P2.4 - Sequences and series

Syllabus
2019
Topic
P2.4
Level
AS

Learning objectives

Generate a sequence directly or recursively

A sequence is an ordered list whose position matters. A rule can give a term directly from its position, or generate each term from the previous one.

Rule type What must be given How a term is found
Explicit: un=f(n)u_n=f(n) the formula and term number nn substitute nn directly
Recursive: xn+1=f(xn)x_{n+1}=f(x_n) the recurrence and a starting value calculate earlier terms in order

For un=3n1u_n=3n-1, the sequence begins 2,5,8,11,2,5,8,11,\ldots and u7=3(7)1=20u_7=3(7)-1=20 without finding terms 5 or 6.

For x1=2x_1=2 and xn+1=xn21x_{n+1}=x_n^2-1, x2=3,x3=8,x4=63.x_2=3,\qquad x_3=8,\qquad x_4=63. Each input is the preceding term, so the starting value is part of the definition.

Keep the indices aligned: xn+1x_{n+1} is the next term produced from xnx_n. Do not replace xnx_n by nn, or apply a recurrence without its required starting value.

Build and sum an arithmetic sequence

An arithmetic sequence has a constant difference dd. Starting from first term aa, moving to term nn adds dd exactly n1n-1 times.

un=a+(n1)d,Sn=n2(2a+(n1)d)=n2(a+l)u_n=a+(n-1)d,\qquad S_n=\frac n2\bigl(2a+(n-1)d\bigr)=\frac n2(a+l)

The sum formula follows by writing the same finite sum forwards and backwards: Sn=a+(a+d)++[a+(n1)d],S_n=a+(a+d)+\cdots+[a+(n-1)d], Sn=[a+(n1)d]++(a+d)+a.S_n=[a+(n-1)d]+\cdots+(a+d)+a. Each of the nn paired columns totals 2a+(n1)d2a+(n-1)d, so doubling and then halving gives the formula.

Sigma notation compresses an addition: r=1nur\sum_{r=1}^{n}u_r means u1+u2++unu_1+u_2+\cdots+u_n. Setting a=1,d=1a=1,d=1 gives 1+2++n=n(n+1)2.1+2+\cdots+n=\frac{n(n+1)}2.

For 7,11,15,7,11,15,\ldots, a=7a=7 and d=4d=4. Thus u20=83u_{20}=83 and S20=202(7+83)=900.S_{20}=\frac{20}{2}(7+83)=900.

Term nn uses n1n-1 differences. In a finite sum, identify the number of terms before substituting; it is not automatically the value of the final term.

Recognise increasing, decreasing and periodic sequences

Classify a sequence by comparing consecutive terms or by identifying an exact repeating cycle.

Type Test
Strictly increasing un+1>unu_{n+1}>u_n for every relevant nn
Strictly decreasing un+1<unu_{n+1}<u_n for every relevant nn
Periodic with period pp un+p=unu_{n+p}=u_n; the order is the smallest positive such pp

For an explicit sequence, inspect un+1unu_{n+1}-u_n: a value always positive proves increasing, while one always negative proves decreasing. For un=n2+1u_n=n^2+1, un+1un=2n+1>0u_{n+1}-u_n=2n+1>0 for positive integer nn, so the sequence is increasing.

The sequence 4,34,13,4,34,13,4,\frac34,-\frac13,4,\frac34,-\frac13,\ldots repeats every three terms, so it is periodic of order 33. With a one-step deterministic recurrence, returning to an earlier value makes the same following cycle repeat.

A few rising or falling terms do not prove the pattern continues. Equality also fails the strict tests; and a stated period is not the order if a smaller positive period works.

Control finite and infinite geometric series

A geometric sequence multiplies by the same common ratio rr each time. With first term aa, its terms and finite sum are

un=arn1,Sn=a(1rn)1r(r1)u_n=ar^{n-1},\qquad S_n=\frac{a(1-r^n)}{1-r}\quad(r\ne1)

Write Sn=a+ar++arn1S_n=a+ar+\cdots+ar^{n-1} and multiply by rr. Subtracting rSnrS_n from SnS_n cancels the middle terms, leaving (1r)Sn=aarn,(1-r)S_n=a-ar^n, which gives the finite-sum formula.

Only when r<1|r|<1 does rn0r^n\to0, so only then does the geometric sum converge to S=a1r.S_\infty=\frac{a}{1-r}. A negative ratio may converge while the terms alternate signs.

For a=5,r=0.8a=5,r=0.8, requiring Sn>24S_n>24 gives 25(10.8n)>2425(1-0.8^n)>24, hence 0.8n<0.040.8^n<0.04. Therefore n>log(0.04)log(0.8)14.43,n>\frac{\log(0.04)}{\log(0.8)}\approx14.43, so the smallest integer is 1515.

Use arn1ar^{n-1} for term nn, not arnar^n. Never use SS_\infty unless r<1|r|<1; when solving an inequality with 0<r<10<r<1, dividing by logr<0\log r<0 reverses its direction.

Expand a positive integer power with binomial coefficients

For a positive integer nn, the binomial theorem gives every term of (a+bx)n(a+bx)^n without repeated multiplication.

(a+bx)n=r=0n(nr)anr(bx)r,(nr)=n!r!(nr)!(a+bx)^n=\sum_{r=0}^{n}\binom nr a^{n-r}(bx)^r,\qquad \binom nr=\frac{n!}{r!(n-r)!}

Here n!=n(n1)21n!=n(n-1)\cdots2\cdot1 and 0!=10!=1. The index rr is the power of xx, so choosing r=0,1,2,r=0,1,2,\ldots produces terms in ascending powers.

For (23x)4(2-3x)^4, (23x)4=24+(41)23(3x)+(42)22(3x)2+(43)2(3x)3+(3x)4=1696x+216x2216x3+81x4.\begin{aligned}(2-3x)^4={}&2^4+\binom41 2^3(-3x)+\binom42 2^2(-3x)^2\\&+\binom43 2(-3x)^3+(-3x)^4\\={}&16-96x+216x^2-216x^3+81x^4.\end{aligned}

The sign belongs inside (bx)r(bx)^r: odd powers keep a negative sign and even powers become positive. This formula here is restricted to positive integer nn and terminates after n+1n+1 terms.