# Calculus problem.

1. Nov 30, 2003

Not my own homework problem, but somebody else's.

Parallel to a tall building runs a 6ft high fence. It is 2ft away from the building. What is the shortest ladder needed to reach the building from the other side of the fence?

I've drawn a diagram for this problem (see attached).

I've called the height above the ground at which the ladder touches the building 'h' and the distance of the base of the ladder from the fence 'x' and the length of the ladder 'L'.

Then

L2 = h2 + (x + 2)2

I think that by expressing L as a function in terms of one of the variables, x or h, and differentiating with respect to that variable, I can find then find the minimum length of the ladder.

By considering similar triangles

6/x = h/(2 + x)

So

x = 12/(h - 6)

And

h = 12/x + 6

But by substituting in either of these values, the differentiation dL/dx or dL/dh becomes horrible.

Is there an easier way to solve this problem? Am I overlooking something? I'd like to see if I could find an easier function to differentiate.

#### Attached Files:

File size:
2 KB
Views:
70
2. Nov 30, 2003

### arcnets

This is the trick: Look for an extremum of L2. You know that L is positive, so an extremum of L2 is also an extremum of L.
I've tried
$$\frac{d}{dh}L^2 = 0$$
It's lengthy but easy.

Edit: I think the rest of your work is OK.

Last edited: Nov 30, 2003
3. Nov 30, 2003

Okay, so I can do dL2/dx or dL2/dh rather than and dL/dx or dL/dh and that will make the maths easier.

You've suggested dL2/dh.

Substituting x = 12/(h - 6) into the equation for L2 means that I would have to differentiate

h2 + [(12/(h - 6)) + 2]2

with respect to h. Right?

4. Nov 30, 2003

### arcnets

Correct.
Simplify before you derive. A lot of things will cancel.

5. Nov 30, 2003

So

L2 = h2 + [(12/(h - 6)) + 2]2

= h2 + 144/[(h - 6)2] + 48/(h - 6) + 4

What's cancelling?

6. Nov 30, 2003

### HallsofIvy

Oh, c'mon. Combine the fractions before you square!

7. Nov 30, 2003

To get

L2 = h2 + 4h2/(h - 6)2

...?

Can I simplify it further?

8. Dec 1, 2003

### arcnets

I think this will do. Next, derive...

9. Dec 1, 2003