> Let T be a triangle of area 6 square inches and perimeter 12 inches. Show
> that the triangle T can be cut into 2006 triangles each of which has perimeter strictly
> larger than 6 inches.
I don't get it, how is that possible...? 2006 smaller triangles? Are
you combing 2 smaller triangles to make another triangle?
William Elliot - 27 Apr 2008 08:14 GMT
> I don't get it, how is that possible...? 2006 smaller triangles? Are
> you combing 2 smaller triangles to make another triangle?
Sure enough,
another of those Googler Gooslers from the house of spam.
Make them thin. If you don't include the problem statement,
eventually, this thread will turn to gibberish. It's important
to keep the essential context. Read about why and other
netiquette manners expected of posters at:
http://oakroadsystems.com/genl/unice.htm
> > Let T be a triangle of area 6 square inches and perimeter 12 inches. Show
> > that the triangle T can be cut into 2006 triangles each of which has perimeter strictly
> > larger than 6 inches.
>
> Hint: a 3-4-5 right - triangle is an example of a T satisfying the
> given.
But not the only one!
> Hint 2: you can make an infinite number of sub-triangles with
> perimeter strictly larger than 10 inches, 6 is too easy.
I don't see that -- all sides of T might be less than 5, thus no
two points of T have a distance of 5: Consider the triangle
with sidelengths a = b = 4.65269812718231...
and c = 2.69460374563538...
(more precisely, a=b is solution of (6-a)^2 * (a-3) = 3 and c=12-2a).
Among very many triangles, at least one must have very small area
(here: area at most 6/2006), hence its perimeter is approximately
twice
its longest side and thus not much longer than about 9.3
> Hint 3: remember: the perimeter of a (non-degenerate) triangle is
> strictly larger than twice the longest side.
To maximize the minimal perimeter of the subtriangles, assume
that AB is the longest side (of length c),
let D be a point on BC with BD=eps,
divide BD into 2006 parts by points P[0] = B, P[1], ..., P[n] = D.
Then the triangles AP[k]P[k+1] have perimeter at least 2*(c-eps),
the same is true for and AP[n]C.
Thus we replace the limit length 6 inches with any value
less than twice the longest side of T.
What can we say about the longest side of T?
As noted above, it may happen to be <4.653.
By Heron, we have 36 = 6(6-a)(6-b)(6-c) or
6 = (6-a)(6-b)(6-c).
At least one of the factors on the RHS is <=cubrt(6) = 1.817...
hence at least one side is >= 4.18...
We can do better:
Wlog. a>=b>=c.
Then c<=4 and hence 6-c>=2, (6-a)(6-b)<=3,
hence at least one of these two factors is <=sqrt(3)=1.732...
and a >= 4.267...
But that implies c<=(12-4.267)/2 = 3.86..., (6-a)(6-b) <=2.811...,
a >=6-sqrt(2.811...) = 4.3232...
We could repeat this; in the limit we will find that
a >= L and c<= 6-L/2 where L is determined by
6-sqrt(6/(6-L/2)) = L
or
12 = (6-L)^2 * (12-L)
sqrt(6/(12-2L)) = 6-L, hence L = 6-cubrt(3) = 4.56775...
Since the triangle with
a = 6-cubrt(3) = 4.56775...
b = c = 3+cubrt(3)/2 = 3.7211...
is indeed possible, we see that the conditions
A=6, p=12, a>=b>=c
allow us to always produce arbitrarily many subtriangles
such that each has perimeter >M
*if and only if* M<12-2cubrt(3) = 9.1155...
hagman