SOLUTION MANUAL FOR BASIC PROBABILITY THEORY Robert AshSOLUTIONS TO PROBLEDS Chapter 1 2, 0, = AB AC + BC anc’ + ace + a°ac B 8 dD, AtB+C apcc® + avec® + PBSC o . e cascy® = ao + AF ec! o a where AB= ANB, A+ B= AUB 4. (a) xE AN GO) iffxeAandxe Be C iffxe ANBandx banc o * o (ANB) = (AN C) (o) ke A -4BUC) iffxeAanexd Ruc iff x ¢ A and x ¢ Bandx $c {ff xe A-Bandx dc iff xe (A- By +a It is true that (AUC) - BC (A- B) UC. For if xe (AUC) - Band x ¢ C then x ¢ A~ B, But che sets need not be equal. For example, if A= B-=C then (AUC) -B*A-A=¢, and (A-RUCHOUAZA, 6. a 1 Bo = (AU B)S, which will not be empty unless AU B= CG. Thus a° and 3° will be disjoint iff AUB*M. (ANC) (AC CAN B=, hence ANC and BNC are disjoint. CE (AUG) MN UO, s0 AU C and BU C are not disjoint if Crd.8 rae : 2 ie A, CAL always. If xe AL then x ¢ A A, since 8 aS Any So SAE hence 1 Ay AL. ALS UL Ay always. If x ¢ some Ay then x ¢ A; since ACA C1... CAL; hence n Ui het Ay ii No. For example, tet A, = (0, db. Note also chat a 2 r Le 1 for all n, but £ i=l 2 t= retion 1.3 P(A) = P(A-B) + P(A) B), 80 any example in which P(AN B) < P(B) will do (e.g., let A= BS). wgtion 1.4 The probabllity that the first digit will be > 5, but the second and third will be <5, is (4)(6)(6)/10? = .144, Thus the destred probability is 3(.144) = .432. The number of outcomes is (24)(18), and the number of favorable cases is 3(S) + 8(7) + 13(6), thus p = 149/432. (a) The first card may’be chosen in 52 ways, the second in 48 since the first face value cannot be repeated, and the third in 44, ete. Thus p = (52) (48)... (20) (16)/52!0 (b) The probability that exactly 9 cards will be of the same suit is Ge )se) a) (First select the suit, and 9 of i3 face values, then the odd card.) Similarly the alae! that all 10 ae will - of Be same suit is aC, 1G) « Thus p = (4( B39 + aca Go « The total number of positions available to the women is GC) The adjacent positions for the women may:be selected in mas w+ Lemtl ways. Thus p= (mt1)/(™")lo. wi. 13. The probability that at least one is defective is l - the Byes Gores probability that none is defective, so l-p = This is an application of the formula P(A B) = P(A) + PB) - P(AN B), Ax eee y 3 kings}, B = {exactly 3 aces}. Thue peau ay = C22 hy l8y + AH - HQC. (a) A sentence of length k must start with a word of length 1 or 2; there is only one possible word of length 1, but there are 2 possible words of length 2. If the first word is of length j, the remainder of the sentence may be completed in N(k-J) ways; the result follows, (b) _ ee = xs c this will be a sojution provided aay a 2, ties Naked = G,'9r ho Zor N= =. Thus A+ B(-1) is a solution. Also N(O) = AtB, N(1) = 2A-B, 60 A and B are determined by N(0) and N(1). Since N(O) and N(1) determine N(k) for all k, any two solutions that agree when k = 0 and 1 agree everywhere, so that a2* +-3(-i)* is the general solution. “In the present case, AtB = 1, 2A-B = 1, 0 A= 2/3, B= 1/3. The total number of outcomes is 365"; the number of favorable cases is (365)(364)...(365-rtl) = (365). Thus p = (365), /365". (a) Let A be a subset of A= {1,2,...,n}. Either 1 ¢ A or 1¢ A; this gives us two possibilities. In general, ecithee keAork $A, k= 1,2,...,n. This gives us 2(2)...(2) = 2" ways of choosing A, Alternately, the number of subsets with exactly k membecs is the number of ways of selecting k distinct integers out of n, namely G ). The total number of subsets is os @ = (4198 = 29, (b) The number of ways of ie subsets A with exactly k members is @- Raving chosen such an A, we have B= A+ a ga-k subset of A°. Since there are subsets of AT, B may be chosen in 2""* ways, ‘The number of pairs of subsets is a veo GD 2K = (142) = 99,(a) Let @= (1,2, the partition, where Ay contains J other elements on}. The integer 1 belongs to a set A, of 1 (J = O,1,...,n-1), Thus A) can be chosen in cp) ways. Having chosen A), we must partition Ass this can be done in g(n-l-j) ways. Thus : 1 LG nel - i) = ss Bod = EOP) sorts) = EG) soi a-l = Ey gee. ko (o) tec hin) ee? £ RVR. Thea x=0 a a-L ° k Ey we Ede gL k=0 ko eo J ae ee i eeb or oce My je0 Ft kw | ; Now h(0) = et Thus g and h satisfy the difference equation of (a), and they agree when n= 0. By the form of the difference equation, they agree everywhere. fon 1.5 "Tf" is immediate (set By = A. ). For the “only if" part, it r rf r suffices to show that if A,,...,A, are independent, then PEASE Ay A vee MAL) = PAS) PCAy).« -P(A,)5 the result thea1. (continued) follows by an induction argument. But P(AYS 1 Ay Meee AQ) = PCA Mee AQ) - PALM AQ Mv AY) = (1-P(A,))P(AQ) «P(A, by independence c = P(A,YP(AD)« PCA). a) kel n-keL s 2G) fe F 7 < pike ae qa : Thue p(kt1)/p(k) is > 1 iff (n-k)p > (ktl)q, toe. Lf : k < (ntl)p - 1 <1 iff ke > (ntl)p = 1 = 1iff k = (ntl)p - 1, The result follows. 3. -(a) Let A= {spade is obtained}, B = {heart is obtained}; P(A B) = 0 $ P(A)P(B). (b) Let A= {spade is obtained}, B = {ace is obtained}. P(A N B) = 1/52, P(A) = 1/4, P(B) = 1/13. (c) 1 A and B are independent and mutually exclusive, then either A or B must have probability zero. For P(A Q B) = 0 by disjointness, and = P(A)P(B) by in- dependence, Sim{larly, if the events A,, i ¢ I, are in- dependent and disjoint, either all or all but one of the events must have probability zero. (If P(A,) # 0, apply the above argument to each Ay, J # 1, to conclude that P(Ay) = 0 for all j # 4.) (4) Let A= {spade}, B = {spade or heart}; P(A NB) = P(A) = 1/4, P(A)P(B) = 1/8. 6. There are as many terms in (1.5.2) as there are unordered samplea of size n out of k, with replacement, t.e. cern} (see 1.4.4).7. (a) For a favorable outcome, we must select ny of the available » 4 = 1,2,...,k. The total number of t, balls for color C; outcomes is the number of ways of selecting n distinct objects from a set of t; the result follows. (b) This {s @ standard multinomial problem. The probability is ' a a, n 1 k 7 Py esa where py» t/t. B. (a) P(A A) = P(A)P(A), hence P(A) © (P{A))?, 80 that P(A) = O or Le (b) Tf P(A) = 0, then since AM B is a subset of A, PCAN B) = 0 also, Thus PCAN B) = P(A)P(B). If P(A) = 1, then p(aS) = 0, hence by the above argument, A° and B are in- dependent. But then A and B are independent (see Remark 1 or Problem 1 of Section 1.5). Section 1.6 i, Let X be the number of successes. Then P{all successes occur consecutively|4 < xX < 6} 6 Z P{X*k and all successes occur consecutively} ker 6 LX P{ Xk} kG 6, 4 10) 10-k° © (7p%g’ + 69°q> + 5p . AG asd Cypa kee 2. Mx > 3[x> 1} = XD 3, X> 1V/PX> I} = MX > 3}/AX2 YY 1 - or Lah = apa - Gye?a™7 5. We may regard’ this problem as one of dealing two 13 card hands t« players 1 and 2 from a deck with 26 cards, of which 6 are spades In each case, we are looking for the probability that (say) player 1 received a particular number of spades. Once the nunber of spades for player 1 te determined, that of player 2 is determined also. Thus, @ digs = .36 oo COED + 9/88 = 26) 4912S = 48 ) 2 GDh = a5 «@) 2G97G% = 01. 6. Let A= {first two balle white}, B = {six white balls in the sample}. If the sampling ie done with replacement, then 13)? &2/3)4 073)" P(AIB) = P(A N B)/P(B) = Bas (SSO SONe a 29) (27398173) If the sampling is done without replacement, P(A|B) ie the number of ways of selecting 4 positions out of 8 for the white balls (the first 2 positions must be occupied by white bails), divided by the number of ways of selecting 6 positions out of 10; i.e. ree = 1/3. Note that the answer is the same with replacement as without replacement. Once it is specified that 6 white and 4 black balls are obtained, the problem ie simply one of counting arrangenents. 8. (a) The probability is P(AB + CD + AED + CED) where A is the event that the switch labeled 'A' is closed, etc, and + stands for union, product for intersection. Using the expansion formuta (1.4.5) for the union of n events, we obtain (writing ab for P(AB), etc.) ab + cd + aed + ceb - abcd ~ abed - abce - cdea - cdeb + abcde + 4abcde - abcde = 2p? + 2p? - sp* + 2p.