Вы находитесь на странице: 1из 18

Mathematics Olympiad

Geometry

Narasimhan R. Chari

m
1. a) Sine Rule: In any triangle ∆ ABC, using standard notation, let l(BC) = a, l(CA) = b,
l(AB) = c and let R be the radius of the circumcircle. Then

o
a b c
= = = 2R

l.c
sin A sin B sin C
1 abc
b) The area of the triangle is ∆ = [ABC] = ab sin C =
2 4R

ai
Proof: Let O be the circumcentre. Let COD be the diameter of the circumcircle.
Case 1: Suppose ∆ ABC is an acute-angled triangle. Then O is inside the triangle.
gm BC a
∠CDB = ∠CAB = ∠ A; ∠CBD = 90◦ ; sin CDB = ; hence sin A = . By symmetry, we
CD 2R
a b c
get the sine rule, = = = 2R
sin A sin B sin C
ri@

Case 2: Suppose ∆ ABC is an obtuse-angled triangle. Then O is outside the triangle.


BC a
∠CDB = π − ∠CAB = π − ∠ A; ∠CBD = 90◦ ; sin CDB = ; hence sin (π − A) = . But
CD 2R
a a b c
sin(π − A) = sin A; sin A = . As before, we get the sine rule, = = = 2R
2R sin A sin B sin C
AD
ha

b) Drop AD perpendicular from A on BC. Then in the triangle ∆ ADC, sin C = , i.e., AD =
AC
1 1 1
AC sin C. Hence, the area of the triangle = (base)(height) = (BC)(AD) = (BC)(AC) sin C,
2 2 2
1 c abc
sic

i.e., ∆ = ab sin C; then by the sine rule, c = 2R sin C ⇒ sin C = ; hence we get ∆ =
2 2R 4R

2. In triangle ∆ ABC, let l(BC) = a, l(CA) = b, l(AB) = c, 2s = a + b + c. Then


ra

i) a = b cos C + c cos B (Projection formula)


ii) c2 = a2 + b2 − 2ab cos C (Cosine Rule)
na

iii) Area of the triangle = [ABC] = s(s − a)(s − b)(s − c) (Heron0 s formula),
p

(Alexandria, Egypt, 60 AD)


Proof: Method 1: Drop AD perpendicular from A on to BC.
CD
Case 1) : Suppose ∠C ≤ 90◦ . Then cos C > 0; cos C = .
AB
Hence the projection of AC on BC is given by CD = b cos C. Similarly, the projection of AB on
BC is given by BD = c cos B. Hence BC = BD + DC ⇒ a = b cos C + c cos B
ii) The Pythagoras theorem gives AB2 = AD2 + BD2 and AC2 = AD2 + DC2 . Also,
BC2 = (BD + DC)2 = BD2 + DC2 + 2BD. DC. Consider
BC2 + AC2 − AB2 = BD2 + DC2 + 2BD. DC + AD2 + DC2 − (AD2 + BD2 )
= 2DC(DC + BD) = 2DC. BD = 2(b cos C)a ⇒ a2 + b2 − c2 = 2ab cos C
⇒ c2 = a2 + b2 − 2ab cos C
Case 2: Suppose ∠C > 90◦ . Then cos C < 0;
cos (ACD) = cos(π −C) = − cos C. Hence CD = −b cos C
Also, BC2 = (BD − DC)2 = BD2 + DC2 − 2BD DC.
BC2 + AC2 − AB2 = BD2 + DC2 − 2BD DC + AD2 + DC2 − (AD2 + BD2 )
= 2DC(DC − BD) = 2DC (−BC) = 2(−b cos C)(−a) ⇒ a2 + b2 − c2 = 2ab cos C
⇒ c2 = a2 + b2 − 2ab cos C
−→ − → − →
Method 2 (Vectors): AB = AC + CB; taking magnitudes on both sides we get,

m

→ −
→ − →
|AB|2 = |AC + CB|2 ; using |→

v |2 = →−
v ·→
−v we get

→2  −
→ − →   −
→ − → 
|AB| = AC + CB · AC + CB

o
−→ − → −
→ − → − → − →
= AC · AC + 2 AC · CB + CB · CB

l.c

→ −→ −
→ − →
= |AC|2 + |CB|2 + 2 |AC| |CB| cos (π −C)
Hence c2 = a2 + b2 − 2 ab cos C

ai
−→ −→
Note that the (directed) angle of anticlockwise rotation from the vector AC to the vector CB is
(π −C). gm
iii) Drop AD perpendicular from A to BC. Then
1 AD
Area of the triangle is equal to ∆ = BC.AD; in ∆ ADB, sin B = ; AD = c sin B.
2 AB
1
Area = a(c sin B)
2
ri@

1 p a2 + c2 − b2
= ac 1 − cos2 B; use the cosine rule, cos B =
2 s 2ac
1 (a2 + c2 − b2 )2
= ac 1 −
2 (2ac)2
ha

s
1 (4a2 c2 − (a2 + c2 − b2 )2
= ac
2 4a2 c2
p
(2ac + a2 + c2 − b2 )(2ac − a2 − c2 + b2 )
sic

1
= ac
2 2ac
1p
= (a + c + b)(a + c − b)(b + a − c)(b − a + c)
4
ra

But a + b + c = 2s; b + c − a = (a + b + c) − 2a = 2s − 2a = 2(s − a); c + a − b = 2(s − b);


a + b − c = 2(s − c). Hence
1p p
Area = [ABC] = 2s(2s − 2a)(2s − 2b)(2s − 2c) = s(s − a)(s − b)(s − c)
na

2
3. Half-angle formulas: We know that A/2, B/2, C/2 are between 0 and 90 degrees; hence all
their trigonometric ratios are positive.
  r   r
A (s − b)(s − c) A s(s − a)
i) sin = ; ii) cos =
2 bc 2 bc

Hint: i) cos A = 1 − 2 sin2 (A/2); hence, using the cosine rule, we get,
b2 + c2 − a2
2 sin2 (A/2) = 1 −

m
2bc
2bc − b − c2 + a2
2
=

o
2bc
a − (b − c)2
2
=

l.c
2bc
(a + b − c)(a − b + c)
=
2bc
(a + b + c − 2c)(a + b + c − 2b)

ai
=
2bc
(2s − 2c)(2s − 2b) gm
=
2bc
2(s − c)(s − b)
=
  rbc
A (s − b)(s − c)
Hence, sin =
ri@

2 bc
q
cos (A/2) = 1 − sin2 (A/2)
r
(s − b)(s − c)
= 1−
bc
ha

s
bc − s2 + bs + cs − bc
= ; (a + b + c = 2s ⇒ bs + cs = 2s2 − as)
bc
s
−s2 + 2s2 − as
sic

=
bc
r
s(s − a)
=
bc
ra
na

3
4. Suppose X, Y, Z are points on the sides BC, CA, AB of a triangle ∆ ABC, respectively. Let
AZ BX CY sin ACZ sin BAX sin CBY
R= ; R∗ = . Then
ZB XC YA sin ZCB sin XAC sin Y BA
a) R = R∗ .
b) Ceva0 s theorem, (Italy, 1678): AX, BY, CZ are concurrent if and only if R = 1
c) Menelaus0 theorem, (Alexandria, Egypt, 110 AD): X, Y, Z are collinear if and only if

m
R = −1.
AZ −
→ −

(A ratio ZB is defined to be positive if Z is between A and B, so that AZ and ZB are in the same

→ −

o
direction and the ratio is negative if Z is outside the line segment AB, i.e., AZ and ZB are in
opposite directions).

l.c
Hint: a) The sine rule applied to triangles ∆ ABX and ∆ ACX gives
BX sin BAX XC sin XAC sin XAC BX XC sin BAX BX AB. sin BAX
= ; = = ; hence . = ; hence =
AB sin AXB CA sin AXB sin AXB AB CA sin XAC XC CA. sin XAC

ai
CY BC. sinCBY AZ CA. sin ACZ
Similarly, = and =
YA AB. sinY BA ZB BC. sin ZCB
gm
Multiplying these three equations we get, R = R∗
b) Suppose AX, BY, CZ are concurrent at P. Let [ABC] denote the area of the triangle ∆ ABC. If
two triangles have a common vertex and collinear bases, their heights from this vertex are equal;
hence the ratio of their areas is equal to the ratio of their bases,
ri@

BX [ABX] [PBX] [ABX] − [PBX] [ABP] CY [BCP] AZ [CAP]


= = = = ; similarly, = and =
XC [AXC] [PXC] [AXC] − [PXC] [CAP] YA [ABP] ZB [BCP]
BX CY AZ
Hence, we get Ceva’s theorem, =1
XC YA ZB
ha

Conversely, suppose this equation is satisfied. Let AX and BY intersect at P. Let CP intersect
BX CY AZ 0 BX CY AZ
AB at Z 0 . By part a), we get 0
= 1. But we have assumed that = 1.
XC YA Z B XC YA ZB
AZ AZ 0
sic

Hence = .
ZB Z 0 B

But A, Z, B are collinear and we know that, given any two points A and B and a given real
number r, there is a unique point which divides the line segment AB in the ratio r : 1. Hence
ra

Z = Z0.
c) Menelaus theorem: Assume X, Y, Z are collinear. Drop AP, BQ, CR perpendiculars from the
na

AZ AP
vertices A, B, C on the line XY Z. We get similar triangles ∆ APZ ∼ ∆ BQZ; hence, = ;
ZB QB
BX BQ CY CR
∆ BXQ ∼ ∆CXR; hence, = ; ∆ APY ∼ ∆CRY ; hence, = .
CX RC AY AP
AZ BX CY
Multiplying these three equations we get Menelaus’ theorem, R = = −1. Note that
ZB XC YA
either exactly one of these three fractions is negative or all three are negative. The converse of
Menelaus’ theorem is proved as in b) above.

4
5. In ∆ ABC, D and E are points on BC and CA respectively such that AE = EC and DC = 2BD.
AP BP
Find and
PD PE
Hint: Let AD ∩ BE = P. By Menelaus’ theorem in ∆ ADC with BPE as transversal, we get
AP DB CE AP 1 1 AP
. . = −1 ⇒ . . =1⇒ = 3; using Menelaus’ theorem in ∆ BCE with APD
PD BC EA PD 3 1 PD
BD CA EP 1 2 EP BP
as transversal, we get . . = −1 ⇒ . . =1⇒ =1

m
DC AE PB 2 1 PB PE

6. BE and CF are the bisectors of ∠ B and ∠C in ∆ ABC. BE ∩CF = I; AY ⊥ BE; AX ⊥ CF. Show

o
b+c−a
that XY =
2

l.c
Hint: Extend AX and AY to intersect BC at X 0 and Y 0 respectively. ∆ ABY ∼ = ∆Y 0 BY , by (AAS);
hence AB = BY , AY = YY ; ∆ ACX ∼
0 0
= ∆ X CX, by (AAS); hence AC = CX 0 , AX = XX 0 . In
0

∆ AX 0Y 0 , XY is the line joining the midpoints of AX 0 and AY 0 ⇒ X 0Y 0 = 2XY ; hence


XY = 12 [X 0C + BY 0 − BC] = 21 (AC + AB − BC) = 12 (b + c − a)

ai
7. D, E and F are points on the sides BC, CA, AB of ∆ ABC. AD, BE and CF are concurrent at K.
BD BF gm
= and ∠ADB = ∠AFC. Prove that ∠ABE = ∠CAD.
DC FA
BD BF
Hint: Given = ⇒ DF k CA ⇒ ∠ FDK = ∠ DAC. Also for the quadrilateral BDKF, the
DC FA
exterior angle ∠ AFC = the interior opposite angle ∠ BDK. Hence BDKF is a cyclic quadrilat-
eral; ∠ ABE = ∠ FDK = ∠ DAC.
ri@

8. BB1 and CC1 are the bisectors of angles ∠ B and ∠C of ∆ ABC. E and F are the feet of the
perpendiculars from A on BB1 and CC1 respectively. If the incircle of the triangle touches AB
at D, prove that AD = EF = s − a
Hint: ∠ ADI = ∠ AFI = ∠ AEI = 90◦ . Hence AI is the diameter of the circle through A, D, F, I, E.
ha

∠ DAI = A/2. Hence ∠ AID = 90 − A/2. ∠ BIC = 180 − B/2 −C/2 = 90 + A/2. Hence
∠ FIE = ∠ BIC = 90 + A/2. Also AFIE is a cyclic quadrilateral. Hence
∠ FAE = 180 − (90 + A/2) = 90 − A/2. Hence the angle subtended by the chord AD at I on the
sic

circumference is equal to the angle subtended by the chord FE at A on the circumference. Hence
AD = FE.
9. In ∆ ABC, points D and E are taken on BC such that BD = DE = EC. F is on CA, CF = FA;
[APQ]
ra

BF intersects AD and AE at P and Q respectively. Find the ratio of the areas .


[PDEQ]
Hint: BC is trisected at D and E; AC is bisected at F. PD is half of EF which is half of AD.
na

AP 3 AQ 3
Hence = . Menelaus’ theorem in ∆ ADE with transversal BPQ gives = ; the area of
AD 4 AE 5
1 1 [APQ] 9
∆ ADE is [ADE] = AD. AE. sin(DAE); similarly, [APQ] = AP. AQ. sin(DAE); = ;
2 2 [ADE] 20
[APQ] 9 9
= = .
[PDEQ] 20 − 9 11

10. In a unit square ABCD draw quadrants of circles centred at ,A and B with radius 1. Find
the radius of the circle with centre M which touches the arc BD internally at Q, the arc AC
externally at P and the side AD.
Hint: AM = AQ − MQ = 1 − r. Draw MN perpendicular from M to AB.
MN 2 = (1 − r)2 − r2 = 1 − 2r; in ∆ AMB we get (1 + r)2 = (1 − r)2 + MN 2 , hence r = 1/6.

5
11. AB = AC in ∆ ABC; Γ is the circumcircle of the triangle ABC. D is on the arc AB of Γ not
containing C and E is on the arc AC of Γ not containing B; AD = CE. Prove that BE k AD.
(RMO 2013)
Hint: The official solution is incomplete. The chords AD and CE are equal. Hence these arcs are
also equal and they subtend equal angles at any point on the circumference. ∠ABD = ∠CAE;
Also ∠B = ∠C; ∠AEC = 180 − B = 180 − C = ∠ADB; hence the triangles ABD and CAE are
congruent (AAS). ∠BAD = ∠ACE = ∠ABE; hence the alternate angles are equal, which implies

m
AD k BE. Note that the congruent triangles also show AE = BD; ADBE is an isosceles trapez-
ium; AB = DE = AC; ADCE is also a cyclic trapezium; AE k CD.

o
12. Points D and E are taken on the sides BC and AC of ∆ ABC such that BD = 3DC and AE = 4EC;

l.c
P is on AD; D is the midpoint of EP; AP ∩ BC = S. Find BS/SD. (RMO 2013)
Hint: Let F = the midpoint of AE (or the midpoint of AP). The midpoint theorem gives
CD CF 3 2 5
AF = FE; = = ; DS = CD; CS = CD; BC = 4CD;

ai
DS FA 2 3 3
5 7 BS 7
BS = 4CD − CD = CD; = .
3 3 SD 2 gm
13. A convex polygon Γ is such that the distance between any two vertices of Γ is at most one. Prove
that
a) the distance between any two points on the sides of the polygon is at most one;
ri@

b) if X and Y are two distinct points inside Γ there exists a point Z on the boundary of Γ such
that XZ +Y Z ≤ 1.
Hint: a) Let AB and CD be any two sides of the polygon. Let P and Q be points on AB and CD
respectively. ∠PQC + ∠PQD = 180◦ ; one of these angles is ≥ 90◦ and the other is ≤ 90◦ . Let
∠PQC ≥ 90; PC > PQ; similarly ∠APC + ∠CPB = 180◦ ; let ∠CPB ≥ 90; BC > PC;
ha

BC > PC > PQ; but BC ≤ 1 (given); hence PQ ≤ 1.


b) Take any 2 points X and Y in the interior of the convex polygon. Then XY lies inside the
polygon Γ. Suppose XY extended on both sides intersects the sides AB and CD at P and Q.
sic

(XP + Y P) + (XQ + Y Q) = XP + XQ + Y P + Y Q = PQ + PQ = 2PQ ≤ 2, (by part a); hence


either XP + Y P ≤ 1 or XQ + Y Q ≤ 1; if XP + Y P ≤ 1, take Z = P on the boundary of the
polygon satisfying XZ +Y Z ≤ 1.
ra

14. I is the incentre of the right-angled triangle ∆ ABC with ∠B = 90◦ . AI extended intersects BC at
D; CI extended intersects AB at E; J is on AC; IJ ⊥ AI; K is on AC; IK ⊥ CI; DJ = EK. Prove
na

that AB = BC. (CRMO 2016(3))


Hint: Let α = A/2; β = C/2; α + β = 45◦ ; A, I, D are collinear; C, I, E are collinear, (but B, I, J
are not collinear). ∠ BDA = α + 2β ; ∠ DIC = α + β ; ∠ BEC = 2α + β ; ∠ AIE = α + β = 45◦ .
∠ EIK = 90◦ , (given); ∠ AIK = 90 − 45 = 45◦ ; ∆ AIE ∼ = ∆AIK, (AAS); EI = IK, AK = AE.
∠ DIJ = 90 , (given); ∠CIJ = 90 − (α + β ) = 90 − 45 = 45◦ ; ∆CID ∼

= ∆CIJ, (AAS);
DI = IJ; hence ∆ DIJ is isosceles. similarly ∆ EIK is isosceles. Also, DJ = EK, (given); hence
∆ DIJ ∼= ∆ EIK; IK = IJ; ∠ IKJ = ∠ IJK; taking these as exterior angles we get
1◦
α + (α + β ) = β + (α + β ); hence α = β = 22 ; A = C = 45◦ ; AB = BC
2

6
15. Find the minimum area of a right-angled triangle whose inradius is equal to one.
a+c−b
Hint: If ∠B = 90◦ , we have r = . r = 1 gives b = a + c − 2. ∆ = rs = 1/2 ac and r = 1
2
2
gives (a − 2)(c − 2) = 2; put a − 2 = x; c − 2 = y; y = ; ∆ = a + c − 1 = x + y + 3 = x + 2/x + 3
x
r !2
√ 2 √ √
∆= x− + (3 + 2 2); hence the minimum area is 3 + 2 2, which is attained when
x
√ √

m
x = y = 2; the triangle is isosceles, a = c = 2 + 2.
16. I is the incentre of ∆ ABC; AI, BI,CI intersect the circumcircle of ∆ ABC at A1 , B1 ,C1

o
respectively. AA1 ∩ B1C1 = P; B1C1 ∩ AC = Q; BB1 ∩ AC = R; IP = IR; QP = QR.
Prove that the triangle ∆ ABC is equilateral.

l.c
Hint: First note that, in any triangle ∆ ABC, AA1 is perpendicular to B1C1 , because,
∠B1 PA1 = ∠PC1C + ∠CC1 A1 + ∠C1 A1 P = B/2 + A/2 +C/2 = 90◦ . The given conditions
QP = QR, IP = IR imply that ∆ PIQ ∼ = ∆ RIQ, (SSS); ∠QRB = 90◦ ; B/2 + C = 90◦ ; also

ai
A + B +C = 180◦ ; hence A + B/2 = 90◦ ; A = C; hence ∆ APQ ∼ = ∆ B1 RQ, (AAS, since
∠APQ = B1 RQ = 90◦ , ∠AQP = ∠RQB1 and QP = QR); hence AQ = B1 Q; ∠QAB1 = ∠QB1 A;
B/2 = C/2; B = C; A = B = C; the triangle is equilateral.
gm
ri@
ha
sic
ra
na

7
17. (Sharygin, Problems in plane geometry, 1988, Page 228, Problem 58) A1 , B1 ,C1 are points on the
sides BC,CA, AB respectively, of ∆ ABC; A2 , B2 ,C2 are points on the sides B1C1 ,C1 A1 , A1 B1 re-
spectively, of ∆ A1 B1C1 . If AA1 , BB1 ,CC1 are concurrent and A1 A2 , B1 B2 ,C1C2 are concurrent,
show that AA2 , BB2 ,CC2 are concurrent.
Hint: First recall Ceva’s theorem: Three Cevians AA1 , BB1 and CC1 are concurrent at a point P
AC1 BA1 CB1
if and only if · · = 1.
C1 B A1C B1 A

m
Step 1: In standard notation [ ABC] denotes the area of ∆ ABC. Let AA1 , BB1 ,CC1 intersect at
AC1 [ACC1 ]
P; using CC1 as the Cevian, = , since these triangles have collinear bases and the
C1 B [BCC1 ]

o
same height. Hence

l.c
AC1 1/2.AC.CC1 . sin ACC1 AC. sin ACC1
= = .
C1 B 1/2.BC.CC1 . sinC1CB BC. sinC1CB

Step 2: Repeat Step 1 in ∆ AC1 B1 with Cevian AA2 ; we get

ai
B1 A2 [AA2 B1 ] 1/2.AA2 .AB1 . sin B1 AA2 sinCAA2 A2 B1 .AC1
= = ; hence = .
C1 A2 [AA2C1 ] 1/2.AA2 .AC1 . sinC1 AA2 gm sin A2 AB A2C1 .AB1
Change (A, B1 ,C1 , A2 ) to (B,C1 , A1 , B2 ) and then cyclically to (C, A1 , B1 ,C2 ) and multiply all
the three equations to get
   
sinCAA2 . sin ABB2 . sin BCC2 B1 A2 C1 B2 A1C2 AC1 BA1 CB1
= · · · · ·
sin BAA2 . sinCBB2 . sin ACC2 A2C1 B2 A1 C2 B1 B1 A C1 B A1C
ri@

By Ceva’s theorem, given, the product of the sine ratios on the left is equal to 1 and one of the
products on the right is 1; hence the other product on the right is also equal to 1; by the converse
of Ceva’s theorem we get the required result.
Note: Ceva’s theorem can be proved by repeating step 1 with AA1 and BB1 as Cevians:
AC1 AC. sin ACC1 BA1 BA. sin BAA1 CB1 CB. sinCBB1
ha

= ; = and = .
C1 B BC. sinC1CB A1C CA. sin A1 AC B1 A BA. sin ABB1
Multiplying these three equations we get
sic

AC1 . BA1 .CB1 sin ACC1 . sin BAA1 . sinCBB1


=
C1 B. A1C. B1 A sin A1 AC. sin B1 BA. sinC1CB
sin ACP sin BAP sinCBP AP BP CP
Then the sine rule simplifies the right to · · = · · =1
sin PAC sin ABP sin PCB CP AP BP
ra

18. If three internal angle bisectors of a convex quadrilateral ABCD are concurrent at X then all the
na

four angle bisectors are concurrent at X; ABCD is a circumscribed quadrilateral and X is the
centre of the inscribed circle of ABCD; AB +CD = AD + BC.
Hint: Suppose the angle bisectors of three angles A, B,C intersect at X; join DX;
draw XP, XQ, XR, XS perpendiculars from X on the sides AB, BC, CD, DA respectively. Trian-
gles XAP and XAS are congruent by AAS; hence XS = XP; similarly XP = XQ and XQ = XR;
hence XR = XS; ∆ XDR ∼ = ∆ XDS, (hypotenuse and one side); this gives XD bisects ∠ ADC.
Hence XP = XQ = XR = XS; X is the centre of the unique circumcircle of the cyclic quadrilat-
eral PQRS; this is the inscribed circle of ABCD, touching the sides at P, Q, R, S; AP and AS are
tangents to the circle through PQRS, hence AP = AS; similarly, BP = BQ; CQ = CR; DR = DS;
adding these we get AD + BC = AB +CD.

8
19. a) P is a point inside an equilateral triangle ∆ ABC. If PA = 3, PB = 4, PC = 5 find [ABC], i.e.,
the area of ∆ ABC. √
36 + 25 3
Ans: [ABC] =
4
b) P is a point inside an equilateral
√ triangle ∆ ABC. If PA = x, PB = y, PC = z show that the
3 2 3
area of ∆ ABC is [ABC] = (x + y2 + z2 ) + [x, y, z]
8 2
c) It is easier to construct a scalene triangle ∆ ABC having integer sides and an interior point P

m
at integral distances from all the three sides. For example,
AB = 19, BC = 22, CA = 8, with a point P such that PA = 4, PB = 17, PC = 6

o
d) (Open problem): Does there exist a square ABCD having integer sides, with a point P inside
the square such that PA, PB, PC, PD are all integers.

l.c
Hint: a) Take Q outside ∆ ABC such that ∠ABQ = ∠PBC and BQ = 4; i.e., rotate BA about
B (in the positive, i.e., anticlockwise direction) through ∠PBC and take BQ = BP = 4. Then
∆ ABQ ∼ = ∆ PBC, (SAS); AQ = PC = 5; ∠PBQ = ∠CBA = 60◦ ; PB = PQ = 4, hence ∆PBQ is

ai
equilateral; PQ = 4; hence AQ2 = 52 = 32 + 42 , hence ∠APQ = 90◦ ; √ ∠APB = 90 + 60 = 150◦ .
2 2 2
The cosine √ rule gives AB = AP √ + PB − 2AP.PB cos 150 = 25 + 12 3;
gm
3 36 + 25 3
[ABC] = (side)2 =
4 4
b) As before, take Q outside the equilateral triangle ∆ ABC such that AQ = PQ = x = AP;
∠QAB = ∠PAC = 60 − ∠BAP;
ri@

∆QAB ∼ = ∆PAC, (SAS); QB = PC = z; [APQ] + [BPQ] = [x, x, x] + [x, y, z]; hence


[APBQ] = [APB] + [AQB] = [APB] + [APC] = [x, x, x] + [x, y, z]; cyclically, we get
[BPC] + [CPA] = [z, z, z] +√[x, y, z] and [CPB] + [APB] = [y, y, y] + [x, y, z]; adding these three equa-
3 2
tions we get 2[ABC] = (x + y2 + z2 ) + 3[x, y, z], hence the answer.
4
In the case of an equilateral triangle with side a and PA = x, PB = y, PC = z, this is actually a
ha

symmetric formula, (Prithwijit De, 2008), (Christopher Bradley, Challenges in Geometry,


Oxford Univ. Press, 2005)
3(x4 + y4 + z4 + a4 ) = (x2 + y2 + z2 + a2 )2 .
sic

The geometric reason for the symmetry can be seen from the following construction: Given an
equilateral triangle ∆ ABC with integer sides and a point P which is at integer distances from
all the vertices; PA = x, PB = y, PC = z. We can construct an equilateral triangle on the side
ra

BP. There are two such triangles, since we can rotate BP, about B, (keeping B fixed), through
∠ 60◦ , either clockwise or anticlockwise. We get two equilateral triangles ∆ BPP0 and ∆ BPP00 ,
with side y; say P0 is opposite C; P00 is opposite A. Then
na

∆ BP00C ∼ = ∆ BPA, (SAS, BP00 = BP = y, BC = BA = a,


∠ P00 BC = 60 − ∠ PBC = ∠ PBA). Hence P00C = PA = x. Hence C is at integer distances a, z, x
from the vertices of ∆ BPP00 . Similarly, ∆ BAP0 ∼ = ∆ BCP; hence A is also at integer distances
0
a, x, z from the vertices of the triangle ∆ BPP . By symmetry, we see that there exist equilateral
triangles with base CP or AP with each side x or y or z and points (which may be outside the
triangles), which are at integer distances from all the three vertices. This explains the symmetry
of the formula (in all the four variables).
Fact: This equation has solutions in positive integers. The smallest example is an equilateral
triangle with each side 112 units and a point P inside the triangle at distances of 57, 65 and 73
from the vertices, (Hugh ApSimon,1990, David Wells, 1997, www.mathafou.free.fr); there are
other solutions, e.g., (a, x, y, z) = (112, 57, 65, 73), (147, 73, 88, 95), (185, 43, 147, 152),

9
(273, 97, 185, 208). In fact there are infinitely many integer solutions, e.g.,
m = 2(u2 − v2 ), n = u2 + 4uv + v2 , x = m2 + n2 , y = m2 − mn + n2 , z = m2 + mn + n2 ,
a = 8(u + v)(u − v)(u2 + uv + v2 ).
If P lies on a side BC of the equilateral triangle ∆ ABC of side a and PA = d, PB = m, PC = n,
then by Stewart’s theorem, na2 + ma2 = (m + n)(d 2 + mn);
m + n = a; d 2 = m2 + mn + n2 ; the last equation corresponds to 120◦ triangles with integer sides;
for example,

m
(m, n, d) = (3, 5, 7), (5, 16, 19), (7, 8, 13), (7, 33, 37) (9, 56, 61), etc.
20. In a convex quadrilateral ABCD, ∠BAC = 50◦ , ∠CAD = 60◦ , ∠CBD = 30◦ , ∠BDC = 25◦ ;

o
AC intersects BD at E. Find ∠AEB
Hint: ∠ BCD = 180 − 30 − 25 = 125◦ ; ∆BCD is an obtuse-angled triangle; hence its

l.c
circumcentre lies outside the triangle. Let O be the circumcentre of ∆ BCD; suppose if possible
O 6= A. O lies on the opposite side of BD as C; ∠BOC = 2∠BDC = 50; also given ∠BAC = 50;
hence OABC is a cyclic quadrilateral. Similarly, ∠COD = 2∠CBD = 60 = ∠CAD; hence OACD

ai
is also cyclic; similarly OBCD is also cyclic; but O is the circumcentre of ∆ BCD; hence O must
lie inside the circumcircle of ∆ BCD; this is a contradiction. Hence O = A; AB = AC = AD = r;
∠ABD = ∠ADB =
180 − 50 − 60
2
gm
= 35◦ ; ∠AEB = 60 + 35 = 95◦ .

21. Morley0 s theorem, (Frank Morley, Anglo-American, 1899): The triangle formed by the trisec-
tors of the angles of a triangle is equilateral.
ri@

Hint: (Naraniyengar, 1909): Let P, Q, R be the vertices opposite A, B,C respectively


 (ofthe pro-
180 − B
posed equilateral triangle). In ∆ AQC, sin AQC = sin (π − A/3 − B/3) = sin ;
  3 

C C
b sin 2R sin B sin
AQ AC 3 3
the sine rule gives  =  ; AQ = =  ....(1)
ha

  
C π −B π −B π −B
sin sin sin sin
3 3 3 3
sin 3x = 3 sin x − 4 sin3 x
sic

 
3 2
= 4 sin x − sin x
4
√ ! √ !
3 3
= 4 sin x + sin x − sin x
ra

2 2
= 4 sin x (sin 60 + sin x) (sin 60 − sin x)
na

  
60 + x 60 − x 60 + x 60 − x
= 4 sin x 2 sin cos 2 cos sin
2 2 2 2
   
60 + x 60 + x 60 − x 60 − x
= 4 sin x. 2 sin cos . 2. sin cos
2 2 2 2
= 4 sin x sin(60 + x) sin(60 − x); hence
   
B π +B π −B
sin B = 4 sin sin sin
3 3   3    
B C π +B
Substitute in eq (1); AQ = 8R sin sin sin
3 3 3
     
B C π +C
Similarly, AR = 8R sin sin sin
3 3 3

10
   
π +C π +C
sin sin
AR 3 sin(AQR) 3
=   ; =  ;
AQ π +B sin(ARQ) π +B
sin sin
3 3
A π + B π +C
Also, ∠ AQR + ∠ ARQ = π − = +
3 3 3
◦ C ◦ B
Hence, ∠ AQR = 60 + ; ∠ ARQ = 60 +
3 3

m
A B

Similarly, ∠ BRP = ∠CQP = 60 + ; ∠CPQ = ∠ ARQ = 60◦ + ;
3 3
C A B

o

∠ BPR = ∠ ARQ = 60 + ; ∠ ARB = 180 − − ; ◦
 3   3 3 
A B A B

l.c

∠ PRQ = 360 − 180 − − ◦ − 60 + ◦ ◦
− 60 + = 60◦
3 3 3 3
By symmetry, ∠ PQR = ∠ QPR = 60◦ .
Hence ∆ PQR is equilateral.

ai
22. a) P is a point inside a parallelogram ABCD. Prove that
gm 1
Area[PAB] + Area[PCD] = Area[PBC] + Area[PAD] = Area[ABCD]
2
b) P is a point inside a rectangle ABCD. Prove that PA2 + PC2 = PB2 + PD2
Hint: a) Draw EPF parallel to AD with E on AB and F on CD. The triangle ∆ BPC and the
parallelogram EBCF have a common base and equal heights. Hence [BPC] = 21 [EBCF].
ri@

Similarly, [APD] = 12 [AEFD]. Adding, we get Area[PBC] + Area[PAD] = 21 Area[ABCD].


Similarly, drawing a line through P parallel to AB we get
Area[PAB] + Area[PCD] = 21 Area[ABCD].
b) Draw EPF parallel to AD with E on AB and F on CD. Then
PA2 + PC2 = PE 2 + EA2 + PF 2 + FC2 = PE 2 + FD2 + PF 2 + EB2 ; hence
ha

(PE 2 + EB2 ) + (FD2 + PF 2 ) = PB2 + PD2


23. In ∆ ABC, ∠ A = 80◦ , ∠ B = 50◦ ; O is the circumcentre. The bisectors of angles ∠ ABO and
∠ ACO intersect CO and BO at X and Y respectively. Show that ∆ AXY is equilateral.
sic

Hint: First method: Let BX and CY intersect at P. ∠ ABX = ∠ XBO = ∠ ACY = ∠YCO = 20◦ ;
∠ OBX = ∠ OCX = 10◦ ; ∠ XOB = ∠Y OC = 20◦ ; ∆ ABP ≡ ∆Y BP, (AAS, since
∠ BAP = 40◦ ⇒ ∠ APB = 120◦ = ∠ APC = ∠ BPY ); hence AB = BY ;
ra

180 − 40
∠ BAY = ∠ BYA = = 70◦ ; ∠ AY X = 70 − 10 = 60◦ ; similarly, ∠ AXY = 60◦ . Hence
2
∆ AXY is equilateral.
na

Second method: ∆ BXC ≡ ∆ BYC, (AAS) ⇒ BY = CX, BX = CY ; also


OB = OC ⇒ BY − BO = CX −CO ⇒ OX = OY ⇒ ∠ OXY = ∠ OY X = 10◦ ;
∠ BYC = ∠ BXC = 140◦ ; XY k BC; BXYC is an isosceles trapezium; OX = OY = BX = CY ;
the cosine rule gives BC2 = AB2 + AC2 − 2 AB AC cos A. Hence
BC BY
BC2 = 2 AB2 (1 − cos 80◦ ) ⇒ BC = 2 AB sin 40◦ ; the sine rule in ∆ BYC gives ◦
= ,
sin 140 sin 30◦
hence BC = 2 BY sin 40◦ ; but sin 140◦ = sin(180 − 40)◦ = sin 40◦ ; hence AB = BY , as before.

11
24. Mollweide’s equations, (German, 1808, famous in geography for Mollweide’s area-preserving
projection of the earth as an ellipse): In any triangle ∆ ABC,
   
A−B A−B
cos sin
a+b 2 a−b 2
=   and =  
c C c C
sin cos
2 2

m
Hint: a) Let AD be the internal bisector of ∠ A, with D on BC. Then the sine rule in triangles
∆ ADB and ∆ ADC gives
AB BD AC DC
=   and =   ; but sin(ADC) = sin(180−ADB) = sin(ADB)

o
sin(ADB) A sin(ADC) A
sin sin
2 2

l.c
         
A B C C−B C−B B −C
= sin C + = sin C + 90 − − = sin 90 + = cos = cos ;
2 2 2 2 2 2
hence

ai
   
A B −C
(AB + AC) sin = (BD + DC) cos ; hence we get
 2  2
B −C gm
cos
b+c 2
=  
a A
sin
2
ri@

b) The sine rule a = 2R sin A, b = 2R sin B, c = 2R sinC can also be used directly:
     
A+B A−B A−B
2 cos sin sin
a − b 2R sin A − 2R sin B 2 2 2
= = =  
c 2R sinC C C C
2 sin cos cos
2 2
ha

25. If three internal angle bisectors of the angles in a convex quadrilateral intersect at a point then
the fourth internal angle bisector also passes through the same point.
sic

Hint: Let the bisectors of angles A, B, C intersect at X. Join DX. Drop perpendiculars
XP, XQ, XR, XS from X on the sides AB, BC, CD, DA respectively. The triangles SXA and
PXA are congruent (AAS); hence XP = XQ; similarly ∆ PXB ∼ = ∆ QXB =⇒ XP = XQ;
∆ QXC ∼ = ∆ RXC =⇒ XQ = XR; hence XP = XS; hence ∆ RXD ∼ = ∆ SXD, (Hypotenuse, side);
ra

∠ XDR = ∠ XDS; hence XD bisects ∠ D


26. (INMO 2000) The incircle of ∆ ABC touches BC, CA, AB at K, L, M respectively. The line
na

through A parallel to LK intersects MK at P. The line through A parallel to MK intersects LK


at Q. Show that PQ bisects AB and AC.
Hint: Tangents to the incircle from the point B must be of equal lengths; BK = BM,
∠ BKM = ∠ BMK; MK k AY ⇒ ∠ BYA = ∠ BAY ; AB = AY ; AB − BM = BY − BK; AM = KY .
Similarly, tangents from A to the incircle must be equal. AM = AL; AMKY is an isosceles
trapezium; AM = AL = KX = KY ; K is the midpoint of XY ; in triangle ∆ AXY , KP k YA;
P = midpoint XA; similarly, Q = midpoint YA; PQ k XY k BC. Similarly, in ∆ ABX,
P = midpoint XA and PZ k BC. Hence Z is the midpoint of AB; similarly, W is the midpoint of
AC.

12
27. (INMO 2000),√ (Excursions in mathematics, printing◦ mistake) In a convex quadrilateral◦ PQRS,
PQ = RS, ( 3 + 1)QR = SP, ∠ RSP − ∠ SPQ = 30 . Prove that ∠ PQR − ∠ QRS = 90 .
Hint: Using areas, [PQRS] = [PQR]√+ [RSP] = [QRS] + [SPQ]; let PQ = p; QR = q;
RS = r; SP = s; given, p = r and ( 3 + 1)q = s;
√ + 1/2rs sin RSP = 1/2qr sin√QRS + 1/2ps sin SPQ;
1/2pq sin PQR
sin PQR + ( 3 + 1) sin RSP = sin QRS + ( 3 + 1) sin SPQ;

sin PQR − sin QRS = ( 3 + 1)(sin SPQ − sin RSP);

m
√ −30◦
       
PQR + QRS PQR − QRS SPQ + RSP
cos . sin = ( 3 + 1) cos . sin ;
2 √2 2 2

o
3−1
sin(15◦ ) = sin(45◦ − 30◦ ) = √ ; the cosine terms on the two sides are negatives of each
2 2

l.c
other,   √ √
PQR − QRS ( 3 + 1)( 3 − 1) 1
since the angles are supplementary; sin = √ =√ ;
2 2 2 2

ai
PQR − QRS ◦ ◦
= 45 ; ∠ PQR − ∠ QRS = 90
2
gm
28. I is the incentre of ∆ ABC; AI, BI, CI are extended to intersect the circumcircle of the triangle at
D, E, F respectively. Lines through I parallel to the sides BC, CA, AB intersect EF, FD, DE at
K, L, M respectively. Show that K, L, M are collinear.
180 − A
Hint: BIC = 180−B/2−C/2 = 180− = 90+A/2; ∠ EIF = 90+A/2; BI = r csc B/2;
2
ri@

BI DI DI r
BD = CD = DI; = = ; DI = ;
sin C sin (A/2 + B/2) cos C/2 2 sin B/2 sin C/2
KF KI
∠ KFI = ∠ KEI + ∠FIE = C/2 + 90 + A/2 = 180 − B/2; = ;
sin C/2 sin (180 − B/2)
KF sin C/2
ha

= ; similarly,
KI sin B/2
EK cos (C/2 + A/2) sin B/2 EK EK/KI sin2 B/2
= = ; hence = = ; cyclically, we get
KF/KI sin2 C/2
sic

KI sin C/2 sin C/2 KF


FL sin2 C/2 DM sin2 A/2 EK FL DM
= 2
; = 2 ; multiplying, we get = −1;
LD sin A/2 ME sin B/2 KF LD ME
ra

hence by Menelaus’ theorem, K, L, M are collinear.


29. ‘Outer Napoleon Triangle’ Let the points A0 , B0 , C0 be taken outside the triangle ∆ ABC oppo-
site the vertices A, B, C respectively, such that the outer triangles ∆ BA0C, ∆CB0 A, ∆ AC0 B are
na

equilateral. Then the centroids G1 , G2 , G3 of these triangles form an equilateral triangle with
the same centroid G as ∆ ABC.
Hint: The vertices A = z1 , B = z2 , C = z3 can be taken as complex numbers in the plane. For
convenience we write z1 = a, z2 = b and z3 = c. They form an equilateral triangle if and only if
the points are not collinear and the three angles are ∠ 60◦ , i.e., each side is obtained by rotating
another side through an angle of ∠60◦ , (either clockwise or anticlockwise); we know that anti-
clockwise rotation through an angle θ corresponds to multiplication by eiθ ; hence we have
b − a = (c − a) e±iπ/3 ; cube both sides and use eiπ = −1; (b − a)3 + (c − a)3 = 0 and b − a is
not equal to a − c, (otherwise B is the midpoint of AC and the three points would be collinear).
Hence this is equivalent to (b − a)2 − (b − a)(c − a) + (c − a)2 = 0; a2 + b2 √ + c2 = ab + bc + ca,
b + c ± i 3 (b − c)
symmetric in a, b, c; solving this as a quadratic in a, we get a = ;
2
13
√ √
−1 + i 3
a + ω 2 b + ωc = 0; (ω = is the complex cube root of unity; ω 2 = −1−i2
3
;
2
1 + ω + ω 2 = 0 and ω 3 = 1); a, b, c can be permuted as we wish.
For Napoleon’s theorem, ∆ AC0 B is equilateral ⇒ a + ωc0 + ω 2 b = 0; hence c0 = −ω 2 a − ωb;
similarly, a0 = −ω 2 b − ωc; b0 = −ω 2 c − ωa; hence the centroids are
(1 − ω 2 )b + (1 − ω)c (1 − ω 2 )c + (1 − ω)a (1 − ω 2 )a + (1 − ω)b
g1 = ; g2 = ; g3 = ;
3 3 3

m
hence g1 + ωg2 + ω 2 g3 = 0, i.e., ∆ G1 G2 G3 is equilateral; further, since ω + ω 2 = −1, we get
g1 + g2 + g3 a + b + c
= ; hence the centroid of ∆ G1 G2 G3 is G.
3 3

o
30. (USAMO 2000) If P, Q, R are  the points of contact
 of the incircle of ∆ ABC with the sides of the

l.c
1 1 1 3 6
triangle of inradius r and 5 + + − = , prove that AB = AC.
AP BQ CR s−a r
r
A r (s − a)(s − b)(s − c)

ai
Hint: Use tan = and r = . Put s − a = x, s − b = y, s − c = z.
2r s − a s
2 5 5 x+y+z gm
+ + =6 ; simplifying and completing the square we get
x y z xyz
sin A 8
(2yz − 4xy − 4xz)2 + 3(xz − xy)2 = 0; y = z, b = c; 5a = 8b = 8c; = ;
sin B 5
4 64b2 −7 4
A + 2B = 180, cos B = ; a2 = > b2 + c2 ; hence ∠ A > 90◦ ; cos A = ; tan A/2 = .
ri@

5 25 25 3
25
The triangle is similar to ∆ ABC with AB = AC = 5, BC = 8, altitude AD = 4; R =
6
31. a) Bretschneider0 s theorem, (German, 1842): Let a, b, c, d be the lengths of the sides
AB, BC,CD, DA respectively, of a quadrilateral ABCD. Let m and n be the lengths of the
ha

diagonals AC and BD respectively. Then


m2 n2 = a2 c2 + b2 d 2 − 2abcd cos(A +C); mn ≤ ac + bd
b) Ptolemy0 s theorem, (Greek astronomer, Alexandria, Egypt, 150 AD) : ABCD, (in order), is
sic

a cyclic quadrilateral if and only if


AB.CD + AD. BC = AC. BD
c) For any three angles A, B, C, prove the identity
ra

sin(A + B) sin(B +C) = sin A sinC + sin B sin(A + B +C)


Using this identity deduce Ptolemy’s theorem.
d) Euler’s identity for vectors: For any four points in the plane A, B, C, D in the plane, we have
na

−→ −→ − → −→ −→ − →
AB. CD + AC. DB + AD. BC = 0
Deduce the inequality: AC. BD ≤ AB.CD + AD. BC; deduce Ptolemy’s theorem.
e) Prove Ptolemy’s theorem using the cosine rule.
f) Prove Ptolemy’s theorem by projecting the cyclic quadrilateral on a straight line.
Hint: a) Construct a triangle ∆ AKB with K outside the quadrilateral ABCD such that
∠ ABK = ∠CAD and ∠ BAK = ∠ ACD. Similarly, take a point M outside the quadrilateral
ABCD such that in ∆ AMD, ∠ DAM = ∠ BCA and ∠ ADM = ∠CAB. Then
we have similar triangles, ∆ ACD ∼ ∆ BAK; ∆ ADM ∼ ∆CAB.
ac ad ad
AK/BA = CD/AC; AK = ; BK/BA = AD/AC; BK = ; AB/CA = DM/AD; DM = ;
m m m
bd
CB/CA = AM/AD; AM = ; hence BK = DM.
m
14
Further, in ∆ ABD,
∠ KBD + ∠ MDB = A + B + D = 180◦ ; hence, BK k DM. Hence KBDM is a parallelogram.
KM = BD = n. Now use the cosine rule in ∆ ABK; MK 2 = AM 2 + AK 2 − 2AM.MK. cos (MAK);
b2 d 2 a2 c2 2abcd
n2 = 2 + 2 − cos (A +C); hence m2 n2 = a2 c2 + b2 d 2 − 2abcd cos (A +C)
m m m2
We know that −1 ≤ cos θ ≤ 1, for all angles θ ; hence the previous formula gives
|ac − bd| ≤ mn ≤ ac + bd

m
b) We know that ABCD is a cyclic quadrilateral if and only if ∠ A + ∠C = 180◦ ; use
cos(180◦ ) = −1 in Bretschneider’s theorem to get m2 n2 = (ac + bd)2 ; hence mn = ac + bd.
Or, the previous construction gives ∠ MAK = ∠ A + ∠C = 180◦ ; MAK is a straight line. As

o
ac bd
before KBDM is a parallelogram; KM = BD; AK + AM = n; + = n; ac + bd = mn. This

l.c
m m
is Ptolemy’s theorem.
c) Consider sin((A + B) +C) sin B − sin(A + B) sin(B +C)
= sin B [sin(A + B) cosC + cos(A + B) sinC] − sin(A + B) [sin B cosC + cos B sinC]

ai
= − sinC[sin(A + B) cos B − cos(A + B) sin B]
= − sinC sin(A + B − B) = − sinC sin A; hence we get the trigonometric identity
gm
sin(A + B) sin(B +C) = sin A sinC + sin B sin(A + B +C)
To deduce Ptolemy’s theorem: Let ABCD be a cyclic quadrilateral with vertices in order, in the
anticlockwise direction. Let R be the radius of the circumcircle of ABCD. Let
∠ BDA = ∠ BCA = α; ∠CAB = ∠CDB = β ; ∠ DAC = ∠ DBC = γ.
ri@

The sine rule gives AB = 2R sin α; CD = 2R sin γ; AD = 2R sin(α + β + γ); BC = 2R sin β ;


AB
AC = 2R sin(α + β ); BD = 2R sin(β + γ). Substitute sin α = , etc., in the previous trigono-
2R
metric identity
sin (α + β ) sin (β + γ) = sin α sin γ + sin β sin (α + β + γ) to get Ptolemy’s theorem
AC. BD = AB.CD + AD. BC
ha


→ −→ − → −→ −→ − →
d) To prove Euler’s identity for vectors: AB. CD + AC. DB + AD. BC = 0
Consider the left side = (~b −~a).(d~ −~c) +~c −~a).(~b − d) ~ + (d~ −~a).(~c −~b)
sic

= ~b.(d~ −~c) −~a.(d~ −~c) +~c.(~b − d)~ −~a.(~b − d)~ + d.(~


~ c −~b) −~a.(~c −~b)
= ~b.d~ −~b.~c −~a.d~ +~a.~c +~c.~b −~c.d~ −~a.~b +~a.d~ + d.~
~ c − d.
~ ~b −~a.~c +~a.~b
= 0, (since the dot product is commutative, i.e., ~u.~v =~v.~u, for all vectors ~u and ~v).
Similarly, for any four complex numbers a, b, c, d, we have the identity
ra

(a − b)(c − d) + (a − c)(d − b) + (a − d)(b − c) = 0


⇒ |(a − c)(b − d)| = |(a − b)(c − d) + (a − d)(b − c)| (1)
na

The triangle inequality gives |z + w| ≤ |z| + |w|


⇒ |(a − c)(b − d)| ≤ |(a − b)(c − d)| + |(a − d)(b − c)|
⇒ AC. BD ≤ AB.CD + AD.BC, for any four points in the plane.
z1
Equality holds in the triangle inequality if and only if the ratio is a positive real number.
z2
Hence equality holds in equation (1) if and only if the cross-ratio satisfies the condition
a−b
(a − b)(c − d)
∈ R+ ⇔ a − d ∈ R−
(a − d)(b − c) c−b
c−d
   
a − b c − b
⇔ arg − arg ≡ π mod 2π
a−d c−d

15


[ arg(a − b) represents the angle needed to reach the  vector  BA starting from the positive real
a−b −
→ −→
axis and turning in the anti-clockwise direction; arg is the angle between AB and AD]
a−d
⇔ A and B lie on opposite sides of BD and ∠ BAD + ∠ BCD = 180◦ . Hence equality holds in
Ptolemy’s inequality if and only if ABCD is a cyclic quadrilateral, (or these points are collinear).
e) Ptolemy’s theorem using the cosine rule: Let the lengths of the sides be AB = a, BC = b,
CD = c, DA = d in the cyclic quadrilateral ABCD.

m
We can calculate the lengths of the diagonals, AC and BD using the cosine rule. First consider
∆ ABC; we get AC2 = AB2 + BC2 − 2AB BC cos B;

o
then the cosine rule in ∆ ADC gives AC2 = AD2 + DC2 − 2AD DC cos D;
but cos D = cos(π − B) = − cos B; hence, multiply the first equation by cd, the second equation

l.c
by ab and add so that cos B is eliminated; we get
(ab + cd)AC2 = (ad + bc)(ac + bd). Hence
r r
(ac + bd)(ad + bc) (ac + bd)(ab + cd)

ai
AC = . Similarly, we get BD = .
ab + cd ad + bc
Multiplying these lengths gives Ptolemy’s theorem: AC. BD = AB.CD + AD. BC
gm
f) Ptolemy’s theorem by projection: Take a point O on the circumcircle of radius R, outside
the given cyclic quadrilateral ABCD (with vertices labelled in order in the anticlockwise direc-
tion). Let l be a straight line outside the quadrilateral such that l intersects the line segments
OA, OB, OC, OD at A0 , B0 , C0 D0 , respectively, (in order). For these four collinear points it is
ri@

easy to verify Euler’s identity:


b(d − c) + d(c − b) = bd − bc + cd − bd = c(d − b); shift the origin to a, i.e., change b, c, d to
b − a, c − a, d − a respectively. We get (b − a)(d − c) + (d − a)(c − b) = (c − a)(d − b). This is
Euler’s identity for collinear points.
(b0 − a0 )(d 0 − c0 ) + (d 0 − a0 )(c0 − b0 ) = (c0 − a0 )(d 0 − b0 ). Now,
ha

−−→ −−→ −−→ −−→ −−→ −−→


∠ (A0 B0 , C0 D0 ) = 0; ∠ (A0 D0 , B0C0 ) = 0; ∠ (C0 A0 , B0 D0 ) = π. Hence,
−− → −−→ −−→ −−→ −−→ −−→
A0 B0 . C0 D0 + A0 D0 . B0C0 = C0 A0 . B0 D0 ........................(1)
sic

Let p be the perpendicular distance from O to l. Multiply both sides by 1/4 p2 to get
(1/2 p A0 B0 )(1/2 pC0 D0 ) + (1/2 p A0 D0 )(1/2 p B0C0 ) = (1/2 pC0 A0 )(1/2 p B0 D0 )
1/2 p (A0 B0 ) = area [OA0 B0 ] = 1/2 OA0 OB0 sin(A0 OB0 ) = 1/2 OA0 OB0 sin(AOB) etc .
Cancelling 1/4 OA0 OB0 OC0 OD0 from all the terms we get the trigonometric form of Ptolemy’s
ra

theorem:
sin(AOB) sin(COD) + sin(AOD) sin(BOC) = sin(COA) sin(BOD)
na

Now multiply both sides by 4R2 and use the sine rule, AB = 2R sin(AOB) etc. to get Ptolemy’s
theorem
(AB)(CD) + (AD)(BC) = (CA)(BD)

16
32. Feuerbach0 s theorem, (German, 1822): The nine-point circle of a triangle is tangent to the
incircle and the three ex-circles of the triangle.
Hint: (Coxeter and Greitzer) The nine-point circle O9 passes through the midpoints P1 , P2 , P3
of the sides BC,CA, AB of the triangle ∆ ABC, respectively. The in-circle, denoted by (I, r),
of radius r, touches the sides BC,CA, AB at A1 , A2 , A3 respectively. The ex-circle opposite A,
denoted by Ca , with centre Ia , of radius ra , touches the sides BC,CA, AB at B1 , B2 , B3 respec-
tively. In standard notation, BC = a, CA = b, AB = c,

m
AB2 = AB3 = s, CB1 = CB2 = s − b; BB1 = BB3 = s − c; BC1 = BA1 = s − b;
If the triangle is equilateral, the centroid coincides with the orthocentre and the incentre; nine-

o
point circle coincides
√ 2 with the incircle, which touches√ the ex-circle at the midpoints of the sides.
3a 3a ∆ a ∆ a 3

l.c
s = ;∆ = ; r = = ; ra = = .
2 4 s 2 s−a 2
Suppose the triangle is non-equilateral; assume c > b.
CA1 = CB1 = s − c; AB1 = AC1 = s − a. Let P1 be the mid point of BC; BB1 = CA1 = s − b;

ai
a c−b
P1 is the midpoint of B1 A1 ; B1C1 = c − b; B1 P1 = P1 A1 = | − (s − b)| = . Let A0 be the
2 2
gm ab a ab a(c − b)
point of intersection of BC and IIa ; A0C = ; P1 A0 = P1C − A0C = − = .
b+c 2 b+c 2(b + c)
a
Let D be the foot of the altitude AD from A on BC. Then P1 D = P1C −CD = − b cosC,
2
a b(a2 + b2 − c2 ) c2 − b2
P1 D = − ; hence P1 D = . Hence we get
ri@

2 2ab 2a
(c − b)2
P1 D. P1 A0 = (1)
4
The incircle I and the ex-circle Ia are disjoint. Hence they have must have two external and two
internal common tangents. The external common tangents are ABB3 and ACB2 , intersecting at
A which is on the line of centres IIa ; BC is one internal common tangent. Let B0C0 be the other
ha

internal common tangent, with B0 on AC and C0 on AB. Then BC intersects B0C0 on the line of
centres, i.e., at A0 .
sic

Invert with respect to the circle on B1 A1 as diameter, i.e., the circle with centre P1 and radius
c−b (c − b)2
ρ = P1 A1 = . Then equation (1) says P1 D. P1 A0 = = ρ 2 , i.e.,
2 4
A0 and D are inverses of each other. (2)
ra

(Or, on the line AIA0 Ia , A0 divides Ia I internally in the ratio ra : r; A divides Ia I externally in the
same ratio ra : r; hence A and A0 are harmonic conjugates; hence the projections of Ia , I, A0 , A
na

on BC also form a harmonic range, i.e., B1 , A1 , A0 , D form a harmonic range.


a) The incircle (I, r) touches BC at A1 ; hence the inverse I −1 is a circle which touches
(BC)−1 = BC at (A1 )−1 = A1 , (since (P1 A1 )2 = ρ 2 ); hence I −1 = I, (i.e., I is orthogonal to the
circle of inversion).
b) The excircle (Ia , ra ) touches BC at B1 ; hence the inverse Ia−1 is a circle which touches
(BC)−1 = BC at (B1 )−1 = B1 , (since (P1 B1 )2 = (P1 A1 )2 = ρ 2 ); hence Ia−1 = Ia , (i.e., Ia is
orthogonal to the circle of inversion).
c) To prove that (B0C0 )−1 is the nine-point circle O9 :
P1 , P2 , P3 are the midpoints of the sides BC, CA, AB respectively. Let B0C0 intersect P1 P2 at S.
1 c B0 P2 B0 A
Then P1 P2 = AB = ; P1 P2 k BA ⇒ SP2 k C0 A; ∆ B0 SP2 ∼ ∆ B0C0 A; = 0 ;
2 2 SP2 CA

17
b
B0 A − P2 A BA BA − P2 A BA c c −
= 0 ; = = ; 2 = c ; SP = b(2c − b) ;
2
SP2 CA SP2 CA b SP2 b 2c
c b(2c − b) (c − b)2
P1 S = P1 P2 − SP2 = = = ; hence
2 2c 2c
(c − b)2 c (c − b)2
P1 S. P1 P2 = . = = ρ 2 ⇒ S and P2 are inverses of each other. (3)
2c 2 4

m
B0C0 is a straight line; hence its inverse, (B0C0 )−1 must be a circle passing through the centre of

o
inversion, P1 .
Also B0C0 passes through S and A0 . hence (B0C0 )−1 is a circle passing through S−1 = P2 and

l.c
A0−1 = D, from equations (2) and (3).
Hence (B0C0 )−1 is a circle passing through P1 , P2 and D, i.e., (B0C0 )−1 is the incircle of the
triangle.
Now B0C0 is tangent to I and Ia ; since inversion preserves angles, it also preserves tangency,

ai
i.e., (B0C0 )−1 is tangent to I −1 = I and Ia−1 = Ia . Hence the nine-point circle is tangent to the
incircle and the three excircles of the triangle.
gm
33. Lester0 s theorem, (June A. Lester, 1996; discovery by computer search, proof using cross-
ratios): In any scalene triangle, the two Fermat points, the circumcentre and the nine-point centre
form a cyclic quadrilateral.
ri@

Hint: (Nikolai Beluhov, 2012): First, let us recall the locations of the Fermat points. Con-
struct equilateral triangles with circumcentres Ta , Tb , Tc externally on the sides of the trian-
gle, BC, CA, AB respectively. The circumcircles of these triangles, denoted by δa , δb , δc , inter-
sect at the first Fermat point denoted by F1 . Similarly, the equilateral triangles with centres
Sa , Sb , Sc constructed internally on the sides of the triangle, BC, CA, AB respectively, have cir-
ha

cumcircles, denoted by σa , σb , σc , intersecting at the second Fermat point denoted by F2 .


Sa Ta is the perpendicular bisector of the side BC; similarly, Sb Tb and Sc Tc are perpedicular bi-
sectors of CA and AB respectively. Hence Sa Ta , Sb Tb , Sc Tc intersect at the circumcentre O. The
sic

triangles Ta Tb Tc and Sa Sb Sc are both equilateral with the same centroid G as ∆ ABC.
ra
na

18

Вам также может понравиться