HMMT February 2023
February 18, 2023
Geometry Round
1. Let ABCDEF be a regular hexagon, and let P be a point inside quadrilateral ABCD. If the area of
triangle P BC is 20, and the area of triangle P AD is 23, compute the area of hexagon ABCDEF .
Proposed by: Ankit Bisain
Answer: 189
Solution: If s is the side length of the hexagon, h1 is the length of the height from P to BC, and h2
is the length of the height from P to AD, we have [P BC] = 21 s · h1 and [P AD] = 12 (2s) · h2 . We also
√
have h1 + h2 = 23 s. Therefore,
√
3 2
2[P BC] + [P AD] = s(h1 + h2 ) = s .
2
√
3 3 2
The area of a hexagon with side length s is 2 s , giving a final answer of
6[P BC] + 3[P AD] = 6 · 20 + 3 · 23 = 189 .
2. Points X, Y , and Z lie on a circle with center O such that XY = 12. Points A and B lie on segment
XY such that OA = AZ = ZB = BO = 5. Compute AB.
Proposed by: Rishabh Das
√
Answer: 2 13
Solution: Let the midpoint of XY be M . Because OAZB is a rhombus, OZ ⊥ AB, so M is the
midpoint of AB as well. Since OM = 21 OX, △OM X is a 30 − 60 − 90 triangle, and since XM = 6,
√ √ √
OM = 2 3. Since OA = 5, the Pythagorean theorem gives AM = 13, so AB = 2 13.
3. Suppose ABCD is a rectangle whose diagonals meet at E. The perimeter of triangle ABE is 10π and
the perimeter of triangle ADE is n. Compute the number of possible integer values of n.
Proposed by: Luke Robitaille
Answer: 47
Solution: For each triangle T , we let p(T ) to denote the perimeter of T .
First, we claim that 12 p(△ABE) < p(△ADE) < 2p(△ABE). To see why, observe that
p(△ADE) = EA + ED + AD < 2(EA + ED) = 2(EA + EB) < 2p(△ABE),
Similarly, one can show that p(△ABE) < 2p(△ADE), proving the desired inequality.
This inequality limits the possibility of n to only those in (5π, 20π) ⊂ (15.7, 62.9), so n could only
range from 16, 17, 18, . . . , 62, giving 47 values. These values are all achievable because
• when AD approaches zero, we have p(△ADE) → 2EA and p(△ABE) → 4EA, implying that
p(△ADE) → 21 p(△ABE) = 5π;
• similarly, when AB approaches zero, we have p(△ADE) → 2p(△ABE) = 20π; and
• by continuously rotating segments AC and BD about E, we have that p(△ADE) can reach any
value between (5π, 20π).
Hence, the answer is 47.
4. Let ABCD be a square, and let M be the midpoint of side BC. Points P and Q lie on segment AM
such that ∠BP D = ∠BQD = 135◦ . Given that AP < AQ, compute AQ AP .
Proposed by: Ankit Bisain, Luke Robitaille
√
Answer: 5
Solution: Notice that ∠BP D = 135◦ = 180◦ − ∠BAD 2 and P lying on the opposite side of BD as C
means that P lies on the circle with center C through B and D. Similarly, Q lies on the circle with
center A through B and D.
Let the side length of the square be 1. We have AB = AQ = AD, so AQ = 1. To compute AP , let E
be the reflection of D across C. We have that E lies both on AM and the circle centered at C through
B and D. Since AB is tangent to this circle,
AB 2 = AP · AE
√ √
by power of a point. Thus, 12 = AP · 5 =⇒ AP = √1 . Hence, the answer is 5.
5
5. Let ABC be a triangle with AB = 13, BC = 14, and CA = 15. Suppose P QRS is a square such that
P and R lie on line BC, Q lies on line CA, and S lies on line AB. Compute the side length of this
square.
Proposed by: Pitchayut Saengrungkongka
√
Answer: 42 2
Solution:
M B X C
R P
A′
Let A′ be the reflection of A across BC. Since Q and S are symmetric across BC, we get that Q ∈ BA′ ,
S ∈ CA′ . Now, let X and M be the midpoints of AA′ and P R. Standard altitude computation gives
BX = 5, CX = 9, AX = 12. Moreover, from similar triangles, CX : CY = AA′ : P R = BX : BM , so
BM : CM = 5 : 9, so we easily get that BM = 35/2. Now, P M = 12 9 · BY = 42, so the side length is
√
42 2.
6. Convex quadrilateral ABCD satisfies ∠CAB = ∠ADB = 30◦ , ∠ABD = 77◦ , BC = CD, and
∠BCD = n◦ for some positive integer n. Compute n.
Proposed by: Pitchayut Saengrungkongka
Answer: 68
Solution: Let O be the circumcenter of △ABD. From ∠ADB = 30◦ , we get that △AOB is equilateral.
Moreover, since ∠BAC = 30◦ , we have that AC bisects ∠BAO, and thus must be the perpendicular
bisector of BO. Therefore, we have CB = CD = CO, so C is actually the circumcenter of △BDO.
Hence,
∠BCD = 2(180◦ − ∠BOD)
= 2(180◦ − 2∠BAD)
= 2(180◦ − 146◦ ) = 68◦
7. Quadrilateral ABCD is inscribed in circle Γ. Segments AC and BD intersect at E. Circle γ passes
through E and is tangent to Γ at A. Suppose that the circumcircle of triangle BCE is tangent to γ at
E and is tangent to line CD at C. Suppose that Γ has radius 3 and γ has radius 2. Compute BD.
Proposed by: Eric Shen, Luke Robitaille
√
9 21
Answer: 7
Solution: The key observation is that △ACD is equilateral. This is proven in two steps.
• From tangency at C, we have
∠DCA = ∠DCE = ∠EBC = ∠DBC = ∠DAC,
implying that CA = CD.
• Consider the common tangent of γ and Γ at A. By homothety at E, this line is parallel to the
tangent of ⊙(EBC) at C, which is line CD. This implies that AC = AD.
Once we have this, compute
√
AC = 2RΓ · sin 60◦ = 3 3
√
AE = 2Rγ · sin 60◦ = 2 3
√
There are now many ways to finish. One way is to√use Stewart’s theorem on △ADC to get ED =√ 21,
then use Power of Point to get EB = AE·EC
ED = 2 721 . The final answer is BD = BE + ED = 9 721 .
8. Triangle ABC with ∠BAC > 90◦ has AB = 5 and AC = 7. Points D and E lie on segment BC such
that BD = DE = EC. If ∠BAC + ∠DAE = 180◦ , compute BC.
Proposed by: Maxim Li
√
Answer: 111
Solution: Let M be the midpoint of BC, and consider dilating about M with ratio − 13 . This takes
B to E, C to D, and A to some point A′ on AM with AM = 3A′ M . Then the angle condition implies
∠DAE + ∠EA′ D = 180◦ , so ADA′ E is cyclic. Then by power of a point, we get
AM 2 BC 2
= AM · A′ M = DM · EM = .
3 36
2 2 2 2 2 2 2
−BC −BC
But we also know AM 2 = 2AB +2AC , so we have 2AB +2AC = BC , which rearranges to
2 3 2 2
4 12 √36
BC = 2 (AB + AC ). Plugging in the values for AB and AC gives BC = 111.
9. Point Y lies on line segment XZ such that XY = 5 and Y Z = 3. Point G lies on line XZ such that
there exists a triangle ABC with centroid G such that X lies on line BC, Y lies on line AC, and Z
lies on line AB. Compute the largest possible value of XG.
Proposed by: Luke Robitaille
20
Answer: 3
1 1 1
Solution: The key claim is that we must have XG + YG + ZG = 0 (in directed lengths).
We present three proofs of this fact.
Proof 1: By a suitable affine transformation, we can assume without loss of generality that ABC
is equilateral. Now perform an inversion about G with radius GA = GB = GC. Then the im-
ages of X, Y, Z (call them X ′ , Y ′ , Z ′ ) lie on (GBC), (GAC), (GAB), so they are the feet of the per-
pendiculars from A1 , B1 , C1 to line XY Z, where A1 , B1 , C1 are the respective antipodes of G on
(GBC), (GAC), (GAB). But now A1 B1 C1 is an equilateral triangle with medial triangle ABC, so its
centroid is G. Now the centroid of (degenerate) triangle X ′ Y ′ Z ′ is the foot of the perpendicular of the
centroid of A1 B1 C1 onto the line, so it is G. Thus X ′ G + Y ′ G + Z ′ G = 0, which yields the desired
claim. ■
Proof 2: Let W be the point on line XY Z such that W G = 2GX (in directed lengths). Now note that
(Y, Z; G, W ) is a harmonic bundle, since projecting it through A onto BC gives (B, C; MBC , ∞BC ).
By harmonic bundle properties, this yields that Y1G + ZG1
= W2G (in directed lengths), which gives the
desired. ■
Proof 3: Let P ̸= G be an arbitrary point on the line XY Z. Now, in directed lengths and signed areas,
[GBP ] [GCP ] [GBP ]−[GCP ] [GBP ]−[GCP ]
GP GP
GX = [GBX] = [GCX] , so GX = [GBX]−[GCX] = [GBC] = 3([GBP ]−[GCP ])
[ABC] . Writing analogous
GP GP GP GP GP
equations for GY and GZ and summing yields GX + GY + GZ = 0, giving the desired. ■
With this lemma, we may now set XG = g and know that
1 1 1
+ + =0
g g−5 g−8
Solving the quadratic gives the solutions g = 2 and g = 20
3 ; the latter hence gives the maximum (it is
not difficult to construct an example for which XG is indeed 20/3).
10. Triangle ABC has incenter I. Let D be the foot of the perpendicular from A to side BC. Let X be
a point such that segment AX is a diameter of the circumcircle of triangle ABC. Given that ID = 2,
IA = 3, and IX = 4, compute the inradius of triangle ABC.
Proposed by: Maxim Li
11
Answer: 12
Solution: Let R and r be the circumradius and inradius of ABC, let AI meet the circumcircle
√ of ABC
again at M , and let J be the A-excenter. We can show that △AID ∼ △AXJ (e.g. by bc inversion),
and since M is the midpoint of IJ and ∠AM X = 90◦ , IX = XJ. Thus, we have IX 2R
= XA IA
XJ = ID , so
2 2 2
2XI +2AI −AX
R = IX·IA 2
2ID = 3. But we also know R − 2Rr = IO =
2
4 . Thus, we have
2IX 2 + 2IA2 − 4R2
1 2 11
r= R − = .
2R 4 12