A family of triangles all have the same area which is numerically the same as their perimeters. The greatest circumradius of any of them is exactly twice the least circumradius of any of them. What's the area of a triangle in this family of triangles (to 3 decimal places)?
Note: An equilateral triangle with an area of 1 2 3 also has a perimeter of 1 2 3 , but this is the only such triangle with this area and perimeter, and hence only one circumradius.
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
The area A of a triangle expressed in terms of the inradius r and perimeter p is just A = r × p / 2 . There's more detail on the formula from the wiki here.
So if the area and perimeter equal each other, after substitution the formula is p = r × p / 2 , that is when r = 2 .
5 8 . 3 1 5 . . . . See below.
Log in to reply
I see that now. Thanks for bringing this to my attention. Let me first check with my own figures, although it's looking like right now that you have the better numbers.
Okay, yes, I worked a little bit more improving the accuracy of the numerical computation of the solution to the set of simultaneous equations I've posted, and I do come up with an area of 58.315..., not 58.316, when rounded off. I've changed the answer from 58.316 to 58.315.
At least I didn't find out that making the assumption that the extrema triangles are isosceles triangles was wrong somehow. That would have been a real puzzler if that had been the case.
Log in to reply
A triple u , v , w > 0 defines one of these triangles, provided that u v w = 4 ( u + v + w ) . The third variable w is determined by the first two, and we can always choose u = v . Thus any family of triangles must contain an isosceles example.
I guess you got lucky that the extreme values of circumradius can be achieved by isosceles triangles, because you did not prove that.
Log in to reply
@Mark Hennings – Right, I did not prove that, what I did was to consider bundling together the family of triangles sharing a common incircle of radius 2 , not quite like the way I have it in my posted graphic, that showed me that for symmetry reasons, the extrema were isosceles, so it wasn't luck. I decided to post this problem once I had realized this. Maybe later I'll post an explanation showing that any infinitesimal departure from either extrema isoceles results in a departure from extrema circumradius.
It told me that it is advance level. 😢
We need 2 s = Δ = s ( s − a ) ( s − b ) ( s − c ) , and hence 4 s = ( s − a ) ( s − b ) ( s − c ) . Writing u = s − a , v = s − b , w = s − c , we deduce that u , v , w > 0 are such that s = u + v + w , a = v + w , b = u + w , c = u + v , and we require 4 ( u + v + w ) = u v w Since 2 1 b c sin A = Δ , we deduce that 2 R = 2 Δ a b c = 2 Δ ( u + v ) ( u + w ) ( v + w ) = 2 Δ u 2 v + u 2 w + u v 2 + v 2 w + u w 2 + v w 2 + 2 u v w = 2 Δ ( u + v + w ) ( u v + u w + v w ) − u v w = 2 Δ ( u + v + w ) ( u v + u w + v w − 4 ) = 4 u v + u w + v w − 4 If we put p = 8 R , then u , v , w are positive real roots of the equation f ( X ) = X 3 − s X 2 + ( p + 4 ) X − 4 s = 0 and so we deduce that s 2 ≥ 3 ( p + 4 ) , f ( x ) has turning points 3 1 [ s ± s 2 − 3 ( p + 4 ) ] , with f ( 2 1 ( s ± s 2 − 3 ( p + 4 ) ) ) = 2 7 1 s ( 9 p − 7 2 − 2 s 2 ) ∓ 2 7 2 ( s 2 − 3 ( p + 4 ) ) 2 3 and so the fact that f ( X ) has three positive real roots can be stated by the condition ∣ ∣ s ( 9 p − 7 2 − 2 s 2 ) ∣ ∣ s 2 ( 2 s 2 − 9 p + 7 2 ) 2 4 p 3 − ( s 2 − 4 8 ) p 2 + ( 1 9 2 − 8 0 s 2 ) p + 1 6 ( s 2 + 4 ) 2 ≤ 2 ( s 2 − 3 ( p + 4 ) ) 2 3 ≤ 4 ( s 2 − 3 ( p + 4 ) ) 3 ≤ 0 If we put s = 2 q , this inequality becomes p 3 − ( q 2 − 1 2 ) p 2 + ( 4 8 − 8 0 q 2 ) p + 6 4 ( q 2 + 1 ) 2 ≤ 0 One of the roots of the cubic equation g ( X ) = X 3 − ( q 2 − 1 2 ) X 2 + ( 4 8 − 8 0 q 2 ) X + 6 4 ( q 2 + 1 ) = 0 must be negative. We want the other roots to be of the form 2 t , t for positive t . Thus we want the roots of the cubic to be 2 t , t , q 2 − 1 2 − 3 t . Matching the other coefficients of the cubic, we deduce that 2 t 2 + 3 t ( q 2 − 1 2 − 3 t ) = 4 8 − 8 0 q 2 2 t 2 ( q 2 − 1 2 − 3 t ) = − 6 4 ( q 2 + 1 ) 2 and hence 7 t 2 − 3 ( q 2 − 1 2 ) t + ( 4 8 − 8 0 q 2 ) = 0 6 t 3 − 2 ( q 2 − 1 2 ) t 2 − 6 4 ( q 2 + 1 ) 2 = 0 Solving these equations simultaneously, we deduce that t = 3 ( q 4 + 2 5 6 q 2 − 2 4 ) 1 6 ( 4 4 q 4 + 1 6 1 q 2 + 1 3 ) Since t is a root of the cubic g ( X ) , we deduce that q must satisfy the polynomial equation 9 q 1 0 − 1 9 5 4 q 8 + 9 6 7 3 q 6 − 1 9 7 8 7 6 q 4 − 1 1 4 4 q 2 − 1 6 = 0 Thus the problem finally reduces to that of solving a quintic equation in q 2 . Solving numerically, we find that we have exactly one positive root q , namely 1 4 . 5 7 8 7 8 7 . This leads to the desired area of 5 8 . 3 1 5 2 . This is 5 8 . 3 1 5 to 3 decimal places.
A fearsome but complete proof that the extrema triangles are isosceles, unlike my solution. Thank you!
and, provided that s 2 ≥ 3 ( p + 4 ) , ...
How did you know this? All I can think of is Weitzenböck's inequality , which doesn't come close at all.
... f ( x ) has turning points 3 1 [ s ± s 2 − 3 ( p + 4 ) ] , with f ( 2 1 ( s ± s 2 − 3 ( p + 4 ) ) ) = 2 7 1 s ( 9 p − 7 2 − 2 s 2 ) ∓ 2 7 2 ( s 2 − 3 ( p + 4 ) ) 2 3 and so the fact that f ( X ) has three positive real roots can be stated by the condition ∣ ∣ s ( 9 p − 7 2 − 2 s 2 ) ∣ ∣ s 2 ( 2 s 2 − 9 p + 7 2 ) 2 4 p 3 − ( s 2 − 4 8 ) p 2 + ( 1 9 2 − 8 0 s 2 ) p + 1 6 ( s 2 + 4 ) 2 ≤ 2 ( s 2 − 3 ( p + 4 ) ) 2 3 ≤ 4 ( s 2 − 3 ( p + 4 ) ) 3 ≤ 0
This is a nice trick (you taught me last time already). But the (final) inequality can be easily deduced via cubic discriminant , so we don't need to know that s 2 ≥ 3 ( p + 4 ) in the first place.
and hence 7 t 2 − 3 ( q 2 − 1 2 ) t + ( 4 8 − 8 0 q 2 ) = 0 6 t 3 − 2 ( q 2 − 1 2 ) t 2 − 6 4 ( q 2 + 1 ) 2 = 0 Solving these equations simultaneously, we deduce that t = 3 ( q 4 + 2 5 6 q 2 − 2 4 ) 1 6 ( 4 4 q 4 + 1 6 1 q 2 + 1 3 )
How did you solve them simultaneously with ease? You got a cubic equation and a quadratic equation of t . Did you do the whole Cardano/depress cubic method? Because that looks ridiculously tedious. Or do you have a better approach?
Solving numerically, we find that we have exactly one positive root q , namely 1 4 . 5 7 8 7 8 7 .
Can we prove that this quintic equation can't be solved algebraically? I'm still not an expert in Galois Theory yet, so maybe there's some trick that I'm not aware of. (Note that WolframAlpha suggests that the roots can't be expressed algebraically)
Log in to reply
I have slightly rephrased the proof. The fact that f ( X ) = 0 has three roots means that f must have turning points, and so the first inequality is necessary. The final inequality also implies that s 2 ≥ 3 ( p + 4 ) , so all is well. I wanted to ensure that we had three positive real roots, and so I did the proof by hand, and did not use the discriminant.
We have t satisfying a cubic and a quadratic equation (both with coefficients involving q ). Thus t must satisfy the linear polynomial which is the remainder obtained when the cubic is divided by the quadratic. That expresses t in terms of q .
I did not take the time to investigate the solubility of the quintic. There are ways of checking whether its Galois group is S 5 or not, but I felt that the solution was long enough.
Thank you, Mark, for your solution. I've worked on this problem for quite a while. Still I can't find the mistake in my approach. Can you please help me spot the mistake? Let's call R the circumradius and r the inradius, p the perimeter and S the area. After easily computing r=2, for Euler's inequality we have R>2r=4 or R=2r=4. That means that the minimum value for R is 4. Since the problem states that the maximum value for R is exactly twice its minimum value, we can say that R<8 or R=8. Now since R=abc/4S and S is the same for all these triangles, than R reaches its maximum value as the product abc reaches its maximum value, which is (a+b+c)^3 ----------------- (for AM GM inequality) 27 So I got an expression for the upper bound of R, which is
27x4S Which equals
108
(because S is numerically equal to (a+b+c)). Now I just equaled the upper bound expression with its known value, getting p^2 -------=8 108 Now the perimeter is equal to the square root of 8x108=864 That is p=29.39387... And now the area simply equals rp/2
S=rp/2=2p/2=29.39387...
Where did I go wrong?
Log in to reply
Sorry for the form, I still have no program to write it clearly (the "------" should have been the line of a fraction... They should be ((a+b+c)^3)/27 Than ((a+b+c)^3)/(27x4s) which equals ((a+b+c)^2)/108 Than p^2/8=108
The maximum circumradius of the family is twice the minimum circumradius of the family . That does not mean that the minimum circumradius of that family has to be 4 .
Log in to reply
Seems like I relied too much on Eulers inequality R>2r. Why doesn't that hold true in here?
Log in to reply
@Alessandro Vici – The inequality R ≥ 2 r still holds, but you are using it too strongly. The conditions on the family of triangles are very exacting - see my proof.
Log in to reply
@Mark Hennings – I see.. thank you
@Mark Hennings – The case of the equilateral triangle seemed to confirm my proof, instead I should have taken the note into more consideration. In fact it is true that R>2r, but 2r is not exactly its minimum value, which is actually bigger than that.
Problem Loading...
Note Loading...
Set Loading...
Here's a graphic of two isosceles triangles in the family of triangles with the greatest and least circumradadii (dark blue circles). All triangles in this family have the same inradius of 2 (red circle), which is the property of any triangle in which its area and its perimeter are the same.
Let ( a 1 = b 1 , c 1 ) and ( a 2 = b 2 , c 2 ) be the sides of the two isosceles triangles, with corresponding base angles θ 1 , θ 2 . Then we have the following (true for both)
c = 4 C o t ( 2 1 θ )
a = 2 1 c S e c ( θ )
P = a + b + c
R = 4 ⋅ 2 ⋅ 2 1 P a b c
Then we solve the simultaneous equations by numerical means
P 1 = P 2
2 R 1 = R 2
with the solution θ 1 = 8 5 . 3 9 3 4 7 … and θ 2 = 1 5 . 9 3 0 8 2 … , which, when worked out, gives us the area of the triangles 5 8 . 3 1 5 …