APPLIED PROBABILITY PAS PNR Os Mario LefebvreMario Lefebvre Département de mathématiques et de génie industriel Ecole Polytechnique de Montréal, Québec C.P. 6070, succ. Centre-ville Montréal H3C 3A7 Canada [email protected] Cover design by Mary Burgess. Mathematics Subject Classification (2000): 60-01. 62-01 Library of Congress Control Number: 2006923352 ISBN-10 0-387-28454-0 eISBN 0-387-28505-9 ISBN-13 978-387-28454-5 Printed on acid-free paper. ©2006 Springer Science + Business Media LLC All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media LLC, 233 Spring Street, New York, NY 10013, USA) and the author, except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology aow known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed in the United States of America, (TXQUEB) 987654321 springercomaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.12 Examples of Applications 3 Figure 1.1, Example of a system. dx(t) = ~ax(t) dt + bu(t) dt, a.) x(0) where a (> 0) and b are constants and the variable u(r) is called the control. Whittle (see [24]) and the author used the equations above in research papers as a rudimentary model for the displacement of an airplane that is preparing to land. When we cannot predict the result of an experiment repeated under the same con- ditions, we say that it is a random experiment. We can then make a list of all possible ‘outcomes and try to compute the probability of each of these possible outcomes. In this textbook, we are only interested in random experiments. Probability is used in practically every pure or applied science. Examples in en- gineering, particularly in electrical engineering, where we must resort to the calculus of probabilities are the following a) Many systems may be represented by a number of components placed in series or in parallel. For example, consider the system described by the diagram in Fig 1.1. In reliability, we must be able to calculate the probability that such a system will function during a certain period of time, or at a given instant. In the latter case, we must then know, for each component, the probability that it functions at this instant and take into account the fact that the components perhaps do not operate independently from one another. In the case of reliability during a given period of time, we must know the distribution of the lifetime of each component. b) In communication, we must often take into account the “noise” present in a sys- tem. For example, suppose that a system transmits either a 0 or a 1, and that there is a risk p that the number transmitted will not be received correctly (see Fig. 1.2). We may be interested in computing the probability that a 0 has been transmitted, given that a 0 has been received, or that a transmission error has occurred, etc. c) In automatic control, if we take the random perturbations into account, then the model (1.1) above becomes dx(t) = —ax(t) dt + bu(t) dt +dWw, (1.2) where W(t) is called a Brownian motion (or a Wiener process). We also say that dW(t)/dt is a Gaussian white noise. Equation (1.2) is an example of a stochastic differential equation. 4) In computer science, probsbility is used, in particular, to help us make decisions in expert systems. We also make use of probability in simulation and in artificial intelligence, as well as in the field of queueing theory.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.12 2 Elementary Probabilities Discrete Sample Spaces If Sis a discrete sample space, we may write that S=ferer--h where e, is a possible outcome (or an elementary event). Let A C S; then the proba- bility of event A can be obtained by making use of the following formula: P(A] = So Pte}. Q.5) kereA If the number n of elementary events eg is finite and if the e,’s are equiprobable (or equally likely), so that P[{ex}] = 1/n Vk, then we may write that P[A) = n(A)/n, (2.6) where n(A) is the number of elementary events in A. Example 2.2.1 i) In the case of the sample space S; in Example 2.1.1, the four elementary events are not equiprobable. If we denote the probability P[{0}] by p, then we may write that PUM) = PLZ) =3p and PU{3}] = p. Moreover, 3 Pu = 1 p= 1/8. i= Hence, we find that P(Ai) = PI{1]] = 3/8 (41/4). On the other hand, the ex's are equally likely in the case of S) and we may write, directly, that P[A2] = PILABB, BAB, BBA)] = 3/8. Note that the probabilities P[A;] and P[A2] must be equal because the events Aj and ‘Az correspond to the same outcome of the three draws in the random experiments E} and E2. Thus, to calculate the probability of getting exactly one brand A transistor in three draws at random and with replacement from a box containing 10 brand A and 10 brand B transistors, it is simpler to consider the sample space S2 for which the outcomes are equally likely. ii) Since the sample space $3 is denumerably infinite, the ex's cannot be equiprobsble. We can show that P{A3] = PU{S, 6}] = PI(S}] + P[{6}] = (10/20)°(10/20) + (1/2)%(1/2). The second equality above is obtained by incompatibility of the events {5} and {6}, while the third equality results from the notion of independence (which will be dis- cussed in Section 2.5).aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.16 2 Elementary Probabilities + nq =n. Then, the number of different permutations of the n objects taken all at once is given by nt! (2.10) ny! ny!no!- ‘The numbers obtained for varying values of n; in this expression are called the multi- nomial coefficients. Example 2.3.3 With the letters a,a,b, b, b and c, we can form 6! 21311! different code words (or access words). We will complete this section with other examples of combinatorial analysis problems, Example 2.3.4 If a letter (or two) may be repeated (only once) in Example 2.3.1, then there are 5! p34.c36c8> Rach = = 7, 893, 600+3, 588, 000-+234, 000 = 11,715, 600 different codes. The first term above corresponds to the case when there are no repe- titions, the second term to the case when there is exactly one repetition and the third one to the case when there are exactly two repetitions. Moreover, we may write that PE = CH .5). Example 2.3.5 With the letters in the word essays, the number of distinct words containing four letters that we can form is given by ar (with 19) +3 “(with2s's) +C74 = (with 38's) = 244364 12 = 72. Example 2.3.6 If we have one one-dollar coin, two quarters and three nickels, then we can pay 2 x 3 x 4 — 1 = 23 different sums exactly. Note that we can employ the principle of multiplication here, because whatever the number of coins of a certain type that we use to pay. the sum will always be different. The problem would be more complicated if we also had a dime, for instance Example 2.3.7 If there are n persons in a class, then the probability that at least two of these n persons have the same birthday is given (disregarding leap years) by 365 Pa 1-3, forn = 2,3,... 365.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.20 2 Elementary Probabilities S 8, 8, By B, Figure 2.5. Example of a partition of a sample space with n = 4 Definition 2.4.2. Let By, B2,... , Bn be events such that i) BB =O F js ii) Ur Be = S. We say that the events By form a partition of the sample space S. Remarks. i) In probability, it is natural to impose in the preceding definition the additional condition: iii) P[Bx] > 0 fork = 1,2,..... ii) A and A® constitute a partition of S. Similarly, ANB, AN BS, ACO B and ACN BY constitute a partition of S. Now, let A C S. We can split up A into n disjoint and exhaustive parts as follows (see Fig. 2.5) A= (ANB) U(AN BU... U(ANBy)- Hence, using Axiom III in the definition of P[A], we obtain the following proposi- tion, Proposition 2.4.3. (Total probability rule) Ler By, B2,... , By be a partition of S. Then, for any AC Swe have: n n PLA] = D0 PLAN Be] = > PLALBeIP LB] if PLBq] > 0 Vk. (2.15) k=l =I Remark. The second equality above follows from the multiplication rule. Finally, making use of Proposition 2.4.2 and the total probability rule, we obtain Bayes’ rule, Proposition 2.4.4. (Bayes' rule) Let A.C S and By, Ba such that P{By] > 0 for k = 1,2,... nm. Then, PLA|B;]PLB)] P[A|By]P[ Ba] Bp be a partition of S P[B;|Al = for j=aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.24 2 Elementary Probabilities rtt}He ER 21H{3 Figure 2.6. Graphical representation of a two-out-of-three system. Example 2.5.2 A box contains 10 brand A diodes and 20 brand B diodes. Ten diodes are taken, at random and with replacement, and the brand of each one is noted. Com- pute the probability of getting at least one brand A diode. Solution. Let the events be as follows: F = at least one brand A diode is obtained and Ay = the kth diode selected is a brand A diode. We want P[F] = PIA, UAQU...U Ajo] = 1 = PLAP NASON... Ato). Now, because the diodes are taken with replacement, the events Ax are (globally) independent. Moreover, P{ Ax] = 10/30 = 1/3 for any k. Then, we have: PIF] = 1 — (2/3)'° ~ 0.9827. Example 2.5.3 (see [14]) In a certain factory, 96% of the computers manufactured conform to the norms. Each computer produced is subjected to two independent quality control operations. We assume that each operation classifies as “good” 98% of the computers that effectively conform to the norms, and 6% of the computers that actually do not conform to the norms. Calculate the probability that a computer sold effectively conforms to the norms. Solution. Let the events be as follows: ‘A = the computer has been classified as “good” twice and B = the computer conforms to the norms. We wantaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.28 2 Elementary Probabilities 8 8 8 Figure 2.7. System in Exercise no. 5. 4) (10)? (26)? puan=(3) (2) (5) = 02415; 4) (10\* (26 PIB = () (33) (2) = 0.0619: 4 P4i= (3) ~ 0.0060. b) P[AN B‘] = P[{0}U{1}U (2}] > 0, which implies that A and B° are not incompatible. Exercise no. 5 (2.5) We have 10 brand A components, denoted by Aj... Ajo, and 20 brand B components, denoted by B1,... , By. We want to build a system made up of two subsystems: three brand B components placed in series and two brand A components placed in parallel. Moreover, the two subsystems are placed in parallel (see Fig. 2.7). a) How many distinct systems can be built? Remark. The disposition of the components inside the subsystems does not matter. b) Assume that two brand A components are defective, while five brand B compo- nents are defective. What is the probability that the system functions if the compo- nents are taken at random and operate independently from one another? Solution a) There are (7?) = 1140 possibilities for the first subsystem and (1?) = 45 pos- sibilities for the second one. Then, by the principle of multiplication, there are 1140 x 45 = 51, 300 distinct systems that can be built. b) We have: 15 x 14x 13 91 P [Subsystem B functions] csi an andaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.32 2. Elementary Probabilities P[A| B] + P[B| Al= a) Calculate P [A] if P[AN B] = & b) Calculate P[B] if P[AU B] = 2. Problem no.7 A communication system transmits three signals: s;,s2 and 53, with equal prob- ability. The reception is sometimes erroneous, because of the “noise.” It was found, experimentally, that the probability pi; of receiving the signal s;, given that the sig- nal s; has been transmitted, is given by the following table: {Reception St 2 ji f08 O01 Of Transmission s2|0.05 090 0.05) sa. 0.02 008 0.90) a) Calculate the probability that the signal ») has been transmitted, given that the signal s2 has been received. b) If we assume that the transmissions are independent, what is the probability of receiving two consecutive 53 signals? Problem no.8 A box contains five brand A, three brand B and two brand C transistors, The transistors are all distinguishable. In how many ways can We pick, at random and without replacement, four transistors if we want to get at least one of each brand? Remark. The order in which the transistors are selected does not matter. Problem no.9 According to the data collected, 40% of all human beings have type A blood, 10% have type B, 45% type O and 5% type AB. Moreover, we think that 90% of people who have type O blood are incorrectly classified, whereas 3% of people with type B, 10% with type AB and 2% with type A blood are classified as having type O blood. a) What is the probability that a person classified as having type O blood really has this type of blood? b) If we assume that each person is classified independently from the others, what is the probability that two given persons classified with type O blood do not have this type of blood? Problem no. 10 A diskette is taken at random from a box containing 10 brand A, 15 brand B and 25 brand C diskettes. We assume that, in general, 95% of the brand A diskettes are perfect. This percentage is 97% for the brand B diskettes and 99% for those of brand c.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.36 2 Elementary Probabilities Problem no. 18 In a certain city, 60% of the calls to the emergency number require the help of the police department, 40% that of the ambulance service, and 20% ask for the fire department. Moreover, 40%, 20% and 5% of the calls require only the help of the police department, the ambulance service and the fire department, respectively. Finally, 5% of the calls require the help of the three services and 10% do not ask for any of these three services. a) Calculate the probability that a given call requires the help of the police department and of the ambulance service, but not of the fire department. Hint. Use a Venn diagram. b) Let the events be as follows: A =acertain call requires the help of the police department and B = acertain call requires the help of the ambulance service. Are the events AU B and A UB* incompatible (or mutually exclusive)? Justify your answer by a numerical computation. c) We consider ten calls to the emergency number. Among these calls, there are ex- actly seven that required the help of the police department. If four calls are taken at random (and without replacement) among the ten calls considered, what is the prob- ability that exactly two of these four calls required the help of the police department? Problem no. 19 Computers A and B exchange information within a network. The probability that the information sent from A to B is incorrectly transmitted is equal to 0.01, while this probability is equal to 0,005 in the case when the information is sent from B to A. We consider 10 messages transmitted from A to B and 15 messages from B to A. We assume that all transmissions are independent. a) What is the probability that a message taken at random among the 25 messages considered has been incorrectly transmitted? b) Two messages are taken at random and without replacement among the 25 mes- sages considered. Given that both messages have been incorrectly transmitted, what is the probability that they were transmitted by the same computer? ) Suppose that exactly two of the 25 messages considered have been incorrectly transmitted. What is the probability that the two erroneous messages were transmit- ted by the same computer? Problem no. 20 An engineer responsible for the control of the quality in the company where she works receives a batch of 200 parts used in the computers that they build. She decides to pick 10, at random and without replacement, and to test them. Let E be the random experiment that consists in counting the number of defective parts amongaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.40 2 Elementary Probabilities 4 5 Figure 2.11. System in Multiple choice question no. 10. Question no. 10 We consider a system constituted of five identical components that operate inde- pendently from one another and that are connected as in the diagram in Fig. 2.11. Knowing that exactly three of the five components work, what is the probability that the system functions? a)0.6 b)0.7 c)0.8 d)0.9 e) we cannot compute it Question no. 11 The license plates in a certain country bear six characters taken at random among the 26 letters of the alphabet and the ten digits {0, 1, 9}. What is the probability that a given license plate bears at least one digit? a) 0.0546 b)0.1419 c)0.3275 d)O.8581 e) 0.9995 Question no. 12 A student gets up at time X and goes to bed at time Y, where 0 < X < Y < 24, Let S = {(x, y): 0 < x < y < 24}. Write in mathematical form the event A = the student is up at least three hours more than he lies in bed. a) ((e, y) € Si ly — x1 23} b) (x, y) € Ss ly — x] < 13.5} o) (x,y) eS: y—x > 13.5) d){(x, y) ES: y—x < 15} e){(x,y) eS: y—x 2 15} Question no. 13 Let A,B and C be events such that A C B, A and C are incompatible, PUB U C)¢] = 1/10, PIB AC] = 3/10, PLAS NB) = 1/2 and P[C] = 13/20. Calculate P[ A]. a)O b)0.05 ¢)0.35 d)05 ¢)09 Question no. 14 Two men and two women sit at a circular table. In how many different ways can they be seated if the two men must be opposite each other? Remark. Two ways are equivalent if each person has the same neighbors to his/her left and to his/her right. a)2 b)3 c)4 d)6 e)24aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.44 2 Elementary Probabilities precedes. A passenger traveling on board this plane is taken at random. What is the probability that the passenger A) arrives at destination or started his/her journey outside the country? a) 0.12 )0.24 c)0.375 d)056 ©) 0.875 B) travels for business or started his/her journey inside the country? 2) 0,064 b)0.424 ¢) 0.676 d)0.74_ e) 0.964 Question no. 27 Customers arrive ata store, at random and independently from one another, from 9:00 a.m. A) Suppose that there have been two customers between 9:00 a.m. and 10:00 a.m. i) Calculate the probability that the second customer arrived before 9:30, given that the first one arrived at 9:15. a) V4 b)W/3 c)7W/IS d) 29/60 e) 1/2 ii) What is the probability that the two customers arrived inside a 15-minute interval (that can be located anywhere inside the hour considered)? a) 4b) 1B c) 7G A)I/2 €) 9/16 B) Suppose now that there have been four customers between 10:00 a.m. and 11:00 am. i) What is the probability that exactly three of these four customers arrived be- tween 10:00 and 10:30? a) VIG b)I/B oc) I/4 A) 1/2 &) 3/4 ii) What is the probability that there has been one customer in each quarter of an hour of the hour considered? a) 0.00391 —b) 0.00586 c) 0.04167 d) 0.06250 e) 0.09375 Question no. 28 A maker of lasers estimates that 20% of the lasers the company produces are of excellent quality, 75% are of acceptable quality, and 5% are defective. Moreover, we assume that the quality of a given laser does not depend on the other lasers. To improve the reliability of the lasers the company sells (those they think are non- defective), they submit their products to a quality control procedure. Suppose that this procedure classifies as excellent 90% of the lasers that are of excellent quality, acceptable 10% of the lasers that are of excellent quality, excellent 10% of the lasers that are of acceptable quality, acceptable 85% of the lasers that are of acceptable quality, acceptable 5% of the lasers that are defective, defective 95% of the lasers that are defective.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.48 2 Elementary Probabilities B) In A), how many distinct arrangements of the energy levels of the particles corre- spond to the energy level (2, 1, 0) of the system or to level (1, 1, 1)? a)9 b)10 c)12 d)l4 e)I5 C) i) Let P; be the probability that a particle has energy level ¢;, for i = 1,2, 3. Suppose that P; = 1/4, P2 = 1/2 and P; = 1/4. What is the probability that the three particles all have a different energy level? a6 b) V8 03/6 a) 14 e)3/8 ii) Given that the three particles have the same energy level, what is the probabil- ity that this energy level is €1 or €? a) 2/3 b) 7/0 4S d)5/6 ©) 9/10 Question no. 38 A) A number is taken at random in the set {1,2,....8}, so that P{{k)] = 1/8 for k = 1.2.....8, We define the events A=(1,2,3,4), B=(2,3,4,5) and C = (4,5,6.7} Which pairs of events are independent? a) (A,B) only b)(B,C) only c) (A, B) and (B,C) only d) no pairs) all the pairs B) Suppose that in A), we have: P[{k}] = cy if k = 1,3,5,7 and PIA} k = 2,4, 6, 8. For what values of c; and c are the events D = {1,2} and F = {2, 3) independent? ade =c2 = 1/16 b)cy =e2 = 3/16 c)ey = 1/16; cn = 3/16 d) cy = 3/16: c2 = 1/16) no values of ¢; and c2 Question no. 39 The probabilities of the events A, B and C are P[A] = 1/2, P[B] = 1/3 and P[C] = 1/4. Moreover, A and C are incompatible, B and C are incompatible, and P[ANB]= PLAN B). A) Calculate PLAS NB). a)I/2 b) 6c) 14 A) IB e) 1/2 B) Calculate P[A| BUC‘). a) 1/4 by) 13 c)1/2 d)I/12_ ¢) 2/3. Question no. 40 A box contains five brand A components, five components of brand B and five of brand C. Five components are taken at random and without replacement. A) What is the probability that the five components taken at random are of the same brand? Dr >) 00°) ram) T0908) 500aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.92 2 Elementary Probabilities A) What is the probability that the system functions? a)716 b) 1/2 9/16 a) 5B e) 11/16 B) Let p be the answer in A). Given that the system functions, what is the probability that component no. | functions? at Da OF OF ef Question no. 49 A hydroelectric station is subject to receive three types of shocks. The probability of a shock of type I is equal to 30%, that of a shock of type II is equal to 50%, and that of a shock of type III is equal to 20%. Moreover, the various types of shocks are incompatible The probability that a shock of type I causes a power failure is equal to 40%. In the case of type IT and type II] shocks, this probability is equal to 20% and 10 90%, respectively. A) What is the probability that th shock, from a given time instant? a)3/7 b)/2 «c)4/7_ d) 5/7 e) 67 tion receives a type II shock before a type III B) Given that a power failure (caused by a shock) has occurred, what is the proba- bility that a type IT shock has caused this failure? a) V/7 by c)VS dj) 1/4 e&) 3 Question no. 50 Manufactured objects are classified into three categories: compliant with the norms (1), containing one or more minor defects only (11), or containing one or more major defects (III) A) Four objects will be taken, at random and with replacement, among those that will be manufactured tomorrow. In how many different ways will it be possible to classify them if i) the order in which the objects will be taken matters? a) 64 b)81 c) 100 d) 121) 196 ii) the order in which the objects will be taken does not matter? a)12 b)13. c) 14 d)IS e) 16 B) Suppose that in a batch of 100 objects, there are 70 objects that comply with the norms and 23 that have one or more minor defects only. i) A sample of three objects is taken, at random and without replacement. What is the probability that the sample contains an object of each category? a) 0.0697 b)0.0797 c) 0.0897 _d)0.0997__e) 0.1097 ii) Four objects are taken, at random and without replacement. What is the prob- ability that at least one object of each category is obtained? a) 0.0952) 0.1052 c)0.1152 d) 0.1252 e) 0.1352aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.2 The Distribution Function ST = {Mir Mi2, Mii M2, Mi2M2, M1) M3, Mj2M3, M2M3}. In this case, we obtain: A= (Mi M2, M12M2, Mii M3, Mi2M3} and, given that the elementary events are now equiprobable, we find directly that P[X = 1) = P[B] = P[A] = 4/6 = 2/3. ii) We can also define the events F = exactly one brand M, component is obtained, and F) =a brand M, component is obtained on the first draw. Then, by the total probability rule, we can write that 22.2 PUR] = PUP | FuJPLPLT+ PUP | FFIPURL= 55+ 5" Bizie of 8 3.2 The Distribution Function Definition 3.2.1. The distribution function of the random variable X is defined, for any real number x, by Fx(x) = PIX S21]. (3.2) Remark. More precisely, we may write that Fx(x) = Pils € S: X(s) 0}. Then we find, using the random variable X of part i) above, that px(z| A) = if P[A] > 0. (3.10) [ x oi x(x | X = 0)/23 173 Indeed, we have: PUX =O}N{X > 0}] _ PLX =0) V/2 PxO\X=9=— “Sys o) ~ PIXE O]~ 1241/4 =2/3aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.3. The Probability Mass and Density Functions 69 £6) elo i 2 3 4 x Figure 38. Probability density function of the random veriable in Example 3.3.5. Definition 3.3.4. Le: X be a continuous random variable with Sx = |a,b]. The number x) for which Fx (ap) = p. (3.19) where 0 < p <1, is called the quantile of order p of X. In particular, x1/2 (or &) is called the median of X Xm Remarks. i) The number x, is unique. However, if X is not a continuous random vari- able defined on [a, 6]. then we must modify the foregoing definition. For example, we define the median xn, in the general case, as follows: PLX < xm) 21/2 and P[X > Xm) > 1/2. (3.20) If X is not a continuous random variable defined on a single interval (a, 4), then the median is not necessarily unique. ii) If 100p = k, an integer, x, is also called the kth percentile of X. The terms decile (if k isa multiple of 10) and quartile are used as well Example 3.35 Let cx if0 0 (see Fig. 3.8). Calculate the constant c, the distribution function, the median and the 99th percentile of X. Solution. We must have: 00 2 4 =f pewoar = [ cxdx + [ e(4—x)dx =4e 4 ¢= 1/4. oo 0 I»aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.4 Important Discrete Random Variables 2B intix) Fy(x) =D px). k=O We can use a pocket calculator to evaluate this function. There are also tables of the functions py and Fy. Table 3.1 (p. 74) gives the value of the function Fy (x) for a few values of n € (2,3,... ,20} and of p € (0.05, 0.50]. Note that for values of p greater than 0.50, we can use the following relationship: P(B(n, p) =k] — P[B(n,1— p) sn—k-—1) fork € (0,1,..., a}. Remark. Suppose that n transistors are taken at random from a box containing m brand A and N ~ m brand B transistors. Then, if the transistors are taken with re- placement, the number X of brand A transistors obtained has a binomial distribution with parameters n and p = m/N. However, if the transistors are taken without re- placement, then the trials are not independent. We can show that, in this case, we have: (=e x(k) = A" for k = max(0,n — N+ m},...,min(n, m). We say that X has a hypergeometric distribution with parameters N, n and m. We write: X ~ Hyp(N,n, m). Geometric distribution. Let X be the number of Bernoulli trials needed to obtain a first success. Then, Sx ....). We say that X has a geometric distribution with parameter p. We write: X ~ Geom(p) (or X ~ Geo(p)). Proposition 3.4.2. Let X ~ Geom(p). We have: px(k) Ip fork =1,2,.... (3.24) Proof. Let FE be the random experiment in which the results of the Bernoulli trials are observed until event A occurs. We have: Kal Xe2 Xa3 aa S=(A | ASA, ACASA, ... Then, we may write that AS Aj) = gS" p. (k=1) times px(k) = PIX =k) = PI{aca® og! Indeed, in order that X = k, there must first be A — 1 consecutive failures, and then a success. The above result is then obtained by independence. Q Properties. i) The function px is non-negative andaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.4 Important Discrete Random Variables 7 Poisson distribution” Poisson’s theorem. Let X ~ B(n, p). Suppose that n > 00 and p | 0 in such a way that nx p remains equal to aw. Then, we find that k e pxtk) > fork =O,1,.... (3.28) ‘a kt Proof. Let px := px(k). For k = 0, we have: Po =(-py=(1-2) en asn 00, Next, we consider the ratio per — (apg! bp iw 2" Pk (i) pkqn-* k+lq k+l n n s Since the two factors inside the square brackets tend to | as n tends to infinity, we can write that . Pk+t a lim PAYL 2 3.29 nano py k+1 Ge) Using this result recurrently, we obtain Equation (3.28). Indeed, we have: Pw no lim — =——_> 1) — “, oe Ter mPa ae so that etc. Oo Definition 3.4.1. If X is a discrete random variable with Sy = {0,1,...} and for which 0k kK then we say that X has a Poisson distribution with parameter a > 0. We write: X ~ Poi(a). e Px(k) fork =0,b.. (3.30) Siméon Denis Poisson, 1781-1840, was born and died in France. He first studied medicine and, from 1798, mathematics at the Ecole Polytechnique de Paris, where he taught from 1802 to 1808. His professors at the Ecole Polytechnique were, among others, Laplace and Lagrange. In mathematics, his main results were his papers on definite integrals and Fourier series. The Poisson distribution appeared in his important book on probability theory pub- lished in 1837. He also published works on mechanics, electricity, magnetism and astron- omy. His name is associated with various mathematical results.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.4 Important Discrete Random Variables 81 (0.8179, 0.8187), (0.3585, 0.3679), (0.1216, 0.1353). We observe that the approximation deteriorates as p increases, the error percentages being equal to about 0.1%, 2.6% and 11.3%, respectively. But when we add many terms, the errors can cancel one another. For example, comparing the numerical val- ues of the corresponding terms in b) and c), we notice that the probability P[W = k] sometimes overestimates and sometimes underestimates the probability P[Y = k], so that the errors partly cancel. That is why it is generally easier to approximate the distribution function Fy than the probability mass function px ,as mentioned above. d) Let U be the number Of programs required to obtain a first program that contains no errors. We assume that there are at least five programs remaining to be received from the time instant considered. We seek P[U <6]=P[V <6], where V ~ Geom(p = P[X = 0] ~ 0.05) = PIV <5] =1—.95)° ~ 0.2262. Remark. The random variable U itself does not have a geometric distribution, since the group of students considered is not infinite. In fact, U is a defective random variable, because, if there are 20 students in the group (and if nobody bas submitted his/her program yet), we can write that eS 20 20 Drv =M = Pt =H +05 > Pv aay = ia im a 1— (0.95) = 0.6415 < 1. Wr Example 3.4.2 In a certain lottery, six balls are picked at random and without re- placement among 49 balls numbered from 1 to 49. We win a prize if the combination that we have chosen contains at least three “good” numbers. A player decides to buy ‘one ticket per drawing until he wins a prize. What is the probability that he must buy less than ten tickets? Solution. Let M be the number of good numbers in the chosen combination, and let X be the number of tickets that the player will have to buy to win a prize for the first time. We have: P(M > 3)=1-P[M <2] 6) ( 43 > G0 19) Bm @) (1 - 6,096, 454) + (6 - 962, 598) + (15 - 123, 410) 13, 983, 816 13, 723, 192 = 1- 0.9814 = 0.0186. ~ 3,983, 816aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.5 Important Continuous Random Variables 85, PIX >t+s|X>t]=P[X>s] Ys,t>0. (3.36) The demonstration of this property of the exponential distribution is similar to that of Proposition 3.4.3, p. 76, for the geometric distribution. This property is used in reliability and in queueing theory. v) The exponential distribution can be considered on the entire real line by defining A all fx(x) = forx € R, where 4, is a positive parameter. This random variable, called the double exponential distribution, is also known as the Laplace distribution> Example 3.5.2 Suppose that the lifetime X (in years) of a machine has an exponential distribution with parameter A = 1/3. What is the probability that a three-year-old machine will still work at the end of three additional years? Solution. We want PIX > 6| X >3)= PIX > 6-3] =e"! ~ 0.3679. Note that the answer would be the same if the machine were brand-new, or six years old, etc. If we apply this property to the lifetime of a car, for instance, it is clear that the assumption that we make is not entirely realistic. Indeed, the probability that a new car lasts at least three years is surely larger than the probability that a ten-year-old car lasts at least three additional years. However, the assumption of an exponential lifetime can be acceptable for a shorter period of time. For example, the probability that a one-year-old car lasts at least two additional years should be almost equal to the probability that a brand-new car lasts at least two years. Gamma distribution Definition 3.5.3. The gamma function, denoted by P'(-), is defined (for « > 0) by T@)= f xt let dx (3.37) ‘0 Doing the above integral by parts, we find that [@)=@—-Dl@-1) ifa>1 (3.38) 3 Pierre Simon (Marquis de) Laplace, 1749-1827, was born and died in France. In addition to being a mathematician and an astronomer, he was also a minister and a senator. He was made count by Napoléon and marquis by Louis XVIII. He participated in the organization of the Ecole Polytechnique de Paris. His main works were on astronomy and on the calculus of probabilities: the Traité de mécanique céleste, published in five volumes, from 1799, and the Théorie analytique des probabilités, the first edition of which appeared in 1812. Many mathematical formulas bear his name.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.5. Important Continuous Random Variables: 89 04 1 rc | ( Leg = | exp|- SoH las exp(—1?/2) dt, Via Jaco P| 207 Vint Joo by making the change of variable + = (x — j2)/o. Then, using polar coordinates, we can write that 1 pe pe 1px pte P= f [ exp[—(1? +5?)/2]dtds = =f f rexp(— 21 Sooo Jo 2n Jo Jo We easily find that / = 1. Then, knowing that J > 0 (because fy(x) > 0), we deduce that 1 = 1. ii) The parameter 1 is a position parameter. while o is a scale parameter. Further- more, all Gaussian distributions have the same general shape, namely that of a bell (see Fig. 3.18). iii) The function fy is symmetrical with respect to j1; that is, /2)dOdr. Sx@+ w= fx(-x +p). Moreover, fx(x) attains its maximum at x = 4 and has two inflection points: at xapto. iv) If ¢ =0 and o = I then we say that X has a standard Gaussian distribution. Its probability density function is given by 1 G(2) = Se exp(-27/2) for—00 < z < 00 (3.44) Vin and its distribution function is denoted by ®. Now, if X ~ N(, 0?), we find that its distribution function can be expressed in terms of @ as follows: Pray =0(2=#), 3.45) 7 Thus, any probability involving a Gaussian distribution can be obtained by using the N(0, 1) distribution. To evaluate the function ®, we can use a statistical softwareaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.6 Transformations 93 Example 3.6.1 Let X ~ Geom(1/2) and ee 1 if X iseven, O if X isodd. We have: ES oo yl) = D> px 2k) = Yd /2)47"(1/2) = 1/4 + 1/4)? + = 1/3. fat it It follows that py(0) = 1 — (1/3) = 2/3. Therefore, we have: [y [o ifz Ler) | 23131 | f » fo 1] (Froy [231 Now, let Z := X — 1. Then, we have: Sz = {0, 1, and and patk) = pxtk +1) = (1/2) FY" (1/2) = (1/2)**"! fork =0,1,... b) If X is a continuous random variable, but g is a staircase function, that is, a fune- tion that takes at most a denumerably infinite number of values, then ¥ is a discrete random variable, We have: Py (yg) = PIY = ye] = P[X € Al, G51) where A is the event equivalent to {Y = yx} in Sy Example 3.6.2 Let X ~ Exp(a) and Y := int(X}+1. We already mentioned in Section 3.5 that ¥ ~ Geom(p = 1 — e~*), Indeed, in this case, the event equivalent to IY =khis{k—1 dr. 0 t 0 b) The proof is similar to the previous one. a Remark. We can also prove the result in a) as follows. By integrating by parts, we find that f Ul — Fx@)] dx = xf1— recog +f xfx (x) dx = Jim x{l — Fx(x)] + EX). Now, by l'Hospital’s rule (H.R.), we have (assuming that ELX?] < 00): 1 = Fx@) ur Snes Jim, x1 — Fx] = lim, lim x? f(x) = 0, since x? f(x) must converge (toward 0) when x tends to infinity (otherwise the integral of x? fy (x) from 0 to infinity would diverge). Example 3.7.1 Let X ~ Poi(a). We have: s a ot ay ak a, SA at! 8 yo BpC= ke ane Em e oe [=e tae" =a, Example 3.7.2 Let X ~ Exp(A). Making use of the preceding proposition, we obtain (see Equation (3.35)): eod= | pix > dx = [ eM dx= 0 'O We can demonstrate the following proposition or use it as a definition, Proposition 3.7.3. Let X be a random variable and let Y := g(X). Then, we have: oo. Vo scudpx(ax) if X is discrete, Ey] =} ii 8(x) fx (x) dx if X is continuous. 00aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.38 Transforms 105 E[X] = * exp(eia)|,,-g = — Je exple!*aye!“aj|,,_9 =a, do which corresponds to the result obtained in Example 3.7.1. Likewise, we find that E[X*]=07 +a. Then, making use of Proposition 3.7.5, we obtain: VAR[X] = E[X*] —(E[X))? = 0? +a — (a)? =a. So, in the case of the Poisson distribution, the parameter w is both the mean and the variance of the distribution. A similar result is obtained with the exponential distribution, for which the mean and the standard deviation are both equal to 1/2. Example 3.82 Let X ~ Nj, 02). Then, 2 éx(w) = asf. et exp { — aoe las 1 f. | le = oo? + WP = 2jwo? n+ (wo?? = exp) — Qo 20% = exp (Jou - yore?) x P[—0e < N(w + jwo?, 0?) < 00] = exp (Jon - 50?) : Hence, we can write that E[X"] = er exp (Jon - yore?) |oo" ‘We find, in particular, that FIX] = —jexp (Jou- joo 2) Gu wo? and 2 1 # EX} = ~exp (Jon = 30 o Vive - wo? 07} oT +e, so that VAR[X] Note that since we have used the symbols jzx and a} to denote the mean and the variance of the random variable X,, in general, then, to be coherent, in the case of a Gaussian distribution with parameters jz and o? the constants jz and a? had to represent the mean and the variance of this Gaussian distribution, respectively. Remark. We have written above that the integral in the computation of the character- istic function of X ~ N(u, 07) can be expressed in terms of the probability that aaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.9 Reliability 109 Remark. The function r(t), multiplied by dt, is approximately equal to the prob- ability that a machine, that is still working at time r, breaks down in the interval (t,1+-dt), because fr(t\T > tdt ~ Plt. Now, the probability that the lifetime of the system is smaller than or equal to s, given that the system is still functioning at time ¢, is 0 ifs <¢, PIT thi=Fr(s |T >) = § Frs Lee 1— Fr) oe It follows that d fr(s) f =< = 3 fr(s|T> N= Pro |T > N= if's>t (3.80) R’(t), we can also write (taking the limit as s decreases to t) frit) _ Re) 1-=Fr) Ri) Thus, since fr (1) that r= 3.81) Example 3.9.1 If X ~ Exp(A), then the failure rate function is constant: aie Re ri) = ap = ~ Proposition 3.9.1. There is a bijective relationship between the functions R(t) and r(t): * R(t) =oo[-f rosas| (3.82) 0 Proof. We have: 1 f[ rs)ds = Now, K(0) = 1, because T is a non-negative continuous random variable. Hence, Equation (3.82) follows at once. Oo T RY(s) i Risy 8 =~ MRO FINO. Remark. We then deduce from Example 3.9.1 that the exponential distribution is the only one for which the failure rate function is constant.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions 113, Exercise no. 3 (3.6) A well-balanced (or non-biased) die is tossed until a “6” is obtained. a) Given that the first three tosses produced a “1.” what is the probability that the die will have to be tossed more than five times in all? b) Let X be the number of tosses required to obtain a first “6.” We define Y = min{X, 5}. That is, Y is the minimum between X and the number 5. Calculate Py (y) Solution a) We seek P[X > 5 | X > 3], where X ~ Geom(p = 1/6). We have: 5 () -% PIX >5|X>3)=P(X>2=(2 6) ~ 36 b) We can write that PIYsyl=PIX=y] ify =1.2.3.4 and 4 Puy =s)=P(x>si=Pix>41= (2) Thus, we have: y [i 2 3 4 5 5) (1 sy(t 3) (1 s)\* vfs (a) ts) (5) GG) (6) (e) (a) Exercise no. 4 (3.7) Let pecs: a ifx>0. elsewhere. a) We define ¥ = 2In X. Calculate fy (y). b) Let Z := 1/X. Calculate E[Z1. Solution » del? e dy fry) = fx @) 2(eryeerr [ser =exp{y—e’} ify € R,since 2In0 = —oo and 2Inco = 0.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions W7 Hint, We have: Q (1.2815) = 0.1 and Q (2.3263) = 0.01. Solution a) We have, by Markov’s inequality: EU) 2b odeh PIX2 615 = 5 PIX <6)21-5=5. _ PIX >5] b) PIX > 51X23 = SO SS. We have: P[X <5]= P[Poi(1-5) > 3] = and P(X <3] =P [Poi (3) = 3) =1— It follows that cont, 18.5¢7 P[X > 5|X >3)°S — |X >3) 35e c)i)Y ~NQ2x 3-4,4 x 1) =N(2, 4) -2 = PLY 1. It follows that eae -PSl a >| + Fx (Vins) | ving iol Winx? —_ vn 1 oe a exp |B any)? tory > 1. Jae Pa | var 7 wena flere Ke Pare [~ Hoe age ‘ © Jeoe co Vim Vaan a. is = Jr [ (0.4) 8] 2 ~ 0.5774. i) bzlo) = Eel?) = ElelOXY) = eI Ele!%) = € 1x Qn) = ele 202 = eed? ii) We can write that d 2 A fg jute (_ to les) = 6 J0- WE j — Aw)? — 4 = (jf)? — 4 =5. ElZ?] = -93(0)|,-9 =aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions 125 b) Let 26 20-Y ify > 1, ro-={ 0 elsewhere. i) Calculate the characteristic function of Y. ii) Use the function calculated in i) to obtain the mean of Y. Problem no. 19 A crate holds two boxes that contain ten items each. Alll the items were made by the same machine. Suppose that the probability that an item made by this machine is defective is equal to 0.05, independently from one item to another. a) What is the probability that there are less than three defective items among the 20 items considered, knowing that there is at least one defective among these 20 items? b) Use a Poisson distribution to compute (approximately) the probability in a) Problem no. 20 Let X be the waiting time (in minutes) before being served at a certain counter. We suppose that the distribution function of X is given by 0 ifr <0, Fx(x) = i ifs =0, 1-3, ifr >0. a) Calculate P[X =0| X < 1]. b) Calculate the probability density function of X c) Let ¥ := X | {X > 0}. Calculate i) fy (y) and ii) FLY]. Problem no. 21 Let X be a discrete random variable whose probability mass function is given by the following table: a) Calculate the variance of VX. b) Let Z := X | (X > 1}. Calculate the probability density function of Z c) i) Calculate the moment generating function of X. ii) Use the function computed in i) to obtain E [X?]. Problem no. 22 A discrete random variable X has the following probability mass function: x |-2 ~-1 0 1 px@) | & 4 4 a) Calculate Fy (0) + Fx (2).aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions 129 5 Be 2v=l+ W Letus write that X = 2/. a) Suppose that § = | and that V has a uniform distribution on the interval (0, 1). Calculate i) the probability density function of X; ii) the expected value of X. b) Suppose that V = 1 and that $ has a uniform distribution on the interval [1, 2]. Calculate the moment generating function of X. Problem no. 34 Let X be the number of different versions of a program that a computer science student has to write to obtain a program that works. We suppose that X has a geo- metric distribution with parameter p = 1/4. We consider a class with 20 students, who must each write a program individually. a) Calculate P[4 < X <6] X > 2]. b) Let N be the number of students, among the 20, who have to write exactly two versions of their programs to obtain a version that works. Calculate P|N < 1]. c) Use Poisson’s approximation to calculate the probability in b). Problem no. 35 In information theory, the differential entropy of a continuous random variable X is defined by Hy = —Efln fy(X)]. Moreover, the relative entropy of the continuous random variables X and Y is given by H( fx; fir) > f fire) In for (2)/ fr (a) dx. Jose a) Calculate Hy if X ~ Exp(). b) Let ¥ := e%, where X ~ U[0, 1]. Find i) the probability density function of the random variable Y; ii) Hy c) Calculate H( fx; fy) if X ~ U[O, 1] and Y ~ Exp(1). Problem no. 36 a) Let X ~ N(O, 1). We define ¥ = |X}. i) Calculate PLY <2]. Hint. We have: Q(1) = 0.159 and Q(2) = 0.0228. ii) Calculate VAR[Y]. b) Let Z ~ G(w = 2,4 = 1). Obtain E[Z>] by making use of the characteristic function of Z Hint. We have: 2 (w) = (1 — jo)~?. Problem no. 37 Let X be the number of unsuccessful interviews that a graduating student must have before getting a first job. According to the data collected, we accept that E[X] = 2.5.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10. Exercises, Problems, and Multiple Choice Questions: 133 Multiple Choice Questions Question no. 1 We suppose that the probability that a telephone call lasts more than five minutes is equal to 0.1, independently from one call to another. A) Calculate the probability that, among 20 calls taken at random, there are more than 18 calls that do not last more than five minutes. a)O b)0.3917 c) 0.6083 d) 0.6769 e)1 B) Calculate approximately the probability in A) using a Poisson distribution. a)O b) 0.3233 c)0.4060 d)0.6767 e)1 C) Calculate the probability that it takes less than five calls to obtain a first call that lasts more than five minutes. a) 0.0001 b) 0.3439 c)0.4095 d)0.6561 _) 0.9999 D) What is the probability that, among five calls taken at random, the longest lasts Jess than five minutes? a) 0.00001 b)0.1 c)04095 d) 0.5905 €)0.9 Question no. 2 Let X be the time (in days) required to repair a machine. We suppose that the average repair time is equal to four days and the standard deviation to two days. A) What is, at most (and with as much accuracy as possible), the probability that the repair time is smaller than one day or greater than seven days? a)4/9 b) 2/3. c)1_ d)wecannot calculate ite) none of these answers B) Suppose that X has a uniform distribution on the interval (a, b]. Find the con- stant @ a)O b)2_ c)4 d)wecannotcalculate ite) none of these answers C) Suppose that X has a gamma distribution with parameters @ = 4 and k = 1 Calculate P[X < 4]. a) 0.2381 b) 0.3528 )0.5665 d)0.5768 _e) 0.6288 D) Suppose that X has a Gaussian distribution. Find the number xo such that PIX —4| < xo] = 0.99. Hint, We have: Q~! (0.01) ~ 2.326 and Q~! (0,005) ~ 2.576. a) 2.326 b)2.576 ¢) 4.652 d)5.152 €) 10.304 Question no.3 Let ; ox ifx <0, e*" ifx > 0, FxQ) = where 4 > 0. Calculate Fx(1 | X > 0). a)je™* bye cc) 1 —e*_— dy 1 — Jee) 2-eaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions 137 must ¢xamine to end the random experiment. What distribution does N have if the probability that an object is defective is equal to 1/10, independently from one object to another? a) Geom(1/10) _b) Geom((9/10)°)_ c) B(20, 1/10) d) B(20, (9/10)”°) e) none of these answers Question no. 19 Calculate P[X? < 9], where X tion with parameters a = 2 and 2 ae? by de“? c) 1 -eF dy t—4e7F_— ee) 1- Ye arandom variable that has a gamma distribu- Question no. 20 The lifetime (in months) of brand A components has (approximately) a Gaussian N(50, 100) distribution and that of brand B components a Gaussian N(60, 100) dis- «ribution. A component is taken at random from a box containing 10 brand A and 20 brand B components. Calculate the probability that the component selected lasts at least 50 months. Hint, We have: Q(1) ~ 0.159 and Q(2) ~ 0.0228. a) 0.6705 b)0.7273_ c) 0.7386 d)0.8181 —_e) none of these answers Question no. 21 We define Y = 1/X?, where X is a random variable having a uniform distribu- tion on the interval {1, 3]. Calculate fy (y). aay if l/9 0. Calculate E[X~?]. a) 16 b) 1/4 c)6 d)oo e)co-co Question no. 23 Let X be a random variable having a gamma distribution with parameters @ = 30 and 4 = 20. According to Markov’s inequality, what is the minimum value of P(X <2)? a) V6 b)1/4 c)1/2 d)3/4 e) 56 Question no. 24 A discrete random variable N has the following generating function: 1 Lg, bee Gu@ = 5 +32 +3e- Calculate VAR]. a)2 6) 8/3 c) 20/3. d) 28/3) none of these answersaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.3.10 Exercises, Problems, and Multiple Choice Questions 141 lifX > 0, OifX <0. ay b)}z g(l+z) dl+z e)l+z+2 Question no. 32 Let X be the utilization time (in months) of a laser printer until its first break- down. A) Suppose that X ~ Exp(1/3). Calculate E{|X —3]]. a)3e~!b)3(1—e7!) c)6e"!_— d)3_— ee) (1 - 7!) B) Suppose that X ~ G(a = 2,4 = 1). We define ¥ = 1/(X + 1). Find fy(1/2). a) fe! bye! c) 27! d) de“! e) Be! C) Suppose that the time Z (in months) required for the 25th breakdown to occur has approximately a Gaussian N(75, 225) distribution. Calculate P(Z < 90 | Z > 60) Hint. We have: Q(1) ~ 0.159. a) 0.1891 b)0.3180 c) 0.6820 d)0.8109 e) 0.8410 Question no. 33, A) Let V be a random variable having a geometric distribution with parameter p. Calculate Fy(n | N > 1) forn = 2,3... ayl—q"' byl—q" cyl—q"*! d)t—p™' e)1—p" B) Let X ~ UO, 1. i) Obtain the characteristic function of ¥ := In(X + 1). Hint. We have: 6x (w) = “=! Jar ie 1 gietl | 2ie ae. aie a) — YS COST SO a je jo+l jotl Qjo jeoln2 ii) Let Lif X <03, ¥ {2 ifX 503. Calculate the generating function of V := 2W a) (0.3 40.72) b) 22(0.3+0.72) c) 203 + 1.42) d) [20.3 + 0.72)P e) 20.3 + 0.72) Question no. 34 A number N is taken at random in the set {0, 1,2, ...,9}. Calculate py (5) + Fy (5.5) a)0.6 b)065 ¢)0.7 d)0.75 e)O08 Question no. 35 Let fx(x)=2x ifO 0]. a6 b)14 c)/3 di 3/4 C) Let ¥ := |X|. Find fy(y) for < y <1. al bg-y c)e+y d)2y e)2-2y Question no. 55 Let X be a random variable having a uniform distribution on the interval (0, 1]. We define ¥ = max{ x. $]. A) Find Fy(y) for } < y < 1. What type of random variable is ¥? b) y;continuous c) 2y — 1; continuous d) y; mixed e) 2y — 1; mixed B) Calculate E[Y]. a) 9/16 b) 5/8 c) 11/16 d) 3/4) 13/16 Question no. 56 We want to find a model for the number NV of cars crossing a certain intersection during a one-minute period. According to the data collected, the most frequent valueaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Index Absorbing state, 227 Autocorrelation function, 223 Autocovariance function, 223 Average power, 223 Bayes, 19 Bayes" formuta, 19 rule, 20 Bernoulli, 70 Bernoulli trials, 71 Best estimator linear, 184 non-linear, 184 Bias, 254 Bienaymé, 102 Bienaymé~Chebyshev's inequality, 102 Brown, 233 Brownian motion, 233 standard, 233 Cardano, 1 Cauchy. 261 Central limit theorem, 189 Chapman-Kolmogorov equations, 227 Characteristic function, 103 joint, 181 Chebychev, 102 Coefficient of determination, 315 Combination, 15 Confidence coefficient, 258 imerval, 258 one-sided, 258 limit, 258 Continuity correction, 193 Correlation, 177 Correlation coefficient, 177 ‘of a stochastic process, 223 sample, 322 Covariance, 177 Critical value, 278 Cross-correlation function, 224 Cross-covariance function, 224 De Moivre-Laplace approximation, 193 Dependent variable, 307 Distribution Bernoulli, 7! beta, 152 binomial, 71 Cauchy, 261 chi-square, 86 double exponential, 85 Erlang, 86 exponential, 83 Fisher, 279 gamma, 86 Gaussian, 88 standard, 89 generalized Pareto, 303 geometric, 73354 Index hypergeometric, 73 Laplace, 85 lognormal, 90 multinomial, 172 multinormal, 179 negative binomial, 76 Pareto, 154 Pascal, 76 Poisson, 77 Rayleigh, 156 sampling, 253 Student, 260 uniform, 82 Weibull, 88 Distribution function, 57 conditional, 63, 166 first-order, 222 joint, 161 marginal, 162 nih-order, 222 second-order, 222 Erlang, 86 Error of type 1, 265 of type 11, 265 Estimation, 182 Estimator, 254 consistent, 255 maximum likelihood, 256 relatively more efficient, 255 unbiased, 254 Event(s), 8.56 elementary, 7 equiprobable, 12 equivalent, 56 incompatible, 7 independent, 21 Expectation, 95, 176 conditional, 96, 168 Failure rate, 109 Feller, 190 Fermat.2 Fisher, 277 Fit index, 315 Fourier, 103 Gamma function, 85 Gauss, 88 Gaussian white noise, 235 Generating function, 106 Goodness-of-fit test, 262 Gosset, 261 Huygens, 2 Independence, 21 conditional. 21 of random variables, 165 by pairs, 23 global. 23 Initial distribution, 226 Khintchin, 188 Kolmogorov, 227 Kurtosis coefficient. 101 Laplace, 85 Law of large numbers strong, 188 weak, 188 Lévy. 190 Likelihood function, 256 Linear combination, 185 Lyapunov, 190 Markov, 101 Markov chain, 225 stationary, 225 Markov’s inequality, 101 property, 225 Matrix doubly stochastic, 226 stochastic, 226 transition, 226 Mean, 95. of a stochastic process, 223 sample, 253 temporal, 238Mean square error, 183, 255, Median, 69. Memoryless property, 76, 85 Method of least squares, 308 Method of maximum likelihood, 256 Method of moments, 257 Mode, 151 Moivre, 193 Moment generating function, 108 Moments, with respect to the mean, 100 with respect to the origin, 100 Multiplication rule, 19 Newton, 23 P-value, 267 Partition, 20 Pascal, 2 Pearson, 262 Percentile, 69 Permutation, 14 of non-distinguishable objects, 15 Point estimation, 254 Poisson, 77 Poisson's theorem, 77 Prediction, 314 Principle of multiplication, 13 Probability, 10 conditional, 18 Probability density function, 66 conditional, 67, 167 first-order, 222 joint, 160 marginal, 163 second-order,222 Probability mass function, 64 conditional, 65, 166 joint, 158 marginal, 159 Quantile, 69 Random experiment, 7 sample, 253 Index 355 Random variable(s), 55 iid. 186 of continuous type, 59 of discrete type, 58 of mixed type, 60 orthogonal, 177 Random vector, 157 continuous, 160 discrete, 158 Random walk, 227 Reflecting state, 227 Regression curvilinear, 318 simple linear, 307 through the origin, 309 Relative frequency, 4 Reliability function, 108 Residuals sample, 315 standardized, 316 theoretical, 308 Response variable, 307 Sample space,7 Skewness coefficient, 101 Standard deviation, 99 Statistic, 253 Stochastic integral, 235 Stochastic process(es), 221 Bernoulli, 222 continuous-state, 221 continuous-time, 221 counting, 228 diffusion, 232 discrete-state, 221 discrete-time, 221 ergodic, 238 mean, 238 Gaussian, 232 independent, 224 orthogonal, 224 Poisson, 228 stationary, 235 strict-sense, 235 wide-sense, 223, 236356 Index uncorrelated, 224 Wiener, 233 with independent increments, 223 with stationary increments, 224 Sum of squares due to regression, 312 of the errors, 310 total, 312 Telegraph signal, 230 random, 231 semi-random, 231 Test(s),266 ofa mean o known, 266 @ unknown, 269 of a variance o?, 271 of the equality of two means known variances, 273 paired observations, 276 unknown variances, 275 of the equality of two variances, 277 of the significance of regression, 311 Total probability rile, 20 Trajectory, 221 Transformations of random variables, 92 of random vectors, 172 Tree diagram, 13 Unbiased estimator, 254 Variance, 99 conditional, 99, 169 of a stochastic process, 223 sample, 253 Weibull, 88 Wiener, 232