We take all of the positive integers that are greater than 100 and color them black or white. Prove that no matter what the coloring is, we can always find two distinct numbers a and b so that a, b, and a+b are all the same color.
-
$\begingroup$ You don't get to choose the colouring so that is also a very wrong answer. The point is that you should not need to, but why? @humn $\endgroup$– NijCommented Feb 19, 2023 at 2:59
-
1$\begingroup$ Did you mean to say "Prove that whatever coloring is chosen, we can always find two distinct numbers a and b so that a, b, and a+b are all the same color."? Because the problem as written is trivial. $\endgroup$– A. I. BreveleriCommented Feb 21, 2023 at 7:38
-
$\begingroup$ @A.I.Breveleri, yes. And the way it is written, it actually isn't trivial. $\endgroup$– NielIGuessCommented Feb 22, 2023 at 23:12
5 Answers
First I'll ignore the requirement that the numbers are greater than 100, and show that the positive integers cannot be coloured black/white without creating a monochrome additive triplet.
Without loss of generality, we can assume the number 1 is white.
The numbers (5,10,15) cannot all be black, so at least one of them is white. Let k be one such white number.
Clearly k-1 and k+1 must be black to avoid white triplets (1,k-1,k) and (1,k,k+1).
Number 2 must be white to avoid black triplet (2,k-1,k+1).
Number k+2 must be black to avoid white triplet (2,k,k+2).
Number 3 must be black to avoid white triplet (1,2,3).
We now have black triplet (3,k-1,k+2).
Note that 3 and k-1 are distinct because k is at least 5.
To accommodate the requirement that the numbers are greater than 100
the same argument can be used but with everything multiplied by 101.
We'll prove the assertion that there are distinct a,b,a+b of the same colour.
Obvious Fact: If there is a threshold above which all numbers are the same colour the assertion is true.
So let's instead make the
Assumption: There are arbitrarily large numbers of either colour.
Using the assumption find k<l<m of colour x such that l>2k and m>2l. Then if any of the (distinct) numbers l-k,m-l,m-k is colour x we are done. Otherwise they are all colour y and we are done, also.
Proof by contradiction: assume there are no such pairs.
We can easily see that there must be infinitely many numbers of both colors, e.g. if there were only finitely many black numbers then any two numbers greater than the largest black number would both be white, as well as their sum.
Let a, b, c be three distinct numbers of the same color. Without loss of generality, let them be black. We'll add/subtract pairs of numbers of the same color to get new numbers that must be the opposite color to satisfy our assumption.
a + b : white
b + c : white
(a + b) + (b + c) = a + 2b + c : black
(a + 2b + c) - c = a + 2b : white
(a + 2b + c) - a = 2b + c : white
(a + 2b) + (2b + c) = a + 4b + c : black
(a + 4b + c) - (a + 2b + c) = 2b : white
Since there's nothing special about b (we can always find two other numbers of the same color as any given number) this means that for any number x, 2x must be the opposite color. So 4b must also be black.
4b - b = 3b : white
But now we have a contradiction: b and 4b are both black, while 2b and 3b are both white. So b + 4b = 2b + 3b = 5b must be both black and white, which is impossible.
-
$\begingroup$ The statement assumed for contradiction only applies to the sum of distinct numbers, so I think care must be taken to show distinctness every time it's applied. For instance, the conclusion that a + 2b is white can only be made if it's not equal to c. $\endgroup$– noedneCommented Aug 24 at 21:04
Yet another proof.
Choose x and y such that x > 100 is black, x+1 is white, y > x+1 is white and y+1 is black.
If you can't find such x and y, that means the numbers stop alternating betwen black and white. All number larger than some n are the same color and you can chose any a and b > n.
If you find such x and y, consider x+y+1. If it is black, then a = x, b = y+1, a+b = x+y+1 are all black. If it is white then a = x+1, b = y, a+b = x+y+1 are all white.
A different proof:
It's clear that neither color can disappear after a given number, so there'll always be both numbers. With the $r$ numbers being red and $b$ numbers blue:
$r_1 + r_2 = b_1$
$r_1 - r_2 = b_2$
$b_1 - b_2 = 2r_2 = r_3$
Meaning multiplying a positive integer by $2$ doesn't change the color.
$3r = 2r+r$
Multiplication by $3$ changes the color. So if we multiply a number by $9=3\times 3 =2^3+1$, it leads to a contradiction.
-
1$\begingroup$ Why should the vertices of such a rectangle (containg 0, say) necessarily be of the form $x,y,x+y$? $\endgroup$ Commented Sep 20 at 8:55
-
$\begingroup$ Each row is a sequence of consecutive integer, and the next row continues from there, so for a rectangle, the difference between two vertices on the same row or column is equal to the one between the remaining two. If the top left vertex equals $0$, the rest will be $x, y, x+y$. $\endgroup$– NautilusCommented Sep 20 at 9:16
-
$\begingroup$ It doesn't matter where the $0$ is as long as it's at the top left corner of a rectangle., because $0$ or negative integers don't change the $x, y, x+y$ relationship. $\endgroup$– NautilusCommented Sep 20 at 9:23
-
$\begingroup$ Ah, I thought you meant to colour the whole grid. But then, how do you decide on an adquate width of the rectangle? As stated, it now sounds like this should work for any width, but that is clearly false (for just length 2 rows, this would amount to always being able to find an integer $x$ such that $1,x$ and $1+x$ are always the same colour, which is not necessarily possible). $\endgroup$ Commented Sep 20 at 9:42
-
$\begingroup$ The whole grid is meant to be colored though. The $0$ is just t the beginning or somewhere in the middle (that can correspond to the top left corner of any rectangle formed in it), and everything that comes before (if any) is a negative integer. $\endgroup$– NautilusCommented Sep 20 at 9:49