Let $ABC$ be a non-equilateral triangle with integer sides. Let $D$ and $E$ be respectively the mid-points of $BC$ and $CA$ ; let $G$ be the centroid of $\Delta{ABC}$. Suppose, $D$, $C$, $E$, $G$ are concyclic. Find the least possible perimeter of $\Delta{ABC}$.
Problem
Source: INMO 2018
Tags: geometry, Law of Sines
21.01.2018 14:15
Looks like we posted at the same time......
21.01.2018 14:16
I will delete my post. Please make a contest collection.
21.01.2018 14:28
jrc1729 wrote: $\mathbf{1)}$ Let $ABC$ be a non-equilateral triangle with integer sides. Let $D$ and $E$ be respectively the mid-points of $BC$ and $CA$ ; let $G$ be the centroid of $\Delta{ABC}$. Suppose, $D$, $C$, $E$, $G$ are concyclic. Find the least possible perimeter of $\Delta{ABC}$. This means $\angle BCA + \angle BGA = 180^{\circ}$, so $G$ is the C-HM point of $\triangle ABC$; i.e. $\odot(ACG)$ and $\odot(BCG)$ are tangent to $\overline{AB}$. It is well-known that this implies $AC^2 + BC^2 = 2AB^2$, but I post the solution here for completeness. Letting $M$ be the midpoint of $\overline{AB}$, we need $\tfrac{1}{3} MA^2 = MG \cdot MA = MB^2 = MC^2$, or $\tfrac{1}{3} \tfrac{2a^2 + 2b^2 - c^2}{4} = \tfrac{c^2}{4}$, so $a^2 + b^2 = 2c^2$. So $(a + b)^2 + (a - b)^2 = (2c)^2$. It is well-known that a necessary and sufficient condition for this to have integer solutions $(a, b)$ is for $c$ to have a prime factor congruent to 1 mod 4. It is clear that in the minimal perimeter triangle $\gcd(a, b, c) = 1$. The smallest possible values of $c$ are 5 and 13. $c = 5$ leads to $\{a, b\} = \{1, 7\}$, not a triangle. $c = 13$ leads to $\{a, b\} = \{7, 17\}$, which indeed works. The answer is $a + b + c = \boxed{37}$.
21.01.2018 16:01
By Power of a point, \[1 = \frac{BG\cdot BE}{BD\cdot BC} = \frac{\frac23 BE^2}{\frac12 BC^2} \implies \frac{BE}{BC}= \frac{\sqrt3}2\]Let $AB = c, BC = a, CA = b$, By Appolonius' theorem. \[ a^2 + c^2 = 2(BE^2 + EC^2) = \frac32 a^2 + \frac12 b^2 \iff a^2 + b^2 = 2c^2\] I couldn't do anything after that in the paper. I just realized this, now I feel sad \[a^2+b^2=2c^2 \iff (a-b)^2+(a+b)^2 = (2c)^2 \]WLOG $a>b$ Now the smallest pythagorean triple is \[3,4,5 \implies a-b = 3, a+b = 4, 2c = 5 \implies a = \frac72, b=\frac12, c=\frac52\]\[\text{But } a + c = \frac62 < \frac72 = b \]Therefore, this can not be a solution Now, the second smallest pythagorean triple is \[(a-b,a+b, 2c) = (5, 12, 13) \implies a = \frac{17}2, b=\frac72, c=\frac{13}2 \]\[\text{and }b + c = 10 > \frac{17}2 = a\]Therefore $(a:b:c) = (17:7:13)$ Thus, the smallest value of $a,b,c$ is $17,7,13$. Therefore $a+b+c = 37$.
21.01.2018 20:23
I got $a^2 + b^2 = 2c^2$. Then I wrote - by putting small values for $c$, we that $(17, 7, 13)$ is a valid solution with the smallest value of $c$. Then, I used that to get a bound on the value of $c$ (using the triangle inequality) till which we have to check for any other possible solutions. The bound was not very large. ($c<=18$). Will they cut any marks for this? (I didn't explicitly check values on the answer sheet. I did it mentally and some of it on rough sheets.)
02.12.2018 07:32
CantonMathGuy wrote: jrc1729 wrote: $\mathbf{1)}$ Let $ABC$ be a non-equilateral triangle with integer sides. Let $D$ and $E$ be respectively the mid-points of $BC$ and $CA$ ; let $G$ be the centroid of $\Delta{ABC}$. Suppose, $D$, $C$, $E$, $G$ are concyclic. Find the least possible perimeter of $\Delta{ABC}$. This means $\angle BCA + \angle BGA = 180^{\circ}$, so $G$ is the C-HM point of $\triangle ABC$; i.e. $\odot(ACG)$ and $\odot(BCG)$ are tangent to $\overline{AB}$. It is well-known that this implies $AC^2 + BC^2 = 2AB^2$, but I post the solution here for completeness. Letting $M$ be the midpoint of $\overline{AB}$, we need $\tfrac{1}{3} MA^2 = MG \cdot MA = MB^2 = MC^2$, or $\tfrac{1}{3} \tfrac{2a^2 + 2b^2 - c^2}{4} = \tfrac{c^2}{4}$, so $a^2 + b^2 = 2c^2$. So $(a + b)^2 + (a - b)^2 = (2c)^2$. It is well-known that a necessary and sufficient condition for this to have integer solutions $(a, b)$ is for $c$ to have a prime factor congruent to 1 mod 4. It is clear that in the minimal perimeter triangle $\gcd(a, b, c) = 1$. The smallest possible values of $c$ are 5 and 13. $c = 5$ leads to $\{a, b\} = \{1, 7\}$, not a triangle. $c = 13$ leads to $\{a, b\} = \{7, 17\}$, which indeed works. The answer is $a + b + c = \boxed{37}$. It is not necessary that $\gcd(a, b, c) =1$ is required for least perimeter. The triplet $(6, 8,10)$ was also possible
26.01.2019 15:16
Hydrohelium, for the minimal perimeter, it is enough to notice that the triplet should be a primitive one.
02.02.2019 18:09
CantonMathGuy wrote: jrc1729 wrote: $\mathbf{1)}$ Let $ABC$ be a non-equilateral triangle with integer sides. Let $D$ and $E$ be respectively the mid-points of $BC$ and $CA$ ; let $G$ be the centroid of $\Delta{ABC}$. Suppose, $D$, $C$, $E$, $G$ are concyclic. Find the least possible perimeter of $\Delta{ABC}$. This means $\angle BCA + \angle BGA = 180^{\circ}$, so $G$ is the C-HM point of $\triangle ABC$; i.e. $\odot(ACG)$ and $\odot(BCG)$ are tangent to $\overline{AB}$. It is well-known that this implies $AC^2 + BC^2 = 2AB^2$, but I post the solution here for completeness. Letting $M$ be the midpoint of $\overline{AB}$, we need $\tfrac{1}{3} MA^2 = MG \cdot MA = MB^2 = MC^2$, or $\tfrac{1}{3} \tfrac{2a^2 + 2b^2 - c^2}{4} = \tfrac{c^2}{4}$, so $a^2 + b^2 = 2c^2$. So $(a + b)^2 + (a - b)^2 = (2c)^2$. It is well-known that a necessary and sufficient condition for this to have integer solutions $(a, b)$ is for $c$ to have a prime factor congruent to 1 mod 4. It is clear that in the minimal perimeter triangle $\gcd(a, b, c) = 1$. The smallest possible values of $c$ are 5 and 13. $c = 5$ leads to $\{a, b\} = \{1, 7\}$, not a triangle. $c = 13$ leads to $\{a, b\} = \{7, 17\}$, which indeed works. The answer is $a + b + c = \boxed{37}$. Does this also imply that $AC^2+BC^2=2AB^2???$ I read the article about the $HM$ points by anantmudgal09 but i did not see this fact.... can you suggest an article with all the properties of $HM$ points please??
02.02.2019 18:45
It is easily obtained by power of point, but is it a standard lemma or theorem?
25.05.2019 05:10
I want to say that instead of achieving this using synthetic results, we can get $a^2+b^2=2c^2$ in 5 to 10 minutes by a barycentric bash.
25.05.2019 06:06
Pi-is-3 wrote: I want to say that instead of achieving this using synthetic results, we can get $a^2+b^2=2c^2$ in 5 to 10 minutes by a barycentric bash. Better: you can guess that beforehand. Read automedian triangles.
01.04.2020 08:32
with out losing generility let assume , $AB=c,BC=a,CA=b$. clearly $\angle BGD=\angle DCE =C$. Again$\angle BDG=\angle BEA=\alpha$(say) From $\triangle EBG$ we get $\angle EBC =\alpha -c$ and $\angle BAE=180-(\alpha+c)$. so we get , $GD=\frac{\sin (\alpha-c) .a}{2\sin C }$. $GE=\frac{\sin(\alpha+c) .b}{2\sin C}$. $ED=\frac{c}{2}$. $GC =\frac{c\sin \alpha}{2\sin B}$, since $G,E,C,D$ are concyclic applying ptolemy's theorem get, $ED.GC=DC.GE+GD.EC $ $\Rightarrow \frac{ab}{4} .(\frac{\sin (\alpha-c) +\sin(\alpha+c)}{\sin C} =\frac{bc\sin\alpha}{4\sin B}$. $\Rightarrow \frac{2a\sin C}{\cos C}=\frac{c}{\sin B}$. $\Rightarrow \frac{2a\cos c}{c}=\frac{c}{b}$ $\Rightarrow 2ab \cos C =c^2$. using cosine rule get, $2ab\frac{a^2+b^2-c^2a}{2ab} =c^2$. $\Rightarrow a^2 +b^2 =2c^2$. $(\frac{a+b}{2})^2+(\frac{a-b}{2})^2=c^2$. we know least pythagorian triplet is $3,4,5$. but in this case we get $a,b=\{1,7\}$. its not possible since $\angle B$ is obtuse. next least pythagorian triplet is $\{12,13,5\}$. we get $a,b=\{17,7\}$. so least possible perimeter is $13+17+7=37$.
29.04.2020 10:48
Yay another INMO Geo solved! (Or should I say NT ) By Apollonius theorem, we have $AD^2=\frac{2(b^2+c^2)-a^2}{4}$ But, $AD=3GD$ So, $GD^2=\frac{2(b^2+c^2)-a^2}{36}$ By PoP, we have $AG\cdot AD=AE\cdot AC\implies 2GD\cdot 3GD=\frac{b}{2}\cdot b$ So, $GD^2=\frac{b^2}{12}$ And hence, $\frac{b^2}{12}=\frac{2(b^2+c^2)-a^2}{36}$ Simplifying gives $a^2+b^2=2c^2$
Since $a,b,c$ are positive integers, we take $d=2$ to get $a=17,b=7,c=13$ and hence, the minimal perimeter is $a+b+c=37$ $\blacksquare$.
29.04.2020 11:46
Math-wiz wrote: Yay another INMO Geo solved! (Or should I say NT ) By Apollonius theorem, we have $AD^2=\frac{2(b^2+c^2)-a^2}{4}$ But, $AD=3GD$ So, $GD^2=\frac{2(b^2+c^2)-a^2}{36}$ By PoP, we have $AG\cdot AD=AE\cdot AC\implies 2GD\cdot 3GD=\frac{b}{2}\cdot b$ So, $GD^2=\frac{b^2}{12}$ And hence, $\frac{b^2}{12}=\frac{2(b^2+c^2)-a^2}{36}$ Simplifying gives $a^2+b^2=2c^2$
Since $a,b,c$ are positive integers, we take $d=2$ to get $a=17,b=7,c=13$ and hence, the minimal perimeter is $a+b+c=37$ $\blacksquare$. I believe that the NT part of your proof is incomplete This is because you have not proved that there is no other triple for which the perimeter is less than 37
04.07.2020 17:17
ftheftics wrote: with out losing generility let assume , $AB=c,BC=a,CA=b$. clearly $\angle BGD=\angle DCE =C$. Again$\angle BDG=\angle BEA=\alpha$(say) From $\triangle EBG$ we get $\angle EBC =\alpha -c$ and $\angle BAE=180-(\alpha+c)$. so we get , $GD=\frac{\sin (\alpha-c) .a}{2\sin C }$. $GE=\frac{\sin(\alpha+c) .b}{2\sin C}$. $ED=\frac{c}{2}$. $GC =\frac{c\sin \alpha}{2\sin B}$, since $G,E,C,D$ are concyclic applying ptolemy's theorem get, $ED.GC=DC.GE+GD.EC $ $\Rightarrow \frac{ab}{4} .(\frac{\sin (\alpha-c) +\sin(\alpha+c)}{\sin C} =\frac{bc\sin\alpha}{4\sin B}$. $\Rightarrow \frac{2a\sin C}{\cos C}=\frac{c}{\sin B}$. $\Rightarrow \frac{2a\cos c}{c}=\frac{c}{b}$ $\Rightarrow 2ab \cos C =c^2$. using cosine rule get, $2ab\frac{a^2+b^2-c^2a}{2ab} =c^2$. $\Rightarrow a^2 +b^2 =2c^2$. $(\frac{a+b}{2})^2+(\frac{a-b}{2})^2=c^2$. we know least pythagorian triplet is $3,4,5$. but in this case we get $a,b=\{1,7\}$. its not possible since $\angle B$ is obtuse. next least pythagorian triplet is $\{12,13,5\}$. we get $a,b=\{17,7\}$. so least possible perimeter is $13+17+7=37$. I guess u have made a small mistake whike writing $\alpha$
17.07.2020 19:09
ftheftics wrote: with out losing generility let assume , $AB=c,BC=a,CA=b$. clearly $\angle BGD=\angle DCE =C$. Again$\angle BDG=\angle BEA=\alpha$(say) From $\triangle EBG$ we get $\angle EBC =\alpha -c$ and $\angle BAE=180-(\alpha+c)$. so we get , $GD=\frac{\sin (\alpha-c) .a}{2\sin C }$. $GE=\frac{\sin(\alpha+c) .b}{2\sin C}$. $ED=\frac{c}{2}$. $GC =\frac{c\sin \alpha}{2\sin B}$, since $G,E,C,D$ are concyclic applying ptolemy's theorem get, $ED.GC=DC.GE+GD.EC $ $\Rightarrow \frac{ab}{4} .(\frac{\sin (\alpha-c) +\sin(\alpha+c)}{\sin C} =\frac{bc\sin\alpha}{4\sin B}$. $\Rightarrow \frac{2a\sin C}{\cos C}=\frac{c}{\sin B}$. $\Rightarrow \frac{2a\cos c}{c}=\frac{c}{b}$ $\Rightarrow 2ab \cos C =c^2$. using cosine rule get, $2ab\frac{a^2+b^2-c^2a}{2ab} =c^2$. $\Rightarrow a^2 +b^2 =2c^2$. $(\frac{a+b}{2})^2+(\frac{a-b}{2})^2=c^2$. we know least pythagorian triplet is $3,4,5$. but in this case we get $a,b=\{1,7\}$. its not possible since $\angle B$ is obtuse. next least pythagorian triplet is $\{12,13,5\}$. we get $a,b=\{17,7\}$. so least possible perimeter is $13+17+7=37$. $EBG$ is not a triangle but a straight line..
06.02.2022 13:16
$a,b,c$ sides tringle.$\implies 2c^2=a^2+b^2$. Remaing easy.
05.03.2022 21:19
Nice combination of NT and GEO Geo part: Claim1 : $a^2 + b^2 = 2c^2$. Proof : we have $GD^2=\frac{2(b^2+c^2)-a^2}{36}$ and also from power of point $A$ we have $6GD^2 = b^2/2$ and they simply prove our claim. NT part: we have $a^2 + b^2 = 2c^2$ or in fact $(a-b)^2 + (a+b)^2 = (2c)^2$ and we know $a,b,c$ are integer. so triple $(a-b,a+b,2c)$ must be pythagorean. we know smallest pythagorean triple is $(3x,4x,5x)$. but our triple can't be in this form because then : $a = \frac{7x}{2}$, $b = \frac{x}{2}$ and $c = \frac{5x}{2}$ and $b+c$ is less than $a$. next smallest pythagorean triple is $(5x,12x,13x)$. with this we have no inequality contradiction so $(a,b,c) = (17,7,13)$ so least possible perimeter of $ABC$ is $17 + 7 + 13 = 37$.
30.12.2022 23:21
By PoP we get $BE = \frac{\sqrt3 a}{2}$ On using appolonius thm and substituting the above value we get \[a^2 + b^2 = 2c^2\]Now we just have to check for pythagorean triplets and we find the smallest solution to be $17,7,13$ So the ans is 37