4. CHAPTER 9 GENERATING FUNCTIONS Section 9.1 ‘The number of integer solutions for the given equations is the coefficient of (a) 2 in (l+etat+...$27), (b) 2” in (lte¢e+.. ena ae -+2™) or eee PO+e t2t+. () in (+2 $2 $e 4... +29), (4) 2 in (+ ata?+...4 2%) (1+ a7 +2t+...42 (@+eta% 4.0.42”) or (+2¢e? +... 2 tetate. Mer tee. (a) (ltztz?+...42%)§ or (1+z+a7+...)% (b) (@+274...42)? or (1 +ata?+.. ©) @ +2 +...42%) or M14 eta? +...) (4) Q+aetart...+278)(o 42 +... + 2%) or (teta? +... (ee +a pcp () GM +e 4. tote tert... ¢2¥) oF ($284... Pte tar+...)® (a) The generating function is either (1+2-+2?+2°+...+2") or (1¢etz?+2°+...)% [The number of ways to select 10 candy bars is the coefficient of 2° in either case.] (b) The generating function is either (1t+e+2?42°+...+2")* or (l+2+2?+294...)%, [The number of selections of r objects is the coefficient of 2” in either case.] (a) The first factor counts the pennies; the nickels are counted by the second factor. (b) f@)= (rte? tah +2. +2 + 2% 4...) ey +0, +63 + 4 = 20, —3 S cy, 2, -BS og 5,0 Sy B+a) + (3-402) + +e) +e = 31 By tte $454 24 = 31, OS 24,23,24 0S 25 S10. Consequently, the answer is the coefficient of 2” in the generating function Gtetetd. Pde tet. $2), (a) (1+ az)(1 + be)(1+c2)---(1+r2)(1 + sx)(1 + tz) (b) (L+ax + a%z? + a%x*)(1 + br + Bo? + P28). (1 +d + 22? + P29), 22910. Section 9.2 (@) Q+a)* (b) 842)" (©) (+2)? (d) 623/142) (e) (Q-2y? (f) 2*/(1-az) (b) 0,0,0,0,1,1,3,1, tar tatg ah + P+ah sequence 0,0,0,1,0,1,0,1,0,1,- (@) f(@) =1(1+32) = 14(—32)+(—82)' +(-32)*+...., 80 f(x) generates the sequence 1,~-3,9?, -39... ()_ fe) = 1/82) = (1/3)[1/( — (2/3))] = 2/3)I1. + (2/8) + (@/3)? + (2/3) +. J, 80 J(2) generates the sequence 1/3,(1/3)?, (1/3)*, (1/3)',... @) f(z) =1/(-2) +32" - 11 = (14242? +2%4...)+327—11, s0 f(z) generates the sequence do, 41,2)... Where a = —10, a7 = 4, and a; = 1 for all i # 0,7. Be +. 80 fle) (2) = f(z) ~ asz® + 32° (2) — age® + 32° — a72? + 727 ( F(z) — Qayz +2 — 2aga? + 32° a ae cree es eae (ee) @) CP) = ary) = (7) o Govr=corP yea = CF) (Sea = (a = (2) (eye - ICH" + )G) = 0) -OO+@ ® +040 © ()+G)+@) © @+(a)+G) @ 0 &) (BD - 5G)" = (2) - 5G) © G-o8+ OE)" + YEO" + GY? + QG)C-p" = )+0G)+ OG) +O) +@) (a) (tat +... = ot eta? +...) = 2!%(1 — 2). The coefficient of 2! in (aay is (Q)y = oer = (i). (b) (®+at+...42°) = 221 +2+27+...+2°)!, The coefficient of 2” in {(1—z2)"/(1— a) = (1-272) 4 = 1407+, +t (Gat Get 230u. 12, 13, 14, 15. 16. ie YD + G2)? = IY") + (12) = (i) -40)- Consider each package of 25 envelopes as one unit. Then the answer to the problem is the coefficient of 2! in (2427 +...42% 42%)! = M1 4¢2+...+2°%)4. This is the same as the coefficient of 2 in ((1~2%)/(1—2)j = (1—2®)4(1 —2)4 = [142% 462... 2™Y(F)4..4(54)(—2)+..+ (GS) (2) +... + (Ga) (—2) 4. Consequently the answer is (5$)(~1)"-4(¢f)(—1)" +6(34)(-1)* = (8) —4(¢4) +6(2). (a) The coefficient of x in (2? +2°+...))= 21 tatz%+...)§= 21-2) = 2 + (a0) + (GPY(-a) +d is (8) = (yay = (8). This is of ways to distribute the 24 ‘bottles of one type of soft drink among the surveyors so that each gets at least two bottles. Since there are two types, the two cases can be distributed according to the given restrictions in (*)” ways. (b) The coefficient of 2% in (2? +24+...)° is (7) and the answer is (18)(1). (at at tat batt att a8) 21 —a8)/(1— 2) = 212) G2) + (P) a) + (P)(-2)?+...], The numerator of the answer is the coefficient of 2!® in (1—°)'[(#2) + (Pes)+ [= (122° + (2) 2 — (1) a8 +... 272) + (BY) 4.4 and this equals Ge 1) — @)G B- 1) + (2) ( oP) 1s— QC ‘ye (@)- (AE By + 1a) (AG J) — (2). The final answer is obtained by dividing the last result by 6", the size of 1¢ sample (48+ ate spe +210) = 2%(1+2 +27 + 2°)(1 +25), 80 we need the coefficient of i in (i —2*)/(1 — z)(1 + 225 + 2) = (1 - ct) — z)9(1 + 225 4+ 2) = 11 (24) + (at? +. eS) + (Ca) + GaP +. at 22° +2), ‘This coefficient is {(G2)(—1)"* +2(°)(-1) + F)(-)41- OIG) ra) GY + OIG) out + ar 0 ~ BIG) = 1G) +26) + CY - QC) 202) + o + Gi(e) +: + 20) (5) (§). This result is then divided by (49)(2*), the size e the sample space, in determine the probability. Here we need the coefficient of 2 in (1+zfz?42°+...)JQl+ 22 424+...) = Q/Q. ~ 2)¥C/ — 27) = 0/0 — 2)°0/0. + 2)). Using partial fraction decomposition, 25+ qty = MUS + BU + Ge 4 where the coefficient of 2 is (—1)*(1/8) + (1/8) + (0/4)(33)(-1 + /2)(3 (4/8) + (-1)"] + /4y("4) + (272)("2?). For the hamburgers we need the coefficient of 2"? in (e+ 27 +...)(2?+2°4...)° = ae ah 231at. 18. 19. 20. z°(1/(1 — z))'. This is the coefficient of 2° in (1-2)-ie, ()(-D8 = For the hot dogs we need the coefficient of x6 in (x? +24+...)(Lte+z?+. : +8) = 2(1/(1—2))((1—2*)/(1—2)). This is the coefficient of 2" in (1—2*)(1—2)~4 = [1— Gt Gen anil) +(Hcae hie or-OAar+OG (3) - @() + @)Q)- By the rule of product the total number of distributions for the prescribed conditions is iG) - OE) + OO: Qae-dt ati afl (et eget ted 428) Sle(rtett.. te totes... to Pt(etett... +24. Cte tte ere eee te one roll ‘two rolls three rolls where the 1 takes care of the case where the die is not rolled. (b= 42)? = (() + (9) (40) + (Y)(-42)? +... The coefficient of 2” is (Acer (21/2) = n+1)((-1/2) —n +2)- nl (L42n—2)(1 fen a8) 042) oy - (1/2) = {U2)_ gyn (2n~1)(2n ~3 nl [2a =n 3) GI VCYA) _ Int SOD gy 2 -() (a) There are 2°-t = 27 compositions of 8 and 218/21 = 24 pelindromes of 8. Assuming each composition of 8 bas the same probability of being generated, the probability a palindrome of 8 is generated is 2"/2" = 1/8. (b) Assuming each composition of n has the same probability of being generated, the probability a palindrome of n is generated is 21/2) /9"-1 = gln/A-ntt a gi-fnfal, (a) We palindrome of 11 starts with 1, then tbat palindrome ends in 1. Upon removing ‘14’ feom the start and ‘+1’ from the end of the palindrome, we find a palindrome of 9. And there are 21°! = 2 = 16 palindromes of 9. Similar arguments tells us that there are 2l’/ = 8 palindromes of 11 that start with 2, 218/11 = 4 palindromes of 11 that start with 3, and 2l°/) = 2 palindromes of 11 that start with 4. 23221, 22, 23. 24, 25. 26, 27. (b) For the palindromes of 12, we find that 2U°/2) = 32 start with 1, 2!8/4 = 16 start with 2, 216/4) = 8 start with 3, and 2I4/4) = 4 start with 4. ‘The number of palindromes of n that start (and end) with ¢ is the number of palindromes of n ~ 2t. This is 21-29/21, Suppose a palindrome of n has an even number, say 2k, of summands. Let « be the sum of the last k summands. Then n = 2s, contradicting n odd. Let n = 2k, The palindromes of n with an even number of summands have a plus sign at the center and their number is the number of compositions of k ~ namely, 2°! = 2(-/)-), Since there are 2°? palindromes in total, the mumber with an odd number of summands ig 200 — (0M) = 9°/3(q — 1) = V/A(A) = 260/)-1, (a) ‘The number of palindromes of 10, where all summands are even, equals the number of palindromes of 5, which is 20/41 = 4, (b) 2071 =8 (ce) teal (a) Pr(¥ = y) = (§)""(4), 9 = 12,3, {b) and (c) Using the general formulas at the end of Example 9.18, with p = } and Cee! eee E(Y) =} = gy = 6, and i ay = War) = Valet = VOTE? = SCG) = VHB = sarr02s, Here we want 5%, Pr(Y = 2i). EE, Pr(Y =) = EEE 1G) = DEED" = GY) + + + OOE+ + OF = Oks = Orby = Bab - OED Let the discrete random variable Y count the number of tosses Leroy makes until he gets the first tail. Then Pr(Y = y) = ea), oe Here we are interested in Pr(¥ = 1) + Pr(¥ = 3) 4+ Pr(¥ =5)+--- GG) +--= Olt OF +O += Make = Odes = VO Since E(Y) = (a) Pry 3) + (30) + 2 2 4, the probability of success for each Bernoulli trial is p = ) = a) = 4 = 0.139942. 1) = (3) and Pr(¥ = 2) = (4) = B, s0 Pr(¥ > 3) =1-@)-(B) = 38 = 0.326531. one aoe 23 elgg = GADLED ++ © Poy. 3 d)=re.(# GP +-J=anG yy YOA+ + 23329, 30. 31. 32. 33. (a) (Leetate a? + o\Ote +2274 82+... _ Pr(¥ 25 and ¥ > 3) (@) Pr¥ 2 SY 23) = “Sosy Ay Gy. (0) Prk 2 OY 24) = (9G = (6) Var (Y) = 4/p*, where g = 1~p Consequently, oy = 28/9 = 1.763834. (s) The differences are 3~1,6—3,8—6,15~8, and 15—15 ~ that is 2,3,2,7, and 0, where2+3424+7+0=14. (>) (35,815) (ce) {lta,lta+bi+etbtel+a+b+c4+d} = Pr(¥ 2 5)/Px(¥ 23) = audy = 4, "4. So Var (¥) = (4/8 = (4)/$) = QB) = Be Using the ideas developed in Example 9.17, we consider one such subset: 1<1<3< 6 < 10 < 15 < 30 < 42 < 50. This subset determines the differences 0,2,3,4,5,15,12,8, which sum to 49. A second such subset is 1 <5 7< 9 < 15 < 21 < 32 < 43 < 50 < 50, which provides the differences 6,2,6,6,11,11,7, 0, which also sum to 49. ‘These observations suggest @ one-to-one correspondence between the subsets and the in- teger solutions of c + cz +c3+...+ce = 49 where c,cg 20 and o > 2 for 2 5. (b) Qa etat— 24-2 pet 284...) = ghp = (1+2)%, the generating function for the sequence (73), (37), (3), (;2),.-. Hence the convolution of the given nai of seauenoe it o Cryo Where (=r) Cis) =Ccnne nen. [Tie dhe alternating sequence 1, 2,3, -4,5, -6,7,... Section 9.3 7; G1; 542; BAL+L; 443; 44-241; 4414144; 34941; $4242; 3424141; BELA HHL; 2424241; 221414; 2TH, 141141 tL (2) f(@) = 1/G ~ 2101/0 ~ 2*)101/0 ~ 2) 9] ++ = TE E/G ~ 27) (b) oz) = (+24) +241 +28) Te,0-+24) () A(z)=( +2) 4+2°)\1 +25). (1+ 21) ‘The number of partitions of 6 into 1’s, 2's, and 3's is 7. @ G/a-*)0/0 -A)N/G -#yB/0 ~ #7) (b) A/G ~ AYE /0 — BA — NEEL 2) (a) and (b) (L+a?4+at+a5+...)L+atta%t.. + e+e 4 = Mi ea (a) fe) stata? t+... +251 eettaett.. ta) Te, teh te% +... 42%) = 12, [0 - 28 - 24) (b) THE + af to 4. + of) = UR (= e)/(2 = 2h] Let f(x) be the generating funcion for the number of partitions of n where no summand appears more than twice, Let g(z) be the generating function for the number of partitions of n where no summand is divisible by 3. deepsea Gerteyits? $21 +22 + 2° + ct + 28) Let f(z) be the generating function for the number of partitions of n where no summand is divisible by 4; g(x) is the generating function for the number of partitions of n where no even summand is repeated (odd summands may be repeated). f@= te te te rear 23510. abe was (1+2")- ps eH ae gen : Bais ‘This result follows from the one-to-one correspondence between the Ferrers graphs with summands (rows) not exceeding m and the transpose graphs (also Ferrers graphs) that have m summands (rows). Consider the Ferrers graph for a partition of 2n into n summands (rows). Remove the first column of dots and the result is a Ferrers graph for a partition of n. This correspondence is one-to-one, from which the result follows. Section 9.4 (e) = (b) (es (d) = (e) ae*= (f) we (a) f(z) = 3c = 30%) 49', so f(x) is the exponential generating function for the sequence 3,3?,3°, (b) f(x) = 6c — ee Gat _ gy, 2 ao f(x) is the exponential generating function for the sequence 3,24,138,...,6(6") ~3(2"),. (e) 1,4,3,1,1,1,1,... (a) 1,9,14, -10,2,25,28,... () f(e)=lertatt et function for the sequence 0!,1!,2!, (f) f(z) = 3[1 + 22+ (2x)? + (22) +...]+Dizo f, 80 f(z) is the exponential generating, function for the sequence 4,7,25,145,...,(3n!)2" +1,... Fla) + [3 —as](2*/3!) (2) + [-1~ ao}(2"/a!) = e — (1262%)/31 (2) + [2 ~ Qar}(2"/1!) + [4 ~ 2aa}(x?/2) Seat & ), so f(x) is the exponential generating (©) 92) = (A) gla) = 2f x) + 3e° 4 [2 — 2a — 3](x4/L!) + [4 — 2aq ~ 3](2?/21) + [8 — 2a ~ 3]{z"/3!) fa) (24 (22/2) + 2/8) += (CF ~ Uh = eo (e+ (Jer - Ger + ()). The coefficient of 21/121 in (e* ~1)* is 41 — (4)3" + (Jo? - (4). (b) How many onto functions are — A= {1,2,3,...,12} to B= {red, white, blue, black}? (fle) = (Lb a + (2/2) + FL + 8/2 + (2A +P = lle + AP = 23610. (1/4)(e*)(e™ + 2 +e) = (1/4)(e** + 2c +1). Here the coefficient of x1?/(12!) is (1/4)[4*? + 2(2")] and this counts the number of signals where the number of blue flags is even and the number of black flags is even. oa) = (L+2+(2?/2!)+...)(4 (29/31) +...)? = e[(e* ~e7*)/2}? = (1/4)(e#* —2e* 41). ‘The coefficient of z!?/(12!) in g(x) is (1/4)[4"? — 2(2!2)], and this counts the signals where the numbers of blue and black flags are both odd. Consequently, the number of signals where the total number of blue and black flags is even is (1/4)[4"? + 2(2")] + (4/4)[4? — 2(2'*)] = (1/2)(4"9). We find that = ie l¢a¢egege- (ONF+ADS+ANE+WNEH, 0 1/(1 ~ 2) is the exponential generating function for the sequence 0!, 1!,2!,3!,.... wil (a) i) (1+ 2)%(1 +24 (27/21)? (i) (+ 2) +2 + (27/21) (0 + 2 + (27/21) + (29/31) + (24/41)? Gi) GQ +2P(Q +24 (27/2) (b) (1+2)-(1+24(2?/2!))-(1 +24 (2?/2!) +(29/3!)+(2*/4!))-((2?/2!) + (29/3!) + (24/41). + The answer is the coefficient of @ in (24+ 4...498)4 A(z) = f(z)g(2) = co + cre + (2/2!) + ca(29/3!) + ..., where en(a" fn!) = Thaolaia'/it)(by-i2"*/(n —i)!) = (Etao(aiby (Cin = a)" = [DtLele!/Cil(n — i)!asdy—s(2"/n!) = [Thao ("aid ~c](2"/n!) - (a) (2/2){3 + 1)/(3) (b) (2/4)[3" + 3]/(3%) (©) (/2)|3# - 1/63") (€) (/2)8" = 1)/8") (©) G/2)[8" + 1/0") (a) f(x) = (x + (2/3!) + (08/51) +.) (at (0/21) + (27/3 +...) (eet = (1/2\(e* — 1)(e2*) = (1/2)(e? — 1)(e* — e*) = (1/2)(e* — c%* ~ ce” + e*). ‘The answer is the coefficient of 2*/(20!) in f(e) which is (1/2)[4% - 3° — 27° 44], (b) gla) = (+ 2 + (29/3!) + (et/4l) +.) = (e ~ (27/2))" = el — (JeH(xt/2) + eed ~ (Set(z?/2)° + (27/2)'. The coefficient of 27°/(20!) in g(z) is 4° ~ {)(4/2)(3")¢20)(29)+ (3) (4/4)(2#*)(20)(29)(18)(17)~ ($) (2/8)(1"4)(20)(19)(18)(47)(16)(15) {o) R(a) = (1 +2 + (2°/3!) + (24/4) +...) = (eF ~ (24/2))* = &* - (Je*(a5/6) + 237(4) e%*(28/6)* ~ (4)e*(25/6)* + (2°/6)'. The coefficient of 27°/(20!) in A(z) is 4° — e 3 ({)a7ey(a"*)(20y(29y(a8) + (2)(4/6)*(2"*)(20)(19)¢18)(1716)(45) ~ ($) (2/6)*(20)/(421). (d) The coefficient of 2%°/(20!) in (e*)%(1 + (22/2!) = e + (27/2!) is 3 + (1/2)(3*)(20)(19). Section 9.5 + (a) 142 +27 is the generating function for the sequence 1,1,1,0,0,0,..., so (1+2+2%)/(1~2) is the generating function for the sequence L1+L1+141,14+1+1+40,... ~ that is, the sequence 1,2,3,3,.... (b) 1+2 +2? + 2° is the generating function for the sequence 1,1,1,1,0,0,0,..., 80 (l+2+2?+2°)/(1—2) is the generating function for the sequence 1,1 +1,14+1+1,1+ T4+141,1414+1+1+40,14+14+1+1+4040,...~- that is, the sequence 1,2,3,4,4,4,.... (c)_142z is the generating function for the sequence 1,2,0, 0,0, 50 (14+2z)/(1~2) js the generating function for the sequence 1,1 +2,1-+2+ 0,1 0+0,... ~ that is, the sequence 1,3,3,3,.... Consequently, (1/(1~2))[(1 + 22)/(1 ~ 2)} = (1+ 22)/(1~2)? is the generating function for the sequence 1,1 + 3,14+3+43,1+3+4343,... — that is, the sequence 1,4,7,10,.... © @= (@) 2/Q-2) Gi) 2/0 -2) iv) 2/0-29 Dhak the coefficient of 2” in 2/(1~z)* the coefficient of 2” in 2(1—2)~* the coefficient of 2*-! in (1 — 2) apr = yr RS) nt) = Mn +1)(n) F(z) = [e(1+2)}/(1—z)* generates 0?,1°,2?,3%,...; [(1t2)|/(1—2)* = 07412242222 Peart. (d/dollats/1—z)| = P42 tbo? 2(d/dz){(2+2?)/(—2)] = OF Pr + %- 2? 439-22 4...; (d/de)(x + 2?)/(1 — 2) (2? + de + 1/11 —2)', so a(z? + 4a + 1)/(2 —2)§ generates 0°,0° + 19,0 + 19+ 2°,..., and the coefficient of 2" is Dhoi®. (e44e4+2)(1—a)* = (8442742) ((Z)+(5 1 (B)i-2)*+...]. Here the coefficient of a® is (B)- ayn? 44(c8)(~ ay? + (8) (aad = (mt) + 4(083) + (43) = (1/4) f(r + Tn )(n — Dn ~ 2) + 4(n + 2}(n + 1)(n){n = 1) + (n + 3)(n + 2X(n + I(n)] = [Gn + 1}()/41](Gn? + 6n) = (1/4)(n + 1(n)(n? + n) = (nr + (n)/2 (b) ee aon . The function (1 + 2)f(x) generates the sequence ap, ao + 41,41 + 43,42 + s,.... For the BEQUENCE Mg, Go + G1, Oo + G1 + G2, 01 + G2 + Gy, G2 + G3 + a4,..., the generating function is 238(l+24+2%)f(z). (1-2) f(2) = (1-2 \ag+aye-+an274 ag2?+...) = apt (a; —a9)2-+(a2~a;)e?4+(ag—an)29-+ +180 (12) f(z) is the generating function for the sequence 49, a1—a9, 42~4},43—a3,.. {f(@) — F@IM@ — 1) = OMe ~ 1))(@o ~ a0) + (re ~ a4) + (az2? — a) +..]. For n> 0, (a,2" — a,)/(2 — 1) = a,(2" — D/(z - 1) = a(z™ $2" +...407 +241), 90 Uf(@) — FDI/(@ ~ 1) = an + az + 1) + age? + 2 +1) + ag(e? +2? +241) +... Hence the coefficient of 2", for n>0,is Day. a it follows that e*/(1— z) Since e* is the generating funetion for 1,1,1/2! generates the sequence a9,),42,..., where a, = D2 (a) fy =1t+2+27 +4 c%+--- is the generating function for the sequence 1, 1, 1,1, .... Applying the summation operator, we then learn that (;4;)? is the generating function for the sequence 1,14+1,1+14+1,1+1+4141, ... — that is, the sequence 1, 2, 3, 4, .... Consequently, z/(1 — z)? is the generating function for 0, 1, 2, 3, 4, ... and «/(1—2)* the generating function for 0,0+1,0+142,0+1+24+3,0414+243+4,...— that is, the sequence 0, 1, 3, 6, 10, ... (where 1, 3, 6, 10, ... are the triangular numbers). (b) The sum of the first n triangular numbers is the coefficient of 2 in the generating function 2/(1— 2) = 2(1—2)-* = af(7') + (7')(—2) + (F)(-2)? +++. So the answer is the coefficient of 2"-? in (1—z2)~ and this is (~4)(—1)""? = (-1)"71 (“B41 = (22) = G/6)n(n + 1}(n +2), as we learned in Example 4.5. Supplementary Exercises (a) 6/—2z)+1/(- 2) (b) 1/(1-az) (©) 1/1-G+a)z] (a) 1/A-2)+1/(-az) Let f(2) = (282% p24 tap cl) = 21425 42% 42% +207), in f(x) equals the coefficient of x™ in ((1 ~2"*)/(1 ~ 29))° = (1 — 219)!9(1 — 28)" = f= (}at + (1) a9 — + 238-129) + (-39)(—29) + (1)(—29? +... J. This coefficient 8 G)CO"- A) CY + EOI = &)- OG) +E) 0). ‘The generating function for each type of bullet is (2? +27 +.. as ae Alte+at+ .-+2°)4, The coefficient of 2 in (1-2°)(1—2)-* = [1 — ({}a* + (9 ome SQN a ‘ye Gear Ghcernd » Gew-GIew+ OG) =0)-OE* 6) By the rule of product the answer is {({3) ~ (1) (2) + (:)?. ‘The coefficient of 2*° fz)= Geet Pee +. iter? +a% tet...) (lta! ta2% 42% 4...) 2398 10. Let f(z) be the generating function for the number of partitions of n where no even summand is repeated (although an odd summand may be repeated); (2) is the generating function for the number of partitions of n in which no summand occurs more than three times. Then g(2) = (14242742°\(1+2?42442%\(1+2%42°42" dz)(+27)N+ HY + 20 +2) +25) (27) /(~ 2) + 2°) - 24/1 ~ 2?) + 2) ~ eY/1-2 +2)... = [1/-)A+2)1/A-2 0 +24 [1/—2*)\(1 +: ater te%4.. \(L +21 te%+ao¢o%4.. lteter +e% 4284...) te! (2). ‘This result is the coefficient of 2"°/(10!) im (1+ (2?/2!) +(x9/31)+...)* = (e—2)* = ef — (Jet + (Sate — (Sater + ({)2*. This coefficient is 4"— (1)(10)(3°) + (2) 20)(9)(28)— ()aoyoys). (0) (12a) 21 4 pe, ALAMBMADUesLayae CHAIN _y gr = 14+ D2, MMe gr, so g(x) is the exponential generating function for 1,5,5(7), 5(7)(9),... (b) (1 az) 14 52, MiMi aibort (gay = 1 ~ abe + b~ 1)a?2?/2! +. Consequently, by comparing coefficients of like powers of 2, we have ~ab = 7,(b—1)a? = 7-1 and a=4,b=-7/4. For exch partition of n, place a row of n+k dots above the top row in its Ferrers graph and the result is a Ferrers graph for a partition of 2n+k where n+k is the largest summand, This one-to-one correspondence yields P, = P3. Taking the transpose of a Ferrers graph for a partition in P; yields the Ferrers graph for a partition in Ps, and vice versa. The result now follows from these two observations. For each n € Zt, (1+2)" = (5) +(")e+ (327+ (s)2*+...+(t)2". Taking the derivative of both sides we find that n(l-fayrt= (") +2(s)era(t)e+ ~+n(") When 2 =1 we obtain (Qo) f@)=C4t2)i+e el ¢e%te%t2" 2 PB Tae) mies ¥ nfl +i)" = (2 sak $ oth poh pp oh) eee 240coms 12, 13. 14. 15. 16. (a) The coefficient of z™ in (c+2?+...)% = 2%1+e+2+...)" is the coefficient of 2 in (Ltztz?+...)!? = (1-2), and thisis (-?)(-1)8 = (-1)8(487)(-1)8 = (¥7). (b) The coefficient of 2° in (e+2?+...)9=2%(1+2427+...)° is (7)(-1)*= (2). The probability for this type of distribution is (4) (2)/(%). Fix m, 0| = &, for k a fixed positive integer, we have a, = k" and the generating function is g(z) = 1+ ke + Kz? ++ = 1+ (ke) + (ka) +-- = ple. {@) 2*f(z) 0) (fhe) = (09 aye bags +o baat)" (a) 1= Deby H(A) = AL + 3+ (2) +] = Bacal = HIL/(8/4)] = (4/3), 20 & = 9/4, 24417. 18. (b) Pr(X = 3) = (3/4)(1/4)° = 3/256 Pr(X < 3) = D8 Pr(X = 2) = (3/4)[1 + (1/4) + (1/4)? + (1/4)9} = (3/4)(85/64) = 255/256 Pr(X > 3) = 1 ~ Pr(X <3) = 1 — (255/256) = 1/256 {Alternately, Pr(X > 3) = Pr(X > 4) = OB, Pr(X = 2) = (8/4)[(1/4)* + (1/4) + a/aye+ = (8/4)(2/4)(0 + (1/4) + /4) 4-1 = oN /8Matral = (1/4)* = 1/256.) Pr(X 29) = S24 Pr(X = 2) = TE 4(8/4)(1/4)" = (8/4)(1/4)[1 + (1/4) + (1/4)? +] = (8/64) g=izql = 1/4)? = 1/16, (c) For n € Zt, Pr(X > n)= DR, Pr(X = 2) = ON, (3/4)(1/4)" = (8/4)(1/4)" Deeo(1/4)! = 1/4)". Consequently, Pr(z > 4|X > 2) = Oe = Pr(X > 4)/Pr(X 22) = (1/4)1/(1/4)? = (1/4). Likewise Pr(X > 104|X > 102) = (1/4). Fork ¢ Zk fined, we find that Pr(¥ > 4) = Tua tp (where q = 1~p) aa ipsatpt gps =o pl +qd ete] Pay tp(t) = gh1. Consequently, Pr(¥ > ml¥ > n) = Pr(¥ > mand ¥ > mips Sn) = Pr(¥ > m)/Pr(¥ > n) = gh¥/g" = g™™". [This property is the reason why a geometric random variable is said to be memoryless. In fact, the geometric random variable is the only discrete random variable with this property.) (a) The car travels the first mile in one hour, the second mile in 1/2 hour, the third mile in 1/4 [= (1/2)] hour, and the fourth mile in 1/8 [= (1/2)°] hour. Consequently, the average velocity for the first four miles is 4/[1-+(1/2)+(1/2)*+(1/2)"] = 4/[[1-(4/2)4J/[1—(1/2)]] = 4/[2(15/16)] = 32/15 = 22 miles per hour. (b) The average velocity for the first n miles is n/[1 + (1/2) + (1/2? +-+-+(1/2] = n/{f — (2/2)"1/1 = 2/2)]] = n/[2{(2" ~ 1)/2"]] = n(2"-*)/[2" = 1] miles per hour. (c) For n= 19 the average velocity is 4980736/524287 = 9.500018120 miles per hour. For n = 20 the average velocity is 2097152/209715 = 10.00000954 miles per hour. Hence the smnallest value of n for which the average velocity for the first n miles exceeds 10 miles per hour is n = 20. 242.