4
$\begingroup$

I have looked through the online literature and there seems to be conflicting answers to this question. Consider the finite sum

$$\sum_{i=0}^{\lfloor n/2 \rfloor} {n-1\choose i}$$

Is this expression considered closed-form?

These two links say that such a finite sum is not considered closed-form:

  1. https://opendsa-server.cs.vt.edu/ODSA/Books/CS3/html/Summations.html#:~:text=This%20is%20known%20as%20a%20closed%2Dform%20solution
  2. http://www3.govst.edu/wrudloff/CPSC438/CPSC438/CH05/Chapter5/Section.5.2.pdf

while this link says that such a finite sum is considered closed-form:

  1. https://en.wikipedia.org/wiki/Closed-form_expression#:~:text=Yes-,Finite%20sum,Yes,-Finite%20product

As far as I know, a closed-form expression has a finite number of standard operations or known functions. Based on this, my hunch is that the sum above is closed-form.

$\endgroup$
5
  • 2
    $\begingroup$ Since you cannot bound the number of terms in that sum when considered over all $n$, some people might argue it is too general to be a finite sum. $\endgroup$
    – Henry
    Commented Apr 5, 2022 at 23:47
  • $\begingroup$ Related: stats.stackexchange.com/questions/70848/… $\endgroup$
    – Dan
    Commented Apr 5, 2022 at 23:49
  • $\begingroup$ @Henry that's an interesting point. is it formally accepted that if you cannot bound the number of terms in an expression, then it isn't closed-form? you say "some people might argue" so it seems like the question of whether or not the above sum is closed-form can't really be answered definitively $\endgroup$
    – SNN
    Commented Apr 5, 2022 at 23:59
  • $\begingroup$ You ask what is the formal definition of closed-form. Why does there have to be a single definition rather than what a particular user defines it to be? $\endgroup$
    – Henry
    Commented Apr 6, 2022 at 0:10
  • $\begingroup$ @Henry It's not that there needs to be a single definition, but math is largely built on unambiguous definitions which gives it its black-and-white characteristic. If such a universal definition exists in the math literature, I would like to follow it. If there truly isn't such a formal definition, then I can live with the ambiguity, but I first need to know that there isn't a universally accepted definition of closed-form. So I guess my question really is, is there a clear-cut definitive answer to the question I have posed above? $\endgroup$
    – SNN
    Commented Apr 6, 2022 at 0:17

1 Answer 1

4
$\begingroup$

Hint: The notion of closed form in mathematics is context dependent. Finite sums of type \begin{align*} \sum_{i=0}^{\lfloor n/2 \rfloor} {n-1\choose i} \end{align*} are usually not considered to be in closed form, since there is bound index variable $i$.

We find in Chapter I: What Is Enumerative Combinatorics? in the classic Enumerative Combinatorics, Vol. I by R. P. Stanley:

The basic problem of enumerative combinatorics is that of counting the number of elements of a finite set. Usually we are given an infinite collection of finite sets $S_i$ where $i$ ranges over some index set $I$ (such as the nonnegative integers $\mathbb{N}$), and we wish to count the number $f(i)$ of elements in each $S_i$ simultaneously. Immediate philosophical difficulties arise. What does it mean to count the number of elements of $S_i$? There is no definite answer to this question. Only through experience does one develop an idea of what is meant by a determination of a counting function $f(i)$. The counting function $f(i)$ can be given in several standard ways:

  • The most satisfactory form of $f(i)$ is a complete explicit closed formula involving only well-known functions, and free from summation symbols.

Note: The crucial aspect is the term bound variable. As long as there are bound variables with limited scope given by $\Sigma$ or $\prod$ symbols, the expression is not considered to be in closed form.

$\endgroup$
2
  • $\begingroup$ This answer was helpful, thank you! I think the most convincing way for me to see that it isn't closed-form is to notice that the number of terms grows to infinity as n goes to infinity, whereas in a truly closed-form algebraic expression, n may grow to infinity but the number of terms is more or less constant, i.e. the set of terms is always finite. $\endgroup$
    – SNN
    Commented Apr 7, 2022 at 21:39
  • 1
    $\begingroup$ @SNN: You're welcome. I've also added a note which might be helpful. $\endgroup$ Commented Apr 8, 2022 at 9:39

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .