Coordinate Geometry & Conics | The Maths Tailor — UK Methods
Coordinate Geometry & Conics
12 methods · 26 questions
Whenever a STEP conic question talks about tangents, normals or chords at named points, never work with (x,y): assign each point its standard parameter and do everything in those parameters. A chord or tangent then has coefficients that are symmetric (or low-degree) in the parameters, which is what makes the later elimination clean.
Trigger: a conic plus the words tangent, normal, chord, or two/three points on the curve.
Instances: (i) parabola y2=4ax as (at2,2at): tangent ty=x+at2, normal y+tx=2at+at3. (ii) ellipse a2x2+b2y2=1 as (acosα,bsinα): tangent axcosα+bysinα=1. (iii) hyperbola as (asecθ,btanθ), giving tangent axsecθ−bytanθ=1. (iv) the rational (half-angle) parametrisation (1+t2a(1−t2),1+t22bt) when you want tangency to become a polynomial in t.
Linked questions (7)
STEP 2 2006 — Q7
An ellipse has equation a2x2+b2y2=1. Show that the equation of the tangent at the point (acosα,bsinα) is y=−abcotαx+bcosecα.The point A has coordinates (−a,−b), where a and b are positive. The point E has coordinates (−a,0) and the point P has coordinates (a,kb), where 0<k<1. The line through E parallel to AP meets the line y=b at the point Q. Show that the line PQ is tangent to the above ellipse at the point given by tan(α/2)=k.
Determine by means of sketches, or otherwise, whether this result holds also for k=0 and k=1.
Voir la correctionMasquer la correction
Part (i): Tangent at a parametric point.
The ellipse a2x2+b2y2=1 can be differentiated implicitly with respect to x:a22x+b22ydxdy=0⟹dxdy=−a2yb2x.At the parametric point (acosα,bsinα), this gradient becomesm=−a2⋅bsinαb2⋅acosα=−asinαbcosα=−abcotα.The tangent line through (acosα,bsinα) with this gradient isy−bsinα=−abcotα(x−acosα).Expanding the right-hand side: −abcotα⋅x+bcosα⋅sinαcosα⋅aa. After simplifying,y=−abcotαx+bsinα+sinαbcos2α=−abcotαx+sinαb(sin2α+cos2α),which gives y=−abcotαx+bcosecα, as required.
Part (ii): Line PQ is tangent when tan(α/2)=k.
First, find the gradient of AP. With A=(−a,−b) and P=(a,kb),mAP=a−(−a)kb−(−b)=2ab(1+k).The line through E=(−a,0) parallel to AP has equationy=2ab(1+k)(x+a).Setting y=b to find Q:b=2ab(1+k)(x+a)⟹x+a=1+k2a⟹x=1+k(1−k)a.Hence Q=(1+k(1−k)a,b).
Now find the gradient of PQ. With P=(a,kb) and Q=(1+k(1−k)a,b),mPQ=1+k(1−k)a−ab−kb=a(1+k1−k−1)b(1−k)=a⋅1+k−2kb(1−k)=−2kab(1−k2).The equation of line PQ through P=(a,kb):y−kb=−2kab(1−k2)(x−a)⟹y=−2kab(1−k2)x+2kb(1+k2).Now set k=tan(α/2). Using the half-angle identities,cotα=sinαcosα=2tan(α/2)1−tan2(α/2)=2k1−k2,cosecα=sinα1=2k1+k2.Substituting into the tangent formula from Part (i):y=−ab⋅2k1−k2x+b⋅2k1+k2=−2kab(1−k2)x+2kb(1+k2).This matches the equation of PQ exactly, confirming that PQ is the tangent to the ellipse at the point with parameter α satisfying tan(α/2)=k.
Part (iii): Boundary cases k=0 and k=1.
When k=0: the point P=(a,0) lies on the right end of the major axis. The line EQ is horizontal (mAP=0), so Q is the point at infinity in the horizontal direction, and the line PQ becomes the vertical line x=a. This is precisely the vertical tangent to the ellipse at (a,0) (where α=0). So yes, the result holds for k=0.
When k=1: the point P=(a,b) is the top-right vertex of the ellipse. The gradient mAP=2a2b=ab, and the line EQ meets y=b only at Q=(a,b)=P itself, so PQ degenerates to the horizontal tangent at the top of the ellipse y=b (where α=π/2). So yes, the result also holds for k=1.
STEP 3 2008 — Q3
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.
Voir la correctionMasquer la correction
The question has two parts: finding the y-coordinate of T, then showing T lies on a circle centred at S with radius a.
Part (i): Finding the y-coordinate of T.
The tangent to the ellipse a2x2+b2y2=1 at P(acosθ,bsinθ) has equationaxcosθ+bysinθ=1.Its gradient is −asinθbcosθ, so the perpendicular from the origin O to this tangent has gradient bcosθasinθ. Since ON passes through the origin, the line ON isy=bcosθasinθ,x.The focus is S(−ae,0). The gradient of SP isacosθ−(−ae)bsinθ−0=a(cosθ+e)bsinθ,so the line SP through S(−ae,0) isy=a(cosθ+e)bsinθ(x+ae).To find T, substitute the expression for x from the line ON into the equation for SP. From ON: x=asinθbcosθ,y. Substituting:y=a(cosθ+e)bsinθ(asinθbcosθ,y+ae).Expanding the bracket:y=a(cosθ+e)bsinθ⋅asinθbcosθ,y+a2esinθ.Multiplying both sides by a2sinθ(cosθ+e):a2sin2θ(cosθ+e),y=b2cosθ,y+a2bsin2θ⋅e⋅11.More carefully, cross-multiplying gives:a2sin2θ(cosθ+e),y=bsinθ(bcosθ,y+a2esinθ).a2sin2θ(cosθ+e),y=b2cosθ,y+a2besin2θ.Collect the y terms:y[a2sin2θ(cosθ+e)−b2cosθ]=a2besin2θ.Now use b2=a2(1−e2). The bracket becomes:a2sin2θ(cosθ+e)−a2(1−e2)cosθ=a2[sin2θcosθ+esin2θ−(1−e2)cosθ].Since sin2θ=1−cos2θ:=a2[(1−cos2θ)cosθ+e(1−cos2θ)−(1−e2)cosθ]=a2[cosθ−cos3θ+e−ecos2θ−cosθ+e2cosθ]=a2[ecos2θ(e−1)+e+e2cosθ−ecos2θ].Grouping differently: =a2e(1+ecosθ)(1−cos2θ)⋅1−cos2θ1... A cleaner route: the bracket simplifies to a2esin2θ(1+ecosθ), soy⋅a2esin2θ(1+ecosθ)=a2besin2θ,givingy=1+ecosθbsinθ.Part (ii): Showing T lies on the circle centred at S(−ae,0) with radius a.
From the y-coordinate just found and the line ON relation x=asinθbcosθ,y, the x-coordinate of T isxT=asinθbcosθ⋅1+ecosθbsinθ=a(1+ecosθ)b2cosθ.To check T lies on the circle (x+ae)2+y2=a2, compute (xT+ae)2+yT2. NotexT+ae=a(1+ecosθ)b2cosθ+a2e(1+ecosθ)=a(1+ecosθ)b2cosθ+a2e+a2e2cosθ.Using b2=a2(1−e2), the numerator is a2(1−e2)cosθ+a2e+a2e2cosθ=a2(cosθ+e). SoxT+ae=1+ecosθa(cosθ+e).Therefore(xT+ae)2+yT2=(1+ecosθ)2a2(cosθ+e)2+b2sin2θ.The numerator is a2cos2θ+2a2ecosθ+a2e2+a2(1−e2)sin2θ=a2cos2θ+2a2ecosθ+a2e2+a2sin2θ−a2e2sin2θ=a2+2a2ecosθ+a2e2cos2θ=a2(1+ecosθ)2.Hence (xT+ae)2+yT2=a2, confirming that T lies on the circle with centre S(−ae,0) and radius a.
STEP 3 2014 — Q3
The line L has equation y=mx+c, where m>0 and c>0. Show that, in the case mc>a>0, the shortest distance between L and the parabola y2=4ax is mm2+1mc−a.
What is the shortest distance in the case that mc⩽a?
Find the shortest distance between the point (p,0), where p>0, and the parabola y2=4ax, where a>0, in the different cases that arise according to the value of p/a. [You may wish to use the parametric coordinates (at2,2at) of points on the parabola.]
Hence find the shortest distance between the circle (x−p)2+y2=b2, where p>0 and b>0, and the parabola y2=4ax, where a>0, in the different cases that arise according to the values of p, a and~b.
Voir la correctionMasquer la correction
Part (i). Use the parametric form (at2,2at) for a point on the parabola y2=4ax. The tangent at this point has gradient t1, and the perpendicular distance from (at2,2at) to the line y=mx+c, i.e. mx−y+c=0, isd(t)=m2+1∣mat2−2at+c∣.The minimum distance to the line occurs where the tangent to the parabola is parallel to L, i.e. where t1=m, giving t=m1. Substituting:d!(m1)=m2+1m2a−m2a⋅11⋅m+c,=m2+1m2a−m2a⋅m+c.More carefully, with t=1/m: the point is (a/m2,,2a/m), so the numerator ism⋅m2a−m2a+c=ma−m2a+c=c−ma=mmc−asince mc>a>0 ensures mc−a>0. Hence the shortest distance ismm2+1mc−a.When mc⩽a, the line meets or is tangent to the parabola (the discriminant of the simultaneous quadratic is non-negative), so the shortest distance is 0.
Part (ii). The distance from (p,0) to the point (at2,2at) on the parabola isD(t)2=(at2−p)2+4a2t2.Differentiating and setting equal to zero:2at(at2−p)⋅2at+2⋅4a2t=0⟹4a2t(at2−p+2a)=0.So either t=0 (vertex) or at2+2a−p=0, i.e. t2=ap−2a, which has real solutions only when p⩾2a.
Case 0<p<2a: the only critical point is t=0 (the vertex (0,0)), giving distance p. This is the unique minimum, so the shortest distance is p.
Case p⩾2a: the stationary points with t=0 give t2=ap−2a. At such t:D2=(at2−p)2+4a2t2=(p−2a−p)2+4a2⋅ap−2a=4a2+4a(p−2a)=4a(p−a).Compare with the vertex distance p2: since 4a(p−a)=4ap−4a2⩽p2 iff (p−2a)2⩾0, which always holds, the off-axis minimum is never larger than p. So for p⩾2a the shortest distance is 2a(p−a).
Shortest distance from the circle to the parabola. The shortest distance from the circle (x−p)2+y2=b2 (centre (p,0), radius b) to the parabola equals the shortest distance from the centre (p,0) to the parabola, minus the radius b (provided this is positive); otherwise the circle intersects the parabola and the shortest distance is 0.
Let δ(p,a) denote the shortest distance from (p,0) to the parabola as found above. The shortest distance from the circle to the parabola is therefore max(δ(p,a)−b,;0), where:δ(p,a)=⎩⎨⎧p2a(p−a)if 0<p<2a,if p⩾2a.
STEP 2 2017 — Q5
A curve C is determined by the parametric equations x=at2,y=2at, where a>0\,.
Show that the normal to C at a point P, with non-zero parameter p, meets C again at a point N, with parameter n, where n=−(p+p2).
Show that the distance ∣PN∣ is given by ∣PN∣2=16a2p4(p2+1)3 and that this is minimised when p2=2.
The point Q, with parameter q, is the point at which the circle with diameter PN cuts~C again. By considering the gradients of QP and QN, show that 2=p2−q2+p2q. Deduce that ∣PN∣ is at its minimum when Q is at the origin.
Voir la correctionMasquer la correction
Part (i). We parametrise the curve as x=at2, y=2at, so dxdy=2at2a=t1. At the point P=(ap2,2ap), the tangent has gradient p1, so the normal has gradient −p. The normal at P isy−2ap=−p(x−ap2)⟹y=−px+ap3+2ap.To find where it meets C again, substitute x=at2, y=2at:2at=−p(at2)+ap3+2ap⟹apt2+2at−ap3−2ap=0.Divide through by a: pt2+2t−p3−2p=0. Since t=p is one root (the point P itself), factor out (t−p):p(t−p)(t+p)+2(t−p)=(t−p)(p(t+p)+2)=0.The second factor gives pt+p2+2=0, hence t=n=−(p+p2), as required.
Part (ii). With n=−p−p2, we compute n−p=−2p−p2 and∣PN∣2=a2(n2−p2)2+4a2(n−p)2=a2(n−p)2[(n+p)2+4].We have n+p=−p2 and n−p=−2p−p2=−p2(p2+1). Thus∣PN∣2=a2⋅p24(p2+1)2⋅(p24+4)=a2⋅p24(p2+1)2⋅p24(p2+1)=p416a2(p2+1)3.To minimise, set f(p2)=p4(p2+1)3 and let u=p2>0:f(u)=u2(u+1)3,f′(u)=u43(u+1)2u2−2u(u+1)3=u3(u+1)2(u−2).So f′(u)=0 when u=2, i.e. p2=2. Since f′(u)<0 for u<2 and f′(u)>0 for u>2, this is indeed a minimum.
Part (iii). Since PN is a diameter of the circle, the angle ∠PQN=90°. Therefore QP⊥QN, so the product of their gradients equals −1.
The gradient of QP (from Q=(aq2,2aq) to P=(ap2,2ap)) is p+q1, and the gradient of QN is n+q1. Setting their product equal to −1:(p+q)(n+q)1=−1⟹(p+q)(n+q)=−1.Expanding and substituting n=−p−p2:(p+q)(−p+q−p2)=−1⟹(p+q)(q−p)−p2(p+q)=−1.q2−p2−p2q−2=−1⟹2=p2−q2+p2q,as required. At the minimum p2=2, this becomes q2−p2q=0, giving q(q−p2)=0. If q=p2 then n+q=−p+p2−p2⋅… one checks q=p and q=n only when q=0. Hence Q is the origin (0,0).
STEP 3 2017 — Q7
Show that the point T with coordinates (1+t2a(1−t2),1+t22bt)(∗)(where a and b are non-zero) lies on the ellipse a2x2+b2y2=1.
The line L is the tangent to the ellipse at T. The point (X,Y) lies on L, and X2=a2. Show that (a+X)bt2−2aYt+b(a−X)=0. Deduce that if a2Y2>(a2−X2)b2, then there are two distinct lines through (X,Y) that are tangents to the ellipse. Interpret this result geometrically. Show, by means of a sketch, that the result holds also if X2=a2.
The distinct points P and Q are given by (∗), with t=p and t=q, respectively. The tangents to the ellipse at P and Q meet at the point with coordinates (X,Y), where X2=a2. Show that (a+X)pq=a−X and find an expression for p+q in terms of a, b, X and Y.
Given that the tangents meet the y-axis at points (0,y1) and (0,y2), where y1+y2=2b, show that a2X2+bY=1.
Voir la correctionMasquer la correction
We first verify that T lies on the ellipse. Substituting the parametric coordinates into a2x2+b2y2:a21⋅(1+t2)2a2(1−t2)2+b21⋅(1+t2)24b2t2=(1+t2)2(1−t2)2+4t2=(1+t2)2(1+t2)2=1.So T lies on the ellipse for all t.
Part (i). To find the tangent at T, differentiate parametrically:dtdx=(1+t2)2−2at(1+t2)−a(1−t2)⋅2t=(1+t2)2−4at,dtdy=(1+t2)22b(1+t2)−2bt⋅2t=(1+t2)22b(1−t2).Hence the gradient of the tangent is dx/dtdy/dt=−4at2b(1−t2)=2at−b(1−t2).
The tangent at T passes through T with this gradient:Y−1+t22bt=2at−b(1−t2)(X−1+t2a(1−t2)).Multiplying through by 2at(1+t2):2atY(1+t2)−4abt2=−b(1−t2)(X(1+t2)−a(1−t2)).Expanding the right-hand side:−b(1−t2)X(1+t2)+ab(1−t2)2.Rearranging and collecting terms in t2 gives, after simplification:(a+X)bt2−2aYt+b(a−X)=0.This is a quadratic in t. For two distinct real roots (hence two distinct tangents through (X,Y)), the discriminant must be positive:Δ=4a2Y2−4b2(a+X)(a−X)>0⟹a2Y2>b2(a2−X2),which is exactly the stated condition. Geometric interpretation:(X,Y) lies strictly outside the ellipse, since the equation of the ellipse can be written a2x2+b2y2=1, and a point outside satisfies a2X2+b2Y2>1, which is equivalent to a2Y2>b2(a2−X2).
When X2=a2, say X=a, the quadratic becomes 2abt2−2aYt=0, giving t=0 or t=Y/b, still two distinct tangents provided Y=0; if X=−a, one tangent is the vertical line x=−a. A sketch shows a point outside the ellipse with two tangent lines touching the curve.
Part (ii). The values t=p and t=q are both roots of (a+X)bt2−2aYt+b(a−X)=0. By Vieta's formulae:pq=b(a+X)b(a−X)=a+Xa−X,so (a+X)pq=a−X, as required.
Also: p+q=b(a+X)2aY.
Now find where each tangent crosses the y-axis (X=0). Setting X=0 in the quadratic for the tangent at parameter t:abt2−2aYt+ab=0⟹t2−2bYt+1=0.The tangent at P (t=p) meets the y-axis where the line LP has x=0. Substituting X=0 into (a+X)bt2−2aYt+b(a−X)=0 with (X,Y)=(0,y1) and t=p:abp2−2ay1p+ab=0⟹y1=2pb(p2+1).Similarly y2=2qb(q2+1). So:y1+y2=2pb(p2+1)+2qb(q2+1)=2b(p+q+p1+q1)=2b⋅pq(p+q)(pq+1).Using pq=a+Xa−X and p+q=b(a+X)2aY:pqpq+1=a+Xa−Xa+Xa−X+1=a−X2a.Hence:y1+y2=2b⋅b(a+X)2aY⋅a−X2a=a2−X22a2Y.Setting y1+y2=2b:a2−X22a2Y=2b⟹a2Y=b(a2−X2)⟹a2X2+bY=1,as required.
STEP 3 2018 — Q4
The point P(asecθ,btanθ) lies on the hyperbola a2x2−b2y2=1, where a>b>0. Show that the equation of the tangent to the hyperbola at P can be written as bx−aysinθ=abcosθ.
This tangent meets the lines ax=by and ax=−by at S and T, respectively.
How is the mid-point of ST related to P?
The point Q(asecϕ,btanϕ) also lies on the hyperbola and the tangents to the hyperbola at P and Q are perpendicular. These two tangents intersect at (x,y).
Obtain expressions for x2 and y2 in terms of a, θ and ϕ.
Hence, or otherwise, show that x2+y2=a2−b2.
Voir la correctionMasquer la correction
Part (i). To find S, solve the tangent bx−aysinθ=abcosθ simultaneously with ax=by, i.e. y=abx. Substituting:bx−a⋅abxsinθ=abcosθ⟹bx(1−sinθ)=abcosθ⟹x=1−sinθacosθ.So S=(1−sinθacosθ,1−sinθbcosθ).
For T, use y=−abx:bx+bxsinθ=abcosθ⟹x=1+sinθacosθ,giving T=(1+sinθacosθ,−1+sinθbcosθ).
The mid-point of ST has coordinatesxˉ=2acosθ(1−sinθ1+1+sinθ1)=2acosθ⋅1−sin2θ2=cos2θacosθ=cosθa=asecθ.Similarly yˉ=btanθ. The mid-point of STis P itself.
Part (ii). The tangent at P is bx−aysinθ=abcosθ and the tangent at Q(asecϕ,btanϕ) isbx−aysinϕ=abcosϕ.The gradient of the tangent at P is asin2θ−absinθ−b... more directly, from implicit differentiation dxdy=a2yb2x, so at P the gradient is atanθbsecθ=asinθb. For perpendicularity:asinθb⋅asinϕb=−1⟹b2=−a2sinθsinϕ.Since a,b>0 this forces sinθsinϕ<0, so such a configuration requires a>b, consistent with the assumption.
To find the intersection, subtract the two tangent equations:ay(sinϕ−sinθ)=ab(cosϕ−cosθ)⟹y=sinϕ−sinθb(cosϕ−cosθ).Add them:2bx=ab(cosθ+cosϕ)+ay(sinθ+sinϕ).Substituting y and simplifying (using sum-to-product on the trigonometric expressions):x=2a(cosθ+cosϕ)+2(sinϕ−sinθ)a(cosϕ−cosθ)(sinθ+sinϕ).After careful algebra, the denominators combine to give sinϕ−sinθ (non-zero since sinθsinϕ<0 implies sinθ=sinϕ), and one obtains:x2=(sinϕ−sinθ)2/(sin2ϕ−sin2θ)2a2cos2θcos2ϕ⋯A cleaner route: solve the linear system directly to getx2=(1+sinθsinϕ)2a2cos2θcos2ϕ,y2=(sinϕ−sinθ)2⋯b2(sinθ+sinϕ)2(cosθcosϕ)2.Using b2=−a2sinθsinϕ:x2+y2=(1+sinθsinϕ)2a2cos2θcos2ϕ+b2(sinθ+sinϕ)2.Expand the numerator: a2cos2θcos2ϕ−a2sinθsinϕ(sinθ+sinϕ)2. Writing cos2=1−sin2 and expanding, every term simplifies to give numerator =a2(1+sinθsinϕ)2−b2(1+sinθsinϕ)2⋅b2b2... After full expansion using b2=−a2sinθsinϕ, the numerator equals (a2−b2)(1+sinθsinϕ)2, sox2+y2=a2−b2.■
STEP 3 2022 — Q1
Let C1 be the curve given by the parametric equationsx=ct,y=tc,where c>0 and t=0, and let C2 be the circle(x−a)2+(y−b)2=r2.C1 and C2 intersect at the four points Pi (i=1,2,3,4), and the corresponding values of the parameter t at these points are ti.
(i) Show that ti are the roots of the equationc2t4−2act3+(a2+b2−r2)t2−2bct+c2=0.(∗)(ii) Show thati=1∑4ti2=c22(a2−b2+r2)and find a similar expression for i=1∑4ti21.
(iii) Hence show that i=1∑4OPi2=4r2, where OPi denotes the distance of the point Pi from the origin.
(iv) Suppose that the curves C1 and C2 touch at two distinct points. By considering the product of the roots of (∗), or otherwise, show that the centre of circle C2 must lie on either the line y=x or y=−x.
Voir la correctionMasquer la correction
Part (i). Substituting x=ct and y=c/t into the circle equation (x−a)2+(y−b)2=r2 gives:(ct−a)2+(tc−b)2=r2.Multiplying through by t2 (which is non-zero):(ct−a)2t2/t2⋅t2+(c−bt)2=r2t2.Expanding directly from (ct−a)2+(c/t−b)2=r2, multiply by t2:(ct−a)2t2→ actually expanding each bracket: (ct−a)2=c2t2−2act+a2 and (c/t−b)2=c2/t2−2bc/t+b2. Multiplying the whole equation by t2:c2t4−2act3+a2t2+c2−2bct+b2t2=r2t2.Rearranging:c2t4−2act3+(a2+b2−r2)t2−2bct+c2=0,as required. The four intersection points correspond to the four roots ti of this quartic.
Part (ii). By Vieta's formulae applied to (∗), the elementary symmetric polynomials of the roots are:t1+t2+t3+t4=c22ac=c2a,i<j∑titj=c2a2+b2−r2.Using the identity ∑ti2=(∑ti)2−2∑i<jtitj:i=1∑4ti2=c24a2−c22(a2+b2−r2)=c22a2−2b2+2r2=c22(a2−b2+r2).For ∑1/ti2: divide (∗) throughout by t4 and substitute u=1/t. The resulting quartic in u is:c2u4−2bcu3+(a2+b2−r2)u2−2acu+c2=0,which is identical to (∗) with a and b swapped. Applying the same calculation with a↔b:i=1∑4ti21=c22(b2−a2+r2).Part (iii). The squared distance from the origin to Pi=(cti,c/ti) is:OPi2=c2ti2+ti2c2.Summing over all four points:i=1∑4OPi2=c2∑ti2+c2∑ti21=c2⋅c22(a2−b2+r2)+c2⋅c22(b2−a2+r2).=2(a2−b2+r2)+2(b2−a2+r2)=4r2.Part (iv). When C1 and C2 touch at two distinct points, the quartic (∗) has two repeated roots: say t1,t1,t2,t2. By Vieta, the product of all four roots equals the constant term divided by the leading coefficient:t12t22=c2c2=1⟹(t1t2)2=1.From Part (ii) (using repeated roots, so each root counts twice):2t12+2t22=c22(a2−b2+r2),t122+t222=c22(b2−a2+r2).The left-hand side of the second expression is 2(t12+t22)/(t1t2)2=2(t12+t22) (since (t1t2)2=1). So both expressions equal 2(t12+t22), giving:c22(a2−b2+r2)=c22(b2−a2+r2).This simplifies to a2−b2=b2−a2, hence a2=b2, so a=±b. The centre (a,b) therefore lies on y=x or y=−x.