Académique Documents
Professionnel Documents
Culture Documents
Dots
By Daniel Zaharopol, for AoPS Infinity Math Jam
February 28, 2007
Welcome!
Learning new mathematics doesnt stop with going to class. You only understand something
deeply when you try out some problems interesting problems, hard problems, which
challenge you to think about just what you learned. It is my hope that the problems below
will do that for you.
But they really can be hard. These wont be like problems in school; you wont be able
to solve them by sitting down and be done in five minutes per problem.
So sit down. Give yourself an hour, minimum. Get some desk space and some scratch
paper, and maybe a cup of tea or juice or coffee or whatever you like to drink. Read over
the first problem and play around with it. Think back to the definitions and how to apply
them. Start to try things until you get a coherent argument a proof for each problem.
The key to enjoying these guys is to enjoy being stuck. Dont get frustrated; keep thinking
and enjoy the challenge. Go to sleep, think about them in the shower the next morning. If
youre still stuck, feel free to private message/e-mail me and ask about them. Im happy to
provide pointers. And also: you will probably not be able to solve every problem!
Difficulty
The problems go from relatively easy to quite difficult conceptually. Most of the problems
require you to pair up two sets (i.e. to show that they have the same size/cardinality), or
else to prove that there is no way to pair them up.
Ill be around 15-20 minutes early for the next Math Jam to discuss your results, and
afterwards, as well.
Enjoy!
(Here F is paired with E; every point of E is matched with exactly one point of F , and
vice-versa.)
Heres how to pair E with G:
In general, two sets X and Y have the same cardinality if you can pair up the elements
of X with the elements of Y . If X and Y have the same cardinality, we write |X| = |Y |.
We came up with a set P which was the set of all paths down an infinite binary tree.
Here are two sample paths:
Problem 1: Two Es
Show that the following set H is countable:
(Hint: This is an example of a problem that asks you to pair up two sets. Its asking you to
pair up E and H.)
Problem 3: Infinite Gs
Here is a set which Ill call J:
Its an infinite number of Gs, all lined up next to each other. Is this set countable?
Problem 4: Coordinates
We can think of E as being an array of dots lined up on an infinite ray.
In this way, weve placed E inside the real numbers. So we can think of E as being all
points with positive integer coordinates. Im also going to call this set G1 , and I write it
mathematically like this:
E = G1 = {x|x is a positive integer}.
Spend a minute making sure that you understand this notation! You can read it as E
equals G1 equals the set of all x where x is a positive integer.
The set G that we described before is really points in a plane (x, y) where both x and y
are positive integers:
4. Is G countable?
G = {(x1 , x2 , x3 , . . .)|xi is a positive integer for each i}
Recall that the data that goes into making a path is just an infinite set of instructions to
move left or right: a path is an infinite string of Ls and Rs. So the path in the picture is
just LLRLR....
1. Suppose that we look instead at a trinary tree a tree that branches three times at
each level. So that looks like
Let P3 be the set of all infinite paths descending the trinary tree. Show that |P2 | = |P3 |.
6
2. Now suppose we look at a tree that branches infinitely many times at each level. That
is, at each stage we can go down branch 1, branch 2, branch 3, etc., as follows:
Study this new tree. What does an infinite path down this tree correspond to? Is the
set of infinite paths down this tree, P , of the same cardinality as P2 , or different?
So what if, instead of getting an L or R for each point of E, we got an L or R for each
point of P2 ? Instead of sequences of Ls and Rs, a point of this set, which Ill write
2P2 , is a choice of L or R for each infinite path down the binary tree (i.e. for each point
of P2 ). How does this compare to the other infinite sets we have so far?
Problem 7: Computers
Lets use cardinality to prove something about what computers can and cant do. (!)
To simplify the problem, we will consider only questions you can pose to a computer
that have a yes/no answer.
We will assume that all inputs to the computer are encoded in binary. So an input
looks like 01101111, for example: all 0s and 1s.
Definition 1. A decision problem is a rule for assigning YES or NO to every finite binary
string. A computer program solves that problem if it gives the correct answer to every binary
string you give as input.
Here are some examples:
One example of a decision problem is PERFECT-SQUARES. In this problem, we give
a computer a number and it should say YES if the number is a perfect square, and NO
if its not. For example, the computer should respond YES to 10, because this is 4 in
base 10, which is a perfect square. It should also respond YES to 100011101000010110,
because this is 145942 in base 10, which is 3822 . On the other hand, it should respond
NO to 11, which is 5 in base 10 and so not a perfect square. It should also respond
NO to 10111110101101010, which is equal to 97642 in base 10 (whose square root is
somewhere between 312 and 313).
Another example might be to find some way to encode addition into binary. So we
might input 1 + 1 = 2 (in binary) and the computer should respond YES, because this
is a correct addition. On the other hand, it would respond NO to 3 + 5 = 9.
There are lots of decision problems, some of which arent easy to write in simple
mathematics.
Were ready to analyze what computers can and cannot do.
1. What is the cardinality of the set of all possible inputs, i.e. the set of all finite binary
strings?
2. What is the cardinality of all possible problems that is, all the ways of assigning YES
or NO to every finite binary string?
8
3. What is the cardinality of all possible computer programs? (Assume that computer
programs can be of any finite length, and that they are composed of characters from a
fixed alphabet you can even assume every computer program is written in 0s and
1s, since thats how a computer stores it.)
Thus, is it possible to write a computer program for every possible problem, or are there
problems no computer can ever solve?1
It turns out that not just an we prove there are problems where you cant write a computer program
to solve them, but we can give examples of these problems. One particularly famous example is the Halting
Problem, which states that you can never have a perfect debugger. Essentially, it says that you cant write
a computer program that tells you if other programs you input run forever or not.