-1

Is there a formula to write a number $N$ as the sum of $M$ natural numbers (not necessarily distinct), where order DOES NOT matter? I know that you can use ${N-1}\choose{M-1}$ to find the number of ways counting order, but I'm looking for a formula that defines, for example, $1+1+2$ and $1+2+1$ as the same.

  1. How many ways are there to write 13 as the sum of 4 natural numbers if order does not matter?

  2. How many ways are there to write 8 as the sum of 4 natural numbers if order does not matter?

  3. How many ways are there to write 11 as the sum of 5 natural numbers if order does not matter?

Thanks!

AsmiK
  • 11
  • You can model this by ordering the numbers and then looking at the differences between consecutive ones. You are looking to count the number of solutions of $x_1+x_2+...+x_M=N$ such that $0\leq x_1\leq x_2\leq...\leq x_M$. Define $y_1=x_1$, $y_2=x_2-x_1$, $y_3=x_3-x_2$,...,$y_{M}=x_M-x_{M-1}$. Then, you want to count the number of solutions of $y_1+...+y_M=X_m\leq N$, with $y_i\geq0$. –  Apr 03 '18 at 19:04
  • I'm not quite sure of what you mean by "differences between consecutive ones". Could you please give me an example? Thanks. – AsmiK Apr 03 '18 at 19:08
  • Added the explanation. Essentially, it is transforming the new problem to a version of the old one. –  Apr 03 '18 at 19:08
  • You are asking about the much-studied subject of partitions. You will find lots of information on that web page. – Rob Arthan Apr 03 '18 at 19:12
  • Ok, I see what you're saying noodlesGroup. Thanks for the link, Rob Arthan. – AsmiK Apr 03 '18 at 19:14
  • @noodlesGroup: I think you may be getting this mixed up with the theory of ordered partitions or compositions where an approach like yours gives a correspondence between ordered partitions of $N$ and subsets of ${1, \ldots, N-1}$ leading to an explicit formula. – Rob Arthan Apr 03 '18 at 19:26
  • @RobArthan That is what the OP is asking. The partitions have a fixed size $M$, and since the order doesn't matter, they can be parts can be considered non-decreasing in size. The solution does have an explicit formula similar to the one that they quoted in the question. –  Apr 03 '18 at 19:33
  • @noodelsGroup: but the transformations you have suggested don't help with unordered partitions. There is no known explicit formula. If you know one, you should state it and become famous. – Rob Arthan Apr 03 '18 at 20:19
  • @RobArthan There is a difference between not being a fixed sum of hypergeometric terms and not a formula. In between you even have a sum of hypergeometric terms, where the size of the summation can depend on the input. –  Apr 03 '18 at 20:33
  • So please give an answer detailing the approach you have in mind. The OP is clearly a beginner in this subject area and will not be able to recreate anything involving the theory of hypergeometric series just from a few hints. – Rob Arthan Apr 03 '18 at 20:35

1 Answers1

3

This is a question about partitions. One is asking how many partitions of $N$ are there into exactly $M$ parts. Call this $P(N,M)$ One answer is that there is a generating function $$\sum_{N=M}^\infty P(N,M)x^N=\frac{x^M}{(1-x)(1-x^2)\cdots(1-x^M)}$$ for each $M$.

Angina Seng
  • 161,540