Community

Looking for a challenging algebra text? Preparing for MATHCOUNTS or the AMC exams?
Check out Art of Problem Solving's Introduction to Algebra by Richard Rusczyk.
Login Register Memberlist Search AoPS Blogs Contests Galleries Forum Index
The time now is Fri Nov 27, 2009 1:51 pm
All times are UTC - 8
View posts since last visit
View unanswered posts
The maximal value of the inradius of a an inscribed triangle
Moderators: High School Olympiad Moderators, darij grinberg, N.T.TUAN, orl, pbornsztein, pohoatza, yetti
Post new topic   Reply to topic View previous topicView next topic
7 Posts • Page 1 of 1
Author Message
vess
Yang-Mills Theory
Yang-Mills Theory

Offline
Joined: 10 May 2004
Posts: 724
Location: Cambridge, MA
Bulgaria

To rate posts you must be logged in
#1
The maximal value of the inradius of a an inscribed triangle
Bulgarian TST's 2004 --- Problem 3.

Let ABC be a triangle whose vertices lie in the interior or on the boundary of the unit square. Determine the maximal possible value of the inradius of ABC.

PostPosted: Thu May 27, 2004 4:50 am  Back to top 
  ProfilePM
iura
Riemann Hypothesis
Riemann Hypothesis

Offline
Joined: 09 Apr 2004
Posts: 461
Location: Chisinau
Moldova, Republic ofUnited States

To rate posts you must be logged in
#2
I think the maximum is reached by the isosceles triangle whose vertices are A,B and midpoint of CD.
Please tell me is my claim true?

PostPosted: Thu May 27, 2004 9:08 am  Back to top 
  ProfilePMYMICQ
harazi
Birch & Swinnerton Dyer
Birch & Swinnerton Dyer

Offline
Joined: 12 Nov 2003
Posts: 5494
Location: Paris
RomaniaFrance

To rate posts you must be logged in
#3
Yes, iura, your claim is true. It seems that this is a classical super-hard problem. I wonder who solved it during the contest without knowing it. I find it simply impossible. Some hints:
Obviously, we must search for traingles with vertices on the sides of the square. Then it's not very difficult (though I don't find it trivial) to show that if we have such a triangle, then we can find another triangle with one vertex a vertex of the square and the other two vertices on the opposite sides of the square (relative to the vertex of the square which was already chosen) and having larger inradius. And so the problem "reduces" to proving the very hard inequality:
\sqrt{1+x^2}+\sqrt{1+y^2}+\sqrt{(1-x)^2+(1-y)^2}\geq(1+\sqrt{5})(1-xy) for any x and y between 0 and 1. This is a hell of an inequality! We fix y and treat this as a function in x, let it be f(x). Then it's easy to see f is convex. We have three cases:
First case:
y is at least \frac{1}{4}. In this case it's easy to show that f'(0)>0 and since f' is increasing we find that f'(x)>0 and so f is increasing. From now on it's easy.
Second case: first, a notation: take r^2=1-\frac{6\sqrt{3}}{(1+\sqrt{5})^2}. The second case is when x is smaller than \frac{1}{4} and y is smaller than r. It is not hard to find that f'(\frac{1}{4})<0 and so f is decreasing on [0,\frac{1}{4}]. Using the first case we find that f(\frac{1}{4})\geq 0 . Due to the symetry we have proved that this is true for all points (x,y) outside the set A of those points with both coordinates between r and \frac{1}{4}.
Third case : (x,y) is in A. In this case we consider the triangle O(0,0), M(1,y) and N(x,1). Then the sum of radicals is exactly its perimeter. And its area is \frac{1-xy}{2}. Using the fact that in any triangle 12\cdot\sqrt{3}\cdot S\leq (perimeter)^2 we find that the sum of radicals is at least \sqrt{6\sqrt{3}}\cdot\sqrt{1-xy} which is at least (1+\sqrt{5})(1-xy) since we have (x,y) in A.
This is not my solution. I'm not so smart. But it shows how extremeley difficult this problem is. And the inequality proved here is one of the most difficult I have ever seen.

PostPosted: Fri May 28, 2004 11:35 am  Back to top 
  ProfilePM
harazi
Birch & Swinnerton Dyer
Birch & Swinnerton Dyer

Offline
Joined: 12 Nov 2003
Posts: 5494
Location: Paris
RomaniaFrance

To rate posts you must be logged in
#4
And to show how hard this problem is, I jus say it was a conjecture of Louis Funar (1984!!!!!!!!!!) one of our best olympics ever. This really says a lot. I wonder if anyone could solve this problem.

PostPosted: Fri May 28, 2004 11:51 am  Back to top 
  ProfilePM
iura
Riemann Hypothesis
Riemann Hypothesis

Offline
Joined: 09 Apr 2004
Posts: 461
Location: Chisinau
Moldova, Republic ofUnited States

To rate posts you must be logged in
#5
thank you. I will try to understand the solution.
I've spent more then 4.5 hours on it for sure!
i have some ideas, but I'm still not sure they are correct.
We get that one vertex is a vertex on a square, the other lying on two adjacent sides.
We use the following: From triangles with given area, the biggest inradius has the isosceles triangle(1), and from isosceles triangles with given area, the smaller the base the bigger the inradius( i'm not sure of the last)(2)
If the triangle is AMN we show MN>AM,AN or we could use (1) to increase inradius.
Now dilate AMN to become triangle with area 1/2, then make it isosceler, then decrease the base to 1. I wonder if all these operations increase the inradius.
If it's true, then we reach our desired triangle and we are done.

PostPosted: Sat May 29, 2004 3:43 am  Back to top 
  ProfilePMYMICQ
ZERO-00
New Member
New Member

Offline
Joined: 10 Aug 2008
Posts: 1

To rate posts you must be logged in
#6
for harazi
i found a simple proof for the inequality

PostPosted: Sun Aug 10, 2008 1:59 pm  Back to top 
  ProfilePM
Marian Dinca
New Member
New Member

Offline
Joined: 04 Sep 2008
Posts: 16

To rate posts you must be logged in
#7
The maximal value of the inradius of a an inscribed triangle

Lema 1:
For triangle ABC to exist two triangle isosceles same perimeter and same area of trangle ABC.
Proof:
Let a, b, c sides of triangle ABC and s = \frac {a + b + c}{2}. Area of triangle ABC: F = \sqrt {s.(s - a)(s - b)(s - c)}, and aply AM-GM rezults:
F\le \sqrt {s.(\frac {\sum{(s - a)}}{3})^3} = \frac {s^2}{3.\sqrt {3}}\Rightarrow 0 < F\le\frac {s^2}{3.\sqrt {3}}.
Let triangle A_1B_1C_1 of sides x, x, 2s - 2x; x + x > 2s - 2x \rightarrow 4x > 2s \rightarrow x > \frac {s}{2} and x > 2s - 2x \Rightarrow x > \frac {2s}{3} > \frac {s}{2} and 2s - 2x > 0 \Rightarrow x < S because F_1 = \sqrt {s.(s - x)(s - x).(s - 2s + 2x)}\le \sqrt {s.(\frac {s - x + s - x + 2x - s}{3})^3} = \frac {s^2}{3.\sqrt {3}}.
Let f: (\frac {2s}{3};s)\rightarrow(0;\frac {s^2}{3.\sqrt {3}}); \ f(x) = \sqrt {s.(s - x)^2.(2x - s)}, because f continue, f Darboux proprerty\Rightarrow (\exists)x_0\in(\frac {2s}{3};s) and f(x_0) = F.
Using same method, for: g: (\frac {s}{2};\frac {2s}{3})\rightarrow (0;\frac {s^2}{3.\sqrt {3}}); g(x) = \sqrt {s.(s - x)^2.(2x - s)}, to exists y_0\in(\frac {s}{2};\frac {2s}{3}) and g(x_0) = F. Lema 1 proved!

Lema 2:
For any triangle ABC, and a, b, c sides, exists triangle A_1B_1C_1 for sides: a_1 = \sqrt {a.(b + c - a)}; b_1 = \sqrt {b.(a + c - b)}; c_1 = \sqrt {c.(a + b - c)}.
The triangle A_1B_1C_1 are same area of triangle ABC and perimeter: 2.s_1 = a _1 + b_1 + c_1 < a + b + c = 2.s.
Proof: let a = y + z, b = x + z, c = x + y\Rightarrow
\Rightarrow a_1 = \sqrt {2x.(y + z)};b_1 = \sqrt {2y.(x + z)};c_1 = \sqrt {2z.(x + y)};
a_1 + b_1 > c_1; a_1 + c_1 > b_1; b_1 + c_1 > a_1.
16.F^2 = (\sum a).\prod (b + c - a)} = 16.xyz.(x + y + z)
and
16.F_1^2 = 2.\sum {a_1^2b_1^2} - \sum{a_1^4} =
= 2.\sum{2x.(y + z).2y.(x + z)} - \sum {(2x.(y + z))^2} =
= 8.\sum{xy.(z + y).(z + x)} - 4.\sum{x^2.(y + z)^2} =
= 8.\sum{xy.(z^2 + xy + yz + xz)} - 4.\sum{x^2.(y^2 + 2.yz + z^2)} =
= 8.\sum{xy.(z^2 + xy + yz + xz)} - 4.\sum{x^2.(y^2 + z^2)} - 8.xyz.(x + y + z) =
= 8.xyz.(x + y + z) + 8.(\sum{xy})^2 - 8.\sum{x^2y^2} - 8.xyz.(x + y + z) =
= 8.(\sum{xy})^2 - 8.\sum{x^2y^2} = 16.\sum{(xy)(xz)} = 16.xyz.(x + y + z) = 16.F^2.
Rezults: 16.F^2 = 16.F_1^2 \Leftrightarrow F = F_1 and using AM-GM
\Rightarrow a_1 + b_1 + c_1 = \sum{\sqrt {a.(b + c - a)}} < \sum{\frac {a + b + c - a}{2}} = \sum{\frac {b + c}{2}} = \sum...
Acording to Lema 1, exists two triangle isosceles same perimeter and same area of triangle A_1B_1C_1, conclude exists two triangle isosceles same area of triangle ABC and perimeter inferior of perimeter triangle ABC, and:
r = \frac {F}{s} < \frac {F}{s_1} = \frac {F_1}{s_1} = r_1, the is inradius of triangle isosceles.
Let \triangle {MNP} =isosceles, M=midpoint of [DC] and |DN|=CP|=x \Rightarrow |MN| = |NP| = \sqrt {x^2 + \frac {1}{4}} \Rightarrow
r_{MNP} = \frac {\frac {x}{2}}{\frac {2.\sqrt {x^2 + \frac {1}{4}} + 1}{2}} = \frac {x}{2.\sqrt {x^2 + \frac {1}{4}} + 1} = \....
Let: f(x) = \frac {1}{2.\sqrt {1 + \frac {1}{4.x^2}} + \frac {1}{x}}; x\in(0;1].
If x_1 < x_2 \Rightarrow f(x_1) < f(x_2), then: f-increasing \Rightarrow max{f(x)} = f(1) \Rightarrow r_{max} = \frac {1}{2.\sqrt {1 + \frac {1}{4}} + 1} = \frac {1}{\sqrt {5} + 1} = \fr...

PostPosted: Sat Oct 18, 2008 9:26 am  Back to top 
  ProfilePM
Display posts from previous:   Sort by:   
7 Posts • Page 1 of 1
Post new topic   Reply to topic View previous topicView next topic
Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum
You cannot post calendar events in this forum


© Copyright 2008 AoPS Incorporated. All Rights Reserved. • FoundationPrivacyContact Us