 0.99983 0.99983 3.6 0.99984 0.99985 0.99985 0.99986 0.99986 0.99987 0.99987 0.99988 0.99988 0.99989 3.7 0.99989 0.99990 0.99990 0.99990 0.9999 0.9999 0.9999 0.9999 0.9999 0.9999 3.8 0.99993 0.99993 0.99993 0.99994 0.99994 0.99994 0.99994 0.99995 0.99995 0.99995 3.9 0.99995 0.99995 0.99996 0.99996 0.99996 0.99996 0.99996 0.99996 0.99997 0.99997 4.0 0.99997 0.99997 0.99997 0.99997 0.99997 0.99997 0.99998 0.99998 0.99998 0.99998