Puzzling Partitions

Let D ( n ) D(n) be the number of ways of adding distinct positive integers to sum to n n .

Let O ( n ) O(n) be the number of ways of adding odd positive integers to sum to n n .

Let D = D ( 2016 ) D = D(2016) and O = O ( 2016 ) O = O(2016) .

Which is bigger?


Examples

For example D ( 5 ) = 3 D(5) = 3 since

5 = 5 = 2 + 3 = 1 + 4 \begin{aligned} 5 &= 5 \\ &= 2+ 3 \\ &= 1 + 4 \end{aligned}

and O ( 4 ) = 2 O(4) = 2 since

4 = 1 + 1 + 1 + 1 = 1 + 3 \begin{aligned} 4 &= 1 + 1 + 1 + 1 \\ &= 1 + 3 \end{aligned}

are the only valid solutions.
D D is greater than O O O O is greater than D D They're the same size!

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

It turns out that D ( n ) = O ( n ) D(n) = O(n) for all n N n \in \mathbb{N} !

I saw a wonderful proof of this using generating functions that I'll re-interpret here.


Let's prove this surprising result using generating functions!

Let

o ( x ) = o 1 + o 2 x 2 + o 3 x 3 + . . . + o n x n + . . . o(x) = o_1 + o_2x^2 + o_3x^3 + ... + o_nx^n+ ...

where we define o n o_n to be O ( n ) O(n) .

Similarly we let

d ( x ) = d 1 + d 2 x 2 + d 3 x 3 + . . . + d n x n + . . . d(x) = d_1 + d_2x^2 + d_3x^3 + ... + d_nx^n+ ...

where we define d n d_n to be D ( n ) D(n) .

We'll reinterpret these two power series.


We must have

o ( x ) = ( 1 + x + x 2 + x 3 + x 4 + . . . ) ( 1 + x 3 + x 6 + x 9 + x 12 + . . . ) ( 1 + x 5 + x 10 + x 15 . . . ) . . . o(x) = (1+x+x^2 + x^3 + x^4 +...)(1+x^3+x^6+x^9+x^{12}+...)(1 + x^5 + x^{10} + x^{15}...) ...

since we can create the bijection between expanding out these brackets term by term and by choosing how many of each odd integer we choose to sum.

To see this we can think of each term in the the first bracket ( 1 + x + x 2 + x 3 + x 4 + . . . ) (1+x+x^2 + x^3 + x^4 +...) as representing how many 1 1 s we put into a given summation and similarly ( 1 + x 3 + x 6 + x 9 + x 12 + . . . ) (1+x^3+x^6+x^9+x^{12}+...) as how many 3 3 s we put into a given summation. For example the sum 5 = 1 + 1 + 3 5 = 1 + 1 + 3 corresponds to picking x 2 x^2 out of the first bracket and x 3 x^3 out of the second.

Since the exponents will sum when we multiply the terms in the expansion we see that x n x^n must occur exactly the same amount of ways we can sum n n using odd numbers!


Now let's look at d ( x ) d(x) in another way!

d ( x ) = ( 1 + x ) ( 1 + x 2 ) ( 1 + x 3 ) . . . ( 1 + x n ) . . . d(x) = (1+x)(1+x^2)(1+x^3)...(1+x^n)...

I won't fill in the details here of why this is true but hopefully with a little thought of how we re-interpreted o ( x ) o(x) we can see why this is true.


But now we see that

\begin{aligned} d(x) &= (1+x)(1+x^2)(1+x^3)...(1+x^n)...\\ &\text{but since \$x^{2m}-1 = (x^m-1)(x^m+1) \implies(x^m+1) = \frac{x^{2m} - 1}{x^m-1}\$ we have}\\ &= \frac{x^2-1}{x-1} \frac{x^4-1}{x^2-1} \frac{x^6-1}{x^3-1} ...\\ &\text{cancelling out even powered terms gives us }\\ &= \frac{\color{#D61F06}{x^2-1}}{x-1} \frac{\color{#3D99F6}{x^4-1}}{\color{#D61F06}{x^2-1}} \frac{\color{#20A900}{x^6-1}}{x^3-1}\frac{\color{#69047E}{x^8-1} }{\color{#3D99F6}{x^4-1}}...\\ &=\frac{1}{1-x} \frac{1}{1-x^3} \frac{1}{1-x^5} ... \\ &= o(x) \end{aligned}

So they are equal and hence their co-coefficients are respectively equal.

So O ( n ) = D ( n ) O(n) = D(n) for all n N n \in \mathbb{N} therefore O ( 2016 ) = D ( 2016 ) O(2016) = D(2016) .

Can any one think of a nice way how to explicitly show the bijection between each partition of

D ( n ) D(n) with O ( n ) O(n) ?

For example if there a natural way to pair each of the sums in D ( 5 ) D(5) and O ( 5 ) O(5) ,

5 = 1 + 4 = 2 + 3 = 5 \begin{aligned} 5 &= 1 + 4 \\ &=2+3 \\ &= 5 \end{aligned}

5 = 1 + 1 + 1 + 1 + 1 = 1 + 1 + 3 = 5 \begin{aligned} 5 &= 1 + 1 + 1 + 1 + 1\\ &=1 + 1 +3 \\ &= 5 \end{aligned}

Can anyone come up with an algorithm of how to pair these together for all n n ?

Let me know :)

Roberto Nicolaides - 5 years, 2 months ago

Log in to reply

Take any partition a \mathbf{a} of N N into distinct numbers: N = a 1 + a 2 + + a n , N \; = \; a_1 + a_2 + \cdots + a_n \;, and write each a j a_j as a power of 2 2 times an odd number: a j = 2 m j b j , 1 j n . a_j \; = \; 2^{m_j} b_j \;, \qquad 1 \le j \le n \;. Let S S be the collection of all odd numbers that appear in the above: S = { b j 1 j n } . S \; = \; \big\{ b_j \, \big| \, 1 \le j \le n \big\} \;. For each m S m \in S , add together all the coefficients of m m that can be found: k m = j : b j = m 2 m j , k_m \; = \; \sum_{j \,:\, b_j = m} 2^{m_j} \;, then the partition f ( a ) f(\mathbf{a}) of N N consisting of k m k_m copies of m m for each m S m \in S is an odd partition of N N .

Since the partition a \mathbf{a} consists of distinct numbers, no two of the powers of 2 2 which are added together to calculate k m k_m are equal (otherwise there would be distinct i i and j j with b i = b j b_i = b_j and m i = m j m_i = m_j , and hence a i = a j a_i = a_j ). This means that the constituent m j m_j which combine to form each k m k_m can be retrieved from each number k m k_m (by expressing k m k_m in binary). Hence the function f f is an injective map from the distinct partitions of N N to the odd partititions of N N .

It is a simple exercise to see that f f is also surjective, and hence is bijective.

Mark Hennings - 5 years, 2 months ago

Log in to reply

Mark, this is great :). I'm not 100% sure on the notation you used but I think this is the gist of what you're saying (please correct me if I've misunderstood!):

For each a j = 2 m j b j a_j = 2^{m_j}b_j in the distinct partition turn it into m j m_j copies of b j b_j and this will make a partition of just odd partitions!

Explicitly, going from the distinct partitions to the odd partitions of 5 5 we get the following correspondence 5 5 1 + 4 1 + 1 + 1 + 1 + 1 2 + 3 1 + 1 + 3 \begin{aligned} \color{#333333}{5} &\rightarrow \color{#333333}{5}\\ \color{#3D99F6}{1}+\color{#D61F06}{4} &\rightarrow \color{#3D99F6}{1} + \color{#D61F06}{1+1+1+1}\\ \color{#69047E}{2}+\color{#EC7300}{3} &\rightarrow \color{#69047E}{1+1} + \color{#EC7300}{3} \end{aligned} To show that its a bijection we could just show what the inverse process is (this is usually ver illuminating!). It's not immediately obvious to me what this is explicitly, so I'll have a little think and post any thoughts here later.

Roberto Nicolaides - 5 years, 2 months ago

Log in to reply

@Roberto Nicolaides ... 2 m j 2^{m_j} copies of b j b_j ... and you have it.

To show surjectivity, suppose your odd partition has 13 13 copies of 7 7 . Write 13 = 8 + 4 + 1 13 = 8+4+1 in binary. These copies of 7 7 contribute 8 × 7 = 56 8\times7=56 , 4 × 7 = 28 4\times7=28 and 1 × 7 = 7 1\times7=7 to the distinct partition that maps to it...

Mark Hennings - 5 years, 2 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...