Another variation in this question which comes to my mind is that instead of dividing the segment Ak−3Ak−2 in a constant ratio, we divide ∠Ak−3Ak−1Ak−2 in a constant ratio for all k≥4. Again Ak's converges to some point J and point E is defined similarly. We have to find the ratio JEA1J.
A new problem what I want to ask is:
Problem - Let there be a triangle A1A2A3 as shown in the figure below. Lines AiAi+1 are drawn with Ai+1 lying on segment Ai−2Ai−1 such that for some real number r in between 0 and 1
∠Ai−2AiAi−1∠Ai−2AiAi+1=ri≥3,i∈N
In this way △Ai−2Ai−1Ai becomes smaller as i is getting larger and ultimately converges to a point, say J. Join A1 and J and extend it to meet A2A3 at E.
Find the ratio JEA1J
Unlike the previous problem, I think this problem is incomplete. Since this problem requires division of angles, angles of △A1A2A3 must be given in order to find the ratio JEA1J.
So for completeness of the problem suppose angles of △A1A2A3 are given as A1=A,A2=B,A3=C.
I have used the approach of solution given by Chris Lewis in my previous problem but not able to proceed after some steps.
Solution - First shift this figure into a cartesian plane with A1 located at (0,0), A2 located at (a,0) and A3 located at (b,c).
Let a=a12= length of side A1A2. So, a12>0.
Coordinates (b,c) can be determined as:
b=A1A3cosA1
c=A1A3sinA1
Now, sinA2A1A3=sinA3A1A2 (sine rule of triangle)
⇒A1A3=sinA3A1A2sinA2=sinA3a12sinA2
So we get,
b=sinA3a12sinA2cosA1
c=sinA3a12sinA2sinA1
There is some nomenclature which I have introduced for easy writing and better understanding the terms.
Length of segment ApAq is denoted by apq.
Length(ApAq)=apq for all p,q∈N
For simplicity in writing we denote
∠Ai−2AiAi−1 with Ai for all i≥3
Ratio of length of one part of segment AiAi+1 to the segment itself is denoted by si.
and sn=sinAn+2sinAn+3sin[(1−r)An+1]sin(rAn+2) for all n≥2
We see that in the sequence sn there is a sequence of angles An. So we need to first solve for general value of An.
A1,A2,A3 are the three initial angles given. It is easy to see that
A4=A1+rA3
A5=A2+rA4
seems like that next term will be
A6=A3+rA5
but this is wrong, if you see in the diagram you will get that
A6=(1−r)A3+rA5
and in general
An=(1−r)An−3+rAn−1 for all n≥6
So we get the sequence An defined as
A1=A,A2=B,A3=C and
An=An−3+rAn−1 for n=4,5
An=(1−r)An−3+rAn−1 for all n≥6
So what we have done from starting is summarised below
Every point An,n≥1 has its coordinate as
An=(xn,yn) for all n∈N
Definition of sequencexn
x1=0,x2=a12,x3=sinA3a12sinA2cosA1
xn=(1−sn−3)xn−3+sn−3xn−2 for all n≥4
Definition of sequenceyn
y1=0,y2=0,y3=sinA3a12sinA2sinA1
yn=(1−sn−3)yn−3+sn−3yn−2 for all n≥4
Definition of sequencesn
s1=sinA3sinA4sinA2sin(rA3)
sn=sinAn+2sinAn+3sin[(1−r)An+1]sin(rAn+2) for all n≥2
Definition of sequenceAn
A1=A,A2=B,A3=C
An=An−3+rAn−1 for n=4,5
An=(1−r)An−3+rAn−1 for all n≥6
We need to find the coordinates of point J to get the required ratio JEA1J. Let coordinates of point J be (xJ,yJ). Then we have,
xJ=n→∞limxn and yJ=n→∞limyn
In order to get the general term of xn,yn we need the general term of sn for which we need the general term of An.
We see that sequence An is the only independent sequence free of any other sequence. We need to find the general term of this sequence. We see that it is homogenous linear recurrence of degree 3 with constant cofficients.
Let An=λn,n≥3 be the solution of the recurrence relation. Then putting this solution in the recurrence relation we get,
λn=(1−r)λn−3+rλn−1
⇒λ3=(1−r)+rλ2
⇒λ3−rλ2−(1−r)=0
There are the following three roots to this equation:
λ1=1,λ2=−21−r+2(3+r)(1−r)i,λ3=−21−r−2(3+r)(1−r)i where i=−1
So we get,
An=α1n+βλ2n+γλ3n for all n≥3
To get the values of α,β,γ we use the values of A3,A4,A5 in terms of A,B,C.
Solving for α,β,γ from these equations in terms of r by writing λ2 and λ3 in terms of r we get, (I am not writing the intermediate steps, it is very tedious for writing here)
α=3−2rA+B+C=3−2r180°[∵A,B,C are the angles of triangle]
An=3−2r180°+β(−21−r+2(3+r)(1−r)i)n+β(−21−r−2(3+r)(1−r)i)n for all n≥3
with β as given above.
I have solved this question till here and now I am not able to proceed further. If anyone has any idea how to proceed now, please contribute your ideas into this. It is going very complex and large expressions everywhere, so one can use some special triangle A1A2A3 and any special value of r to reduce the expressions and solve further. Please help if anyone has any idea how to proceed further or any other approach to solve this problem.
This discussion board is a place to discuss our Daily Challenges and the math and science
related to those challenges. Explanations are more than just a solution — they should
explain the steps and thinking strategies that you used to obtain the solution. Comments
should further the discussion of math and science.
When posting on Brilliant:
Use the emojis to react to an explanation, whether you're congratulating a job well done , or just really confused .
Ask specific questions about the challenge or the steps in somebody's explanation. Well-posed questions can add a lot to the discussion, but posting "I don't understand!" doesn't help anyone.
Try to contribute something new to the discussion, whether it is an extension, generalization or other idea related to the challenge.
Stay on topic — we're all here to learn more about math and science, not to hear about your favorite get-rich-quick scheme or current world events.
Markdown
Appears as
*italics* or _italics_
italics
**bold** or __bold__
bold
- bulleted - list
bulleted
list
1. numbered 2. list
numbered
list
Note: you must add a full line of space before and after lists for them to show up correctly
Wow, interesting question! I've mostly looked at this numerically so far, but aside from the angle relation you found, I haven't actually found any examples where a "nice" starting triangle (integer side lengths, integer coordinate vertices, equilateral, right-angled) leads to a "nice" (recognisable rational or algebraic) endpoint.
It's certainly a nice property that the angles converge as per your argument; in fact we have
n→∞lim∠An−2AnAn−1n→∞lim∠AnAn−1An−2n→∞lim∠AnAn−2An−1=3−2rπ=3−2r(1−r)π=3−2r(1−r)π
which come from your expression for An (note that ∣λ2∣=∣λ3∣<1, so only the first term matters in the limit). I was surprised the triangle tends towards an isosceles, non-equilateral one (perhaps I shouldn't have been, though).
I wonder if a different coordinate system might be helpful - something more natural to a triangle, like trilinear or barycentric? (I'm not very used to these but it might be worth investigating.) It made sense to use Cartesian coordinates in the first problem as the operations involved could easily be resolved independently in the x and y directions, but I'm not sure that's the case here.
This is a nice result you have deduced. I haven't noticed much on behaviour of angles might be because I was focused on finding general term of xn. These limiting angles sums to 180° without using the angle sum property of triangles in finding those limiting angles, so I think it is a good indication that I haven't made any mathematical mistake in any expression.
After solving this question with cartesian coordinates, I have tried this a bit with trilinear but in that case also, sequence of coordinate of An was coming to be recurrence with variable coefficients as in this case. I don't know barycentric. I will search for it to see whether it is useful to this question or not.
I started with the specific case of r=21, which means each angle is bisected, and chose a right angle isosceles triangle as a convenient triangle to use with A1(0,0), A2(2,0), and A3(1,1), because then the new points are always midpoints of previous segments.
A recursion relationship can be set up with the midpoints so that (xn,yn)=(21(xn−3+xn−2),21(xn−3+xn−2)) for n≥3 and (x1,y1)=(0,0), (x2,y2)=(2,0), (x3,y3)=(1,1). After some computation, J is J(56,52), E is E(23,21), A1J=5210, JE=1010, and JEA1J=4 .
I still need to work on a general case for r, though.
Brilliant, those are definitely "nice" parameters!
I'd been looking at r=21 as well (the bisection certainly helps), but your example just made me realise something else: if the starting triangle is the right "shape" (that is, isosceles with the angles in my earlier post), all the triangles formed are similar.
Another nice(ish) example I've found this way is with r=43 and initial points (in order) (0,0),(2,0),(1,tan30∘), which heads to a final point at (79,73).
With a little bit of calculation, I found that the coordinates A1(0,0), A2(2,0), and A3(1,tan(3−2r1−r⋅180°)) will lead to triangles that are similar and will cut each side at a ratio of r′=2⋅cos(3−2r1−r⋅180°)⋅sin(3−2r1⋅180°)sin(3−2rr⋅180°), then making it a similar problem to the original "converging triangle problem".
I was trying to find some result with r=21 (without any special triangle) but haven't got much. It was nice idea to take right isosceles triangle with r=21. This has reduced the question to simply bisecting previous segment. But this actually converts this question to my previous question.
I always want to solve this question using basic elementary geometry as the solution of previous question. But I think that will not work here.
I noticed that the ratio JEA1J does change depending on where its vertices are, in both this question and its original question. For example, taking my right isosceles triangle with A1(0,0), A2(2,0), and A3(1,1) above with r=21 makes JEA1J=4, however, using A1(0,0), A2(1,0), and A3(0,1) instead (like in the original question's solution) with r=21 makes JEA1J=3. I believe that means that the original question must also include the extra parameter that the height of △A1A2A3 must equal the base of △A1A2A3 to make the given solution correct.
Edit: Nevermind, I had mixed up some variables, so what I said here is not true. Shikhar Srivastava is correct, JEA1J=4 for r=21 irregardless of the size and shape of the triangle.
If you are using r=21 in this question then the ratio will change with respect to vertices of triangle. But if you are r=21 in the previous question then that ratio does not depend on the vertices of triangle, not even on size and shape of triangle. It just depend on r as
JEA1J=1−r2 for any starting triangle.
You can see my solution to that question, the solution is purely based on elementary geometry and no where used the size and shape of triangle. If you put 1−rr instead of 73 in that question and solve further whether by algebraic sequential method or by geometry, you will get the ratio as
JEA1J=1−r2.
I have just solved that question by algebraic method using cartesian coordinates and got the following results:
When A1=(0,0),A2=(a,0),A3=(b,c), with a,c>0 and b∈R, it represents all triangle possible (in order). For ex - triangle with A1=30°,A2=60°,A3=90° and another triangle with A1=60°,A2=90°,A3=30° with same size are considered different.
We have
x1=y1=0,x2=a,y2=0,x3=b,y3=c and
xn=(1−r)xn−3+rxn−2 for all n≥4
And we get the solution of general term of xn as:
Combining some results the equations r′=2⋅cos(3−2r1−r⋅180°)⋅sin(3−2r1⋅180°)sin(3−2rr⋅180°) and JEA1J=1−r′2 found in this thread and simplifying, if r=∠Ai−2AiAi−1∠Ai−2AiAi+1, then:
Thanks for sharing the question. I like this as a forum to discuss these sort of meta-questions.
By the way, have you heard of the Encyclopedia of Triangle Centres? It has a search facility which might be of interest here. I haven't had a chance to look in detail yet but the idea is that if you calculate the trilinear coordinates of a point you're interested in, you can check if it's a "known" triangle centre. There are about 32000 centres on this site so there's a decent chance of finding a match! It could well help with the r=21 case (and perhaps others).
Easy Math Editor
This discussion board is a place to discuss our Daily Challenges and the math and science related to those challenges. Explanations are more than just a solution — they should explain the steps and thinking strategies that you used to obtain the solution. Comments should further the discussion of math and science.
When posting on Brilliant:
*italics*
or_italics_
**bold**
or__bold__
paragraph 1
paragraph 2
[example link](https://brilliant.org)
> This is a quote
\(
...\)
or\[
...\]
to ensure proper formatting.2 \times 3
2^{34}
a_{i-1}
\frac{2}{3}
\sqrt{2}
\sum_{i=1}^3
\sin \theta
\boxed{123}
Comments
@Zakir Husain @Alak Bhattacharya @Chew-Seong Cheong @Yajat Shamji @Atomsky Jahid @David Vreken @Wesley Low @Aryan Sanghi @Jeff Giff @zico quintina @Vinayak Srivastava @Mark Hennings @Hosam Hajjir
Log in to reply
Don't ask me!
Wow, interesting question! I've mostly looked at this numerically so far, but aside from the angle relation you found, I haven't actually found any examples where a "nice" starting triangle (integer side lengths, integer coordinate vertices, equilateral, right-angled) leads to a "nice" (recognisable rational or algebraic) endpoint.
It's certainly a nice property that the angles converge as per your argument; in fact we have n→∞lim∠An−2AnAn−1n→∞lim∠AnAn−1An−2n→∞lim∠AnAn−2An−1=3−2rπ=3−2r(1−r)π=3−2r(1−r)π
which come from your expression for An (note that ∣λ2∣=∣λ3∣<1, so only the first term matters in the limit). I was surprised the triangle tends towards an isosceles, non-equilateral one (perhaps I shouldn't have been, though).
I wonder if a different coordinate system might be helpful - something more natural to a triangle, like trilinear or barycentric? (I'm not very used to these but it might be worth investigating.) It made sense to use Cartesian coordinates in the first problem as the operations involved could easily be resolved independently in the x and y directions, but I'm not sure that's the case here.
Log in to reply
This is a nice result you have deduced. I haven't noticed much on behaviour of angles might be because I was focused on finding general term of xn. These limiting angles sums to 180° without using the angle sum property of triangles in finding those limiting angles, so I think it is a good indication that I haven't made any mathematical mistake in any expression.
After solving this question with cartesian coordinates, I have tried this a bit with trilinear but in that case also, sequence of coordinate of An was coming to be recurrence with variable coefficients as in this case. I don't know barycentric. I will search for it to see whether it is useful to this question or not.
I started with the specific case of r=21, which means each angle is bisected, and chose a right angle isosceles triangle as a convenient triangle to use with A1(0,0), A2(2,0), and A3(1,1), because then the new points are always midpoints of previous segments.
A recursion relationship can be set up with the midpoints so that (xn,yn)=(21(xn−3+xn−2),21(xn−3+xn−2)) for n≥3 and (x1,y1)=(0,0), (x2,y2)=(2,0), (x3,y3)=(1,1). After some computation, J is J(56,52), E is E(23,21), A1J=5210, JE=1010, and JEA1J=4 .
I still need to work on a general case for r, though.
Log in to reply
Brilliant, those are definitely "nice" parameters!
I'd been looking at r=21 as well (the bisection certainly helps), but your example just made me realise something else: if the starting triangle is the right "shape" (that is, isosceles with the angles in my earlier post), all the triangles formed are similar.
Another nice(ish) example I've found this way is with r=43 and initial points (in order) (0,0),(2,0),(1,tan30∘), which heads to a final point at (79,73).
Log in to reply
do you mean (1,tan30°)?
Log in to reply
With a little bit of calculation, I found that the coordinates A1(0,0), A2(2,0), and A3(1,tan(3−2r1−r⋅180°)) will lead to triangles that are similar and will cut each side at a ratio of r′=2⋅cos(3−2r1−r⋅180°)⋅sin(3−2r1⋅180°)sin(3−2rr⋅180°), then making it a similar problem to the original "converging triangle problem".
Log in to reply
I was trying to find some result with r=21 (without any special triangle) but haven't got much. It was nice idea to take right isosceles triangle with r=21. This has reduced the question to simply bisecting previous segment. But this actually converts this question to my previous question.
I always want to solve this question using basic elementary geometry as the solution of previous question. But I think that will not work here.
Log in to reply
Yes, you're right, it does just convert this question to your previous question.
I noticed that the ratio JEA1J does change depending on where its vertices are, in both this question and its original question. For example, taking my right isosceles triangle with A1(0,0), A2(2,0), and A3(1,1) above with r=21 makes JEA1J=4, however, using A1(0,0), A2(1,0), and A3(0,1) instead (like in the original question's solution) with r=21 makes JEA1J=3. I believe that means that the original question must also include the extra parameter that the height of △A1A2A3 must equal the base of △A1A2A3 to make the given solution correct.
Edit: Nevermind, I had mixed up some variables, so what I said here is not true. Shikhar Srivastava is correct, JEA1J=4 for r=21 irregardless of the size and shape of the triangle.
Log in to reply
If you are using r=21 in this question then the ratio will change with respect to vertices of triangle. But if you are r=21 in the previous question then that ratio does not depend on the vertices of triangle, not even on size and shape of triangle. It just depend on r as
JEA1J=1−r2 for any starting triangle.
You can see my solution to that question, the solution is purely based on elementary geometry and no where used the size and shape of triangle. If you put 1−rr instead of 73 in that question and solve further whether by algebraic sequential method or by geometry, you will get the ratio as
JEA1J=1−r2.
I have just solved that question by algebraic method using cartesian coordinates and got the following results:
When A1=(0,0),A2=(a,0),A3=(b,c), with a,c>0 and b∈R, it represents all triangle possible (in order). For ex - triangle with A1=30°,A2=60°,A3=90° and another triangle with A1=60°,A2=90°,A3=30° with same size are considered different.
We have
x1=y1=0,x2=a,y2=0,x3=b,y3=c and
xn=(1−r)xn−3+rxn−2 for all n≥4
And we get the solution of general term of xn as:
When r=43,
xn=94(a+b)+(9−4a+12an+32b−24bn)(2−1)n
yn=94c+(932c−24cn)(2−1)n
So, J=(94(a+b),94c)
Using A1 and J we get,
E=(2a+b,2c)
So, A1J=94(a+b)2+c2
and JE=(21−94)(a+b)2+c2=181(a+b)2+c2
Hence JEA1J=8=1−432=1−r2
When r=43,
xn=3−ra+b+β1(−21+214r−3)n+γ1(−21−214r−3)n
yn=3−rc+β2(−21+214r−3)n+γ2(−21−214r−3)n
It is not needed to calculate β and γ, because ∣∣∣∣−21±214r−3∣∣∣∣<1
So, J=(3−ra+b,3−rc)
Using A1 and J we get,
E=(2a+b,2c)
So, A1J=3−r1(a+b)2+c2
and JE=(21−3−r1)(a+b)2+c2=2(3−r)1−r(a+b)2+c2
Hence JEA1J=2(3−r)1−r3−r1=1−r2
Hence we see that for any triangle A1A2A3 we get the same result of the ratio JEA1J.
Log in to reply
I'm confused. If JEA1J=1−r2, then for the original question where r=73, JEA1J=1−732=27, but the answer is JEA1J=720.
Nevermind, I'm mixing up variables. The original question actually has r=3+73=103, which does come out to JEA1J=1−r2=720.
Combining some results the equations r′=2⋅cos(3−2r1−r⋅180°)⋅sin(3−2r1⋅180°)sin(3−2rr⋅180°) and JEA1J=1−r′2 found in this thread and simplifying, if r=∠Ai−2AiAi−1∠Ai−2AiAi+1, then:
JEA1J=2(sin(3−2r2−r⋅180°)sin(3−2rr⋅180°)+1)
.
One point need to be added in this is that it is true for starting triangle with angles 3−2r180°,3−2r180°(1−r),3−2r180°(1−r).
Thank you both of you @Chris Lewis @David Vreken for contributing your ideas and deriving some wonderful results out of this question.
Log in to reply
Thanks for sharing the question. I like this as a forum to discuss these sort of meta-questions.
By the way, have you heard of the Encyclopedia of Triangle Centres? It has a search facility which might be of interest here. I haven't had a chance to look in detail yet but the idea is that if you calculate the trilinear coordinates of a point you're interested in, you can check if it's a "known" triangle centre. There are about 32000 centres on this site so there's a decent chance of finding a match! It could well help with the r=21 case (and perhaps others).