1

Q) Let T be the line passing through the points P(–2, 7) and Q(2, –5). Let $F_{1}$ be the set of all pairs of circles $(S_{1}$, $S_{2}$) such that T is tangent to $S_{1}$ at P and tangent to $S_{2}$ at Q, and also such that $S_{1}$ and $S_{2}$ touch each other at a unique point, say, M. Let $E_{1}$ be the set representing the locus of M as the pair ($S_{1}$, $S_{2}$) varies in $F_{1}$. Let the set of all straight line segments joining a pair of distinct points of $E_{1}$ and passing through the point R(1, 1) be $F_{2}$. Let $E_{2}$ be the set of the mid-points of the line segments in the set $F_{2}$. Let $C$ be the circle $x^2+y^2+6(2y+7x)=53$. The number of times $C$ intersects $E_{1}$ and $E_{2}$ is (are):

Sid
  • 1,234

3 Answers3

1

As the diameter of the first is $PQ$ $$E_1=\{(x,y)| (x+2)(x-2)+(y-7)(y+5)=0\}-\{P,Q\}$$

and the diameter of the second is $DR$, with $D$ the midpoint of $PQ$ $$E_2=\{(x,y)| x(x-1)+(y-1)^2=0\}-\{(4/5,7/5),(36/37,43/37)\}$$

where the disallowed points stem from the intersections with $l_{PR}$ and $l_{QR}$.

Now

$$\#((V((x+2)(x-2)+(y-7)(y+5))\cup V(x(x-1)+(y-1)^2))\cap C)=4$$

but three points are disallowed, leaving the other intersection $(400/409,349/409)$ so

$$\#((E_1\cup E_2)\cap C)=1$$

E1cupE2capC

Edit

$E_2$ is part of a circle: let $y-1=m(x-1)$ be lines through $R$, then together with the equation for the circle $E_1$ is a part of, we get the relation $(m^2+1)y^2+(-2m^2+2m-2)y-38m^2-2m+1=0$ that gives us solutions $y_{1,2}$ with corresponding $x_{1,2}$. What we're looking for are points $(\frac{x_1+x_2}{2},\frac{y_1+y_2}{2})=(\frac{1}{m^2+1},\frac{m^2-m+1}{m^2+1})$ which implicitizes to $x^2-x+y^2-2y+1=0.$

1

$E_{1}:$

Let $(C_{1},C_{2})$ be the centers of $(S_{1},S_{2})$ respectively. Notice $\overline{PC_{1}} \perp \overline{PQ}$ and $\overline{QC_{2}} \perp \overline{PQ}$.

$M(h_{1},k_{1})$ is the unique meeting point of circles $(S_{1},S_{2})$ where $M,C_{1}$ and $C_{2}$ are co-linear.

Notice: $\triangle PC_{1}M$ and $\triangle QC_{2}M$ are isosceles triangles.

Let $\angle C_{1}PM = \angle C_{1}MP = \phi \implies \angle PC_{1}M = \pi-2\phi$

Let $\angle C_{2}QM = \angle C_{2}MQ = \theta \implies \angle QC_{2}M = \pi-2\theta$

Due to perpendicularity: $\angle QPM + \angle MPC_{1} = \frac{\pi}{2} \implies \angle QPM= \frac{\pi}{2} - \phi$

Similarly: $\angle PQM + \angle MQC_{2} = \frac{\pi}{2} \implies \angle PQM= \frac{\pi}{2} - \theta$

All four vertices shapes have property:

$\angle PC_{1}C_{2}+ \angle C_{1}C_{2}Q + \angle C_{2}QP + \angle QPC_{1} = 360^{\circ}$ $\iff (\pi - 2\phi) + (\pi - 2\theta)+ \frac{\pi}{2}+\frac{\pi}{2} = 2\pi$ $\implies \phi + \theta = \frac{\pi}{2}$

Consider: $\triangle PMQ \implies \angle PMQ +\angle MPQ+ \angle MQP = 180^{\circ}$ $\implies \angle PMQ = 180^{\circ}- (90^{\circ} - \phi) - (90^{\circ}- \theta) \implies \angle PMQ = \frac{\pi}{2}$

$$\text{Fig_1: One instance of set: $F_{1}$:}$$

F_{1}

Notice: $\angle PMQ=\frac{\pi}{2}$ i.e: a right angle.

Hence $$m(\overline{MP})*m(\overline{MQ)}=-1 \iff (\frac{h_{1}+2}{k_{1}-7})(\frac{h_{1}-2}{k_{1}+5})=-1$$

Where small 'm' denotes the slope of a line segment.

Consider G (as of now it seems the locus of M is G) $$G= \{(x,y) \in \mathbb{R^2}:(x-2)(x+2) + (y-7)(y+5) = 0\}$$

However, one fact (that normally goes unnoticed) is one can not include points P and Q in the locus M as if this does occur points M and P (or M and Q) coincide and the radius of $S_{1}$ (or $S_{2}$) tends to zero and radius of $S_{2}$ (or $S_{1}$) tends to infinity making $S_{1}$ (or $S_{2}$) a point and $S_{2}$ (or $S_{1}$) a straight line. Along with this a tangent to a circle can only meet that circle at one point $S_{2}$ (or $S_{1}$) . If P or Q is included in the locus this property is also violated.

Hence: $$E_{1}= \{(x,y) \in \mathbb{R^2}:(x-2)(x+2) + (y-7)(y+5) = 0, (x,y)\notin \{(-2,7),(2,-5)\}\}$$

Note: $E_1$ or $M$ are not circles as circles are defined as the set of all points equidistant from a point.


$E_{2}:$

Let $A:(0,1)$ be the center of $G$ ($G$ is $E_{1}$ without excluded points, see above).

Let $B(h_{2},k_{2})$ be the midpoint of a line segment joining a pair of distinct points of $E_{1}$ and passing through $R$.

Notice: the line segment drawn from the center of any circle to the midpoint of a chord of that circle is perpendicular to the chord. $$\text{Fig_2: One instance of set: $F_{2}$:}$$

F_2 Note: $\angle ABR=\frac{\pi}{2}$ i.e: a right angle.

Hence: $$m(\overline{AB})m(\overline{BR)}=-1 \iff(\frac{h_{2}}{k_{2}-1})(\frac{h_{2}-1}{k_{2}-1})=-1$$

Consider H (as of now it seems the locus of midpoints is H) $$H= \{(x,y) \in \mathbb{R^2}:(x)(x-1) + (y-1)^2= 0\}$$

However, one fact (that normally goes unnoticed) is that since $P$ and $Q$ is not included in the set $E_{1}$ one can not draw a chord passing through $P$ and $R$ and the midpoint found after joining the line PR (and QR) is not valid so must be excluded from the locus $H$. Again: if one does draw a line segment passing through P and R it will pass through only a single point in $E_{1}$ but the question clearly states that: $F_{2}$ = set of all straight line segments joining a pair of distinct points.

Now the task is to find these points to exclude. We must find the midpoint of the line segment generated by joining $P$ and $R$ and equate it to $H$. This midpoint must be removed. Same must be done for $Q$ and $R$.

The line P(-2,7) and R(1,1) is: $y+2x=3$. Substituting this line in $H$ and finding the value will give the midpoint generated by $PR$.

$(x)(x-1) + (2-2x)^2= 0 \implies x=\frac{4}{5},1 \implies (x,y) = (\frac{4}{5},\frac{7}{5}), (1,1)$

However R(1,1) is not the midpoint as $P$, $A$ and $R$ are not co-linear.

The line Q(2,-5) and R(1,1) is: $y+6x=7$. Substituting this line in $H$ and finding the value will give the midpoint generated by $QR$.

$(x)(x-1) + (6-6x)^2= 0 \implies x=\frac{36}{37},1 \implies (x,y) = (\frac{36}{37},\frac{43}{37}), (1,1)$

However R(1,1) is not the midpoint as $Q$, $A$ and $R$ are not co-linear.

Now upon excluding these points from $H$ we get: $$E_{2}= \{(x,y) \in \mathbb{R^2}:x(x-1)+(y-1)^2=0, (x,y)\notin \{(\frac{4}{5},\frac{7}{5}), (\frac{36}{37},\frac{43}{37}\}\}$$


SETS:

$E_{1}= \{(x,y) \in \mathbb{R^2}:(x-2)(x+2) + (y-7)(y+5) = 0, (x,y)\notin \{(-2,7),(2,-5)\}\}$

$E_{2}= \{(x,y) \in \mathbb{R^2}:x(x-1)+(y-1)^2=0, (x,y)\notin \{(\frac{4}{5},\frac{7}{5}), (\frac{36}{37},\frac{43}{37}\}\}$

$C= \{(x,y) \in \mathbb{R^2}:(x+21)^2+(y+6)^2=530\}$


INTERSECTIONS:

$$\text{$C$ and $E_{1}$:}$$

Upon expanding:

$E_{1}: x^2+y^2 = 2y + 39 - \{(-2,7),(2,-5)\}$

$C: x^2 +y^2 = 53 - 12y - 42x $

Solving system of equations:

$C-E_{1}\iff 0 = 14 - 14y -42x \iff y+3x=1 \iff L_{1}$

Upon putting $L_{1}$ in $C$ we get:

$x^2 +(1-3x)^2 = 53 - 12(1-3x) - 42x \implies x = \{-2,2\}$

After finding the corresponding values of $y$ we get:

$(x,y) = \{(-2,7),(2,-5)\}$

However these points are excluded from $E_{1}$, hence no valid intersections.

One should notice $L_{1} = T$.

$$\text{$C$ and $E_{2}$:}$$

Upon expanding:

$E_{1}: x^2+y^2 = 2y + x - 1 - \{(-2,7),(2,-5)\}$

$C: x^2 +y^2 = 53 - 12y - 42x $

Solving system of equations:

$C-E_{2}\iff 0 = 54 - 14y -43x \iff L_{2}$

Upon putting $L_{2}$ in $C$ we get:

$(x,y) = \{(\frac{4}{5},\frac{7}{5}), (\frac{400}{409},\frac{349}{409}\}$

The point $(\frac{4}{5},\frac{7}{5})$ is excluded from $E_{2}$ so this is not a valid intersection.

Hence number of intersections points is: ONE i.e:

$$(\frac{400}{409},\frac{349}{409})$$


Sid
  • 1,234
  • $(\frac{x+2}{y-7})(\frac{x-2}{y+5})=-1$ and $(\frac{x}{y-1})(\frac{x-1}{y-1})=-1$ (not $0$) – Jan-Magnus Økland Jun 28 '19 at 08:29
  • Hi, Sid , I have noticed that you are doing JEE advanced problems since 2017, and you are still posting JEE related problems, so will you be appearing for JEE this year or you have given it already? – V.G Mar 11 '21 at 15:42
-1

Let the centers of $S_1$ and $S_2$ be $O_1$ and $O_2$ respectively. Note that $O_1P\perp PQ$ and also $O_2Q\perp PQ$.

Since $O_1M=O_1P$ and $O_2M=O_2Q$ then both $O_1PM$ and $O_2QM$ are isosceles triangles and $$\begin{align}\angle MPO_1&=\angle PMO_1\\ \angle MQO_2&=\angle QMO_2\end{align}$$ This implies $$\angle PMQ=\angle MPQ+\angle QPM$$ which means that $\angle PMQ$ is a right angle. So $M$ lies on a circle with diameter $PQ$ (that is called the set $E_1$).

The center of $E_1$ is located at $D(0,1)$ and its radius is $2\sqrt{10}$: $$E_1=\{(x,y):x^2+(y-1)^2=40\}$$

Now imagine the set of all the line segments passing through $R(1,1)$ and crossing $E_1$ at two points $A$ and $B$, and note that the point $R$ is inside $E_1$ and $AB$ is in fact a chord of the $E_1$ circle.

Can you show that the loci of the midpoints of all these chords is a circle whose diameter is $RD$?

polfosol
  • 9,245
  • The locus you are claiming (E1) involves the point P and Q in it. If M coincides with P or Q then then the radius of one circle becomes zero which means it is a point no a circle... Also if M coincides with P or Q then it would mean that the tangent of a circle meets that circle at two points... whereas a tangent to a circle can only meet a circle at a single point. – Sid Jun 24 '19 at 23:09
  • @Sid aside from the fact that the locus excludes those points and we normally don't mention it, which other part of the answer you didn't understand? – polfosol Jun 25 '19 at 02:35
  • @Sid: Consider this image. As the diameter of the first circle is $PQ$ $$E_1={(x,y)| (x+2)(x-2)+(y-7)(y+5)=0}-{P,Q}$$

    and the diameter of the second circle is $DR$, $$E_2={(x,y)| x(x-1)+(y-1)^2=0}-{(4/5,7/5),(36/37,43/37)}$$

    where the disallowed points stem from the intersections with $l_{PR}$ and $l_{QR}$.

    Now

    $$#((V((x+2)(x-2)+(y-7)(y+5))\cup V(x(x-1)+(y-1)^2))\cap C)=4$$

    but three points are disallowed, leaving the other intersection $(400/409,349/409)$ so

    $$#((E_1\cup E_2)\cap C)=1.$$

    – Jan-Magnus Økland Jun 25 '19 at 06:37
  • 1
    @Sid As far as I recall, in your original question you said you are having trouble in finding the loci of $M$. So I thought once you learn about that, you can proceed with the rest. I didn't think the excluded points are such a big deal – polfosol Jun 25 '19 at 07:20
  • 1
    You really have a way of making people regret helping you. – polfosol Jun 25 '19 at 07:24
  • @Sid no problem. I misunderstood your intention. By the way, if Jan-Magnus doesn't post an answer, I'll edit mine to the proper form (if that's okay with you). – polfosol Jun 25 '19 at 07:43