Verification:ax+by=31 can be checked by direct substitution (the algebra confirms this).
Summary:x and y are 3515±230, and with x taking the + sign, a=3618−30, b=3618+30 (with a and b swapped if x and y are interchanged). ■
Examiner Notes
This was the most popular question on the paper, and many earned good marks on it. Nearly all the candidates followed the hint, and most then applied the same trick with the third equation. Subsequent success depended on a candidate realising that they had simultaneous equations in xy and x+y, although very rarely some managed to solve directly in x and y.
Obtain simplified expressions for S3(n) and S4(n).
(ii) Explain, using (∗), why Sk(n) is a polynomial of degree k+1 in n. Show that in this polynomial the constant term is zero and the sum of the coefficients is 1.
(ii) Applying induction, with the assumption that St(n) is a polynomial of degree t+1 in n for t<r for some r, and then considering (*),
(n+1)r+1−(n+1) is a polynomial of degree r+1 in n,
and each of the terms −(jr+1)Sr+1−j(n) is a polynomial of degree r+2−j in
n where j≥2, i.e. the degree is ≤r. A sum of polynomials of degree ≤r+1 in n, is a polynomial of degree ≤r+1 in n, and there is a single non-zero term in nr+1 from just (n+1)r+1 so the degree of the polynomial is not reduced to <r+1, i.e. it is r+1. (The initial case is true to complete the proof.)
If, Sk(n)=∑i=0k+1aini=∑r=0nrk then Sk(0)=a0+∑i=1k+1ai0i=∑r=00rk=0 and so a0=0
Sk(1)=∑i=0k+1ai1i=∑r=01rk=1 and so ∑i=0k+1ai=1 as required.
We proceed by strong induction. S1(n)=21n2+21n is a polynomial of degree 2. Suppose St(n) is a polynomial of degree t+1 in n for all t<k. From (∗):
The term (n+1)k+1−(n+1) is a polynomial of degree k+1 in n. Each Sk+1−j(n) for j≥2 is a polynomial of degree k+2−j≤k by the inductive hypothesis. Therefore Sk(n) is a polynomial of degree k+1 in n.
Constant term is zero. Setting n=0: Sk(0)=∑r=00rk=0k=0 (for k≥1). So the constant term is 0.
Sum of coefficients is 1. Setting n=1: Sk(1)=∑r=01rk=0+1=1. Since Sk(1) equals the sum of the coefficients of Sk(n), this sum is 1. ■
Examiner Notes
About three fifths attempted this question, often obtaining the starred result and the familiar S3(n) successfully, but with S4(n) tripping up many. Any that made progress on part (ii) tended to be able to complete the whole question.
3 The point P(acosθ,bsinθ), where a>b>0, lies on the ellipse
a2x2+b2y2=1.
The point S(−ea,0), where b2=a2(1−e2), is a focus of the ellipse. The point N is the foot of the perpendicular from the origin, O, to the tangent to the ellipse at P. The lines SP and ON intersect at T. Show that the y-coordinate of T is
1+ecosθbsinθ.
Show that T lies on the circle with centre S and radius a.
Hint
dxdy=−asinθbcosθ
So the line ON is y=bcosθasinθx
SP is y=a(cosθ+e)bsinθ(x+ae)
Solving simultaneously by substituting for x to find the y coordinate of T,
y=a(cosθ+e)bsinθ(asinθbcosθy+ae)
and using b2=a2(1−e2) to eliminate a2 gives the required result.
Then the x coordinate of T is a(1+ecosθ)b2cosθ.
Eliminating θ using secθ+e=axb2 and tanθ=axby,
(x,y) satisfies (axb2−e)2=1+(axby)2
and again using b2=a2(1−e2), this time to eliminate b2, gives, following simplifying algebra
(x+ae)2+y2=a2, as required.
Model Solution
Finding the tangent at P and the line ON.
The ellipse is a2x2+b2y2=1. Differentiating: a22x+b22ydxdy=0, so dxdy=−a2yb2x.
At P(acosθ,bsinθ): dxdy=−a2⋅bsinθb2⋅acosθ=−asinθbcosθ.
The tangent at P has gradient −asinθbcosθ, so ON (perpendicular from origin to tangent) has gradient bcosθasinθ.
Therefore ON: y=bcosθasinθx, i.e.\ x=asinθbcosθy.
Finding the line SP.
S=(−ea,0) and P=(acosθ,bsinθ). The gradient of SP is:
acosθ−(−ea)bsinθ−0=a(cosθ+e)bsinθ
So SP: y=a(cosθ+e)bsinθ(x+ea).
Finding the y-coordinate of T.
Substituting x=asinθbcosθy from ON into the equation of SP:
Just under half attempted this. Most were reluctant to use parametric differentiation. Some found T‘s coordinates successfully and got not further, but most either made very little progress on the whole question, or got right through it.
Dividing by (x+1)>0 and noting x+12x2−1=x2−12(x−1):
arcoshx>2x2−1x−1■
Part (iii): arcoshx>3x+2x2−1
The result of part (ii) gives arcosht>2t2−1t−1 for t>1. Integrate from 1 to x:
∫1xarcoshtdt>2∫1xt2−1t−1dt
From part (ii), the left side is xarcoshx−x2−1 and the right side is 2(x2−1−arcoshx):
xarcoshx−x2−1>2x2−1−2arcoshx
(x+2)arcoshx>3x2−1
Dividing by (x+2)>0:
arcoshx>3x+2x2−1■
Examiner Notes
Almost exactly the same number attempted this as question 3, but with much less success. The initial inequality was frequently poorly justified, but some managed to apply it correctly to obtain the starred result, and went on to do part (ii) respectably. However, for most, it was a case of all or nothing.
5 The functions Tn(x), for n=0,1,2,…, satisfy the recurrence relation
Tn+1(x)−2xTn(x)+Tn−1(x)=0(n⩾1).(*)
Show by induction that
(Tn(x))2−Tn−1(x)Tn+1(x)=f(x),
where f(x)=(T1(x))2−T0(x)T2(x).
In the case f(x)≡0, determine (with proof) an expression for Tn(x) in terms of T0(x) (assumed to be non-zero) and r(x), where r(x)=T1(x)/T0(x). Find the two possible expressions for r(x) in terms of x.
Hint
There are a number of correct routes to proving the induction, though the simplest is to consider ((Tk+1(x))2−Tk(x)Tk+2(x))−((Tk(x))2−Tk−1(x)Tk+1(x))
For f(x)=0, (Tn(x))2−Tn−1(x)Tn+1(x)=0
and so Tn(x)Tn+1(x)=Tn−1(x)Tn(x) provided that neither denominator is zero, leading to
Tn−1(x)Tn(x)=T0(x)T1(x)=r(x),
and so Tn−1(x)Tn(x)×Tn−2(x)Tn−1(x)×⋯×T0(x)T1(x)=(r(x))n
Thus Tn(x)=(r(x))nT0(x)
Substituting this result into (*) for n=1,
((r(x))2−2xr(x)+1)T0(x)=0, and as T0(x)=0, solving the quadratic gives
r(x)=x±x2−1
Model Solution
Induction proof that (Tn(x))2−Tn−1(x)Tn+1(x)=f(x).
Define g(n)=(Tn(x))2−Tn−1(x)Tn+1(x).
Base case (n=1): g(1)=(T1(x))2−T0(x)T2(x)=f(x) by definition. ✓
Inductive step: Suppose g(k)=f(x) for some k⩾1. We show g(k+1)=f(x).
Consider the difference:
g(k+1)−g(k)=(Tk+1)2−TkTk+2−(Tk)2+Tk−1Tk+1
=Tk+1(Tk+1+Tk−1)−Tk(Tk+2+Tk)
From the recurrence Tn+1+Tn−1=2xTn, we have Tk+1+Tk−1=2xTk and Tk+2+Tk=2xTk+1. Substituting:
g(k+1)−g(k)=Tk+1⋅2xTk−Tk⋅2xTk+1=0
Therefore g(k+1)=g(k)=f(x).
By the principle of mathematical induction, g(n)=f(x) for all n⩾1. ■
Expression for Tn(x) when f(x)≡0.
When f(x)=0, we have (Tn)2=Tn−1Tn+1 for all n⩾1.
Rearranging (and assuming Tn=0):
TnTn+1=Tn−1Tn
This ratio is constant for all n⩾1, so:
Tn−1Tn=Tn−2Tn−1=⋯=T0T1=r(x)
Telescoping:
T0Tn=Tn−1Tn⋅Tn−2Tn−1⋯T0T1=(r(x))n
Therefore:
Tn(x)=(r(x))nT0(x)(1)
Finding r(x) in terms of x.
Substituting (1) into the recurrence T2−2xT1+T0=0 (i.e.\ the case n=1):
(r)2T0−2x⋅r⋅T0+T0=0
Since T0=0, divide through:
r2−2xr+1=0
By the quadratic formula:
r(x)=22x±4x2−4=x±x2−1
Examiner Notes
In terms of attempts and success, this resembled question 2. Apart from some that made no progress at all, the induction was accessible to many, as was the expression for Tn(x). In both of these there were frequent gaps or inaccuracies even though the solutions were understood in essence.
Hence show that x=−32p+Ap−2, where A is an arbitrary constant. Find y in terms of x if p=−3 when x=2.
(ii) Given instead that
y=2xp+plnp,
and that p=1 when x=−41, show that x=−21lnp−41 and find y in terms of x.
Hint
(i) Differentiating y=p2+2xp with respect to x gives
p=2pdxdp+2xdxdp+2p which can be rearranged suitably.
The differential equation dpdx+p2x=−2 has an integrating factor p2 and integrating will give the required general solution.
Substituting x=2,p=−3, leads to A=0, i.e. p=−23x which can be substituted in the original equation and so y=−43x2.
(ii) The same approach as in part (i) generates dpdx+p2x=−p(lnp+1), which with the same integrating factor has general solution
x=−41−21lnp+Bp−2
and particular solution
x=−21lnp−41
Again, substitution of lnp (and p) in the original equation leads to the solution which is y=−21e−2x−21
Model Solution
Part (i)
We are given y=p2+2xp where p=dxdy.
Differentiate both sides with respect to x:
dxdy=2pdxdp+2xdxdp+2p
Since dxdy=p:
p=2pdxdp+2xdxdp+2p
0=2pdxdp+2xdxdp+p
−p=dxdp(2p+2x)
Provided p=0, divide both sides by p:
−1=dxdp(2+p2x)
dpdx=−2−p2x✓
This is a linear first-order ODE in x as a function of p:
dpdx+p2x=−2
The integrating factor is e∫p2dp=p2. Multiplying through:
p2dpdx+2px=−2p2
dpd(p2x)=−2p2
Integrate with respect to p:
p2x=−32p3+A
x=−32p+Ap−2■
Finding y in terms of x when p=−3 at x=2.
Substitute x=2, p=−3:
2=−32(−3)+(−3)2A=2+9A
So A=0, giving x=−32p, hence p=−23x.
Substituting into y=p2+2xp:
y=(−23x)2+2x(−23x)=49x2−3x2=−43x2
Part (ii)
We are given y=2xp+plnp where p=dxdy.
Differentiate both sides with respect to x:
dxdy=2p+2xdxdp+dxdplnp+p⋅p1dxdp
Since dxdy=p:
p=2p+2xdxdp+(lnp)dxdp+dxdp
−p=dxdp(2x+lnp+1)
Divide through by p:
−1=dxdp(p2x+plnp+1)
dpdx=−p2x−plnp+1
This is a linear ODE:
dpdx+p2x=−plnp+1
The integrating factor is again p2. Multiplying through:
dpd(p2x)=−p(lnp+1)
Integrate the right-hand side. Since dpd(plnp)=lnp+1:
∫p(lnp+1)dp=∫p⋅dpd(plnp)dp
Using integration by parts with u=p, dv=(lnp+1)dp (so v=plnp):
∫p(lnp+1)dp=p⋅plnp−∫plnpdp
Alternatively, note that dpd(2p2lnp)=plnp+2p, so
∫p(lnp+1)dp=∫plnpdp+2p2
and ∫plnpdp=2p2lnp−4p2 (by parts with u=lnp, dv=pdp).
Therefore:
∫p(lnp+1)dp=2p2lnp−4p2+2p2=2p2lnp+4p2
So:
p2x=−2p2lnp−4p2+B
x=−21lnp−41+Bp−2
Substitute the condition p=1, x=−41:
−41=−21ln1−41+B=−41+B
So B=0, giving:
x=−21lnp−41✓
Finding y in terms of x.
From x=−21lnp−41, we get lnp=−2x−21, so p=e−2x−1/2.
Substituting into y=2xp+plnp:
y=2x⋅e−2x−1/2+e−2x−1/2(−2x−21)
=e−2x−1/2[2x−2x−21]=−21e−2x−1/2
Examiner Notes
More than 80% attempted this, and with more success than any other question. Having obtained the relation between x and p in each part, quite a few attempts then treated these as differential equations rather than merely substituting back to find expressions for y, and consequent inaccuracies lost marks.
7 The points A, B and C in the Argand diagram are the vertices of an equilateral triangle described anticlockwise. Show that the complex numbers a, b and c representing A, B and C satisfy
2c=(a+b)+i3(b−a).
Find a similar relation in the case that A, B and C are the vertices of an equilateral triangle described clockwise.
(i) The quadrilateral DEFG lies in the Argand diagram. Show that points P, Q, R and S can be chosen so that PDE, QEF, RFG and SGD are equilateral triangles and PQRS is a parallelogram.
(ii) The triangle LMN lies in the Argand diagram. Show that the centroids U, V and W of the equilateral triangles drawn externally on the sides of LMN are the vertices of an equilateral triangle.
[Note: The centroid of a triangle with vertices represented by the complex numbers x, y and z is the point represented by 31(x+y+z).]
Hint
The starting point c−a=21(1+i3)(b−a) leads to the given result.
Interchanging a and b gives 2c=(a+b)+i3(a−b) if A, B, C are described clockwise.
(i) The clue to this is the phrase “can be chosen” and a sketch demonstrates that a pair of the equilateral triangles need to be clockwise, and the other pair anti-clockwise
Applying the results in the stem of the question to this configuration,
and so 2PS=(g−e)+i3(g−e)=−2RQ, PSQR is a parallelogram.
(The pairs could have been chosen with opposite parity leading to very similar working.)
(ii) Supposing LMN is clockwise, U is the centroid of equilateral triangle LMH, V of MNJ, and W of NLK, then
3u=l+m+h where 2h=(l+m)+i3(m−l) with similar results for v and w.
Both 6w, and 3[(u+v)+i3(u−v)] can be shown to equal 3(n+l)+i3(l−n) and so UVW is a clockwise equilateral triangle.
Model Solution
Establishing the relation for anticlockwise equilateral triangles
Let A, B, C be vertices of an equilateral triangle described anticlockwise. The vector from A to B is b−a. Rotating this by 60° anticlockwise (multiplying by eiπ/3) gives the vector from A to C:
c−a=(b−a)eiπ/3
Since eiπ/3=21+2i3:
c−a=(b−a)(21+2i3)
2(c−a)=(b−a)(1+i3)
2c=2a+(b−a)+i3(b−a)=(a+b)+i3(b−a)
For a clockwise equilateral triangle, we rotate by −60° instead:
c−a=(b−a)e−iπ/3=(b−a)(21−2i3)
2c=(a+b)−i3(b−a)
Part (i): Constructing the parallelogram PQRS
Given quadrilateral DEFG, we construct equilateral triangles on each side. For each side, we choose the triangle to be on the outside of the quadrilateral (consistent orientation).
Let P be the third vertex of the equilateral triangle on DE (described anticlockwise), Q on EF, R on FG, S on GD.
These are equal if and only if f−2e+d=d−2g+f, i.e., e=g. That’s not generally true, so let me reconsider the orientation choices.
Actually, we can choose the equilateral triangles with alternating orientations. Let P on DE be clockwise, Q on EF anticlockwise, R on FG clockwise, S on GD anticlockwise:
So 2(q−p)=(f−d)(1+i3) and 2(s−r)=(d−f)(1+i3)=−(f−d)(1+i3).
This gives q−p=−(s−r), so PSRQ is a parallelogram (with PQRS having PQ∥SR and PQ=SR). More precisely, p+r=q+s confirms PQRS is a parallelogram.
Part (ii): Napoleon’s theorem (centroids of external equilateral triangles)
Let triangle LMN be given. Construct equilateral triangles externally on sides LM, MN, NL, with third vertices H, J, K respectively. Their centroids are:
u=3l+m+h,v=3m+n+j,w=3n+l+k
Using the anticlockwise relation (external to triangle LMN):
To show UVW is equilateral, we verify 6w=6u+6v+i3(6v−6u)… no, we use the relation 2w′=(u′+v′)+i3(v′−u′) where u′,v′,w′ are the centroids. Let me check:
So RHS =12w= LHS. Therefore UVW is an equilateral triangle (described clockwise). ■
Examiner Notes
Less than a fifth attempted this and frequently with little success except for obtaining the initial result. The configuration for part (i) tripped up many, although some skipped that to do part (ii) successfully.
8 (i) The coefficients in the series
S=31x+61x2+121x3+⋯+arxr+…
satisfy a recurrence relation of the form ar+1+par=0. Write down the value of p. By considering (1+px)S, find an expression for the sum to infinity of S (assuming that it exists). Find also an expression for the sum of the first n+1 terms of S.
(ii) The coefficients in the series
T=2+8x+18x2+37x3+⋯+arxr+…
satisfy a recurrence relation of the form ar+2+par+1+qar=0. Find an expression for the sum to infinity of T (assuming that it exists). By expressing T in partial fractions, or otherwise, find an expression for the sum of the first n+1 terms of T.
Hint
(i) p=−21(1+px)S=31x with all other terms cancelling and so S=31x/(1−21x)=3(2−x)2x
(ii) Using similar working to part (i)
18+8p+2q=037+18p+8q=0
so p=−25,q=1
and so (1+px+qx2)T=2+3x
giving T=(2+3x)/(1−25x+x2)=2−5x+2x24+6x=(2−x)(1−2x)4+6x
By partial fractions T=314(1−2x)−1−38(1−2x)−1
and so Tn+1=314(1+2x+(2x)2+...+(2x)n)−38(1+2x+(2x)2+...+(2x)n)=3141−2x(1−(2x)n+1)−381−2x(1−(2x)n+1)
Model Solution
Part (i)
The series is S=31x+61x2+121x3+⋯, so a1=31, a2=61, a3=121.
Checking: a1a2=1/31/6=21 and a2a3=1/61/12=21.
So ar+1=21ar, giving p=−21.
Sum to infinity: Consider (1−21x)S:
(1−21x)S=31x+(61−61)x2+(121−121)x3+⋯=31x
since each coefficient of xr for r≥2 is ar−21ar−1=0. Therefore:
S=1−21x31x=3(2−x)2x
Sum of first n+1 terms: The partial sum Sn+1=∑r=1n+1arxr where ar=3⋅2r−11.
Three fifths attempted this with most scoring about two thirds of the marks. Apart from minor errors, the last part (expressing T in partial fractions etc.) was the pitfall for most.