HMMT February 2019: Geometry

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

HMMT February 2019

February 16, 2019


Geometry
1. Let d be a real number such that every non-degenerate quadrilateral has at least two interior angles
with measure less than d degrees. What is the minimum possible value for d?
Proposed by: James Lin
Answer: 120
The sum of the internal angles of a quadrilateral triangle is 360◦ . To find the minimum d, we note
the limiting case where three of the angles have measure d and the remaining angle has measure
approaching zero. Hence, d ≥ 360◦ /3 = 120. It is not difficult to see that for any 0 < α < 120, a
quadrilateral of which three angles have measure α degrees and fourth angle has measure (360 − 3α)
degrees can be constructed.
2. In rectangle ABCD, points E and F lie on sides AB and CD respectively such that both AF and CE
are perpendicular to diagonal BD. Given that BF and DE separate ABCD into three polygons with
equal area, and that EF = 1, find the length of BD.
Proposed by: Yuan Yao

Answer: 3
Observe that AECF is a parallelogram. The equal area condition gives that BE = DF = 31 AB.

Let CE ∩ BD = X, then EX BE 1 2 2
CX = CD = 3 , so that BX = EX · CX = 3EX ⇒ BX = 3EX ⇒
∠EBX = 30◦ . Now, CE = 2BE = CF , so CEF is an equilateral triangle and CD = 23 CF = 23 .

Hence, BD = √23 · 32 = 3.

3. Let AB be a line segment with length 2, and S be the set of points P on the plane such that there
exists point X on segment AB with AX = 2P X. Find the area of S.
Proposed by: Yuan Yao

Answer: 3 + 2π
3

Observe that for any X on segment AB, the locus of all points P such that AX = 2P X is a circle
centered at X with radius 12 AX. Note that the point P on this circle where P A forms the largest angle
with AB is where P A is tangent to the circle at P , such that ∠P AB = arcsin(1/2) = 30◦ . Therefore,
if we let Q and Q0 be the tangent points of the tangents from A to the circle centered at B (call it
ω) with radius 12 AB, we have that S comprises the two 30-60-90 triangles AQB andAQ0 B, each with

area 21 3 and the 240◦ sector of ω bounded by BQ and BQ0 with area 23 π. Therefore the total area

is 3 + 2π 3 .

4. Convex hexagon ABCDEF is drawn in the plane such that ACDF and ABDE are parallelograms
with area 168. AC and BD intersect at G. Given that the area of AGB is 10 more than the area of
CGB, find the smallest possible area of hexagon ABCDEF .
Proposed by: Andrew Lin
Answer: 196
Since ACDF and ABDE have area 168, triangles ABD and ACD (which are each half a parallelogram)
both have area 84. Thus, B and C are the same height away from AD, and since ABCDEF is convex,
B and C are on the same side of AD. Thus, BC is parallel to AD, and ABCD is a trapezoid.
In particular, we have that the area of ABG equals the area of CDG. Letting this quantity be
x, we have that the area of BCG is x − 10, and the area of ADG is 84 − x. Then notice that
[ABG] AG [ADG] x 84−x 2
[CBG] = GC = [CDG] . This means that x−10 = x . Simplifying, we have x − 47x + 420 = 0; this
has solutions x = 12 and x = 35. The area of ABCDEF is twice the area of trapezoid ABCD, or
2[x + (x − 10) + (84 − x) + x] = 4x + 148; choosing x = 12, we get that the smallest possible area is
48 + 148 = 196.
5. Isosceles triangle ABC with AB = AC is inscribed
√ in a unit circle Ω with center O. Point D is the
reflection of C across AB. Given that DO = 3, find the area of triangle ABC.
Proposed by: Lillian Zhang
√ √
2+1 2−1
Answer: 2 OR 2

Solution 1. Observe that


1 1 1
∠DBO = ∠DBA + ∠ABO = ∠CBA + ∠BAO = (∠CBA + ∠BCA) + (∠BAC) = (180◦ ) = 90◦ .
2 2 2

Thus BC = BD =√ 2 by the Pythagorean Theorem on 4DBO. Then ∠BOC = 90◦ , and the distance
from O to BC is 22 . Depending on whether A is on the same side of BC as O, the height from A to
√ √ √ √ √
BC is either 1 + 22 or 1 − 22 , so the area is ( 2 · (1 ± 22 ))/2 = 2±1
2 .
Solution 2. One can observe that ∠DBA = ∠CBA = ∠ACB by property of reflection and ABC
being isosceles,
√ hence DB
√ is tangent to Ω and Power of a Point (and reflection property) gives BC =
BD = OD2 − OB 2 = 2. Proceed as in Solution 1.
Note. It was intended, but not specified in the problem statement that triangle ABC is acute, so we
accepted either of the two possible answers.
6. Six unit disks C1 , C2 , C3 , C4 , C5 , C6 are in the plane such that they don’t intersect each other and Ci
is tangent to Ci+1 for 1 ≤ i ≤ 6 (where C7 = C1 ). Let C be the smallest circle that contains all six
disks. Let r be the smallest possible radius of C, and R the largest possible radius. Find R − r.
Proposed by: Daniel Liu

Answer: 3−1
The minimal configuration occurs when the six circles are placed with their centers at the vertices of
a regular hexagon of side length 2. This gives a radius of 3.
The maximal configuration occurs when four of the circles are placed at the vertices of a square of side
length 2. Letting these circles be C1 , C3 , C4 , C6 in order, we place the last two so that C2 is tangent to
C1 and C3 and C5 is tangent to C4 and C6 . (Imagine pulling apart the √ last two circles on the
√ plane;
this is the configuration you end up with.) The resulting radius is 2 + 3, so the answer is 3 − 1.
Now we present the proofs for these configurations being optimal. First, we rephrase the problem: given
an equilateral hexagon of side length 2, let r be the minimum radius of a circle completely containing
the vertices of the hexagon. Find the difference between the minimum and maximum values in r.
(Technically this r is off by one from the actual problem, but since we want R − r in the actual
problem, this difference doesn’t matter.)
Proof of minimality. We claim the minimal configuration stated above cannot be covered by a cir-
cle with radius r < 2. If r < 2 and all six vertices O1 , O2 , . . . , O6 are in the circle, then we have
that ∠O1 OO2 > 60◦ since O1 O2 is the largest side of the triangle O1 OO2 , and similar for other an-
gles ∠O2 OO3 , ∠O3 OO4 , . . . , but we cannot have six angles greater than 60◦ into 360◦ , contradiction.
Therefore r ≥ 2.
Proof of maximality. Let ABCDEF√be the hexagon, and choose the covering circle to be centered at
O, the midpoint of AD, and radius 3 + 1. We claim the other vertices are inside this covering circle.
B. Let M be the midpoint of AC. Since ABC is isosceles and AM ≥ 1,
First, we will show the√claim for √
we must have BM ≤ 4 − 1 = 3. Furthermore, M O is a midline of ACD, so M O = CD 2 = 1. Thus

by the triangle inequality, OB ≤ M B + OM = 3 + 1, proving the claim. A similar argument proves
the claim for C, E, F √
. Finally, an analogous
√ argument to above shows if we define P √ as the midpoint
of BE, then √ AP ≤ 3 + 1 and DP ≤ 3 + 1, so by triangle inequality AD ≤ 2( 3 + 1). Hence
OA = OD ≤ 3 + 1, proving the claim for A and D. Thus the covering circle contains all six vertices
of ABCDEF .
7. Let ABC be a triangle with AB = 13, BC = 14, CA = 15. Let H be the orthocenter of ABC. Find
the radius of the circle with nonzero radius tangent to the circumcircles of AHB, BHC, CHA.
Proposed by: Michael Ren
65
Answer: 4

Solution 1. We claim that the circle in question is the circumcircle of the anticomplementary triangle
of ABC, the triangle for which ABC is the medial triangle.
Let A0 B 0 C 0 be the anticomplementary triangle of ABC, such that A is the midpoint of B 0 C 0 , B is the
midpoint of A0 C 0 , and C is the midpoint of A0 B 0 . Denote by ω the circumcircle of A0 B 0 C 0 . Denote by
ωA the circumcircle of BHC, and similarly define ωB , ωC .
Since ∠BA0 C = ∠BAC = 180◦ − ∠BHC, we have that ωA passes through A0 . Thus, ωA can be
redefined as the circumcircle of A0 BC. Since triangle A0 B 0 C 0 is triangle A0 BC dilated by a factor of 2
from point A0 , ω is ωA dilated by a factor of 2 from point A0 . Thus, circles ω and ωA are tangent at
A0 .
By a similar logic, ω is also tangent to ωB and ωC . Therefore, the circumcircle of the anticomplementary
triangle of ABC is indeed the circle that the question is asking for.
Using the formula R = abc
4A , we can find that the circumradius of triangle ABC is
65
8 . The circumradius
of the anticomplementary triangle is double of that, so the answer is 65
4 .
Solution 2. It is well-known that the circumcircle of AHB is the reflection of the circumcircle of
ABC over AB. In particular, the circumcircle of AHB has radius equal to the circumradius R = 65 8 .
Similarly, the circumcircles of BHC and CHA have radii R. Since H lies on all three circles (in the
question), the circle centered at H with radius 2R = 65
4 is tangent to each circle at the antipode of H
in that circle.

8. In triangle ABC with AB < AC, let H be the orthocenter and O be the circumcenter. Given that the
midpoint of OH lies on BC, BC = 1, and the perimeter of ABC is 6, find the area of ABC.
Proposed by: Andrew Lin
6
Answer: 7

Solution 1. Let A0 B 0 C 0 be the medial triangle of ABC, where A0 is the midpoint of BC and so
on. Notice that the midpoint of OH, which is the nine-point-center N of triangle ABC, is also the
circumcircle of A0 B 0 C 0 (since the midpoints of the sides of ABC are on the nine-point circle). Thus, if
N is on BC, then N A0 is parallel to B 0 C 0 , so by similarity, we also know that OA is parallel to BC.
Next, AB < AC, so B is on the minor arc AC. This means that ∠OAC = ∠OCA = ∠C, so ∠AOC =
180 − 2∠C. This gives us the other two angles of the triangle in terms of angle C: ∠B = 90 + ∠C and
∠A = 90 − 2∠C. To find the area, we now need to find the height of the triangle from A to BC, and
this is easiest by finding the circumradius R of the triangle.
We do this by the Extended Law of Sines. Letting AC = x and AB = 5 − x,
1 x 5−x
= = = 2R,
sin(90 − 2C) sin(90 + C) sin C

which can be simplified to


1 x 5−x
= = = 2R.
cos 2C cos C sin C
This means that
1 (x) + (5 − x) 5
= =
cos 2C (cos C) + (sin C) cos C + sin C
and the rest is an easy computation:

cos C + sin C = 5 cos 2C = 5(cos2 C − sin2 C)


1
= cos C − sin C
5
Squaring both sides,
1
= cos2 C − 2 sin C cos C + sin2 C = 1 − sin 2C
25
24 7
so sin 2C = 25 , implying that cos 2C = 25 . Therefore, since cos12C = 2R from above, R = 25
14 . Finally,
viewing triangle ABC with BC = 1 as the base, the height is
s  2
2
BC 12
R − =
2 7
1 12
by the Pythagorean Theorem, yielding an area of 2 ·1· 7 = 67 .
Solution 2. The midpoint of OH is the nine-point center N . We are given N lies on BC, and we also
know N lies on the perpendicular bisector of EF , where E is the midpoint of AC and F is the midpoint
of AB. The main observation is that N is equidistant from M and F , where M is the midpoint of BC.
Translating this into coordinates, we pick B(−0.5, 0) and C(0.5, 0), and arbitrarily set A(a, b) where
(without loss of generality) b > 0. We get E( a+0.5 b a−0.5 b
2 , 2 ), F ( 2 , 2 ), M (0, 0). Thus N must have
x-coordinate equal to the average of those of E and F , or 2 . Since N lies on BC, we have N ( a2 , 0).
a

a2 1 2
Since M N = EN , we have 4 = 16 + b4 . Thus a2 = b2 + 41 . The other equation is AB + AC = 5,
which is just p p
(a + 0.5)2 + b2 + (a − 0.5)2 + b2 = 5.
This is equivalent to p p
2a2 + a + 2a2 − a = 5
p p
2a2 + a = 5 − 2a2 − a
p
2a2 + a = 25 − 10 2a2 − a + 2a2 − a
p
25 − 2a = 10 2a2 − a
625 − 100a + 4a2 = 200a2 − 100a
196a2 = 625
625 576 24 12 b
Thus a2 = 196 , so b2 = 196 . Thus b = 14 = 7 , so [ABC] = 2 = 67 .
9. In a rectangular box ABCDEF GH with edge lengths AB = AD = 6 and AE = 49, a plane slices
through point A and intersects edges BF, F G, GH, HD at points P, Q, R, S respectively. Given that
AP = AS and P Q = QR = RS, find the area of pentagon AP QRS.
Proposed by: Yuan Yao

141 11
Answer: 2

Let AD be the positive x-axis, AB be the positive y-axis, and AE be the positive z-axis, with A the
origin. The plane, which passes through the origin, has equation k1 x + k2 y = z for some undetermined
parameters k1 , k2 . Because AP = AS and AB = AD, we get P B = SD, so P and S have the same
z-coordinate. But P (0, 6, 6k2 ) and S(6, 0, 6k1 ), so k1 = k2 = k for some k. Then Q and R both have
z-coordinate 49, so Q( 49 49 2 2
k − 6, 6, 49) and R(6, k − 6, 49). The equation QR = RS then gives
 2  2
49 49
− 6 + (49 − 6k)2 = 2 12 − − 12 .
k k
This is equivalent to
(49 − 6k)2 (k 2 + 1) = 2(49 − 12k)2 ,
which factors as
(k − 7)(36k 3 − 336k 2 − 203k + 343) = 0.
This gives k = 7 as a root. Note that for Q and R to actually lie on F G and GH respectively, we
must have 49 49
6 ≥ k ≥ 12 . Via some estimation, one can show that the cubic factor has no roots in this
range (for example, it’s easy to see that when k = 1 and k = 336 28
36 = 3 , the cubic is negative, and it
also remains negative between the two values), so we must have k = 7.
Now consider projecting AP QRS onto plane ABCD. The projection is ABCD save for a triangle
Q0 CR0 with side length 12 − 49 25 47
k = 5. Thus the projection has area 36 − 2 = 2 . Since the area
of the projection equals [AP QRS] · cos θ, where θ is the (smaller) angle between planes AP QRS
and ABCD, and since the planes have normal vectors (k, k, −1) and (0, 0, 1) respectively, we get
cos θ = (k,k,−1)·(0,0,1)

k2 +k2 +1
= √2k12 +1 = √199 and so
√ √
47 99 141 11
[AP QRS] = = .
2 2

10. In triangle ABC, AB = 13, BC = 14, CA = 15. Squares ABB1 A2 , BCC1 B2 , CAA1 C2 are constructed
outside the triangle. Squares A1 A2 A3 A4 , B1 B2 B3 B4 , C1 C2 C3 C4 are constructed outside the hexagon
A1 A2 B1 B2 C1 C2 . Squares A3 B4 B5 A6 , B3 C4 C5 B6 , C3 A4 A5 C6 are constructed outside the hexagon
A4 A3 B4 B3 C4 C3 . Find the area of the hexagon A5 A6 B5 B6 C5 C6 .
Proposed by: Yuan Yao
Answer: 19444
Solution 1.
We can use complex numbers to find synthetic observations. Let A = a, B = b, C = c. Notice that B2 is
a rotation by −90◦ (counter-clockwise) of C about B, and similarly C1 is a rotation by 90◦ of B about C.
Since rotation by 90◦ corresponds to multiplication by i, we have B2 = (c − b) · (−i) + b = b(1 + i) − ci
and C1 = (b − c) · i + c = bi + c(1 − i). Similarly, we get C2 = c(1 + i) − ai, A1 = ci + a(1 − i),
A2 = a(1 + i) − bi, B1 = ai + b(1 − i). Repeating the same trick on B1 B2 B3 B4 et. al, we get
C4 = −a + b(−1 + i) + c(3 − i), C3 = a(−1 − i) − b + c(3 + i), A4 = −b + c(−1 + i) + a(3 − i),
A3 = b(−1 − i) − c + a(3 + i), B4 = −c + a(−1 + i) + b(3 − i), B3 = c(−1 − i) − a + b(3 + i). Finally,
repeating the same trick on the outermost squares, we get B6 = −a + b(3 + 5i) + c(−3 − 3i), C5 =
−a + b(−3 + 3i) + c(3 − 5i), C6 = −b + c(3 + 5i) + a(−3 − 3i), A5 = −b + c(−3 + 3i) + a(3 − 5i), A6 =
−c + a(3 + 5i) + b(−3 − 3i), B5 = −c + a(−3 + 3i) + b(3 − 5i).
From here, we observe the following synthetic observations.

S1. B2 C1 C4 B3 , C2 A1 A4 C3 , A2 B1 B4 A3 are trapezoids with bases of lengths BC, 4BC; AC, 4AC; AB, 4AB
and heights ha , hb , hc respectively (where ha is the length of the altitude from A to BC, and like-
wise for hb , hc )
S2. If we extend B5 B4 and B6 B3 to intersect at B7 , then B7 B4 B3 ∼ = BB1 B2 ∼ B7 B5 B6 with scale
factor 1:5. Likewise when we replace all B’s with A’s or C’s.

Proof of S1. Observe C1 − B2 = c − b and C4 − B3 = 4(c − b), hence B2 C1 kB3 C4 and B3 C4 = 4B2 C1 .
Furthermore, since translation preserves properties of trapezoids, we can translate B2 C1 C4 B3 such
that B2 coincides with A. Being a translation of a − B2 , we see that B3 maps to B30 = 2b − c and
C4 maps to C40 = −2b + 3c. Both 2b − c and −2b + 3c lie on the line determined by b and c (since
−2 + 3 = 2 − 1 = 1), so the altitude from A to BC is also the altitude from A to B30 C40 . Thus ha
equals the length of the altitude from B2 to B3 C4 , which is the height of the trapezoid B2 C1 C4 B3 .
This proves S1 for B2 C1 C4 B3 ; the other trapezoids follow similarly.
Proof of S2. Notice a translation of −a + 2b − c maps B1 to B4 , B2 to B3 , and B to a point
B8 = −a + 3b − c. This means B8 B3 B4 ∼ = BB1 B2 . We can also verify that 4B8 + B6 = 5B3 and
4B8 + B5 = 5B4 , showing that B8 B5 B6 is a dilation of B8 B4 B3 with scale factor 5. We also get B8
lies on B3 B6 and B5 B4 , so B8 = B7 . This proves S2 for B3 B4 B5 B6 , and similar arguments prove the
likewise part.
Now we are ready to attack the final computation. By S2, [B3 B4 B5 B6 ] + [BB1 B2 ] = [B7 B5 B6 ] =
[BB1 B2 ]. But by the 21 ac sin B formula, [BB1 B2 ] = [ABC] (since ∠B1 BB2 = 180◦ − ∠ABC). Hence,
[B3 B4 B5 B6 ] + [BB1 B2 ] = 25[ABC]. Similarly, [C3 C4 C5 C6 ] + [CC1 C2 ] = 25[ABC] and [A3 A4 A5 A6 ] +
[AA1 A2 ] = 25[ABC]. Finally, the formula for area of a trapezoid shows [B2 C1 C4 B3 ] = 5BC 2 · ha =
5[ABC], and similarly the other small trapezoids have area 5[ABC]. The trapezoids thus contribute
area (75 + 3 · 5) = 90[ABC]. Finally, ABC contributes area [ABC] = 84.
By S1, the outside squares have side lengths 4BC, 4CA, 4AB, so the sum of areas of the outside squares
is 16(AB 2 + AC 2 + BC 2 ). Furthermore, a Law of Cosines computation shows A1 A22 = AB 2 + AC 2 +
2 · AB · AC · cos ∠BAC = 2AB 2 + 2AC 2 − BC 2 , and similarly B1 B22 = 2AB 2 + 2BC 2 − AC 2 and
C1 C22 = 2BC 2 + 2AC 2 − AB 2 . Thus the sum of the areas of A1 A2 A3 A4 et. al is 3(AB 2 + AC 2 + BC 2 ).
Finally, the small squares have area add up to AB 2 + AC 2 + BC 2 . Aggregating all contributions from
trapezoids, squares, and triangle, we get

[A5 A6 B5 B6 C5 C6 ] = 91[ABC] + 20(AB 2 + AC 2 + BC 2 ) = 7644 + 11800 = 19444.

Solution 2. Let a = BC, b = CA, c = AB. We can prove S1 and S2 using some trigonometry instead.
Proof of S1. The altitude from B3 to B2 C1 has length B2 B3 sin ∠BB2 B1 = B1 B2 sin ∠BB2 B1 =
BB1 sin ∠B1 BB2 = AB sin ∠ABC = ha using Law of Sines. Similarly, we find √ the altitude from
C4 to B2 C1 equals ha , thus proving B2 C1 C4 B3 is a trapezoid. Using B1 B2 = 2a2 + 2c2 − b2 from
end of Solution 1, we get the length of the projection of B2 B3 onto B3 C4 is B2 B3 cos BB2 B1 =
(2a2 +2c2 −b2 )+a2 −c2 2 2
−b2 2 2
−c2
2a = 3a +c2a , and similarly the projection of C1 C4 onto B3 C4 has length 3a +b2a .
3a2 +c2 −b2 3a2 +b2 −c2
It follows that B3 C4 = 2a +a+ 2a = 4a, proving S1 for B2 C1 C4 B3 ; the other cases
follow similarly.
Proof of S2. Define B8 to be the image of B under the translation taking B1 B2 to B4 B3 . We claim B8
lies on B3 B6 . Indeed, B8 B4 B3 ∼
= BB1 B2 , so ∠B8 B3 B4 = ∠BB2 B1 = 180◦ − ∠B3 B2 C1 = ∠B2 B3 C4 .
Thus ∠B8 B3 C4 = ∠B4 B3 B2 = 90◦ . But ∠B6 B3 C4 = 90◦ , hence B8 , B3 , B6 are collinear. Similarly we
B7 B3
can prove B5 B4 passes through B8 , so B8 = B7 . Finally, B 7 B6
=B 7 B4 1
B7 B5 = 5 (using B3 B6 = 4a, B4 B5 =
4c, B7 B3 = a, B7 B4 = c) shows B7 B4 B3 ∼ B7 B5 B6 with scale factor 1:5, as desired. The likewise part
follows similarly.

You might also like