# Lines, lines, lines galore

Let $$n$$ be a positive integer. On each side of a square $$n$$ points are chosen. (No point is at a vertex of the square.)

(a) How many segments are there whose endpoints are two of the above points and such that no segment lies along a side of the square?

(b) What is the maximal possible number of intersection points of the segment from part (a)?

Give proof.

Note by Sharky Kesa
3 years ago

MarkdownAppears as
*italics* or _italics_ italics
**bold** or __bold__ bold
- bulleted- list
• bulleted
• list
1. numbered2. list
1. numbered
2. list
Note: you must add a full line of space before and after lists for them to show up correctly
paragraph 1paragraph 2

paragraph 1

paragraph 2

[example link](https://brilliant.org)example link
> This is a quote
This is a quote
    # I indented these lines
# 4 spaces, and now they show
# up as a code block.

print "hello world"
# I indented these lines
# 4 spaces, and now they show
# up as a code block.

print "hello world"
MathAppears as
Remember to wrap math in $$...$$ or $...$ to ensure proper formatting.
2 \times 3 $$2 \times 3$$
2^{34} $$2^{34}$$
a_{i-1} $$a_{i-1}$$
\frac{2}{3} $$\frac{2}{3}$$
\sqrt{2} $$\sqrt{2}$$
\sum_{i=1}^3 $$\sum_{i=1}^3$$
\sin \theta $$\sin \theta$$
\boxed{123} $$\boxed{123}$$

Sort by:

@Julian Poon As I said, do it in a similar way for a). Of course, that assumes that you did a) in a "smart" way.

So, my way of doing a is: There are $${4n \choose 2 }$$ segments, of which $$4 \times { n \choose 2 }$$ of them lie on a side of the square. Hence, there are $$\frac{ 4n (4n-1) } { 2} - 4 \times \frac{ n(n-1) } { 2} = 6n^2$$ pairs of such lines.

Use a similar approach to do b:

Hint: Given any 4 points, there is at most 1 way to get an intersection point.
Hint: How many sets of 4 points are there, which involve 3 or more points on the same line segment?
Hint: If in a set of 4 points, no 3 of them lie on the same line segment, then there is exactly 1 way to get an intersection point, and this doesn't involve a line segment on the side of the square.

Hence, the answer is

${ 4n \choose 4 } - 5 \times { n \choose 3 } \times 3n - 4 \times { n \choose 4 } = \frac{1}{2} n^2 ( 17n^2 - 18n + 3 ).$

Staff - 2 years, 12 months ago

Oh... that's a great approach!

- 2 years, 12 months ago

I'm getting $$6n^{2}$$ for a) and as for b), I'm getting overly complicated formulas which are hopefully correct.

- 3 years ago

UPDATE: I'm so done with this question. Through systematic counting, the solution for b) is

$\left(9n^2\sum _{k=1}^{n-1}k\right)-\left(\sum _{k=1}^{n-1}\sum _{j=1}^{3n}kj\right) + \left(n\sum _{k=1}^n\sum _{j=1}^{2n}\left(j-1+3\left(k-1\right)\right)\right)-\left(\sum _{k=1}^n\sum _{j=1}^{2n}\left(k-1\right)j\right) + \left(n\sum _{k=1}^n\sum _{j=1}^n\left(2j+\left(3k-5\right)\right)\right)-\left(\sum _{k=1}^n\sum _{j=1}^n\left(k-1\right)j\right)$

Goodbye and good luck.

I can elaborate on how I counted it if you want, but there are many other ways to do so.

- 3 years ago

Elaborate please. Your formula might be able to be simplified via peer help.

- 3 years ago

Upon simplification, the answer becomes $\frac{1}{2}n^2\left(17n^2-18n+3\right)$

To get the answer, I first split the question up into $$3$$ different groups, group A, B and C. For each group, the numbers represent the order in which I would draw the lines. I would draw the lines in Group A first, then group B and finally group C. When drawing these lines, I would count the lines it would intersect.

So for example, the first point I am going to entertain would be Point 1 in group A. After drawing all the lines that are connected to that point, I would get 0 intersections.

Next, I would entertain the second point in group A. After drawing all the lines connecting to that point, I would get <Some function in terms of n> intersections.

And so on.

After doing this for Group A, B and C, I get that the number of intersections for

Group A: $$\left(9n^2\sum _{k=1}^{n-1}k\right)-\left(\sum _{k=1}^{n-1}\sum _{j=1}^{3n}kj\right)$$

Group B: $$\left(n\sum _{k=1}^n\sum _{j=1}^{2n}\left(j-1+3\left(k-1\right)\right)\right)-\left(\sum _{k=1}^n\sum _{j=1}^{2n}\left(k-1\right)j\right)$$

And Group C: $$\left(n\sum _{k=1}^n\sum _{j=1}^n\left(2j+\left(3k-5\right)\right)\right)-\left(\sum _{k=1}^n\sum _{j=1}^n\left(k-1\right)j\right)$$

I have "verified" the formula for accuracy by drawing out and manually counting the number of intersections for case $$n=1,2,3$$. The numbers matched.

- 2 years, 12 months ago

Note that b) is maximized when any two segments that don't share an endpoint intersect.

- 3 years ago

Don't over complicate b. Do it in a similar way to a.

Staff - 3 years ago