# For $P$ an arbitrary point in $\triangle ABC$, show that $\sum_{cyc}c(\sin \angle CAP+\sin\angle CBP)\leq a+b+c$

In the interior of $$\triangle ABC$$ we take the arbitrary point $$P$$. Prove that the following inequality holds: $$\small c(\sin\angle CAP + \sin\angle CBP) + a(\sin\angle ABP +\sin\angle ACP) + b(\sin\angle BCP + \sin\angle BAP)\\ \le a + b + c$$

I tried to draw the distances from $$P$$ to the sides of the triangle and write the sines from the right angled triangles, but I couldn't get anything out of it. I saw that a case of equality is when the triangle is equilateral and $$P$$ is its center.

Can you help me?

• Interesting inequality. Where does it come from? – ΑΘΩ Mar 14 '19 at 13:14
• It is from the romanian publishment “Gazeta Matematica”. – Michael Goldberg Mar 14 '19 at 16:17
• If you would be as kind as to tell us what edition of the Gazette that might be, we might be able to look things up ;) – ΑΘΩ Mar 14 '19 at 16:37
• It is from the edition 10/2018, the number of the problem being S:L18.249 :) – Michael Goldberg Mar 14 '19 at 16:53

To simplify derivation introduce the notation: $$BC=a,\quad AC=b,\quad AB=c,\quad \widehat{CAB}=\alpha, \widehat{ABC}=\beta,\quad \widehat{BCA}=\gamma;\\ \widehat{ABP}=\alpha_B,\quad\widehat{ACP}=\alpha_C,\quad \widehat{BAP}=\beta_A,\quad\widehat{BCP}=\beta_C,\quad \widehat{CAP}=\gamma_A,\quad\widehat{CBP}=\gamma_B.$$ Then the initial inequality can be written as: $$\frac{a(\sin\alpha_B+\sin\alpha_C) +b(\sin\beta_A+\sin\beta_C)+c(\sin\gamma_A+\sin\gamma_B)}{a+b+c}\le1.\tag{*}$$ which can be proved as follows: \begin{align} &\frac{a(\sin\alpha_B+\sin\alpha_C) +b(\sin\beta_A+\sin\beta_C)+c(\sin\gamma_A+\sin\gamma_B)}{a+b+c}\\ &=\frac{\sin\alpha(\sin\alpha_B+\sin\alpha_C) +\sin\beta(\sin\beta_A+\sin\beta_C)+\sin\gamma(\sin\gamma_A+\sin\gamma_B)} {\sin\alpha+\sin\beta+\sin\gamma}\tag1\\ &=\frac{(\sin\beta\sin\beta_A+\sin\gamma\sin\gamma_A) +(\sin\alpha\sin\alpha_B+\sin\gamma\sin\gamma_B) +(\sin\alpha\sin\alpha_C+\sin\beta\sin\beta_C)} {\sin\alpha+\sin\beta+\sin\gamma}\tag2\\ &\le\frac{\sin\alpha+\sin\beta+\sin\gamma}{\sin\alpha+\sin\beta+\sin\gamma}=1\tag3. \end{align}

$$(1)$$: Substitution: $$a=2R\sin\alpha, \quad b=2R\sin\beta, \quad c=2R\sin\gamma.$$

$$(2)$$: Rearrangement.

$$(3)$$: Transformation of sine products to cosine differences and then from cosine sums to cosine products.
\begin{align} &\sin\beta\sin\beta_A+\sin\gamma\sin\gamma_A\\ &=\frac{\cos(\beta-\beta_A)-\cos(\beta+\beta_A)}2 +\frac{\cos(\gamma-\gamma_A)-\cos(\gamma+\gamma_A)}2\\ &=\frac{\cos(\beta-\beta_A)+\cos(\gamma-\gamma_A)}2 -\frac{\cos(\beta+\beta_A)+\cos(\gamma+\gamma_A)}2\\ &= \underbrace{\cos\frac{\beta+\gamma-\beta_A-\gamma_A}2}_{\ge0} \underbrace{\cos\frac{\beta-\gamma-\beta_A+\gamma_A}2}_{\le1} - \underbrace{\cos\frac{\beta+\gamma+\beta_A+\gamma_A}2}_{=0}\cos\frac{\beta-\gamma+\beta_A-\gamma_A}2\tag4\\ &\le \cos\frac{\beta+\gamma-\alpha}2=\cos\left(\frac\pi2-\alpha\right)=\sin\alpha, \end{align} where we used $$\beta_A+\gamma_A=\alpha$$.

There may arise an interesting question when does the inequality $$(\text{*})$$ degenerate to equality?

It is easily seen from $$(4)$$ that this happens if and only if the following equations hold: $$\begin {cases} \beta-\beta_A=\gamma-\gamma_A\\ \gamma-\gamma_B=\alpha-\alpha_B\\ \alpha-\alpha_C=\beta-\beta_C \end {cases},$$ which together with $$\begin {cases} \beta_A+\gamma_A=\alpha\\ \gamma_B+\alpha_B=\beta\\ \alpha_C+\beta_C=\gamma \end {cases}$$ implies: $$\begin {cases} \beta_C=\gamma_B=\frac\pi2-\alpha\\ \gamma_A=\alpha_C=\frac\pi2-\beta\\ \alpha_B=\beta_A=\frac\pi2-\gamma \end {cases}.$$

Thus the equality holds if and only if the triangle $$ABC$$ is acute and $$P$$ is its circumcenter.

Note:

With the convention that the angles oriented oppositely to the direction towards the adjacent side are treated as negative, the inequality $$(\text{*})$$ becomes valid for the whole plane. Correspondingly the equality holds at the circumcenter of a triangle, regardless of its shape.

• Thank you very much! – Michael Goldberg Mar 15 '19 at 14:58
• @MichaelGoldberg You're welcome! – user Mar 15 '19 at 15:09
• Great presentation! – Allawonder Mar 16 '19 at 10:28
• @Allawonder Thank you! – user Mar 16 '19 at 10:38
• A lovely solution! Just one remark though: the case of equality entails indeed the equations displayed in the final curly brace, however the conclusion on $P$ is not entirely correct. From $\sphericalangle{ABP}=\sphericalangle{BAP}$ one infers $|PA|=|PB|$ and the analogues, to the conclusion that $P$ is the circumcenter. The requirement that the triangle be acute is indeed valid. – ΑΘΩ Mar 21 '19 at 11:06