Received 25/10/11

A NEW SHARPENING OF THE ERDÖS-MORDELL

INEQUALITY AND RELATED INEQUALITIES

JIAN LIU

Abstract. In this paper we point out a new sharpening of the Erdös-Mordell

inequality. We also prove the weighted generalization of this result and its anal-
ogous. Some related conjectures for the polygon are put forward.

1. Introduction
The famous Erdös-Mordell inequality in geometric inequalities states the follow-
ing: Let P be an interior point of a triangle ABC. Let R1 , R2 , R3 be the distances
from P to the vertices A, B, C, and let r1 , r2 , r3 be the distance from P to the sides
BC, CA, AB, respectively. Then
R1 + R2 + R3 ≥ 2(r1 + r2 + r3 ), (1.1)
with equality if and only if △ABC is equilateral and P is its center. This equality
was conjectured by Erdös [1] in 1935 and was first proved by Mordell [2] in the
same year. Since then, a lot of proofs have been given, various generalizations,
refinements and variations were also studied (see [3]-[29]). We recall here some
related results.
In [2], Barrow proved the stronger inequality:
R1 + R2 + R3 ≥ 2(w1 + w2 + w3 ), (1.2)
where w1 , w2 , w3 are the internal angle-bisectors of ∠BP C, ∠CP A, ∠AP B respec-
tively.
The authors of the monograph [13] gave the following weighted generalization:
x2 R1 + y 2R2 + z 2 R3 ≥ 2(yzr1 + zxr2 + xyr3 ), (1.3)
with equality if and only if P is the circumcenter of △ABC and x : y : z = sin A :
sin B : sin C, where A, B, C denote the angles of △ABC.
In a recent paper [31], the author pointed out the following refinements:
q 
1
q q
2 2 2
R1 + R2 + R3 ≥ a + 4r1 + b + 4r2 + c + 4r3
2 2 2
2
   
c b a c  b a
≥ + r1 + + r2 + + r3
b c c a a b
p
≥ 2 ha r1 + hb r2 + hc r3
≥ 2(r1 + r2 + r3 ), (1.4)

2000 Mathematics Subject Classification. 51M16.

Key words and phrases. triangle, interior point, Erdös-Mordell inequality, conjecture.
1
2 JIAN LIU

where a, b, c denote the sides BC, CA, AB respectively and ha , hb , hc the corre-
sponding altitudes of △ABC.
The purpose of this paper is to point out a new sharpening of the Erdös-Mordell
inequality. In fact, we will establish a more general weighted inequality and prove
another similar result.
2. Main results and their proofs
For Erdös-Mordell inequality (1.1), we have the following sharpening:
Theorem 2.1. For any interior point P of △ABC, we have
p p p
(r2 + r3 )R1 + (r3 + r1 )R2 + (r1 + r2 )R3 ≥ 2(r1 + r2 + r3 ), (2.1)
with equality if and only if △ABC is equilateral and P is its center.
p
Note that R1 + r2 + r3 ≥ 2 R1 (r2 + r3 ) and its two analogues, we know that
1 p p p
(R1 + R2 + R3 ) + r1 + r2 + r3 ≥ (r2 + r3 )R1 + (r3 + r1 )R2 + (r1 + r2 )R3 .
2
From this and (2.1), Erdös-Mordell inequality (1.1) follows at once. Also, according
to (1.1) and above inequality we see that
p p p
R1 + R2 + R3 ≥ (r2 + r3 )R1 + (r3 + r1 )R2 + (r1 + r2 )R3 , (2.2)
which shows that (2.1) is a sharpening of the Erdoös-Mordell inequality.
Actually, we have the following general result:
Theorem 2.2. Let x, y, z be arbitrary real numbers and let u, v, w be positive num-
bers. Then for any interior point P of △ABC holds:
p p p
x2 (v + w)R1 + y 2 (w + u)R2 + z 2 (u + v)R3
√ √ √
≥ 2 (yz ur1 + zx vr2 + xy wr3 ) , (2.3)
with equality if and only if x = y = z, u = v = w, △ABC is equilateral and P is
its center.
In (2.3), put u = r1 , v = r2 , w = r3 , we get
p p p
x2 (r2 + r3 )R1 +y 2 (r3 + r1 )R2 +z 2 (r1 + r2 )R3 ≥ 2(yzr1 +zxr2 +xyr3 ), (2.4)
which is the weighted generalization of (2.1).
In (2.3) put x = y = z = 1, then we have the following weighted inequality with
three positive real numbers u, v, w:
p p p √ √ √
R1 (v + w) + R2 (w + u) + R3 (u + v) ≥ 2( ur1 + vr2 + wr3 ). (2.5)
Clearly, inequality (2.1) is a special case of this inequality.
When the author obtained inequality (2.3), the following analogues was also
found.
Theorem 2.3. Let x, y, z be arbitrary real numbers and let u, v, w be positive num-
bers. Then for any interior point P of △ABC holds:
r r r
2 v+w 2 w+u 2 u+v √ √ √
x R1 + y R2 + z R3 ≥ 2 (yz r1 + zx r2 + xy r3 ) .
u v w
(2.6)
A NEW SHARPENING OF THE ERDÖS-MORDELL INEQUALITY AND RELATED INEQUALITIES
3

with equality if and only if x = y = z, u = v = w, △ABC is equilateral and P is

its center.
In particular, we have
r r r
v+w w+u u+v √ √ √
R1 + R2 + R3 ≥ 2 ( r1 + r2 + r3 ) . (2.7)
u v w
In (2.6) put u = R1 , v = R2 , w = R3 , then
p p p √ √ √
x2 R2 + R3 + y 2 R3 + R1 + z 2 R1 + R2 ≥ 2 (yz r1 + zx r2 + xy r3 ) . (2.8)
which is posed as a conjecture by the author in [29].
To prove our theorems, we will use the weighted Erdös-Mordell inequality (1.3)
and the following useful lemma:
Lemma 2.1. Let α1 , β1 , γ1 , λ1 , µ1 , ν1 , α2 , β2 , γ2 , λ2 , µ2 , ν2 be positive real numbers.
If the following inequality:
α1 x2 + β1 y 2 + γ1 z 2 > λ1 yz + µ1 zx + ν1 xy, (2.9)
α2 x2 + β2 y 2 + γ2 z 2 > λ2 yz + µ2 zx + ν2 xy, (2.10)
hold for any real numbers x, y, z. Then
√ p √ p √ √
α1 α2 x2 + β1 β2 y 2 + γ1 γ2 z 2 > λ1 λ2 yz + µ1 µ2 zx + ν1 ν2 xy. (2.11)
In [32], the author established “ The Composite Theorem ” of the ternary qua-
dratic inequality by using the classical Hölder inequality. The above lemma is only
a consequence of it.
Next, we prove the previous theorems at the same time.
Proof. We first prove Theorem 2.2. For arbitrary real numbers x, y, z and positive
real numbers u, v, w, we have
u(y − z)2 + v(z − x)2 + w(x − y)2 ≥ 0,
which implies that
(v + w)x2 + (w + u)y 2 + (u + v)z 2 ≥ 2(yzu + zxv + xyw), (2.12)
with equality if and only if x = y = z.
Now, applying Lemma 2.1 to inequality (1.3) and (2.12), we obtain the inequality
of Theorem 2.2 at once. From [32], it is easy to know that the equality in (2.11)
holds if and only if both (2.9) and (2.10) become equality and α1 : α2 = β1 : β2 : γ1 :
γ2 . According to this conclusion and the equality conditions of (1.3) and (2.12), we
conclude easily that the equality in (2.3) occurs if and only if x = y = z, u = v = w,
△ABC is equilateral and P is its center. This completes the proof of Theorem
2.2.
Because Theorem 2.1 is an evident corollary of Theorem 2.2, hence Theorem 2.1
is also proved.
By replacing u → vw, v → wu, w → uv in (2.12) and then dividing both sides
by uvw, we get again:
v+w 2 w+u 2 u+v 2
x + y + z ≥ 2(yz + zx + xy), (2.13)
u v w
with equality if and only if x : y : z = u : v : w.
4 JIAN LIU

Applying Lemma 2.1 to (1.3) and (2.13), we immediately obtain inequality (2.6),
and we easily know equality in (2.6) is just the same as in (2.3). The proof of the
Theorem 2.2 is completed. 

The author has failed to prove Theorem 2.1 directly for many times. It keeps to
be an interesting problem.
Remark 2.1. According to the generalization of Lemma 2.1 (see [32]), we can give
the more general result of Theorem 2.2.
Remark 2.2. Since we have the known quadratic inequality:
x2 R1 + y 2 R2 + z 2 R3 ≥ 2(yzw1 + zxw2 + xyw3), (2.14)
which similar to (1.3). By the same way to prove (2.3), we also have the following
inequality:
p p p
x2 (v + w)R1 + y 2 (w + u)R2 + z 2 (u + v)R3
√ √ √
≥ 2 (yz uw1 + zx vw2 + xy ww3 ) , (2.15)
which is similar to Theorem 2.2. Particularly, we have the following stronger
version of Barrow’s inequality:
p p p
(w2 + w3 )R1 + (w3 + w1 )R2 + (w1 + w2 )R3 ≥ 2(w1 + w2 + w3 ). (2.16)
Remark 2.3. From the proof of Theorem 2.3, we can see that the following inter-
esting conclusion is valid:
If p1 , p2 , p3 , q1 , q2 , q3 are positive and the following inequality:
p1 x2 + p2 y 2 + p3 z 2 ≥ q1 yz + q2 zx + q3 xy. (2.17)
holds for any real numbers x, y, z. Then holds:
√ √ √ √ √ √
x2 p2 + p3 + y 2 p3 + p1 + z 2 p1 + p2 ≥ 2( q1 yz + q2 zx + q3 xy). (2.18)
Form this and known inequality (see [33])
x2 Ra + y 2Ra + z 2 Rc ≥ 2(yzr1 + zxr2 + xyr3 ) (2.19)
where Ra , Rb , Rc are the circumradius of △P BC, △P CA, △P AB. We know again
that
p p p √ √ √
x2 Rb + Rc +y 2 Rc + Ra +z 2 Ra + Rb ≥ 2 (yz r1 + zx r2 + xy r3 ) , (2.20)
which was posed by the author in [29].
Remark 2.4. If we apply geometric transformations (see e.g. [10],[13]) to our
theorems, then we can get some new results. For example, make transformation by
Theorem 2.2 we can get
p p p
x2 (v + w)Ra + y 2 (w + u)Rb + z 2 (u + v)Rc
√  p p p 
≥ 2 yz uR1 + zx vR2 + xy wR3 . (2.21)
A NEW SHARPENING OF THE ERDÖS-MORDELL INEQUALITY AND RELATED INEQUALITIES
5

3. Some conjectures
In this section, we will propose some conjecture inequalities for the polygon.
In 1953, L.Fejes Tóth [4] conjectured Erdös-Mordell inequality can be generalized
to the polygon:
n n
X  πX
Ri ≥ sec ri . (3.1)
i=1
n i=1
where Ri are the distances from the interior point P to the vertices Ai (i =
1, 2, · · · , n) and ri the distances from P to the sides Ai Ai+1 (i = 1, 2, · · · , n.An+1 =
A1 ). This generalization has first proved by N.Ozeki [7] in 1957. Here, we give a
stronger conjecture which generalizes Theorem 2.1. Namely,
Conjecture 3.1. Let P be an arbitrary interior point of the polygon A1 A2 · · · An ,
then we have n r n
X p πX
(ri + ri+1 )Ri+1 ≥ 2 sec ri , (3.2)
i=1
n i=1
where An+1 = A1 , rn+1 = r1 , Rn+1 = R1 .
Clearly, if the above conjecture holds true, then it would be showed easily that
(3.1) follows from (3.2) by using AM-GM inequality.
Inequality (2.1) and the refinements (1.4) of the Erdös-Mordell inequality prompt
the author to conjecture that
q q q
a + 4r1 + b + 4r2 + c2 + 4r32
2 2 2 2
hp p p i
≥2 (r2 + r3 )R1 + (r3 + r1 )R2 + (r1 + r2 )R3 . (3.3)
Further, we conjecture that the stronger inequality holds:
q q q
a + 4r1 + b + 4r2 + c2 + 4r32 ≥ R1 + R2 + R3 + 2(r1 + r2 + r3 ).
2 2 2 2
(3.4)
Generally, we propose the following conjecture for the polygon.
Conjecture 3.2. For any interior point P of the polygon A1 A2 · · · An , we have
n q n n
X
2 2
X  πX
ai + 4ri ≥ Ri + sec ri , (3.5)
i=1 i=1
n i=1

where ai denote the lengths of sides Ai Ai+1 (i = 1, 2, · · · , n).

Remark 3.1. It is easy to prove that
Xn n q
X
2 Ri ≥ a2i + 4ri2 . (3.6)
i=1 i=1

Therefore, inequality (3.5 is stronger than Erdös-Mordell inequality (3.1) of the

polygon. The weaker inequality of (3.5):
n q n
X  πX
a2i + 4ri2 ≥ 2 sec ri (3.7)
i=1
n i=1
has not been yet proven.
6 JIAN LIU

Considering the generalization of the following inequality:

p p p √ √ √
R2 + R3 + R3 + R1 + R1 + R2 ≥ 2( r1 + r2 + r3 ), (3.8)
which is the special case of Theorem 2.2. We put forward
Conjecture 3.3. Let 0 < k < 1. For any interior point P of the polygon
A1 A2 · · · An , we have
n
X  π k X k
(Ri + Ri+1 )k ≥ 2 sec ri . (3.9)
i=1
n
The inequality is reversed when −1 ≤ k < 0.
By inequality (1.3), we have
y 2 R1 + z 2 R2 + x2 R3 ≥ 2(zxr1 + xyr2 + yzr3 ),
z 2 R1 + x2 R2 + y 2R3 ≥ 2(xyr1 + yzr2 + zxr3 ).
Adding up them gives
x2 (R2 +R3 )+y 2(R3 +R1 )+z 2 (R1 +R2 ) ≥ 2 [yz(r2 + r3 ) + zx(r3 + r1 ) + xy(r1 + r2 )] .
(3.10)
Applying Lemma 2.1 to inequality (2.12) and (3.10), one obtains
p p p
x2 (v + w)(R2 + R3 ) + y 2 (w + u)(R3 + R1 ) + z 2 (u + v)(R1 + R2 )
√ h p p p i
≥ 2 yz u(r2 + r3 ) + zx v(r3 + r1 ) + xy w(r1 + r2 ) , (3.11)
which is similar to the inequality of Theorem 2.2. In particular, we have
p p p √ √ √ √

R2 + R3 + R3 + R1 + R1 + R2 ≥ 2 r2 + r3 + r3 + r1 + r1 + r2 .
(3.12)
For this inequality, we pose the following generalization:
Conjecture 3.4. Let 0 < k < 1. For any interior point P of the polygon
A1 A2 · · · An , we have
n n
X
k
 π k X
(Ri + Ri+1 ) ≥ 2 sec (ri + ri+1 )k . (3.13)
i=1
n i=1
The inequality is reversed when −1 ≤ k < 0.
Remark 3.2. Let wi denote the internal angle-bisectors of ∠Ai P Ai+1 (i = 1, 2, · · · , n
and An+1 = A1 ). We also suppose that the inequalities of the previous four con-
jectures still hold after replacing ri by wi.
In (2.21), put x = y = z = 1, u = R1 , v = R2 , w = R3 , we get
p p p √
Ra (R2 + R3 ) + Rb (R3 + R1 ) + Rc (R1 + R2 ) ≥ 2(R1 + +R2 + R3 ). (3.14)
For this inequality, we put forward
Conjecture 3.5. For any interior point P of the polygon A1 A2 · · · An , we have
n r n
X p πX
(Ri + Ri+1 )ρi ≥ sec Ri . (3.15)
i=1
n i=1

where ρi denote the circumcenters of △AiP Ai+1 (i = 1, 2, · · · n and An+1 = A1 ).

A NEW SHARPENING OF THE ERDÖS-MORDELL INEQUALITY AND RELATED INEQUALITIES
7

Inequality (3.15) even holds for any point P in the plane. But we can not prove
the weaker linear inequality now:
n n
X 1 πX
ρi ≥ sec Ri . (3.16)
i=1
2 n i=1
We also think this inequality has the following exponential generalization:
Conjecture 3.6. Let 0 < k ≤ 1. For any interior point P of the polygon
A1 A2 · · · An , we have
n  k n
X
k 1 π X k
ρi ≥ sec Ri . (3.17)
i=1
2 n i=1
The inequality is reversed when −1 ≤ k < 0.

References
[1] P.Erdös, Problem 3740, Amer.Math.Monthly, 42(1935), 396.
[2] L.J.Mordell, Egy geometriai, probléma megoldása(Solution of a geometrical problem),
Középiskolai Matematikai és Fizikai Lapok 11(1935), 145-146.
[3] Solution of Problem 3740, Amer.Math.Monthly, 44(1937), 252-254.
[4] L.Fejes Tóth, Lagerungen in der Ebene, auf der Kugel und in Raum, 2nd ed.Berlin:
Springer-Verlag, 1953, 33.
[5] D.K.Kazarinoff, A simple proof of the Erdös-Mordell inequality for triangles, Michigan
Mathematical Journal, 4(1957), 97-98.
[6] G.R.Veldkamp, The Erdös-Mordell Inequality, Nieuw Tijdschr.Wisk, 45(1957/1958),193-
196.
[7] N.Ozeki, On P.Erdös’ inequality for the triangle, J.College Arts Sci. Chiba Univ. 2(1957),
247-250.
[8] L.Bankoff, An elementary proof of the Erdöos-Mordell theorem, Amer.Math.Monthly,
65(1958), 521.
[9] H.C.Lenhard, Verallgemeinerung and Verschärfung der Erdoös-Mordell schen Ungleichung
für polygone, Arch.Math.,12(1961), 311-314.
[10] A.Oppenheim, The Erdös-Mordell Inequality and and other inequalities for a triangle,
Amer.Math.Monthly, 68(1961), 226-230.
[11] A.Oppenheim, New inequalities for a triangle and an internal point, Ann. Math. Pura Appl.
(4)53(1961), 157-164.
[12] D.S.Mitrinović, J.E.Pečarić, On the Erdös-Mordell’s inequality for a polygons, J.College
Arts Sci.,Chiba Univ.,B-19(1986), 3-6.
[13] D.S.Mitrinović, J.E.Pečarić and V. Volenec, Recent Advances in Geometric Inequalities,
Kluwer Academic Publishers, Dordrecht-Boston-London, 1989.
[14] J.Chen, Z.Wang, On the conjectures of Erdös and Fejes Tóth. Bull Math(Wuhan), 7(1989),
3-4. (in Chinese)
[15] Z.Wang, J.Chen, On the Erdös-Mordell inequality. Bull Math(Wuhan), 7(1991), 28-29. (in
Chinese)
[16] A.Avez, A short proof of the Erdös and Mordell theorem, Amer.Math.Monthly, 100(1993),
60-62.
[17] V.Komornik, A short proof of the Erdös-Mordell theorem, Amer.Math.Monthly, 104(1997),
57-60.
[18] M.Dincǎ Generalization of the inequality of Erdös-Mordell, Octogon Math.Magazine,
1(1999), 143-147.
[19] F.Abi-Khuzam, A trigonometric inequality and its geometric applications,
Math.Inequal.Appl., 3(3)(2000), 437-442.
[20] H.Lee, Another proof of the Erdös-Mordell theorem, Forum Geom, 1(2001), 7-8.
8 JIAN LIU

[21] S.Dar, S.Gueron, A weighted Erdös-Mordell inequality, Amer.Math.Monthly, 108(2001),

165-168.
[22] C.Alsina, R.B.Nelsen, A visual proof of the Erdös-Mordell inequality, Forum Geom,
7(2001), 99-102.
[23] C.J.Zhao, M.Beucze, On imporovement of a weighted Erdös-Mordell inequality, Octogon
Math.Magazine. 10(2002), 48-51.
[24] J.Liu, Applications of a corollary of the Wolstenholme inequality, Journal of luoyang teach-
ers college, 22(5)(2003), 11-13.(in Chinese)
[25] N.Dergiades, Signed distances and the Erdös-Mordell inequality, Forum Geom, 4(2004),
67-68.
[26] W.Janous, Further inequalities of Erdös-Mordell type, Forum Geom., 4(2004), 203-206.
[27] S.Gueron, I.Shafrir, A weighted Erdös-Mordell inequality for polygons,
Amer.Math.Monthly, 112(2005),257-263.
[28] J.Liu, A sharpening of the Erdös-Mordell inequality and its applications, Journal of
Chongqing Normal University(Natural Science Edition),22(2)(2005),12-14.(in Chinese)
[29] J.Liu, A new extension of the Erdös-Mordell’s type inequality, Journal of Jilin Normal
University(Natural Science Edition),26(4)(2005), 8-11.(in Chinese)
[30] M.Bombardelli and S.H.Wu, Reverse inequalities of Erdös-Mordell type,
Math.Inequal.Appl., 12(2)(2009), 403-411.
[31] J.Liu, A new proof of the Erdös-Mordell inequality, Int.Electron.J.Geom, 4(2)(2011), 114-
119.
[32] J.Liu, The composite theorem of ternary quadratic inequalities and its applications,
RGMIA Research Report Collection, (11)(2008), art.13.
[33] J.Liu, A generalization of a weighted trigonometric inequality and its equivalent form,
Secondary education(Math),4(1996), 29-32.(in Chinese)

East China Jiaotong University, Jiangxi province Nanchang City, 330013, China
E-mail address: [email protected]