IMO problem 1961, different proofs of the Weitzenbock inequality












7














I'm collecting proof's for Weitzenbock inequality, i made three proofs for this,
See these proofs below(Whoever has a more cool proof please share).



Let a, b, c be the sides of the triangle and A its area, prove that:



$$a^{2}+b^{2}+c^{2}geq4sqrt{3}A$$



$$Proof 1$$



Let $R$ the circunradius.Suppose, by contradiction, that:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}<frac{3}{Rsqrt{3}} tag{1}
end{equation}

Using that:
begin{equation}
a+b+cleq 3Rsqrt{3} tag{2}
end{equation}

Multiplying (1) e (2):
$$(a+b+c)left(frac{1}{a}+frac{1}{b}+frac{1}{c}right)<9$$
What is absurd, this can be seen using the inequality that relates the arithmetic mean to the harmonic mean, or by the inequality of Cauchy-Schwarz. From where we conclude:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}geqfrac{3}{Rsqrt{3}}
end{equation}

Using $displaystyle A=frac{abc}{4R}$, we get:
$$ab+ac+bcgeq frac{sqrt{3}abc}{R}$$
$$ab+ac+bcgeq 4Asqrt{3}$$
This result follows using that $displaystyle a^{2}+b^{2}+c^{2}geq ab+bc+ac$



$$Proof 2$$



By Heron we get:



begin{equation*}
a^2+b^2+c^2 geq 4sqrt{3} sqrt{S(S-a)(S-b)(S-c)}
end{equation*}



Multiplying both sides of the inequality above by $displaystyle sqrt{left(frac{a+b+c}{2abc}right)^2}=sqrt{left(frac{S}{abc}right)^2}$, :
begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} sqrt{frac{S(S-a)}{bc}frac{S(S-b)}{ac}frac{S(S-c)}{ab}}
end{equation*}

By the law of cosine and law of sine we have:



begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{a^2}{bc}+frac{a}{b}+frac{a}{c}right)+ left(frac{b^2}{ac}+frac{b}{a}+frac{b}{c}right)+ left(frac{c^2}{ab}+frac{c}{a}+frac{c}{b}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{sen^2alpha}{senbeta sengamma}+frac{senalpha}{senbeta}+frac{senalpha}{sengamma}right)+left(frac{sen^2beta}{senalpha sengamma}+frac{senbeta}{senalpha}+frac{senbeta}{sengamma}right)+left(frac{sen^2gamma}{senalpha senbeta}+frac{sengamma}{senalpha}+frac{sengamma}{senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
(senalpha+senbeta+sengamma)left(frac{senalpha}{senbeta sengamma}+frac{senbeta}{senalpha sengamma}+frac{sengamma}{senalpha senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
cotalpha+cotbeta+cotgammageq sqrt{3}
end{equation*}

And this inequality follows from the Jensen inequality.



$$Proof 3$$



Let $displaystyle R$ the circunradius, $displaystyle alpha, beta, gamma$ the opposite acute angles to the sides of $displaystyle a, b, c$, so:



$\ \ displaystyle a^2+b^2-c^2=frac{4Rabc(a^2+b^2-c^2)}{4Rabc}=frac{4AR(a^2+b^2-c^2)}{abc}=4Atimesfrac{2R}{c}timesfrac{a^2+b^2-c^2}{2ab}=4Acotgamma=4Asqrt{cot^2gamma}=4Asqrt{csc^2gamma-1}=4Asqrt{frac{4R^2}{c^2}-1}=4Asqrt{frac{4a^2b^2R^2}{a^2b^2c^2}-1}=4Asqrt{frac{a^2b^2}{4left(frac{abc}{4R}right)^2}-1}=4Asqrt{frac{a^2b^2}{4A^2}-1}=2sqrt{4A^2left(frac{a^2b^2}{4A^2}-1right)}=2sqrt{a^2b^2-4A^2}Rightarrow a^2+b^2-c^2=2sqrt{a^2b^2-4A^2}\ \$



By symmetry, we get:



begin{equation*}
2sqrt{a^2b^2-4A^2}=a^2+b^2-c^2
end{equation*}



begin{equation*}
2sqrt{a^2c^2-4A^2}=a^2+c^2-b^2
end{equation*}



begin{equation*}
2sqrt{b^2c^2-4A^2}=b^2+c^2-a^2
end{equation*}



Multiplying these equalities two by two we will have:



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2c^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{a^2c^2-4A^2}=(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



We know that $displaystyle m+ngeq 2sqrt{mn}$, it folows that:



begin{equation*}
2(a^2b^2-4A^2+b^2c^2-4A^2)geq (a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2c^2-4A^2+b^2c^2-4A^2)geq(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2b^2-4A^2+a^2c^2-4A^2)geq(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



Adding all the above inequalities we will have:
begin{equation*}
4a^2b^2+4b^2c^2+4a^2c^2-16times 3 A^2geq (a^2+b^2-c^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



Let $ b^2+c^2-a^2=x,y=a^2+c^2-b^2,z=a^2+b^2-c^2$.Observe that



$$ 2a^2=(a^2+c^2-b^2)+(a^2+b^2-c^2)=y+z$$



$$ 2b^2=(a^2+b^2-c^2)+(b^2+c^2-a^2)=x+z$$



$$ 2c^2=(a^2+c^2-b^2)+(b^2+c^2-a^2)=x+y$$



We get:



$$ 4a^2b^2=(y+z)(x+z)$$



$$ 4a^2c^2=(y+z)(x+y)$$



$$ 4b^2c^2=(x+z)(x+y)$$



Replacing in the inequality above, we have:



begin{equation*}
(x+z)(x+y)+(y+z)(x+y)+(y+z)(x+z)-16times 3 A^2geq xy+xz+yz
end{equation*}



begin{equation*}
x^2+y^2+z^2+2xy+2xz+2yz-16times 3 A^2geq 0
end{equation*}



begin{equation*}
(x+y+z)^2-16times 3 A^2geq 0
end{equation*}



begin{equation*}
x+y+zgeq 4sqrt{3}A
end{equation*}










share|cite|improve this question




















  • 1




    1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
    – bof
    Feb 14 '17 at 5:36










  • Ops That's not what I meant.
    – Israel Meireles Chrisostomo
    Feb 14 '17 at 5:39










  • You can see some methods here.
    – mathlove
    Feb 14 '17 at 7:43
















7














I'm collecting proof's for Weitzenbock inequality, i made three proofs for this,
See these proofs below(Whoever has a more cool proof please share).



Let a, b, c be the sides of the triangle and A its area, prove that:



$$a^{2}+b^{2}+c^{2}geq4sqrt{3}A$$



$$Proof 1$$



Let $R$ the circunradius.Suppose, by contradiction, that:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}<frac{3}{Rsqrt{3}} tag{1}
end{equation}

Using that:
begin{equation}
a+b+cleq 3Rsqrt{3} tag{2}
end{equation}

Multiplying (1) e (2):
$$(a+b+c)left(frac{1}{a}+frac{1}{b}+frac{1}{c}right)<9$$
What is absurd, this can be seen using the inequality that relates the arithmetic mean to the harmonic mean, or by the inequality of Cauchy-Schwarz. From where we conclude:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}geqfrac{3}{Rsqrt{3}}
end{equation}

Using $displaystyle A=frac{abc}{4R}$, we get:
$$ab+ac+bcgeq frac{sqrt{3}abc}{R}$$
$$ab+ac+bcgeq 4Asqrt{3}$$
This result follows using that $displaystyle a^{2}+b^{2}+c^{2}geq ab+bc+ac$



$$Proof 2$$



By Heron we get:



begin{equation*}
a^2+b^2+c^2 geq 4sqrt{3} sqrt{S(S-a)(S-b)(S-c)}
end{equation*}



Multiplying both sides of the inequality above by $displaystyle sqrt{left(frac{a+b+c}{2abc}right)^2}=sqrt{left(frac{S}{abc}right)^2}$, :
begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} sqrt{frac{S(S-a)}{bc}frac{S(S-b)}{ac}frac{S(S-c)}{ab}}
end{equation*}

By the law of cosine and law of sine we have:



begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{a^2}{bc}+frac{a}{b}+frac{a}{c}right)+ left(frac{b^2}{ac}+frac{b}{a}+frac{b}{c}right)+ left(frac{c^2}{ab}+frac{c}{a}+frac{c}{b}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{sen^2alpha}{senbeta sengamma}+frac{senalpha}{senbeta}+frac{senalpha}{sengamma}right)+left(frac{sen^2beta}{senalpha sengamma}+frac{senbeta}{senalpha}+frac{senbeta}{sengamma}right)+left(frac{sen^2gamma}{senalpha senbeta}+frac{sengamma}{senalpha}+frac{sengamma}{senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
(senalpha+senbeta+sengamma)left(frac{senalpha}{senbeta sengamma}+frac{senbeta}{senalpha sengamma}+frac{sengamma}{senalpha senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
cotalpha+cotbeta+cotgammageq sqrt{3}
end{equation*}

And this inequality follows from the Jensen inequality.



$$Proof 3$$



Let $displaystyle R$ the circunradius, $displaystyle alpha, beta, gamma$ the opposite acute angles to the sides of $displaystyle a, b, c$, so:



$\ \ displaystyle a^2+b^2-c^2=frac{4Rabc(a^2+b^2-c^2)}{4Rabc}=frac{4AR(a^2+b^2-c^2)}{abc}=4Atimesfrac{2R}{c}timesfrac{a^2+b^2-c^2}{2ab}=4Acotgamma=4Asqrt{cot^2gamma}=4Asqrt{csc^2gamma-1}=4Asqrt{frac{4R^2}{c^2}-1}=4Asqrt{frac{4a^2b^2R^2}{a^2b^2c^2}-1}=4Asqrt{frac{a^2b^2}{4left(frac{abc}{4R}right)^2}-1}=4Asqrt{frac{a^2b^2}{4A^2}-1}=2sqrt{4A^2left(frac{a^2b^2}{4A^2}-1right)}=2sqrt{a^2b^2-4A^2}Rightarrow a^2+b^2-c^2=2sqrt{a^2b^2-4A^2}\ \$



By symmetry, we get:



begin{equation*}
2sqrt{a^2b^2-4A^2}=a^2+b^2-c^2
end{equation*}



begin{equation*}
2sqrt{a^2c^2-4A^2}=a^2+c^2-b^2
end{equation*}



begin{equation*}
2sqrt{b^2c^2-4A^2}=b^2+c^2-a^2
end{equation*}



Multiplying these equalities two by two we will have:



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2c^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{a^2c^2-4A^2}=(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



We know that $displaystyle m+ngeq 2sqrt{mn}$, it folows that:



begin{equation*}
2(a^2b^2-4A^2+b^2c^2-4A^2)geq (a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2c^2-4A^2+b^2c^2-4A^2)geq(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2b^2-4A^2+a^2c^2-4A^2)geq(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



Adding all the above inequalities we will have:
begin{equation*}
4a^2b^2+4b^2c^2+4a^2c^2-16times 3 A^2geq (a^2+b^2-c^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



Let $ b^2+c^2-a^2=x,y=a^2+c^2-b^2,z=a^2+b^2-c^2$.Observe that



$$ 2a^2=(a^2+c^2-b^2)+(a^2+b^2-c^2)=y+z$$



$$ 2b^2=(a^2+b^2-c^2)+(b^2+c^2-a^2)=x+z$$



$$ 2c^2=(a^2+c^2-b^2)+(b^2+c^2-a^2)=x+y$$



We get:



$$ 4a^2b^2=(y+z)(x+z)$$



$$ 4a^2c^2=(y+z)(x+y)$$



$$ 4b^2c^2=(x+z)(x+y)$$



Replacing in the inequality above, we have:



begin{equation*}
(x+z)(x+y)+(y+z)(x+y)+(y+z)(x+z)-16times 3 A^2geq xy+xz+yz
end{equation*}



begin{equation*}
x^2+y^2+z^2+2xy+2xz+2yz-16times 3 A^2geq 0
end{equation*}



begin{equation*}
(x+y+z)^2-16times 3 A^2geq 0
end{equation*}



begin{equation*}
x+y+zgeq 4sqrt{3}A
end{equation*}










share|cite|improve this question




















  • 1




    1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
    – bof
    Feb 14 '17 at 5:36










  • Ops That's not what I meant.
    – Israel Meireles Chrisostomo
    Feb 14 '17 at 5:39










  • You can see some methods here.
    – mathlove
    Feb 14 '17 at 7:43














7












7








7


3





I'm collecting proof's for Weitzenbock inequality, i made three proofs for this,
See these proofs below(Whoever has a more cool proof please share).



Let a, b, c be the sides of the triangle and A its area, prove that:



$$a^{2}+b^{2}+c^{2}geq4sqrt{3}A$$



$$Proof 1$$



Let $R$ the circunradius.Suppose, by contradiction, that:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}<frac{3}{Rsqrt{3}} tag{1}
end{equation}

Using that:
begin{equation}
a+b+cleq 3Rsqrt{3} tag{2}
end{equation}

Multiplying (1) e (2):
$$(a+b+c)left(frac{1}{a}+frac{1}{b}+frac{1}{c}right)<9$$
What is absurd, this can be seen using the inequality that relates the arithmetic mean to the harmonic mean, or by the inequality of Cauchy-Schwarz. From where we conclude:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}geqfrac{3}{Rsqrt{3}}
end{equation}

Using $displaystyle A=frac{abc}{4R}$, we get:
$$ab+ac+bcgeq frac{sqrt{3}abc}{R}$$
$$ab+ac+bcgeq 4Asqrt{3}$$
This result follows using that $displaystyle a^{2}+b^{2}+c^{2}geq ab+bc+ac$



$$Proof 2$$



By Heron we get:



begin{equation*}
a^2+b^2+c^2 geq 4sqrt{3} sqrt{S(S-a)(S-b)(S-c)}
end{equation*}



Multiplying both sides of the inequality above by $displaystyle sqrt{left(frac{a+b+c}{2abc}right)^2}=sqrt{left(frac{S}{abc}right)^2}$, :
begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} sqrt{frac{S(S-a)}{bc}frac{S(S-b)}{ac}frac{S(S-c)}{ab}}
end{equation*}

By the law of cosine and law of sine we have:



begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{a^2}{bc}+frac{a}{b}+frac{a}{c}right)+ left(frac{b^2}{ac}+frac{b}{a}+frac{b}{c}right)+ left(frac{c^2}{ab}+frac{c}{a}+frac{c}{b}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{sen^2alpha}{senbeta sengamma}+frac{senalpha}{senbeta}+frac{senalpha}{sengamma}right)+left(frac{sen^2beta}{senalpha sengamma}+frac{senbeta}{senalpha}+frac{senbeta}{sengamma}right)+left(frac{sen^2gamma}{senalpha senbeta}+frac{sengamma}{senalpha}+frac{sengamma}{senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
(senalpha+senbeta+sengamma)left(frac{senalpha}{senbeta sengamma}+frac{senbeta}{senalpha sengamma}+frac{sengamma}{senalpha senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
cotalpha+cotbeta+cotgammageq sqrt{3}
end{equation*}

And this inequality follows from the Jensen inequality.



$$Proof 3$$



Let $displaystyle R$ the circunradius, $displaystyle alpha, beta, gamma$ the opposite acute angles to the sides of $displaystyle a, b, c$, so:



$\ \ displaystyle a^2+b^2-c^2=frac{4Rabc(a^2+b^2-c^2)}{4Rabc}=frac{4AR(a^2+b^2-c^2)}{abc}=4Atimesfrac{2R}{c}timesfrac{a^2+b^2-c^2}{2ab}=4Acotgamma=4Asqrt{cot^2gamma}=4Asqrt{csc^2gamma-1}=4Asqrt{frac{4R^2}{c^2}-1}=4Asqrt{frac{4a^2b^2R^2}{a^2b^2c^2}-1}=4Asqrt{frac{a^2b^2}{4left(frac{abc}{4R}right)^2}-1}=4Asqrt{frac{a^2b^2}{4A^2}-1}=2sqrt{4A^2left(frac{a^2b^2}{4A^2}-1right)}=2sqrt{a^2b^2-4A^2}Rightarrow a^2+b^2-c^2=2sqrt{a^2b^2-4A^2}\ \$



By symmetry, we get:



begin{equation*}
2sqrt{a^2b^2-4A^2}=a^2+b^2-c^2
end{equation*}



begin{equation*}
2sqrt{a^2c^2-4A^2}=a^2+c^2-b^2
end{equation*}



begin{equation*}
2sqrt{b^2c^2-4A^2}=b^2+c^2-a^2
end{equation*}



Multiplying these equalities two by two we will have:



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2c^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{a^2c^2-4A^2}=(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



We know that $displaystyle m+ngeq 2sqrt{mn}$, it folows that:



begin{equation*}
2(a^2b^2-4A^2+b^2c^2-4A^2)geq (a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2c^2-4A^2+b^2c^2-4A^2)geq(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2b^2-4A^2+a^2c^2-4A^2)geq(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



Adding all the above inequalities we will have:
begin{equation*}
4a^2b^2+4b^2c^2+4a^2c^2-16times 3 A^2geq (a^2+b^2-c^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



Let $ b^2+c^2-a^2=x,y=a^2+c^2-b^2,z=a^2+b^2-c^2$.Observe that



$$ 2a^2=(a^2+c^2-b^2)+(a^2+b^2-c^2)=y+z$$



$$ 2b^2=(a^2+b^2-c^2)+(b^2+c^2-a^2)=x+z$$



$$ 2c^2=(a^2+c^2-b^2)+(b^2+c^2-a^2)=x+y$$



We get:



$$ 4a^2b^2=(y+z)(x+z)$$



$$ 4a^2c^2=(y+z)(x+y)$$



$$ 4b^2c^2=(x+z)(x+y)$$



Replacing in the inequality above, we have:



begin{equation*}
(x+z)(x+y)+(y+z)(x+y)+(y+z)(x+z)-16times 3 A^2geq xy+xz+yz
end{equation*}



begin{equation*}
x^2+y^2+z^2+2xy+2xz+2yz-16times 3 A^2geq 0
end{equation*}



begin{equation*}
(x+y+z)^2-16times 3 A^2geq 0
end{equation*}



begin{equation*}
x+y+zgeq 4sqrt{3}A
end{equation*}










share|cite|improve this question















I'm collecting proof's for Weitzenbock inequality, i made three proofs for this,
See these proofs below(Whoever has a more cool proof please share).



Let a, b, c be the sides of the triangle and A its area, prove that:



$$a^{2}+b^{2}+c^{2}geq4sqrt{3}A$$



$$Proof 1$$



Let $R$ the circunradius.Suppose, by contradiction, that:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}<frac{3}{Rsqrt{3}} tag{1}
end{equation}

Using that:
begin{equation}
a+b+cleq 3Rsqrt{3} tag{2}
end{equation}

Multiplying (1) e (2):
$$(a+b+c)left(frac{1}{a}+frac{1}{b}+frac{1}{c}right)<9$$
What is absurd, this can be seen using the inequality that relates the arithmetic mean to the harmonic mean, or by the inequality of Cauchy-Schwarz. From where we conclude:
begin{equation}
frac{1}{a}+frac{1}{b}+frac{1}{c}geqfrac{3}{Rsqrt{3}}
end{equation}

Using $displaystyle A=frac{abc}{4R}$, we get:
$$ab+ac+bcgeq frac{sqrt{3}abc}{R}$$
$$ab+ac+bcgeq 4Asqrt{3}$$
This result follows using that $displaystyle a^{2}+b^{2}+c^{2}geq ab+bc+ac$



$$Proof 2$$



By Heron we get:



begin{equation*}
a^2+b^2+c^2 geq 4sqrt{3} sqrt{S(S-a)(S-b)(S-c)}
end{equation*}



Multiplying both sides of the inequality above by $displaystyle sqrt{left(frac{a+b+c}{2abc}right)^2}=sqrt{left(frac{S}{abc}right)^2}$, :
begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} sqrt{frac{S(S-a)}{bc}frac{S(S-b)}{ac}frac{S(S-c)}{ab}}
end{equation*}

By the law of cosine and law of sine we have:



begin{equation*}
left(a^2+b^2+c^2right)left(frac{a+b+c}{2abc}right) geq 4sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{a^2}{bc}+frac{a}{b}+frac{a}{c}right)+ left(frac{b^2}{ac}+frac{b}{a}+frac{b}{c}right)+ left(frac{c^2}{ab}+frac{c}{a}+frac{c}{b}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
left(frac{sen^2alpha}{senbeta sengamma}+frac{senalpha}{senbeta}+frac{senalpha}{sengamma}right)+left(frac{sen^2beta}{senalpha sengamma}+frac{senbeta}{senalpha}+frac{senbeta}{sengamma}right)+left(frac{sen^2gamma}{senalpha senbeta}+frac{sengamma}{senalpha}+frac{sengamma}{senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
(senalpha+senbeta+sengamma)left(frac{senalpha}{senbeta sengamma}+frac{senbeta}{senalpha sengamma}+frac{sengamma}{senalpha senbeta}right) geq 8sqrt{3} cosfrac{alpha}{2}cosfrac{beta}{2}cosfrac{gamma}{2}
end{equation*}



begin{equation*}
cotalpha+cotbeta+cotgammageq sqrt{3}
end{equation*}

And this inequality follows from the Jensen inequality.



$$Proof 3$$



Let $displaystyle R$ the circunradius, $displaystyle alpha, beta, gamma$ the opposite acute angles to the sides of $displaystyle a, b, c$, so:



$\ \ displaystyle a^2+b^2-c^2=frac{4Rabc(a^2+b^2-c^2)}{4Rabc}=frac{4AR(a^2+b^2-c^2)}{abc}=4Atimesfrac{2R}{c}timesfrac{a^2+b^2-c^2}{2ab}=4Acotgamma=4Asqrt{cot^2gamma}=4Asqrt{csc^2gamma-1}=4Asqrt{frac{4R^2}{c^2}-1}=4Asqrt{frac{4a^2b^2R^2}{a^2b^2c^2}-1}=4Asqrt{frac{a^2b^2}{4left(frac{abc}{4R}right)^2}-1}=4Asqrt{frac{a^2b^2}{4A^2}-1}=2sqrt{4A^2left(frac{a^2b^2}{4A^2}-1right)}=2sqrt{a^2b^2-4A^2}Rightarrow a^2+b^2-c^2=2sqrt{a^2b^2-4A^2}\ \$



By symmetry, we get:



begin{equation*}
2sqrt{a^2b^2-4A^2}=a^2+b^2-c^2
end{equation*}



begin{equation*}
2sqrt{a^2c^2-4A^2}=a^2+c^2-b^2
end{equation*}



begin{equation*}
2sqrt{b^2c^2-4A^2}=b^2+c^2-a^2
end{equation*}



Multiplying these equalities two by two we will have:



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2c^2-4A^2}sqrt{b^2c^2-4A^2}=(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
4sqrt{a^2b^2-4A^2}sqrt{a^2c^2-4A^2}=(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



We know that $displaystyle m+ngeq 2sqrt{mn}$, it folows that:



begin{equation*}
2(a^2b^2-4A^2+b^2c^2-4A^2)geq (a^2+b^2-c^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2c^2-4A^2+b^2c^2-4A^2)geq(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



begin{equation*}
2(a^2b^2-4A^2+a^2c^2-4A^2)geq(a^2+c^2-b^2)(a^2+b^2-c^2)
end{equation*}



Adding all the above inequalities we will have:
begin{equation*}
4a^2b^2+4b^2c^2+4a^2c^2-16times 3 A^2geq (a^2+b^2-c^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)+(a^2+c^2-b^2)(b^2+c^2-a^2)
end{equation*}



Let $ b^2+c^2-a^2=x,y=a^2+c^2-b^2,z=a^2+b^2-c^2$.Observe that



$$ 2a^2=(a^2+c^2-b^2)+(a^2+b^2-c^2)=y+z$$



$$ 2b^2=(a^2+b^2-c^2)+(b^2+c^2-a^2)=x+z$$



$$ 2c^2=(a^2+c^2-b^2)+(b^2+c^2-a^2)=x+y$$



We get:



$$ 4a^2b^2=(y+z)(x+z)$$



$$ 4a^2c^2=(y+z)(x+y)$$



$$ 4b^2c^2=(x+z)(x+y)$$



Replacing in the inequality above, we have:



begin{equation*}
(x+z)(x+y)+(y+z)(x+y)+(y+z)(x+z)-16times 3 A^2geq xy+xz+yz
end{equation*}



begin{equation*}
x^2+y^2+z^2+2xy+2xz+2yz-16times 3 A^2geq 0
end{equation*}



begin{equation*}
(x+y+z)^2-16times 3 A^2geq 0
end{equation*}



begin{equation*}
x+y+zgeq 4sqrt{3}A
end{equation*}







geometry trigonometry inequality contest-math






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 8 at 23:29









user376343

2,7982822




2,7982822










asked Feb 14 '17 at 5:27









Israel Meireles Chrisostomo

604413




604413








  • 1




    1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
    – bof
    Feb 14 '17 at 5:36










  • Ops That's not what I meant.
    – Israel Meireles Chrisostomo
    Feb 14 '17 at 5:39










  • You can see some methods here.
    – mathlove
    Feb 14 '17 at 7:43














  • 1




    1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
    – bof
    Feb 14 '17 at 5:36










  • Ops That's not what I meant.
    – Israel Meireles Chrisostomo
    Feb 14 '17 at 5:39










  • You can see some methods here.
    – mathlove
    Feb 14 '17 at 7:43








1




1




1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
– bof
Feb 14 '17 at 5:36




1961 different proofs of the Weitzenbock inequality? Wow, that's a lot of proofs. I have a book with 344 different proofs of the Pythagorean theorem, but that's nothing compared to this.
– bof
Feb 14 '17 at 5:36












Ops That's not what I meant.
– Israel Meireles Chrisostomo
Feb 14 '17 at 5:39




Ops That's not what I meant.
– Israel Meireles Chrisostomo
Feb 14 '17 at 5:39












You can see some methods here.
– mathlove
Feb 14 '17 at 7:43




You can see some methods here.
– mathlove
Feb 14 '17 at 7:43










4 Answers
4






active

oldest

votes


















9














What about a proof (almost) without words?



enter image description here



Let us consider a triangle $ABC$ and its Fermat-Torricelli-Steiner point, such that $widehat{CFA}=widehat{AFC}=widehat{CFB}=120^circ$. It is well-known that $F$ lies on each line joining a vertex of $ABC$ with the free vertex of the equilateral triangle built on the opposite side. By replicating the (colored) triangles $AFC,CFB,BFA$ four times we reach the configuration depicted above. Due to the triangular holes
$$ [ABV_C]+[ACV_B]+[BCV_A]geq 3[ABC] $$
that is equivalent to the claim.





There also is an efficient variational approach. If $ABC$ is a triangle in the plane with $AB=c,AC=b$, the locus of points $P$ such that $PB^2+PC^2=b^2+c^2$ is a circle through $A$ centered at the midpoint of $BC$. It follows that by moving the vertex $A$ along such circle, till meeting the perpendicular bisector of $BC$, the area of $ABC$ increases but $AB^2+BC^2+AC^2$ stays the same. It follows that for a fixed $a^2+b^2+c^2$, the maximum area is achieved only by the equilateral triangle, and $a^2+b^2+c^2geq 4sqrt{3}Delta$ follows.





Yet another way is to exploit the Cauchy-Schwarz inequality. We have
$$ (1+1+1)(a^4+b^4+c^4)geq (a^2+b^2+c^2)^2 $$
hence:
$$ 4Delta = sqrt{(a^2+b^2+c^2)^2-2(a^4+b^4+c^4)} leq frac{1}{sqrt{3}}(a^2+b^2+c^2). $$





Update. Yet another way is to prove a stronger inequality, namely
$$ ab+ac+bc geq 4Deltasqrt{3} $$
which follows from $ab=frac{2Delta}{sin C}$ and the convexity of $frac{1}{sintheta}$ over $(0,pi)$.






share|cite|improve this answer



















  • 1




    Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    May 16 '17 at 18:34










  • @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
    – Jack D'Aurizio
    May 16 '17 at 18:36










  • Wow, this is crisp and pretty!
    – NivPai
    Aug 13 at 17:24



















3














There are some proofs at https://artofproblemsolving.com/community/c6h14388 and a generalization at https://artofproblemsolving.com/community/c6h2958 . Let me actually reproduce my post in the latter thread in full here (with minor modifications -- neither my exposition nor my LaTeX were particularly good back in 2005), which proves several generalizations:




Theorem 1. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $XYZ$ be a triangle with sidelengths $x = YZ$, $y = ZX$, $z = XY$ and area $T$.



(a) (Pedoe inequality) Then, we have the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar.



(b) (Bottema inequality) If $P$ is an arbitrary point in the plane of triangle $ABC$, then we have the inequality



$xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



(c) (Equivalent version of (a)) Let $A = measuredangle CAB$, $B = measuredangle ABC$, $C = measuredangle BCA$ be the angles of triangle $ABC$. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of triangle $XYZ$. Then,



$cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y geq 2$.



This inequality becomes an equality if and only if the triangles $ABC$ and $XYZ$ are similar.




Proof of Theorem 1.



(a) First note that the inequality which we must prove,



$a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$,



is homogeneous in the sidelengths $x$, $y$, $z$ of the triangle $XYZ$ (in fact, these sidelengths occur in the power $2$ on the left hand side, and on the right hand side they occur in the power $2$ as well, since the area of a triangle depends quadratically on its sidelengths). Hence, this inequality is invariant under any similitude transformation applied to triangle $XYZ$; in other words, we can move, reflect, rotate and stretch the triangle $XYZ$ as we wish, and the inequality remains equivalent. But, of course, by applying similitude transformations to triangle $XYZ$, we can always achieve a situation when $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. Hence, in order to prove Theorem 1 (a) for every two triangles $ABC$ and $XYZ$, it is enough to prove it for two triangles $ABC$ and $XYZ$ in this special situation.



So, let us WLOG assume that the triangles $ABC$ and $XYZ$ are in this special situation, i. e. that we have $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. We have to prove the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



By the cosine law in triangle $ABX$, we have



$AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$.



Let's figure out what this equation means. At first, $AB = c$. Then, since $B = Y$, we have $XB = XY = z$. Finally, we have either $measuredangle ABX = measuredangle ABC - measuredangle XBC$ or $measuredangle ABX = measuredangle XBC - measuredangle ABC$ (depending on the arrangement of the points), but in both cases $cos measuredangle ABX = cos left(measuredangle ABC - measuredangle XBCright)$. Since $B = Y$ and $C = Z$, we can rewrite the angle $measuredangle XBC$ as $measuredangle XYZ$. Thus,



$cosmeasuredangle ABX=cosleft(measuredangle ABC-measuredangle XYZright)=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$.



By the cosine law in triangles $ABC$ and $XYZ$, we have



$cosmeasuredangle ABC=dfrac{c^2+a^2-b^2}{2ca}$ and
$cosmeasuredangle XYZ=dfrac{z^2+x^2-y^2}{2zx}$.



Also, by a well-known formula, the area $S$ of triangle $ABC$ equals $S=dfrac12cdot casinmeasuredangle ABC$; thus, $sinmeasuredangle ABC=dfrac{2S}{ca}$. Similarly, $sinmeasuredangle XYZ=dfrac{2T}{zx}$.



Hence,



$cosmeasuredangle ABX=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$
$=dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}$.



Together with $AB = c$ and $XB = z$, this makes the equality



$AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$



transform into



$AX^2=c^2+z^2-2cdot ccdot zcdotleft(dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}right)$.



This immediately simplifies to



$AX^2=c^2+z^2-2left(dfrac{left(c^2+a^2-b^2right)left(z^2+x^2-y^2right)}{4ax}+dfrac{4ST}{ax}right)$.



After some algebra, this becomes



$AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)-2Qright)-16ST}{2ax}$,



where $Q = left(x-aright)left(c^2x-z^2aright)$.



Since $Y = B$ and $Z = C$, we have $YZ = BC$, or, in other words, $x = a$. Thus, $x - a = 0$, so that $Q = 0$. Hence, the above formula for $AX^2$ simplifies to



$AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16ST}{2ax}$.



Since (obviously) $AX^2geq 0$, we thus have



$left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16STgeq 0$,



and thus $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$. So the inequality of Theorem 1 (a) is proven. Equality holds if and only if equality holds in the inequality $AX^2geq 0$; but this is clearly the case if and only if the points $A$ and $X$ coincide, i. e. if the triangles $ABC$ and $XYZ$ are congruent. Now, of course, since the triangle $XYZ$ we are dealing with is not the initial triangle $XYZ$, but just its image under a similitude transformation, the general equality condition is that the triangles $ABC$ and $XYZ$ are similar (not necessarily being congruent). So Theorem 1 (a) is proven.



(b) In our proof of Theorem 1 (a), we WLOG assumed that $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. This time we will assume the same situation, but with a little modification: The point $X$ should lie not in the same halfplane with respect to the line $BC$ as the point $A$, but exactly in the other halfplane. In other words, we have $Y = B$ and $Z = C$ and the points $A$ and $X$ lie in different halfplanes with respect to the line $BC$. Then, by a calculation similar to the one we did in the proof of Theorem 1 (a), we find



$AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2ax}$



(note the $+$ sign instead of the $-$). Since $x = a$, we have $ax=a^2$, and thus we can rewrite this as



$AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2a^2}$.



Hence,



$a^2cdot AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2}$
${=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



Taking the square root of this, we get



$acdot AX=sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



Hence, the inequality which is to be proven,



$xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$,



is equivalent to $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



So we have to prove $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



Well, first, by the Ptolemy inequality for the quadrilateral $PBXC$, we have $CXcdot BP+XBcdot CPgeq BCcdot PX$. Since $B = Y$ and $C = Z$, we have $CX = ZX = y$, $XB = XY = z$ and $BC = YZ = x$, and therefore this inequality rewrites as $ycdot BP+zcdot CPgeq xcdot PX$. Also, by the triangle inequality in the triangle $APX$, we have $AP+PXgeq AX$. Thus,



$xcdot AP+ycdot BP+zcdot CP=xcdot AP+left(ycdot BP+zcdot CPright)geq xcdot AP+xcdot PX$
$=xcdotleft(AP+PXright)geq xcdot AX=acdot AX$,



where we have used $x = a$ in the last step. This proves Theorem 1 (b).



(c) By the Cosine Law in triangle $ABC$, we have $cos A=dfrac{b^2+c^2-a^2}{2bc}$. On the other hand, by a well-known formula for the area of a triangle, the area $S$ of triangle $ABC$ equals $S=dfrac12bcsin A$. Thus, $sin A=dfrac{2S}{bc}$. Hence,



$cot A=cos A / sin A=dfrac{b^2+c^2-a^2}{2bc} / dfrac{2S}{bc}=dfrac{b^2+c^2-a^2}{4S}$.



Similarly, $cot B=dfrac{c^2+a^2-b^2}{4S}$. Thus, $cot A+cot B=dfrac{b^2+c^2-a^2}{4S}+dfrac{c^2+a^2-b^2}{4S}=dfrac{2c^2}{4S}=dfrac{c^2}{2S}$.



Similarly to how we proved $cot A=dfrac{b^2+c^2-a^2}{4S}$, we can find that $cot Z=dfrac{x^2+y^2-z^2}{4T}$. Thus,



$cot Acot Z+cot Bcot Z=left(cot A+cot Bright)cdotcot Z=dfrac{c^2}{2S}cdotdfrac{x^2+y^2-z^2}{4T}$
$=dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$.



Similarly,



$cot Bcot X+cot Ccot X=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}$ and
$cot Ccot Y+cot Acot Y=dfrac{b^2left(z^2+x^2-y^2right)}{8ST}$.



Thus,



$cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
$=left(cot Bcot X+cot Ccot Xright)+left(cot Ccot Y+cot Acot Yright)+left(cot Acot Z+cot Bcot Zright)$
$=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}+dfrac{b^2left(z^2+x^2-y^2right)}{8ST}+dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$
$=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{8ST}$.



But after Theorem 1 (a), we have $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar. Hence,



$cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
$geqdfrac{16ST}{8ST}=2$,



with equality if and only if the triangles $ABC$ and $XYZ$ are similar. And Theorem 1 (c) is proven.



Note that Theorem 1 (a) yields the following geometric inequality:




Corollary 2. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$.




Before proving the Corollary, we show two useful facts:




Corollary 3 (a well-known formula). Let $XYZ$ be a triangle. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of this triangle. Then, $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$.




Proof of Corollary 3. Theorem 1 (c) (applied to $XYZ$, $x$, $y$ and $z$ instead of $ABC$, $a$, $b$ and $c$) shows that $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y geq 2$, and that this inequality becomes an equality if and only if the triangles $XYZ$ and $XYZ$ are similar. Thus, of course, this inequality becomes an equality (since the triangles $XYZ$ and $XYZ$ are similar). Hence, $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y = 2$. Therefore,



$2 = cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y$
$= 2 left(cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Yright)$.



Dividing both sides of this equality by $2$, we obtain $1 = cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y$. This proves Corollary 3.




Theorem 4 (Conway substitution theorem). Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. This triangle satisfies $y^2+z^2-x^2=2u$, $z^2+x^2-y^2=2v$, $x^2+y^2-z^2=2w$. The area $T$ of this triangle equals $T=dfrac12sqrt{vw+wu+uv}$. If $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ are the angles of this triangle, then $cot X=dfrac{u}{2T}$, $cot Y=dfrac{v}{2T}$ and $cot Z=dfrac{w}{2T}$.




Proof of Theorem 4. Since the numbers $v + w$, $w + u$, $u + v$ are nonnegative, their square roots $sqrt{v+w}$, $sqrt{w+u}$, $sqrt{u+v}$ exist, and, of course, are nonnegative as well. Now, we have $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w}$, since



$sqrt{w+u}+sqrt{u+v}geqsqrt{v+w} Longleftarrow left(sqrt{w+u}+sqrt{u+v}right)^2geq v+w$



$Longleftarrow left(w+uright)+left(u+vright)+2sqrt{left(w+uright)left(u+vright)}geq v+w$



$Longleftarrow 2sqrt{left(w+uright)left(u+vright)}geq -2u$ [this is not trivial, since $u$ needs not be $geq 0$!]



$Longleftarrow sqrt{left(w+uright)left(u+vright)}geq-u Longleftarrow left(w+uright)left(u+vright)geqleft(-uright)^2$



$Longleftarrow u^2+left(vw+wu+uvright)geq u^2 Longleftarrow vw+wu+uvgeq 0$.



Similarly, $sqrt{u+v}+sqrt{v+w}geqsqrt{w+u}$ and $sqrt{v+w}+sqrt{w+u}geqsqrt{u+v}$. Thus, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. Denote by $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ the angles of this triangle. We have



$y^2+z^2-x^2=left(sqrt{w+u}right)^2+left(sqrt{u+v}right)^2-left(sqrt{v+w}right)^2$
$=left(w+uright)+left(u+vright)-left(v+wright)=2u$



and similarly $z^2+x^2-y^2=2v$ and $x^2+y^2-z^2=2w$. Now, denoting the area of triangle $XYZ$ by $T$, we note that, as shown during the proof of Theorem 1 (c), we have $cot Z=dfrac{x^2+y^2-z^2}{4T}$, so that $cot Z=dfrac{2w}{4T}=dfrac{w}{2T}$, and similarly $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$.



Corollary 3 yields $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$. Using the relations $cot Z=dfrac{w}{2T}$, $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$, this equality becomes $dfrac{v}{2T}cdotdfrac{w}{2T}+dfrac{w}{2T}cdotdfrac{u}{2T}+dfrac{u}{2T}cdotdfrac{v}{2T}=1$. Hence, $vw+wu+uv=4T^2$, and thus $T=dfrac12sqrt{4T^2}=dfrac12sqrt{vw+wu+uv}$. This completes the proof of Theorem 4.



Proof of Corollary 2. Construct a triangle $XYZ$ for our values of $u$, $v$, $w$ according to Theorem 4. Then, applying Theorem 1 (a) to the triangles $ABC$ and $XYZ$, we get



$a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



Using the formulas given in Theorem 4, we can rewrite this as



$a^2cdot 2u+b^2cdot 2v+c^2cdot 2wgeq 16Scdotdfrac12sqrt{vw+wu+uv}$.



This simplifies to $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$. This proves Corollary 2.




Corollary 5 (Weitzenböck inequality). Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Then, $a^2 + b^2 + c^2 geq 4sqrt{3} S$.




Proof of Corollary 5. Corollary 5 follows from Corollary 2 (applied to $u=1$, $v=1$ and $w=1$).



[Postscriptum: I have been atrocious with attribution and citation back in 2005. I am not sure where the ideas in the above proofs come from; I might have been just the scribe. Theorem 1 (a) was discovered in 1943 by Daniel Pedoe and much earlier (without the equality case) by Neuberg. It also appears as 10.8 in O. Bottema, R. Z. Djordjevic, R. R. Janic, D. S. Mitrinovic, P. M. Vasic, Geometric Inequalities, Wolters-Noordhoff 1969. Theorem 1 (b) appears as 12.56 in the same book and is credited to O. Bottema, Hoofdstukken uit de elementaire meetkunde, Den Haag 1944 (I don't have this book, but interestingly it is not out of print). I might have picked up the Conway substitution from posts by John Conway on the Hyacinthos newsgroup long ago.]






share|cite|improve this answer























  • Uow thank's @darij grinberg!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    Mar 3 '17 at 2:59



















2














using the Heron formula for $A$ we have $$a^2+b^2+c^2geq 4sqrt{3}sqrt{s(s-a)(s-b)(s-c)}$$ after squaring and rearanging we obtain
$$a^4+b^4+c^4geq a^2b^2+b^2c^2+c^2a^2$$ and this is true and follows from
$$x^2+y^2+z^2geq xy+yz+zx$$






share|cite|improve this answer

















  • 2




    after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
    – dxiv
    Feb 14 '17 at 6:37












  • it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:49










  • why the $-1$ i don't see it?
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:50






  • 3




    My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
    – dxiv
    Feb 14 '17 at 7:02








  • 2




    Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
    – dxiv
    Feb 14 '17 at 7:09





















0














Let $P$ be the first Brocard point of $triangle ABC$ and $varphi$ be the Brocard angle. Then, Ceva's theorem implies:
$$sin^3varphi = sin(alpha - varphi)sin(beta-varphi)sin(gamma-varphi).$$
Now an immediate usage of the Jensen's inequality gives $sinvarphileqfrac 12$ or equivalently:
$$varphileq 30^circ.$$
On the other hand, one can prove that $$cotvarphi = cotalpha+cotbeta+cotgamma = dfrac{a^2+b^2+c^2}{4S}geqsqrt{3}.$$



Another way is simply to state Finsler-Hadwiger inequality:
$$a^2+b^2+c^2geq 4sqrt{3}+(a-b)^2+(b-c)^2+(c-a)^2.$$






share|cite|improve this answer





















    Your Answer





    StackExchange.ifUsing("editor", function () {
    return StackExchange.using("mathjaxEditing", function () {
    StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
    StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
    });
    });
    }, "mathjax-editing");

    StackExchange.ready(function() {
    var channelOptions = {
    tags: "".split(" "),
    id: "69"
    };
    initTagRenderer("".split(" "), "".split(" "), channelOptions);

    StackExchange.using("externalEditor", function() {
    // Have to fire editor after snippets, if snippets enabled
    if (StackExchange.settings.snippets.snippetsEnabled) {
    StackExchange.using("snippets", function() {
    createEditor();
    });
    }
    else {
    createEditor();
    }
    });

    function createEditor() {
    StackExchange.prepareEditor({
    heartbeatType: 'answer',
    autoActivateHeartbeat: false,
    convertImagesToLinks: true,
    noModals: true,
    showLowRepImageUploadWarning: true,
    reputationToPostImages: 10,
    bindNavPrevention: true,
    postfix: "",
    imageUploader: {
    brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
    contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
    allowUrls: true
    },
    noCode: true, onDemand: true,
    discardSelector: ".discard-answer"
    ,immediatelyShowMarkdownHelp:true
    });


    }
    });














    draft saved

    draft discarded


















    StackExchange.ready(
    function () {
    StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2143589%2fimo-problem-1961-different-proofs-of-the-weitzenbock-inequality%23new-answer', 'question_page');
    }
    );

    Post as a guest















    Required, but never shown

























    4 Answers
    4






    active

    oldest

    votes








    4 Answers
    4






    active

    oldest

    votes









    active

    oldest

    votes






    active

    oldest

    votes









    9














    What about a proof (almost) without words?



    enter image description here



    Let us consider a triangle $ABC$ and its Fermat-Torricelli-Steiner point, such that $widehat{CFA}=widehat{AFC}=widehat{CFB}=120^circ$. It is well-known that $F$ lies on each line joining a vertex of $ABC$ with the free vertex of the equilateral triangle built on the opposite side. By replicating the (colored) triangles $AFC,CFB,BFA$ four times we reach the configuration depicted above. Due to the triangular holes
    $$ [ABV_C]+[ACV_B]+[BCV_A]geq 3[ABC] $$
    that is equivalent to the claim.





    There also is an efficient variational approach. If $ABC$ is a triangle in the plane with $AB=c,AC=b$, the locus of points $P$ such that $PB^2+PC^2=b^2+c^2$ is a circle through $A$ centered at the midpoint of $BC$. It follows that by moving the vertex $A$ along such circle, till meeting the perpendicular bisector of $BC$, the area of $ABC$ increases but $AB^2+BC^2+AC^2$ stays the same. It follows that for a fixed $a^2+b^2+c^2$, the maximum area is achieved only by the equilateral triangle, and $a^2+b^2+c^2geq 4sqrt{3}Delta$ follows.





    Yet another way is to exploit the Cauchy-Schwarz inequality. We have
    $$ (1+1+1)(a^4+b^4+c^4)geq (a^2+b^2+c^2)^2 $$
    hence:
    $$ 4Delta = sqrt{(a^2+b^2+c^2)^2-2(a^4+b^4+c^4)} leq frac{1}{sqrt{3}}(a^2+b^2+c^2). $$





    Update. Yet another way is to prove a stronger inequality, namely
    $$ ab+ac+bc geq 4Deltasqrt{3} $$
    which follows from $ab=frac{2Delta}{sin C}$ and the convexity of $frac{1}{sintheta}$ over $(0,pi)$.






    share|cite|improve this answer



















    • 1




      Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      May 16 '17 at 18:34










    • @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
      – Jack D'Aurizio
      May 16 '17 at 18:36










    • Wow, this is crisp and pretty!
      – NivPai
      Aug 13 at 17:24
















    9














    What about a proof (almost) without words?



    enter image description here



    Let us consider a triangle $ABC$ and its Fermat-Torricelli-Steiner point, such that $widehat{CFA}=widehat{AFC}=widehat{CFB}=120^circ$. It is well-known that $F$ lies on each line joining a vertex of $ABC$ with the free vertex of the equilateral triangle built on the opposite side. By replicating the (colored) triangles $AFC,CFB,BFA$ four times we reach the configuration depicted above. Due to the triangular holes
    $$ [ABV_C]+[ACV_B]+[BCV_A]geq 3[ABC] $$
    that is equivalent to the claim.





    There also is an efficient variational approach. If $ABC$ is a triangle in the plane with $AB=c,AC=b$, the locus of points $P$ such that $PB^2+PC^2=b^2+c^2$ is a circle through $A$ centered at the midpoint of $BC$. It follows that by moving the vertex $A$ along such circle, till meeting the perpendicular bisector of $BC$, the area of $ABC$ increases but $AB^2+BC^2+AC^2$ stays the same. It follows that for a fixed $a^2+b^2+c^2$, the maximum area is achieved only by the equilateral triangle, and $a^2+b^2+c^2geq 4sqrt{3}Delta$ follows.





    Yet another way is to exploit the Cauchy-Schwarz inequality. We have
    $$ (1+1+1)(a^4+b^4+c^4)geq (a^2+b^2+c^2)^2 $$
    hence:
    $$ 4Delta = sqrt{(a^2+b^2+c^2)^2-2(a^4+b^4+c^4)} leq frac{1}{sqrt{3}}(a^2+b^2+c^2). $$





    Update. Yet another way is to prove a stronger inequality, namely
    $$ ab+ac+bc geq 4Deltasqrt{3} $$
    which follows from $ab=frac{2Delta}{sin C}$ and the convexity of $frac{1}{sintheta}$ over $(0,pi)$.






    share|cite|improve this answer



















    • 1




      Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      May 16 '17 at 18:34










    • @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
      – Jack D'Aurizio
      May 16 '17 at 18:36










    • Wow, this is crisp and pretty!
      – NivPai
      Aug 13 at 17:24














    9












    9








    9






    What about a proof (almost) without words?



    enter image description here



    Let us consider a triangle $ABC$ and its Fermat-Torricelli-Steiner point, such that $widehat{CFA}=widehat{AFC}=widehat{CFB}=120^circ$. It is well-known that $F$ lies on each line joining a vertex of $ABC$ with the free vertex of the equilateral triangle built on the opposite side. By replicating the (colored) triangles $AFC,CFB,BFA$ four times we reach the configuration depicted above. Due to the triangular holes
    $$ [ABV_C]+[ACV_B]+[BCV_A]geq 3[ABC] $$
    that is equivalent to the claim.





    There also is an efficient variational approach. If $ABC$ is a triangle in the plane with $AB=c,AC=b$, the locus of points $P$ such that $PB^2+PC^2=b^2+c^2$ is a circle through $A$ centered at the midpoint of $BC$. It follows that by moving the vertex $A$ along such circle, till meeting the perpendicular bisector of $BC$, the area of $ABC$ increases but $AB^2+BC^2+AC^2$ stays the same. It follows that for a fixed $a^2+b^2+c^2$, the maximum area is achieved only by the equilateral triangle, and $a^2+b^2+c^2geq 4sqrt{3}Delta$ follows.





    Yet another way is to exploit the Cauchy-Schwarz inequality. We have
    $$ (1+1+1)(a^4+b^4+c^4)geq (a^2+b^2+c^2)^2 $$
    hence:
    $$ 4Delta = sqrt{(a^2+b^2+c^2)^2-2(a^4+b^4+c^4)} leq frac{1}{sqrt{3}}(a^2+b^2+c^2). $$





    Update. Yet another way is to prove a stronger inequality, namely
    $$ ab+ac+bc geq 4Deltasqrt{3} $$
    which follows from $ab=frac{2Delta}{sin C}$ and the convexity of $frac{1}{sintheta}$ over $(0,pi)$.






    share|cite|improve this answer














    What about a proof (almost) without words?



    enter image description here



    Let us consider a triangle $ABC$ and its Fermat-Torricelli-Steiner point, such that $widehat{CFA}=widehat{AFC}=widehat{CFB}=120^circ$. It is well-known that $F$ lies on each line joining a vertex of $ABC$ with the free vertex of the equilateral triangle built on the opposite side. By replicating the (colored) triangles $AFC,CFB,BFA$ four times we reach the configuration depicted above. Due to the triangular holes
    $$ [ABV_C]+[ACV_B]+[BCV_A]geq 3[ABC] $$
    that is equivalent to the claim.





    There also is an efficient variational approach. If $ABC$ is a triangle in the plane with $AB=c,AC=b$, the locus of points $P$ such that $PB^2+PC^2=b^2+c^2$ is a circle through $A$ centered at the midpoint of $BC$. It follows that by moving the vertex $A$ along such circle, till meeting the perpendicular bisector of $BC$, the area of $ABC$ increases but $AB^2+BC^2+AC^2$ stays the same. It follows that for a fixed $a^2+b^2+c^2$, the maximum area is achieved only by the equilateral triangle, and $a^2+b^2+c^2geq 4sqrt{3}Delta$ follows.





    Yet another way is to exploit the Cauchy-Schwarz inequality. We have
    $$ (1+1+1)(a^4+b^4+c^4)geq (a^2+b^2+c^2)^2 $$
    hence:
    $$ 4Delta = sqrt{(a^2+b^2+c^2)^2-2(a^4+b^4+c^4)} leq frac{1}{sqrt{3}}(a^2+b^2+c^2). $$





    Update. Yet another way is to prove a stronger inequality, namely
    $$ ab+ac+bc geq 4Deltasqrt{3} $$
    which follows from $ab=frac{2Delta}{sin C}$ and the convexity of $frac{1}{sintheta}$ over $(0,pi)$.







    share|cite|improve this answer














    share|cite|improve this answer



    share|cite|improve this answer








    edited Aug 14 at 14:45

























    answered May 12 '17 at 21:24









    Jack D'Aurizio

    286k33279655




    286k33279655








    • 1




      Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      May 16 '17 at 18:34










    • @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
      – Jack D'Aurizio
      May 16 '17 at 18:36










    • Wow, this is crisp and pretty!
      – NivPai
      Aug 13 at 17:24














    • 1




      Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      May 16 '17 at 18:34










    • @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
      – Jack D'Aurizio
      May 16 '17 at 18:36










    • Wow, this is crisp and pretty!
      – NivPai
      Aug 13 at 17:24








    1




    1




    Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    May 16 '17 at 18:34




    Very good!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    May 16 '17 at 18:34












    @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
    – Jack D'Aurizio
    May 16 '17 at 18:36




    @IsraelMeirelesChrisostomo: there are quite a few exclamation marks, thank you! :D
    – Jack D'Aurizio
    May 16 '17 at 18:36












    Wow, this is crisp and pretty!
    – NivPai
    Aug 13 at 17:24




    Wow, this is crisp and pretty!
    – NivPai
    Aug 13 at 17:24











    3














    There are some proofs at https://artofproblemsolving.com/community/c6h14388 and a generalization at https://artofproblemsolving.com/community/c6h2958 . Let me actually reproduce my post in the latter thread in full here (with minor modifications -- neither my exposition nor my LaTeX were particularly good back in 2005), which proves several generalizations:




    Theorem 1. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $XYZ$ be a triangle with sidelengths $x = YZ$, $y = ZX$, $z = XY$ and area $T$.



    (a) (Pedoe inequality) Then, we have the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar.



    (b) (Bottema inequality) If $P$ is an arbitrary point in the plane of triangle $ABC$, then we have the inequality



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    (c) (Equivalent version of (a)) Let $A = measuredangle CAB$, $B = measuredangle ABC$, $C = measuredangle BCA$ be the angles of triangle $ABC$. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of triangle $XYZ$. Then,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y geq 2$.



    This inequality becomes an equality if and only if the triangles $ABC$ and $XYZ$ are similar.




    Proof of Theorem 1.



    (a) First note that the inequality which we must prove,



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$,



    is homogeneous in the sidelengths $x$, $y$, $z$ of the triangle $XYZ$ (in fact, these sidelengths occur in the power $2$ on the left hand side, and on the right hand side they occur in the power $2$ as well, since the area of a triangle depends quadratically on its sidelengths). Hence, this inequality is invariant under any similitude transformation applied to triangle $XYZ$; in other words, we can move, reflect, rotate and stretch the triangle $XYZ$ as we wish, and the inequality remains equivalent. But, of course, by applying similitude transformations to triangle $XYZ$, we can always achieve a situation when $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. Hence, in order to prove Theorem 1 (a) for every two triangles $ABC$ and $XYZ$, it is enough to prove it for two triangles $ABC$ and $XYZ$ in this special situation.



    So, let us WLOG assume that the triangles $ABC$ and $XYZ$ are in this special situation, i. e. that we have $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. We have to prove the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    By the cosine law in triangle $ABX$, we have



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$.



    Let's figure out what this equation means. At first, $AB = c$. Then, since $B = Y$, we have $XB = XY = z$. Finally, we have either $measuredangle ABX = measuredangle ABC - measuredangle XBC$ or $measuredangle ABX = measuredangle XBC - measuredangle ABC$ (depending on the arrangement of the points), but in both cases $cos measuredangle ABX = cos left(measuredangle ABC - measuredangle XBCright)$. Since $B = Y$ and $C = Z$, we can rewrite the angle $measuredangle XBC$ as $measuredangle XYZ$. Thus,



    $cosmeasuredangle ABX=cosleft(measuredangle ABC-measuredangle XYZright)=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$.



    By the cosine law in triangles $ABC$ and $XYZ$, we have



    $cosmeasuredangle ABC=dfrac{c^2+a^2-b^2}{2ca}$ and
    $cosmeasuredangle XYZ=dfrac{z^2+x^2-y^2}{2zx}$.



    Also, by a well-known formula, the area $S$ of triangle $ABC$ equals $S=dfrac12cdot casinmeasuredangle ABC$; thus, $sinmeasuredangle ABC=dfrac{2S}{ca}$. Similarly, $sinmeasuredangle XYZ=dfrac{2T}{zx}$.



    Hence,



    $cosmeasuredangle ABX=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$
    $=dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}$.



    Together with $AB = c$ and $XB = z$, this makes the equality



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$



    transform into



    $AX^2=c^2+z^2-2cdot ccdot zcdotleft(dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}right)$.



    This immediately simplifies to



    $AX^2=c^2+z^2-2left(dfrac{left(c^2+a^2-b^2right)left(z^2+x^2-y^2right)}{4ax}+dfrac{4ST}{ax}right)$.



    After some algebra, this becomes



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)-2Qright)-16ST}{2ax}$,



    where $Q = left(x-aright)left(c^2x-z^2aright)$.



    Since $Y = B$ and $Z = C$, we have $YZ = BC$, or, in other words, $x = a$. Thus, $x - a = 0$, so that $Q = 0$. Hence, the above formula for $AX^2$ simplifies to



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16ST}{2ax}$.



    Since (obviously) $AX^2geq 0$, we thus have



    $left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16STgeq 0$,



    and thus $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$. So the inequality of Theorem 1 (a) is proven. Equality holds if and only if equality holds in the inequality $AX^2geq 0$; but this is clearly the case if and only if the points $A$ and $X$ coincide, i. e. if the triangles $ABC$ and $XYZ$ are congruent. Now, of course, since the triangle $XYZ$ we are dealing with is not the initial triangle $XYZ$, but just its image under a similitude transformation, the general equality condition is that the triangles $ABC$ and $XYZ$ are similar (not necessarily being congruent). So Theorem 1 (a) is proven.



    (b) In our proof of Theorem 1 (a), we WLOG assumed that $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. This time we will assume the same situation, but with a little modification: The point $X$ should lie not in the same halfplane with respect to the line $BC$ as the point $A$, but exactly in the other halfplane. In other words, we have $Y = B$ and $Z = C$ and the points $A$ and $X$ lie in different halfplanes with respect to the line $BC$. Then, by a calculation similar to the one we did in the proof of Theorem 1 (a), we find



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2ax}$



    (note the $+$ sign instead of the $-$). Since $x = a$, we have $ax=a^2$, and thus we can rewrite this as



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2a^2}$.



    Hence,



    $a^2cdot AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2}$
    ${=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Taking the square root of this, we get



    $acdot AX=sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Hence, the inequality which is to be proven,



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$,



    is equivalent to $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    So we have to prove $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    Well, first, by the Ptolemy inequality for the quadrilateral $PBXC$, we have $CXcdot BP+XBcdot CPgeq BCcdot PX$. Since $B = Y$ and $C = Z$, we have $CX = ZX = y$, $XB = XY = z$ and $BC = YZ = x$, and therefore this inequality rewrites as $ycdot BP+zcdot CPgeq xcdot PX$. Also, by the triangle inequality in the triangle $APX$, we have $AP+PXgeq AX$. Thus,



    $xcdot AP+ycdot BP+zcdot CP=xcdot AP+left(ycdot BP+zcdot CPright)geq xcdot AP+xcdot PX$
    $=xcdotleft(AP+PXright)geq xcdot AX=acdot AX$,



    where we have used $x = a$ in the last step. This proves Theorem 1 (b).



    (c) By the Cosine Law in triangle $ABC$, we have $cos A=dfrac{b^2+c^2-a^2}{2bc}$. On the other hand, by a well-known formula for the area of a triangle, the area $S$ of triangle $ABC$ equals $S=dfrac12bcsin A$. Thus, $sin A=dfrac{2S}{bc}$. Hence,



    $cot A=cos A / sin A=dfrac{b^2+c^2-a^2}{2bc} / dfrac{2S}{bc}=dfrac{b^2+c^2-a^2}{4S}$.



    Similarly, $cot B=dfrac{c^2+a^2-b^2}{4S}$. Thus, $cot A+cot B=dfrac{b^2+c^2-a^2}{4S}+dfrac{c^2+a^2-b^2}{4S}=dfrac{2c^2}{4S}=dfrac{c^2}{2S}$.



    Similarly to how we proved $cot A=dfrac{b^2+c^2-a^2}{4S}$, we can find that $cot Z=dfrac{x^2+y^2-z^2}{4T}$. Thus,



    $cot Acot Z+cot Bcot Z=left(cot A+cot Bright)cdotcot Z=dfrac{c^2}{2S}cdotdfrac{x^2+y^2-z^2}{4T}$
    $=dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$.



    Similarly,



    $cot Bcot X+cot Ccot X=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}$ and
    $cot Ccot Y+cot Acot Y=dfrac{b^2left(z^2+x^2-y^2right)}{8ST}$.



    Thus,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $=left(cot Bcot X+cot Ccot Xright)+left(cot Ccot Y+cot Acot Yright)+left(cot Acot Z+cot Bcot Zright)$
    $=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}+dfrac{b^2left(z^2+x^2-y^2right)}{8ST}+dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$
    $=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{8ST}$.



    But after Theorem 1 (a), we have $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar. Hence,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $geqdfrac{16ST}{8ST}=2$,



    with equality if and only if the triangles $ABC$ and $XYZ$ are similar. And Theorem 1 (c) is proven.



    Note that Theorem 1 (a) yields the following geometric inequality:




    Corollary 2. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$.




    Before proving the Corollary, we show two useful facts:




    Corollary 3 (a well-known formula). Let $XYZ$ be a triangle. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of this triangle. Then, $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$.




    Proof of Corollary 3. Theorem 1 (c) (applied to $XYZ$, $x$, $y$ and $z$ instead of $ABC$, $a$, $b$ and $c$) shows that $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y geq 2$, and that this inequality becomes an equality if and only if the triangles $XYZ$ and $XYZ$ are similar. Thus, of course, this inequality becomes an equality (since the triangles $XYZ$ and $XYZ$ are similar). Hence, $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y = 2$. Therefore,



    $2 = cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y$
    $= 2 left(cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Yright)$.



    Dividing both sides of this equality by $2$, we obtain $1 = cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y$. This proves Corollary 3.




    Theorem 4 (Conway substitution theorem). Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. This triangle satisfies $y^2+z^2-x^2=2u$, $z^2+x^2-y^2=2v$, $x^2+y^2-z^2=2w$. The area $T$ of this triangle equals $T=dfrac12sqrt{vw+wu+uv}$. If $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ are the angles of this triangle, then $cot X=dfrac{u}{2T}$, $cot Y=dfrac{v}{2T}$ and $cot Z=dfrac{w}{2T}$.




    Proof of Theorem 4. Since the numbers $v + w$, $w + u$, $u + v$ are nonnegative, their square roots $sqrt{v+w}$, $sqrt{w+u}$, $sqrt{u+v}$ exist, and, of course, are nonnegative as well. Now, we have $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w}$, since



    $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w} Longleftarrow left(sqrt{w+u}+sqrt{u+v}right)^2geq v+w$



    $Longleftarrow left(w+uright)+left(u+vright)+2sqrt{left(w+uright)left(u+vright)}geq v+w$



    $Longleftarrow 2sqrt{left(w+uright)left(u+vright)}geq -2u$ [this is not trivial, since $u$ needs not be $geq 0$!]



    $Longleftarrow sqrt{left(w+uright)left(u+vright)}geq-u Longleftarrow left(w+uright)left(u+vright)geqleft(-uright)^2$



    $Longleftarrow u^2+left(vw+wu+uvright)geq u^2 Longleftarrow vw+wu+uvgeq 0$.



    Similarly, $sqrt{u+v}+sqrt{v+w}geqsqrt{w+u}$ and $sqrt{v+w}+sqrt{w+u}geqsqrt{u+v}$. Thus, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. Denote by $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ the angles of this triangle. We have



    $y^2+z^2-x^2=left(sqrt{w+u}right)^2+left(sqrt{u+v}right)^2-left(sqrt{v+w}right)^2$
    $=left(w+uright)+left(u+vright)-left(v+wright)=2u$



    and similarly $z^2+x^2-y^2=2v$ and $x^2+y^2-z^2=2w$. Now, denoting the area of triangle $XYZ$ by $T$, we note that, as shown during the proof of Theorem 1 (c), we have $cot Z=dfrac{x^2+y^2-z^2}{4T}$, so that $cot Z=dfrac{2w}{4T}=dfrac{w}{2T}$, and similarly $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$.



    Corollary 3 yields $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$. Using the relations $cot Z=dfrac{w}{2T}$, $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$, this equality becomes $dfrac{v}{2T}cdotdfrac{w}{2T}+dfrac{w}{2T}cdotdfrac{u}{2T}+dfrac{u}{2T}cdotdfrac{v}{2T}=1$. Hence, $vw+wu+uv=4T^2$, and thus $T=dfrac12sqrt{4T^2}=dfrac12sqrt{vw+wu+uv}$. This completes the proof of Theorem 4.



    Proof of Corollary 2. Construct a triangle $XYZ$ for our values of $u$, $v$, $w$ according to Theorem 4. Then, applying Theorem 1 (a) to the triangles $ABC$ and $XYZ$, we get



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    Using the formulas given in Theorem 4, we can rewrite this as



    $a^2cdot 2u+b^2cdot 2v+c^2cdot 2wgeq 16Scdotdfrac12sqrt{vw+wu+uv}$.



    This simplifies to $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$. This proves Corollary 2.




    Corollary 5 (Weitzenböck inequality). Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Then, $a^2 + b^2 + c^2 geq 4sqrt{3} S$.




    Proof of Corollary 5. Corollary 5 follows from Corollary 2 (applied to $u=1$, $v=1$ and $w=1$).



    [Postscriptum: I have been atrocious with attribution and citation back in 2005. I am not sure where the ideas in the above proofs come from; I might have been just the scribe. Theorem 1 (a) was discovered in 1943 by Daniel Pedoe and much earlier (without the equality case) by Neuberg. It also appears as 10.8 in O. Bottema, R. Z. Djordjevic, R. R. Janic, D. S. Mitrinovic, P. M. Vasic, Geometric Inequalities, Wolters-Noordhoff 1969. Theorem 1 (b) appears as 12.56 in the same book and is credited to O. Bottema, Hoofdstukken uit de elementaire meetkunde, Den Haag 1944 (I don't have this book, but interestingly it is not out of print). I might have picked up the Conway substitution from posts by John Conway on the Hyacinthos newsgroup long ago.]






    share|cite|improve this answer























    • Uow thank's @darij grinberg!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      Mar 3 '17 at 2:59
















    3














    There are some proofs at https://artofproblemsolving.com/community/c6h14388 and a generalization at https://artofproblemsolving.com/community/c6h2958 . Let me actually reproduce my post in the latter thread in full here (with minor modifications -- neither my exposition nor my LaTeX were particularly good back in 2005), which proves several generalizations:




    Theorem 1. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $XYZ$ be a triangle with sidelengths $x = YZ$, $y = ZX$, $z = XY$ and area $T$.



    (a) (Pedoe inequality) Then, we have the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar.



    (b) (Bottema inequality) If $P$ is an arbitrary point in the plane of triangle $ABC$, then we have the inequality



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    (c) (Equivalent version of (a)) Let $A = measuredangle CAB$, $B = measuredangle ABC$, $C = measuredangle BCA$ be the angles of triangle $ABC$. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of triangle $XYZ$. Then,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y geq 2$.



    This inequality becomes an equality if and only if the triangles $ABC$ and $XYZ$ are similar.




    Proof of Theorem 1.



    (a) First note that the inequality which we must prove,



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$,



    is homogeneous in the sidelengths $x$, $y$, $z$ of the triangle $XYZ$ (in fact, these sidelengths occur in the power $2$ on the left hand side, and on the right hand side they occur in the power $2$ as well, since the area of a triangle depends quadratically on its sidelengths). Hence, this inequality is invariant under any similitude transformation applied to triangle $XYZ$; in other words, we can move, reflect, rotate and stretch the triangle $XYZ$ as we wish, and the inequality remains equivalent. But, of course, by applying similitude transformations to triangle $XYZ$, we can always achieve a situation when $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. Hence, in order to prove Theorem 1 (a) for every two triangles $ABC$ and $XYZ$, it is enough to prove it for two triangles $ABC$ and $XYZ$ in this special situation.



    So, let us WLOG assume that the triangles $ABC$ and $XYZ$ are in this special situation, i. e. that we have $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. We have to prove the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    By the cosine law in triangle $ABX$, we have



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$.



    Let's figure out what this equation means. At first, $AB = c$. Then, since $B = Y$, we have $XB = XY = z$. Finally, we have either $measuredangle ABX = measuredangle ABC - measuredangle XBC$ or $measuredangle ABX = measuredangle XBC - measuredangle ABC$ (depending on the arrangement of the points), but in both cases $cos measuredangle ABX = cos left(measuredangle ABC - measuredangle XBCright)$. Since $B = Y$ and $C = Z$, we can rewrite the angle $measuredangle XBC$ as $measuredangle XYZ$. Thus,



    $cosmeasuredangle ABX=cosleft(measuredangle ABC-measuredangle XYZright)=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$.



    By the cosine law in triangles $ABC$ and $XYZ$, we have



    $cosmeasuredangle ABC=dfrac{c^2+a^2-b^2}{2ca}$ and
    $cosmeasuredangle XYZ=dfrac{z^2+x^2-y^2}{2zx}$.



    Also, by a well-known formula, the area $S$ of triangle $ABC$ equals $S=dfrac12cdot casinmeasuredangle ABC$; thus, $sinmeasuredangle ABC=dfrac{2S}{ca}$. Similarly, $sinmeasuredangle XYZ=dfrac{2T}{zx}$.



    Hence,



    $cosmeasuredangle ABX=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$
    $=dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}$.



    Together with $AB = c$ and $XB = z$, this makes the equality



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$



    transform into



    $AX^2=c^2+z^2-2cdot ccdot zcdotleft(dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}right)$.



    This immediately simplifies to



    $AX^2=c^2+z^2-2left(dfrac{left(c^2+a^2-b^2right)left(z^2+x^2-y^2right)}{4ax}+dfrac{4ST}{ax}right)$.



    After some algebra, this becomes



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)-2Qright)-16ST}{2ax}$,



    where $Q = left(x-aright)left(c^2x-z^2aright)$.



    Since $Y = B$ and $Z = C$, we have $YZ = BC$, or, in other words, $x = a$. Thus, $x - a = 0$, so that $Q = 0$. Hence, the above formula for $AX^2$ simplifies to



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16ST}{2ax}$.



    Since (obviously) $AX^2geq 0$, we thus have



    $left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16STgeq 0$,



    and thus $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$. So the inequality of Theorem 1 (a) is proven. Equality holds if and only if equality holds in the inequality $AX^2geq 0$; but this is clearly the case if and only if the points $A$ and $X$ coincide, i. e. if the triangles $ABC$ and $XYZ$ are congruent. Now, of course, since the triangle $XYZ$ we are dealing with is not the initial triangle $XYZ$, but just its image under a similitude transformation, the general equality condition is that the triangles $ABC$ and $XYZ$ are similar (not necessarily being congruent). So Theorem 1 (a) is proven.



    (b) In our proof of Theorem 1 (a), we WLOG assumed that $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. This time we will assume the same situation, but with a little modification: The point $X$ should lie not in the same halfplane with respect to the line $BC$ as the point $A$, but exactly in the other halfplane. In other words, we have $Y = B$ and $Z = C$ and the points $A$ and $X$ lie in different halfplanes with respect to the line $BC$. Then, by a calculation similar to the one we did in the proof of Theorem 1 (a), we find



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2ax}$



    (note the $+$ sign instead of the $-$). Since $x = a$, we have $ax=a^2$, and thus we can rewrite this as



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2a^2}$.



    Hence,



    $a^2cdot AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2}$
    ${=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Taking the square root of this, we get



    $acdot AX=sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Hence, the inequality which is to be proven,



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$,



    is equivalent to $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    So we have to prove $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    Well, first, by the Ptolemy inequality for the quadrilateral $PBXC$, we have $CXcdot BP+XBcdot CPgeq BCcdot PX$. Since $B = Y$ and $C = Z$, we have $CX = ZX = y$, $XB = XY = z$ and $BC = YZ = x$, and therefore this inequality rewrites as $ycdot BP+zcdot CPgeq xcdot PX$. Also, by the triangle inequality in the triangle $APX$, we have $AP+PXgeq AX$. Thus,



    $xcdot AP+ycdot BP+zcdot CP=xcdot AP+left(ycdot BP+zcdot CPright)geq xcdot AP+xcdot PX$
    $=xcdotleft(AP+PXright)geq xcdot AX=acdot AX$,



    where we have used $x = a$ in the last step. This proves Theorem 1 (b).



    (c) By the Cosine Law in triangle $ABC$, we have $cos A=dfrac{b^2+c^2-a^2}{2bc}$. On the other hand, by a well-known formula for the area of a triangle, the area $S$ of triangle $ABC$ equals $S=dfrac12bcsin A$. Thus, $sin A=dfrac{2S}{bc}$. Hence,



    $cot A=cos A / sin A=dfrac{b^2+c^2-a^2}{2bc} / dfrac{2S}{bc}=dfrac{b^2+c^2-a^2}{4S}$.



    Similarly, $cot B=dfrac{c^2+a^2-b^2}{4S}$. Thus, $cot A+cot B=dfrac{b^2+c^2-a^2}{4S}+dfrac{c^2+a^2-b^2}{4S}=dfrac{2c^2}{4S}=dfrac{c^2}{2S}$.



    Similarly to how we proved $cot A=dfrac{b^2+c^2-a^2}{4S}$, we can find that $cot Z=dfrac{x^2+y^2-z^2}{4T}$. Thus,



    $cot Acot Z+cot Bcot Z=left(cot A+cot Bright)cdotcot Z=dfrac{c^2}{2S}cdotdfrac{x^2+y^2-z^2}{4T}$
    $=dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$.



    Similarly,



    $cot Bcot X+cot Ccot X=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}$ and
    $cot Ccot Y+cot Acot Y=dfrac{b^2left(z^2+x^2-y^2right)}{8ST}$.



    Thus,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $=left(cot Bcot X+cot Ccot Xright)+left(cot Ccot Y+cot Acot Yright)+left(cot Acot Z+cot Bcot Zright)$
    $=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}+dfrac{b^2left(z^2+x^2-y^2right)}{8ST}+dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$
    $=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{8ST}$.



    But after Theorem 1 (a), we have $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar. Hence,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $geqdfrac{16ST}{8ST}=2$,



    with equality if and only if the triangles $ABC$ and $XYZ$ are similar. And Theorem 1 (c) is proven.



    Note that Theorem 1 (a) yields the following geometric inequality:




    Corollary 2. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$.




    Before proving the Corollary, we show two useful facts:




    Corollary 3 (a well-known formula). Let $XYZ$ be a triangle. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of this triangle. Then, $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$.




    Proof of Corollary 3. Theorem 1 (c) (applied to $XYZ$, $x$, $y$ and $z$ instead of $ABC$, $a$, $b$ and $c$) shows that $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y geq 2$, and that this inequality becomes an equality if and only if the triangles $XYZ$ and $XYZ$ are similar. Thus, of course, this inequality becomes an equality (since the triangles $XYZ$ and $XYZ$ are similar). Hence, $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y = 2$. Therefore,



    $2 = cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y$
    $= 2 left(cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Yright)$.



    Dividing both sides of this equality by $2$, we obtain $1 = cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y$. This proves Corollary 3.




    Theorem 4 (Conway substitution theorem). Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. This triangle satisfies $y^2+z^2-x^2=2u$, $z^2+x^2-y^2=2v$, $x^2+y^2-z^2=2w$. The area $T$ of this triangle equals $T=dfrac12sqrt{vw+wu+uv}$. If $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ are the angles of this triangle, then $cot X=dfrac{u}{2T}$, $cot Y=dfrac{v}{2T}$ and $cot Z=dfrac{w}{2T}$.




    Proof of Theorem 4. Since the numbers $v + w$, $w + u$, $u + v$ are nonnegative, their square roots $sqrt{v+w}$, $sqrt{w+u}$, $sqrt{u+v}$ exist, and, of course, are nonnegative as well. Now, we have $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w}$, since



    $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w} Longleftarrow left(sqrt{w+u}+sqrt{u+v}right)^2geq v+w$



    $Longleftarrow left(w+uright)+left(u+vright)+2sqrt{left(w+uright)left(u+vright)}geq v+w$



    $Longleftarrow 2sqrt{left(w+uright)left(u+vright)}geq -2u$ [this is not trivial, since $u$ needs not be $geq 0$!]



    $Longleftarrow sqrt{left(w+uright)left(u+vright)}geq-u Longleftarrow left(w+uright)left(u+vright)geqleft(-uright)^2$



    $Longleftarrow u^2+left(vw+wu+uvright)geq u^2 Longleftarrow vw+wu+uvgeq 0$.



    Similarly, $sqrt{u+v}+sqrt{v+w}geqsqrt{w+u}$ and $sqrt{v+w}+sqrt{w+u}geqsqrt{u+v}$. Thus, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. Denote by $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ the angles of this triangle. We have



    $y^2+z^2-x^2=left(sqrt{w+u}right)^2+left(sqrt{u+v}right)^2-left(sqrt{v+w}right)^2$
    $=left(w+uright)+left(u+vright)-left(v+wright)=2u$



    and similarly $z^2+x^2-y^2=2v$ and $x^2+y^2-z^2=2w$. Now, denoting the area of triangle $XYZ$ by $T$, we note that, as shown during the proof of Theorem 1 (c), we have $cot Z=dfrac{x^2+y^2-z^2}{4T}$, so that $cot Z=dfrac{2w}{4T}=dfrac{w}{2T}$, and similarly $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$.



    Corollary 3 yields $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$. Using the relations $cot Z=dfrac{w}{2T}$, $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$, this equality becomes $dfrac{v}{2T}cdotdfrac{w}{2T}+dfrac{w}{2T}cdotdfrac{u}{2T}+dfrac{u}{2T}cdotdfrac{v}{2T}=1$. Hence, $vw+wu+uv=4T^2$, and thus $T=dfrac12sqrt{4T^2}=dfrac12sqrt{vw+wu+uv}$. This completes the proof of Theorem 4.



    Proof of Corollary 2. Construct a triangle $XYZ$ for our values of $u$, $v$, $w$ according to Theorem 4. Then, applying Theorem 1 (a) to the triangles $ABC$ and $XYZ$, we get



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    Using the formulas given in Theorem 4, we can rewrite this as



    $a^2cdot 2u+b^2cdot 2v+c^2cdot 2wgeq 16Scdotdfrac12sqrt{vw+wu+uv}$.



    This simplifies to $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$. This proves Corollary 2.




    Corollary 5 (Weitzenböck inequality). Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Then, $a^2 + b^2 + c^2 geq 4sqrt{3} S$.




    Proof of Corollary 5. Corollary 5 follows from Corollary 2 (applied to $u=1$, $v=1$ and $w=1$).



    [Postscriptum: I have been atrocious with attribution and citation back in 2005. I am not sure where the ideas in the above proofs come from; I might have been just the scribe. Theorem 1 (a) was discovered in 1943 by Daniel Pedoe and much earlier (without the equality case) by Neuberg. It also appears as 10.8 in O. Bottema, R. Z. Djordjevic, R. R. Janic, D. S. Mitrinovic, P. M. Vasic, Geometric Inequalities, Wolters-Noordhoff 1969. Theorem 1 (b) appears as 12.56 in the same book and is credited to O. Bottema, Hoofdstukken uit de elementaire meetkunde, Den Haag 1944 (I don't have this book, but interestingly it is not out of print). I might have picked up the Conway substitution from posts by John Conway on the Hyacinthos newsgroup long ago.]






    share|cite|improve this answer























    • Uow thank's @darij grinberg!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      Mar 3 '17 at 2:59














    3












    3








    3






    There are some proofs at https://artofproblemsolving.com/community/c6h14388 and a generalization at https://artofproblemsolving.com/community/c6h2958 . Let me actually reproduce my post in the latter thread in full here (with minor modifications -- neither my exposition nor my LaTeX were particularly good back in 2005), which proves several generalizations:




    Theorem 1. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $XYZ$ be a triangle with sidelengths $x = YZ$, $y = ZX$, $z = XY$ and area $T$.



    (a) (Pedoe inequality) Then, we have the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar.



    (b) (Bottema inequality) If $P$ is an arbitrary point in the plane of triangle $ABC$, then we have the inequality



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    (c) (Equivalent version of (a)) Let $A = measuredangle CAB$, $B = measuredangle ABC$, $C = measuredangle BCA$ be the angles of triangle $ABC$. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of triangle $XYZ$. Then,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y geq 2$.



    This inequality becomes an equality if and only if the triangles $ABC$ and $XYZ$ are similar.




    Proof of Theorem 1.



    (a) First note that the inequality which we must prove,



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$,



    is homogeneous in the sidelengths $x$, $y$, $z$ of the triangle $XYZ$ (in fact, these sidelengths occur in the power $2$ on the left hand side, and on the right hand side they occur in the power $2$ as well, since the area of a triangle depends quadratically on its sidelengths). Hence, this inequality is invariant under any similitude transformation applied to triangle $XYZ$; in other words, we can move, reflect, rotate and stretch the triangle $XYZ$ as we wish, and the inequality remains equivalent. But, of course, by applying similitude transformations to triangle $XYZ$, we can always achieve a situation when $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. Hence, in order to prove Theorem 1 (a) for every two triangles $ABC$ and $XYZ$, it is enough to prove it for two triangles $ABC$ and $XYZ$ in this special situation.



    So, let us WLOG assume that the triangles $ABC$ and $XYZ$ are in this special situation, i. e. that we have $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. We have to prove the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    By the cosine law in triangle $ABX$, we have



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$.



    Let's figure out what this equation means. At first, $AB = c$. Then, since $B = Y$, we have $XB = XY = z$. Finally, we have either $measuredangle ABX = measuredangle ABC - measuredangle XBC$ or $measuredangle ABX = measuredangle XBC - measuredangle ABC$ (depending on the arrangement of the points), but in both cases $cos measuredangle ABX = cos left(measuredangle ABC - measuredangle XBCright)$. Since $B = Y$ and $C = Z$, we can rewrite the angle $measuredangle XBC$ as $measuredangle XYZ$. Thus,



    $cosmeasuredangle ABX=cosleft(measuredangle ABC-measuredangle XYZright)=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$.



    By the cosine law in triangles $ABC$ and $XYZ$, we have



    $cosmeasuredangle ABC=dfrac{c^2+a^2-b^2}{2ca}$ and
    $cosmeasuredangle XYZ=dfrac{z^2+x^2-y^2}{2zx}$.



    Also, by a well-known formula, the area $S$ of triangle $ABC$ equals $S=dfrac12cdot casinmeasuredangle ABC$; thus, $sinmeasuredangle ABC=dfrac{2S}{ca}$. Similarly, $sinmeasuredangle XYZ=dfrac{2T}{zx}$.



    Hence,



    $cosmeasuredangle ABX=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$
    $=dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}$.



    Together with $AB = c$ and $XB = z$, this makes the equality



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$



    transform into



    $AX^2=c^2+z^2-2cdot ccdot zcdotleft(dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}right)$.



    This immediately simplifies to



    $AX^2=c^2+z^2-2left(dfrac{left(c^2+a^2-b^2right)left(z^2+x^2-y^2right)}{4ax}+dfrac{4ST}{ax}right)$.



    After some algebra, this becomes



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)-2Qright)-16ST}{2ax}$,



    where $Q = left(x-aright)left(c^2x-z^2aright)$.



    Since $Y = B$ and $Z = C$, we have $YZ = BC$, or, in other words, $x = a$. Thus, $x - a = 0$, so that $Q = 0$. Hence, the above formula for $AX^2$ simplifies to



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16ST}{2ax}$.



    Since (obviously) $AX^2geq 0$, we thus have



    $left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16STgeq 0$,



    and thus $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$. So the inequality of Theorem 1 (a) is proven. Equality holds if and only if equality holds in the inequality $AX^2geq 0$; but this is clearly the case if and only if the points $A$ and $X$ coincide, i. e. if the triangles $ABC$ and $XYZ$ are congruent. Now, of course, since the triangle $XYZ$ we are dealing with is not the initial triangle $XYZ$, but just its image under a similitude transformation, the general equality condition is that the triangles $ABC$ and $XYZ$ are similar (not necessarily being congruent). So Theorem 1 (a) is proven.



    (b) In our proof of Theorem 1 (a), we WLOG assumed that $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. This time we will assume the same situation, but with a little modification: The point $X$ should lie not in the same halfplane with respect to the line $BC$ as the point $A$, but exactly in the other halfplane. In other words, we have $Y = B$ and $Z = C$ and the points $A$ and $X$ lie in different halfplanes with respect to the line $BC$. Then, by a calculation similar to the one we did in the proof of Theorem 1 (a), we find



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2ax}$



    (note the $+$ sign instead of the $-$). Since $x = a$, we have $ax=a^2$, and thus we can rewrite this as



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2a^2}$.



    Hence,



    $a^2cdot AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2}$
    ${=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Taking the square root of this, we get



    $acdot AX=sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Hence, the inequality which is to be proven,



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$,



    is equivalent to $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    So we have to prove $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    Well, first, by the Ptolemy inequality for the quadrilateral $PBXC$, we have $CXcdot BP+XBcdot CPgeq BCcdot PX$. Since $B = Y$ and $C = Z$, we have $CX = ZX = y$, $XB = XY = z$ and $BC = YZ = x$, and therefore this inequality rewrites as $ycdot BP+zcdot CPgeq xcdot PX$. Also, by the triangle inequality in the triangle $APX$, we have $AP+PXgeq AX$. Thus,



    $xcdot AP+ycdot BP+zcdot CP=xcdot AP+left(ycdot BP+zcdot CPright)geq xcdot AP+xcdot PX$
    $=xcdotleft(AP+PXright)geq xcdot AX=acdot AX$,



    where we have used $x = a$ in the last step. This proves Theorem 1 (b).



    (c) By the Cosine Law in triangle $ABC$, we have $cos A=dfrac{b^2+c^2-a^2}{2bc}$. On the other hand, by a well-known formula for the area of a triangle, the area $S$ of triangle $ABC$ equals $S=dfrac12bcsin A$. Thus, $sin A=dfrac{2S}{bc}$. Hence,



    $cot A=cos A / sin A=dfrac{b^2+c^2-a^2}{2bc} / dfrac{2S}{bc}=dfrac{b^2+c^2-a^2}{4S}$.



    Similarly, $cot B=dfrac{c^2+a^2-b^2}{4S}$. Thus, $cot A+cot B=dfrac{b^2+c^2-a^2}{4S}+dfrac{c^2+a^2-b^2}{4S}=dfrac{2c^2}{4S}=dfrac{c^2}{2S}$.



    Similarly to how we proved $cot A=dfrac{b^2+c^2-a^2}{4S}$, we can find that $cot Z=dfrac{x^2+y^2-z^2}{4T}$. Thus,



    $cot Acot Z+cot Bcot Z=left(cot A+cot Bright)cdotcot Z=dfrac{c^2}{2S}cdotdfrac{x^2+y^2-z^2}{4T}$
    $=dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$.



    Similarly,



    $cot Bcot X+cot Ccot X=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}$ and
    $cot Ccot Y+cot Acot Y=dfrac{b^2left(z^2+x^2-y^2right)}{8ST}$.



    Thus,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $=left(cot Bcot X+cot Ccot Xright)+left(cot Ccot Y+cot Acot Yright)+left(cot Acot Z+cot Bcot Zright)$
    $=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}+dfrac{b^2left(z^2+x^2-y^2right)}{8ST}+dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$
    $=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{8ST}$.



    But after Theorem 1 (a), we have $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar. Hence,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $geqdfrac{16ST}{8ST}=2$,



    with equality if and only if the triangles $ABC$ and $XYZ$ are similar. And Theorem 1 (c) is proven.



    Note that Theorem 1 (a) yields the following geometric inequality:




    Corollary 2. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$.




    Before proving the Corollary, we show two useful facts:




    Corollary 3 (a well-known formula). Let $XYZ$ be a triangle. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of this triangle. Then, $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$.




    Proof of Corollary 3. Theorem 1 (c) (applied to $XYZ$, $x$, $y$ and $z$ instead of $ABC$, $a$, $b$ and $c$) shows that $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y geq 2$, and that this inequality becomes an equality if and only if the triangles $XYZ$ and $XYZ$ are similar. Thus, of course, this inequality becomes an equality (since the triangles $XYZ$ and $XYZ$ are similar). Hence, $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y = 2$. Therefore,



    $2 = cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y$
    $= 2 left(cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Yright)$.



    Dividing both sides of this equality by $2$, we obtain $1 = cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y$. This proves Corollary 3.




    Theorem 4 (Conway substitution theorem). Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. This triangle satisfies $y^2+z^2-x^2=2u$, $z^2+x^2-y^2=2v$, $x^2+y^2-z^2=2w$. The area $T$ of this triangle equals $T=dfrac12sqrt{vw+wu+uv}$. If $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ are the angles of this triangle, then $cot X=dfrac{u}{2T}$, $cot Y=dfrac{v}{2T}$ and $cot Z=dfrac{w}{2T}$.




    Proof of Theorem 4. Since the numbers $v + w$, $w + u$, $u + v$ are nonnegative, their square roots $sqrt{v+w}$, $sqrt{w+u}$, $sqrt{u+v}$ exist, and, of course, are nonnegative as well. Now, we have $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w}$, since



    $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w} Longleftarrow left(sqrt{w+u}+sqrt{u+v}right)^2geq v+w$



    $Longleftarrow left(w+uright)+left(u+vright)+2sqrt{left(w+uright)left(u+vright)}geq v+w$



    $Longleftarrow 2sqrt{left(w+uright)left(u+vright)}geq -2u$ [this is not trivial, since $u$ needs not be $geq 0$!]



    $Longleftarrow sqrt{left(w+uright)left(u+vright)}geq-u Longleftarrow left(w+uright)left(u+vright)geqleft(-uright)^2$



    $Longleftarrow u^2+left(vw+wu+uvright)geq u^2 Longleftarrow vw+wu+uvgeq 0$.



    Similarly, $sqrt{u+v}+sqrt{v+w}geqsqrt{w+u}$ and $sqrt{v+w}+sqrt{w+u}geqsqrt{u+v}$. Thus, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. Denote by $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ the angles of this triangle. We have



    $y^2+z^2-x^2=left(sqrt{w+u}right)^2+left(sqrt{u+v}right)^2-left(sqrt{v+w}right)^2$
    $=left(w+uright)+left(u+vright)-left(v+wright)=2u$



    and similarly $z^2+x^2-y^2=2v$ and $x^2+y^2-z^2=2w$. Now, denoting the area of triangle $XYZ$ by $T$, we note that, as shown during the proof of Theorem 1 (c), we have $cot Z=dfrac{x^2+y^2-z^2}{4T}$, so that $cot Z=dfrac{2w}{4T}=dfrac{w}{2T}$, and similarly $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$.



    Corollary 3 yields $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$. Using the relations $cot Z=dfrac{w}{2T}$, $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$, this equality becomes $dfrac{v}{2T}cdotdfrac{w}{2T}+dfrac{w}{2T}cdotdfrac{u}{2T}+dfrac{u}{2T}cdotdfrac{v}{2T}=1$. Hence, $vw+wu+uv=4T^2$, and thus $T=dfrac12sqrt{4T^2}=dfrac12sqrt{vw+wu+uv}$. This completes the proof of Theorem 4.



    Proof of Corollary 2. Construct a triangle $XYZ$ for our values of $u$, $v$, $w$ according to Theorem 4. Then, applying Theorem 1 (a) to the triangles $ABC$ and $XYZ$, we get



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    Using the formulas given in Theorem 4, we can rewrite this as



    $a^2cdot 2u+b^2cdot 2v+c^2cdot 2wgeq 16Scdotdfrac12sqrt{vw+wu+uv}$.



    This simplifies to $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$. This proves Corollary 2.




    Corollary 5 (Weitzenböck inequality). Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Then, $a^2 + b^2 + c^2 geq 4sqrt{3} S$.




    Proof of Corollary 5. Corollary 5 follows from Corollary 2 (applied to $u=1$, $v=1$ and $w=1$).



    [Postscriptum: I have been atrocious with attribution and citation back in 2005. I am not sure where the ideas in the above proofs come from; I might have been just the scribe. Theorem 1 (a) was discovered in 1943 by Daniel Pedoe and much earlier (without the equality case) by Neuberg. It also appears as 10.8 in O. Bottema, R. Z. Djordjevic, R. R. Janic, D. S. Mitrinovic, P. M. Vasic, Geometric Inequalities, Wolters-Noordhoff 1969. Theorem 1 (b) appears as 12.56 in the same book and is credited to O. Bottema, Hoofdstukken uit de elementaire meetkunde, Den Haag 1944 (I don't have this book, but interestingly it is not out of print). I might have picked up the Conway substitution from posts by John Conway on the Hyacinthos newsgroup long ago.]






    share|cite|improve this answer














    There are some proofs at https://artofproblemsolving.com/community/c6h14388 and a generalization at https://artofproblemsolving.com/community/c6h2958 . Let me actually reproduce my post in the latter thread in full here (with minor modifications -- neither my exposition nor my LaTeX were particularly good back in 2005), which proves several generalizations:




    Theorem 1. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $XYZ$ be a triangle with sidelengths $x = YZ$, $y = ZX$, $z = XY$ and area $T$.



    (a) (Pedoe inequality) Then, we have the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar.



    (b) (Bottema inequality) If $P$ is an arbitrary point in the plane of triangle $ABC$, then we have the inequality



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    (c) (Equivalent version of (a)) Let $A = measuredangle CAB$, $B = measuredangle ABC$, $C = measuredangle BCA$ be the angles of triangle $ABC$. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of triangle $XYZ$. Then,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y geq 2$.



    This inequality becomes an equality if and only if the triangles $ABC$ and $XYZ$ are similar.




    Proof of Theorem 1.



    (a) First note that the inequality which we must prove,



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$,



    is homogeneous in the sidelengths $x$, $y$, $z$ of the triangle $XYZ$ (in fact, these sidelengths occur in the power $2$ on the left hand side, and on the right hand side they occur in the power $2$ as well, since the area of a triangle depends quadratically on its sidelengths). Hence, this inequality is invariant under any similitude transformation applied to triangle $XYZ$; in other words, we can move, reflect, rotate and stretch the triangle $XYZ$ as we wish, and the inequality remains equivalent. But, of course, by applying similitude transformations to triangle $XYZ$, we can always achieve a situation when $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. Hence, in order to prove Theorem 1 (a) for every two triangles $ABC$ and $XYZ$, it is enough to prove it for two triangles $ABC$ and $XYZ$ in this special situation.



    So, let us WLOG assume that the triangles $ABC$ and $XYZ$ are in this special situation, i. e. that we have $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. We have to prove the inequality $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    By the cosine law in triangle $ABX$, we have



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$.



    Let's figure out what this equation means. At first, $AB = c$. Then, since $B = Y$, we have $XB = XY = z$. Finally, we have either $measuredangle ABX = measuredangle ABC - measuredangle XBC$ or $measuredangle ABX = measuredangle XBC - measuredangle ABC$ (depending on the arrangement of the points), but in both cases $cos measuredangle ABX = cos left(measuredangle ABC - measuredangle XBCright)$. Since $B = Y$ and $C = Z$, we can rewrite the angle $measuredangle XBC$ as $measuredangle XYZ$. Thus,



    $cosmeasuredangle ABX=cosleft(measuredangle ABC-measuredangle XYZright)=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$.



    By the cosine law in triangles $ABC$ and $XYZ$, we have



    $cosmeasuredangle ABC=dfrac{c^2+a^2-b^2}{2ca}$ and
    $cosmeasuredangle XYZ=dfrac{z^2+x^2-y^2}{2zx}$.



    Also, by a well-known formula, the area $S$ of triangle $ABC$ equals $S=dfrac12cdot casinmeasuredangle ABC$; thus, $sinmeasuredangle ABC=dfrac{2S}{ca}$. Similarly, $sinmeasuredangle XYZ=dfrac{2T}{zx}$.



    Hence,



    $cosmeasuredangle ABX=cosmeasuredangle ABCcosmeasuredangle XYZ+sinmeasuredangle ABCsinmeasuredangle XYZ$
    $=dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}$.



    Together with $AB = c$ and $XB = z$, this makes the equality



    $AX^2=AB^2+XB^2-2cdot ABcdot XBcdotcosmeasuredangle ABX$



    transform into



    $AX^2=c^2+z^2-2cdot ccdot zcdotleft(dfrac{c^2+a^2-b^2}{2ca}cdotdfrac{z^2+x^2-y^2}{2zx}+dfrac{2S}{ca}cdotdfrac{2T}{zx}right)$.



    This immediately simplifies to



    $AX^2=c^2+z^2-2left(dfrac{left(c^2+a^2-b^2right)left(z^2+x^2-y^2right)}{4ax}+dfrac{4ST}{ax}right)$.



    After some algebra, this becomes



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)-2Qright)-16ST}{2ax}$,



    where $Q = left(x-aright)left(c^2x-z^2aright)$.



    Since $Y = B$ and $Z = C$, we have $YZ = BC$, or, in other words, $x = a$. Thus, $x - a = 0$, so that $Q = 0$. Hence, the above formula for $AX^2$ simplifies to



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16ST}{2ax}$.



    Since (obviously) $AX^2geq 0$, we thus have



    $left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)-16STgeq 0$,



    and thus $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$. So the inequality of Theorem 1 (a) is proven. Equality holds if and only if equality holds in the inequality $AX^2geq 0$; but this is clearly the case if and only if the points $A$ and $X$ coincide, i. e. if the triangles $ABC$ and $XYZ$ are congruent. Now, of course, since the triangle $XYZ$ we are dealing with is not the initial triangle $XYZ$, but just its image under a similitude transformation, the general equality condition is that the triangles $ABC$ and $XYZ$ are similar (not necessarily being congruent). So Theorem 1 (a) is proven.



    (b) In our proof of Theorem 1 (a), we WLOG assumed that $Y = B$ and $Z = C$ and the point $X$ lies in the same halfplane with respect to the line $BC$ as the point $A$. This time we will assume the same situation, but with a little modification: The point $X$ should lie not in the same halfplane with respect to the line $BC$ as the point $A$, but exactly in the other halfplane. In other words, we have $Y = B$ and $Z = C$ and the points $A$ and $X$ lie in different halfplanes with respect to the line $BC$. Then, by a calculation similar to the one we did in the proof of Theorem 1 (a), we find



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2ax}$



    (note the $+$ sign instead of the $-$). Since $x = a$, we have $ax=a^2$, and thus we can rewrite this as



    $AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2a^2}$.



    Hence,



    $a^2cdot AX^2=dfrac{left(a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)right)+16ST}{2}$
    ${=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Taking the square root of this, we get



    $acdot AX=sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$.



    Hence, the inequality which is to be proven,



    $xcdot AP+ycdot BP+zcdot CPgeq sqrt{dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{2}+8ST}$,



    is equivalent to $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    So we have to prove $xcdot AP+ycdot BP+zcdot CPgeq acdot AX$.



    Well, first, by the Ptolemy inequality for the quadrilateral $PBXC$, we have $CXcdot BP+XBcdot CPgeq BCcdot PX$. Since $B = Y$ and $C = Z$, we have $CX = ZX = y$, $XB = XY = z$ and $BC = YZ = x$, and therefore this inequality rewrites as $ycdot BP+zcdot CPgeq xcdot PX$. Also, by the triangle inequality in the triangle $APX$, we have $AP+PXgeq AX$. Thus,



    $xcdot AP+ycdot BP+zcdot CP=xcdot AP+left(ycdot BP+zcdot CPright)geq xcdot AP+xcdot PX$
    $=xcdotleft(AP+PXright)geq xcdot AX=acdot AX$,



    where we have used $x = a$ in the last step. This proves Theorem 1 (b).



    (c) By the Cosine Law in triangle $ABC$, we have $cos A=dfrac{b^2+c^2-a^2}{2bc}$. On the other hand, by a well-known formula for the area of a triangle, the area $S$ of triangle $ABC$ equals $S=dfrac12bcsin A$. Thus, $sin A=dfrac{2S}{bc}$. Hence,



    $cot A=cos A / sin A=dfrac{b^2+c^2-a^2}{2bc} / dfrac{2S}{bc}=dfrac{b^2+c^2-a^2}{4S}$.



    Similarly, $cot B=dfrac{c^2+a^2-b^2}{4S}$. Thus, $cot A+cot B=dfrac{b^2+c^2-a^2}{4S}+dfrac{c^2+a^2-b^2}{4S}=dfrac{2c^2}{4S}=dfrac{c^2}{2S}$.



    Similarly to how we proved $cot A=dfrac{b^2+c^2-a^2}{4S}$, we can find that $cot Z=dfrac{x^2+y^2-z^2}{4T}$. Thus,



    $cot Acot Z+cot Bcot Z=left(cot A+cot Bright)cdotcot Z=dfrac{c^2}{2S}cdotdfrac{x^2+y^2-z^2}{4T}$
    $=dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$.



    Similarly,



    $cot Bcot X+cot Ccot X=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}$ and
    $cot Ccot Y+cot Acot Y=dfrac{b^2left(z^2+x^2-y^2right)}{8ST}$.



    Thus,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $=left(cot Bcot X+cot Ccot Xright)+left(cot Ccot Y+cot Acot Yright)+left(cot Acot Z+cot Bcot Zright)$
    $=dfrac{a^2left(y^2+z^2-x^2right)}{8ST}+dfrac{b^2left(z^2+x^2-y^2right)}{8ST}+dfrac{c^2left(x^2+y^2-z^2right)}{8ST}$
    $=dfrac{a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)}{8ST}$.



    But after Theorem 1 (a), we have $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$, with equality if and only if the triangles $ABC$ and $XYZ$ are similar. Hence,



    $cot Acot Y+cot Acot Z+cot Bcot Z+cot Bcot X+cot Ccot X+cot Ccot Y$
    $geqdfrac{16ST}{8ST}=2$,



    with equality if and only if the triangles $ABC$ and $XYZ$ are similar. And Theorem 1 (c) is proven.



    Note that Theorem 1 (a) yields the following geometric inequality:




    Corollary 2. Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$.




    Before proving the Corollary, we show two useful facts:




    Corollary 3 (a well-known formula). Let $XYZ$ be a triangle. Let $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ be the angles of this triangle. Then, $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$.




    Proof of Corollary 3. Theorem 1 (c) (applied to $XYZ$, $x$, $y$ and $z$ instead of $ABC$, $a$, $b$ and $c$) shows that $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y geq 2$, and that this inequality becomes an equality if and only if the triangles $XYZ$ and $XYZ$ are similar. Thus, of course, this inequality becomes an equality (since the triangles $XYZ$ and $XYZ$ are similar). Hence, $cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y = 2$. Therefore,



    $2 = cot Xcot Y+cot Xcot Z+cot Ycot Z+cot Ycot X+cot Zcot X+cot Zcot Y$
    $= 2 left(cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Yright)$.



    Dividing both sides of this equality by $2$, we obtain $1 = cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y$. This proves Corollary 3.




    Theorem 4 (Conway substitution theorem). Let $u$, $v$, $w$ be three reals such that the numbers $v + w$, $w + u$, $u + v$ and $vw + wu + uv$ are all nonnegative. [Of course, these conditions are fulfilled if the reals $u$, $v$, $w$ themselves are nonnegative, but also in some other cases.] Then, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. This triangle satisfies $y^2+z^2-x^2=2u$, $z^2+x^2-y^2=2v$, $x^2+y^2-z^2=2w$. The area $T$ of this triangle equals $T=dfrac12sqrt{vw+wu+uv}$. If $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ are the angles of this triangle, then $cot X=dfrac{u}{2T}$, $cot Y=dfrac{v}{2T}$ and $cot Z=dfrac{w}{2T}$.




    Proof of Theorem 4. Since the numbers $v + w$, $w + u$, $u + v$ are nonnegative, their square roots $sqrt{v+w}$, $sqrt{w+u}$, $sqrt{u+v}$ exist, and, of course, are nonnegative as well. Now, we have $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w}$, since



    $sqrt{w+u}+sqrt{u+v}geqsqrt{v+w} Longleftarrow left(sqrt{w+u}+sqrt{u+v}right)^2geq v+w$



    $Longleftarrow left(w+uright)+left(u+vright)+2sqrt{left(w+uright)left(u+vright)}geq v+w$



    $Longleftarrow 2sqrt{left(w+uright)left(u+vright)}geq -2u$ [this is not trivial, since $u$ needs not be $geq 0$!]



    $Longleftarrow sqrt{left(w+uright)left(u+vright)}geq-u Longleftarrow left(w+uright)left(u+vright)geqleft(-uright)^2$



    $Longleftarrow u^2+left(vw+wu+uvright)geq u^2 Longleftarrow vw+wu+uvgeq 0$.



    Similarly, $sqrt{u+v}+sqrt{v+w}geqsqrt{w+u}$ and $sqrt{v+w}+sqrt{w+u}geqsqrt{u+v}$. Thus, there exists a triangle $XYZ$ with sidelengths $x=YZ=sqrt{v+w}$, $y=ZX=sqrt{w+u}$, $z=XY=sqrt{u+v}$. Denote by $X = measuredangle ZXY$, $Y = measuredangle XYZ$, $Z = measuredangle YZX$ the angles of this triangle. We have



    $y^2+z^2-x^2=left(sqrt{w+u}right)^2+left(sqrt{u+v}right)^2-left(sqrt{v+w}right)^2$
    $=left(w+uright)+left(u+vright)-left(v+wright)=2u$



    and similarly $z^2+x^2-y^2=2v$ and $x^2+y^2-z^2=2w$. Now, denoting the area of triangle $XYZ$ by $T$, we note that, as shown during the proof of Theorem 1 (c), we have $cot Z=dfrac{x^2+y^2-z^2}{4T}$, so that $cot Z=dfrac{2w}{4T}=dfrac{w}{2T}$, and similarly $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$.



    Corollary 3 yields $cot Ycdotcot Z+cot Zcdotcot X+cot Xcdotcot Y=1$. Using the relations $cot Z=dfrac{w}{2T}$, $cot X=dfrac{u}{2T}$ and $cot Y=dfrac{v}{2T}$, this equality becomes $dfrac{v}{2T}cdotdfrac{w}{2T}+dfrac{w}{2T}cdotdfrac{u}{2T}+dfrac{u}{2T}cdotdfrac{v}{2T}=1$. Hence, $vw+wu+uv=4T^2$, and thus $T=dfrac12sqrt{4T^2}=dfrac12sqrt{vw+wu+uv}$. This completes the proof of Theorem 4.



    Proof of Corollary 2. Construct a triangle $XYZ$ for our values of $u$, $v$, $w$ according to Theorem 4. Then, applying Theorem 1 (a) to the triangles $ABC$ and $XYZ$, we get



    $a^2left(y^2+z^2-x^2right)+b^2left(z^2+x^2-y^2right)+c^2left(x^2+y^2-z^2right)geq 16ST$.



    Using the formulas given in Theorem 4, we can rewrite this as



    $a^2cdot 2u+b^2cdot 2v+c^2cdot 2wgeq 16Scdotdfrac12sqrt{vw+wu+uv}$.



    This simplifies to $ua^2+vb^2+wc^2geq 4sqrt{vw+wu+uv}cdot S$. This proves Corollary 2.




    Corollary 5 (Weitzenböck inequality). Let $ABC$ be a triangle with sidelengths $a = BC$, $b = CA$, $c = AB$ and area $S$. Then, $a^2 + b^2 + c^2 geq 4sqrt{3} S$.




    Proof of Corollary 5. Corollary 5 follows from Corollary 2 (applied to $u=1$, $v=1$ and $w=1$).



    [Postscriptum: I have been atrocious with attribution and citation back in 2005. I am not sure where the ideas in the above proofs come from; I might have been just the scribe. Theorem 1 (a) was discovered in 1943 by Daniel Pedoe and much earlier (without the equality case) by Neuberg. It also appears as 10.8 in O. Bottema, R. Z. Djordjevic, R. R. Janic, D. S. Mitrinovic, P. M. Vasic, Geometric Inequalities, Wolters-Noordhoff 1969. Theorem 1 (b) appears as 12.56 in the same book and is credited to O. Bottema, Hoofdstukken uit de elementaire meetkunde, Den Haag 1944 (I don't have this book, but interestingly it is not out of print). I might have picked up the Conway substitution from posts by John Conway on the Hyacinthos newsgroup long ago.]







    share|cite|improve this answer














    share|cite|improve this answer



    share|cite|improve this answer








    edited Mar 2 '17 at 4:03

























    answered Mar 2 '17 at 3:47









    darij grinberg

    10.2k33061




    10.2k33061












    • Uow thank's @darij grinberg!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      Mar 3 '17 at 2:59


















    • Uow thank's @darij grinberg!!!!!!!!!!!
      – Israel Meireles Chrisostomo
      Mar 3 '17 at 2:59
















    Uow thank's @darij grinberg!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    Mar 3 '17 at 2:59




    Uow thank's @darij grinberg!!!!!!!!!!!
    – Israel Meireles Chrisostomo
    Mar 3 '17 at 2:59











    2














    using the Heron formula for $A$ we have $$a^2+b^2+c^2geq 4sqrt{3}sqrt{s(s-a)(s-b)(s-c)}$$ after squaring and rearanging we obtain
    $$a^4+b^4+c^4geq a^2b^2+b^2c^2+c^2a^2$$ and this is true and follows from
    $$x^2+y^2+z^2geq xy+yz+zx$$






    share|cite|improve this answer

















    • 2




      after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
      – dxiv
      Feb 14 '17 at 6:37












    • it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:49










    • why the $-1$ i don't see it?
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:50






    • 3




      My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
      – dxiv
      Feb 14 '17 at 7:02








    • 2




      Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
      – dxiv
      Feb 14 '17 at 7:09


















    2














    using the Heron formula for $A$ we have $$a^2+b^2+c^2geq 4sqrt{3}sqrt{s(s-a)(s-b)(s-c)}$$ after squaring and rearanging we obtain
    $$a^4+b^4+c^4geq a^2b^2+b^2c^2+c^2a^2$$ and this is true and follows from
    $$x^2+y^2+z^2geq xy+yz+zx$$






    share|cite|improve this answer

















    • 2




      after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
      – dxiv
      Feb 14 '17 at 6:37












    • it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:49










    • why the $-1$ i don't see it?
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:50






    • 3




      My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
      – dxiv
      Feb 14 '17 at 7:02








    • 2




      Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
      – dxiv
      Feb 14 '17 at 7:09
















    2












    2








    2






    using the Heron formula for $A$ we have $$a^2+b^2+c^2geq 4sqrt{3}sqrt{s(s-a)(s-b)(s-c)}$$ after squaring and rearanging we obtain
    $$a^4+b^4+c^4geq a^2b^2+b^2c^2+c^2a^2$$ and this is true and follows from
    $$x^2+y^2+z^2geq xy+yz+zx$$






    share|cite|improve this answer












    using the Heron formula for $A$ we have $$a^2+b^2+c^2geq 4sqrt{3}sqrt{s(s-a)(s-b)(s-c)}$$ after squaring and rearanging we obtain
    $$a^4+b^4+c^4geq a^2b^2+b^2c^2+c^2a^2$$ and this is true and follows from
    $$x^2+y^2+z^2geq xy+yz+zx$$







    share|cite|improve this answer












    share|cite|improve this answer



    share|cite|improve this answer










    answered Feb 14 '17 at 6:27









    Dr. Sonnhard Graubner

    73k42865




    73k42865








    • 2




      after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
      – dxiv
      Feb 14 '17 at 6:37












    • it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:49










    • why the $-1$ i don't see it?
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:50






    • 3




      My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
      – dxiv
      Feb 14 '17 at 7:02








    • 2




      Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
      – dxiv
      Feb 14 '17 at 7:09
















    • 2




      after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
      – dxiv
      Feb 14 '17 at 6:37












    • it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:49










    • why the $-1$ i don't see it?
      – Dr. Sonnhard Graubner
      Feb 14 '17 at 6:50






    • 3




      My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
      – dxiv
      Feb 14 '17 at 7:02








    • 2




      Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
      – dxiv
      Feb 14 '17 at 7:09










    2




    2




    after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
    – dxiv
    Feb 14 '17 at 6:37






    after squaring and rearanging I think you glossed over too summarily over that step. It's not obvious how that follows, and not everybody has a CAS at their fingertips to verify.
    – dxiv
    Feb 14 '17 at 6:37














    it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:49




    it is simple expanding and you will get what i wrote this is a proof by our math circle from Leipzig, the shortest one
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:49












    why the $-1$ i don't see it?
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:50




    why the $-1$ i don't see it?
    – Dr. Sonnhard Graubner
    Feb 14 '17 at 6:50




    3




    3




    My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
    – dxiv
    Feb 14 '17 at 7:02






    My $-1$ is for the hand-waving. You make it look like that's a $1$-line proof, but omit the $1+$ pages of tedious calculations required to actually achieve that squaring and rearanging. The OP went to the pain of actually spelling out their $3$ proofs step by step. An answer that claims to compare (favourably) against those proofs should do the same.
    – dxiv
    Feb 14 '17 at 7:02






    2




    2




    Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
    – dxiv
    Feb 14 '17 at 7:09






    Your notion of "simple" must be different from mine. Yes, it may be plain algebra, and something that, say, Wolfram can do easily. Yet, that doesn't make it "simple" in my books. Sorry, guess we'll just have to agree to disagree on this point.
    – dxiv
    Feb 14 '17 at 7:09













    0














    Let $P$ be the first Brocard point of $triangle ABC$ and $varphi$ be the Brocard angle. Then, Ceva's theorem implies:
    $$sin^3varphi = sin(alpha - varphi)sin(beta-varphi)sin(gamma-varphi).$$
    Now an immediate usage of the Jensen's inequality gives $sinvarphileqfrac 12$ or equivalently:
    $$varphileq 30^circ.$$
    On the other hand, one can prove that $$cotvarphi = cotalpha+cotbeta+cotgamma = dfrac{a^2+b^2+c^2}{4S}geqsqrt{3}.$$



    Another way is simply to state Finsler-Hadwiger inequality:
    $$a^2+b^2+c^2geq 4sqrt{3}+(a-b)^2+(b-c)^2+(c-a)^2.$$






    share|cite|improve this answer


























      0














      Let $P$ be the first Brocard point of $triangle ABC$ and $varphi$ be the Brocard angle. Then, Ceva's theorem implies:
      $$sin^3varphi = sin(alpha - varphi)sin(beta-varphi)sin(gamma-varphi).$$
      Now an immediate usage of the Jensen's inequality gives $sinvarphileqfrac 12$ or equivalently:
      $$varphileq 30^circ.$$
      On the other hand, one can prove that $$cotvarphi = cotalpha+cotbeta+cotgamma = dfrac{a^2+b^2+c^2}{4S}geqsqrt{3}.$$



      Another way is simply to state Finsler-Hadwiger inequality:
      $$a^2+b^2+c^2geq 4sqrt{3}+(a-b)^2+(b-c)^2+(c-a)^2.$$






      share|cite|improve this answer
























        0












        0








        0






        Let $P$ be the first Brocard point of $triangle ABC$ and $varphi$ be the Brocard angle. Then, Ceva's theorem implies:
        $$sin^3varphi = sin(alpha - varphi)sin(beta-varphi)sin(gamma-varphi).$$
        Now an immediate usage of the Jensen's inequality gives $sinvarphileqfrac 12$ or equivalently:
        $$varphileq 30^circ.$$
        On the other hand, one can prove that $$cotvarphi = cotalpha+cotbeta+cotgamma = dfrac{a^2+b^2+c^2}{4S}geqsqrt{3}.$$



        Another way is simply to state Finsler-Hadwiger inequality:
        $$a^2+b^2+c^2geq 4sqrt{3}+(a-b)^2+(b-c)^2+(c-a)^2.$$






        share|cite|improve this answer












        Let $P$ be the first Brocard point of $triangle ABC$ and $varphi$ be the Brocard angle. Then, Ceva's theorem implies:
        $$sin^3varphi = sin(alpha - varphi)sin(beta-varphi)sin(gamma-varphi).$$
        Now an immediate usage of the Jensen's inequality gives $sinvarphileqfrac 12$ or equivalently:
        $$varphileq 30^circ.$$
        On the other hand, one can prove that $$cotvarphi = cotalpha+cotbeta+cotgamma = dfrac{a^2+b^2+c^2}{4S}geqsqrt{3}.$$



        Another way is simply to state Finsler-Hadwiger inequality:
        $$a^2+b^2+c^2geq 4sqrt{3}+(a-b)^2+(b-c)^2+(c-a)^2.$$







        share|cite|improve this answer












        share|cite|improve this answer



        share|cite|improve this answer










        answered Dec 9 at 2:38









        dezdichado

        6,1781929




        6,1781929






























            draft saved

            draft discarded




















































            Thanks for contributing an answer to Mathematics Stack Exchange!


            • Please be sure to answer the question. Provide details and share your research!

            But avoid



            • Asking for help, clarification, or responding to other answers.

            • Making statements based on opinion; back them up with references or personal experience.


            Use MathJax to format equations. MathJax reference.


            To learn more, see our tips on writing great answers.





            Some of your past answers have not been well-received, and you're in danger of being blocked from answering.


            Please pay close attention to the following guidance:


            • Please be sure to answer the question. Provide details and share your research!

            But avoid



            • Asking for help, clarification, or responding to other answers.

            • Making statements based on opinion; back them up with references or personal experience.


            To learn more, see our tips on writing great answers.




            draft saved


            draft discarded














            StackExchange.ready(
            function () {
            StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2143589%2fimo-problem-1961-different-proofs-of-the-weitzenbock-inequality%23new-answer', 'question_page');
            }
            );

            Post as a guest















            Required, but never shown





















































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown

































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown







            Popular posts from this blog

            Bressuire

            Cabo Verde

            Gyllenstierna