
Re: Crossed Ladders Problem (again)
Posted:
Aug 23, 2001 12:46 PM


On 23 Aug 2001, Randall L. Rathbun wrote:
> My GPPari program will solve for the roots, using some type of numerical > algorithm to hundreds of thousands of digits. But what I want to know, is > the very nature of this problem intrinsically involve a quartic so that no > matter how hard we try, it will always pop up?
I think the answer is "yes", even though I'm not quite sure exactly what your question means.
Let me say what my "yes" means. If you're given the lengths a and b of the two ladders, and the height h of their intersection, then the distance x between the two vertical walls is given by a quartic equation that is in general irreducible, although for particular values of a,b,h it might just happen to have a rational root.
It's easy to find such values: we can suppose the bottom of the picture looks like this:
\ / X /\ 15/  \13 / 12 \ /______\ 9 5 for instance. Then obviously we have x = 14 and since the top half of the figure will look like:
\  \13u / 12u \ 5v/ 4v  \ /   5u X 3v
we must have
a = 13u + 13, where 5u = 9 and b = 15 + 5v, where 3v = 5,
giving a = 182/5, b = 70/3, h = 12, (for x = 14).
I rescale this to get a simpler case: multiplying by 15/14 gives
a = 39, b = 25, h = 90/7
for a problem whose answer is x = 15.
I wonder what the simplest rational solution is?
John Conway

