# In written mathematics, is $f(x)$a function or a number?

I often see notation/wording like "let $f(x)$ be a continuous function" or "let $f(x) \in C^0(\mathbb{R})$".

I would say that $\sin$ and $x \mapsto \sin(x)$ are functions, while $\sin(x)$ is a real number.

Are there any correct or best practices in this regard?

-
Mathematics doesn't work that way. f(x) is anything the writer wants. –  anon Oct 15 '10 at 15:31
@muad: I find many of your comments not really helpful. Actually, sometimes -like this time-, I think you could avoid commenting at all: how does your remark relate to the question, which is clear and everything? In every subject anything can mean anything the writer wants it to mean... –  Andy Oct 15 '10 at 16:27
@Andy: What muad is trying to say is that it depends on the context and that is where you should be looking for answers. (At least that is what I interpreted it as) –  Aryabhata Oct 15 '10 at 16:33
@muad: (and @Andy: ) Please don't continue this conversation here, it takes away from the question and the discussions PERTAINING to the question. –  jericson Oct 15 '10 at 17:59
See also this question on MO: mathoverflow.net/questions/40709 –  Matt E Oct 16 '10 at 4:28

Saying that $f$ or $x\mapsto f(x)$ is the function rather than $f(x)$ is the more common modern perspective in higher mathematics. I have even seen a precalculus book that stresses that $f$ is a function, whereas $f(x)$ is a value of the function. The notation $f(x)$ to denote a function remains because it is often more convenient, and it is especially prominent from high school mathematics up to calculus because it is psychologically easier to become accustomed to. This may lead to ambiguity, because in any case $f(x)$ sometimes does refer to a value of the function, but the strictest logical correctness of notation isn't for everyone, or for all situations.

I find it off-putting when I read mathematics that was written many decades ago and come across something like, "Let $\varphi(x)$ be a bounded linear functional on $X$....". I think to myself, "No, $\varphi$ is the functional!" Then I take a deep breath and relax, and find that this notation did not hinder the authors from doing and writing great mathematics.

Ahlfors's complex analysis text has a footnote on page 21 of the 2nd edition, 1966:

Modern students are well aware that $f$ stands for the function and $f(z)$ for a value of the function. However, analysts are traditionally minded and continue to speak of "the function $f(z)$."

-
One place where the abuse of notation in f vs f(x) caused me confusion for so many years is in derivation of Euler-Lagrange equations in calculus of variations. Like all beginners, I couldnt understand how both y and y' are treated as independent variables in L(y,y',t). Turned out, its better written as L(y(t), y'(t), t). y and y' are certainly related, but their values at t arent. When defining a functional, again, it pays to write carefully as S[y] rather than S[y(t)]. –  me10240 Jul 24 at 20:44

You are right. Properly speaking, the function is called just $f$, and its value at the point $x$ is denoted by $f(x)$. Speaking of "the function $f(x)$" is what mathematicians call "abuse of notation", but it is of course often very practical.

(By the way, one usually writes $x \mapsto \sin x$, not $x \to \sin x$; it's "\mapsto" in TeX.)

-
Thanks for your TeX remark - it has now been corrected. –  janmr Oct 15 '10 at 15:28

You are quite right: $f$ by itself should denote a function, $f(x)$ by itself should denote the element in the codomain of $f$ (in this case, the real number) that results when you evaluate $f$ on the element $x$ of its domain, where $x$ should previously have been defined.

However, this rule is honored as much in the breach as the observance: there are many situations where it is convenient to break it. When defining a function by a formula, it's hard to avoid a dummy variable, and so one likes to say "let $f(x)=e^{-x}$" instead of the more correct but awkward "let $f$ be the function $x \mapsto e^{-x}$". In particular, with functions that have multi-letter symbols like $\sin$, I find that people generally prefer to avoid writing them without an argument like $\sin x$. One does not like to talk about $\sin$ as a function in itself, so instead of writing something like $\sin'' = -\sin$, one would rather say "if $f(x)=\sin x$, then $f'' = -f$".

An alternative to a dummy variable that's sometimes used is a dot: $\cdot$. People sometimes write "let $f=g(\cdot + 5)$" to avoid the less correct "let $f(x)=g(x+5)$".

When working with functions of several variables, using dummy variables often helps keep track of which variable is which. One often writes something like: "let $u(x,t)$ be a solution of the heat equation $\frac{\partial u}{\partial t} - \frac{\partial^2 u}{\partial x^2} = 0$". Of course, one is really talking about the function $u$ and not any particular real number of the form $u(x,t)$, but it would be much more awkward to write otherwise. It also reminds you that the first argument of $u$ should be interepreted as space and the second one as time.

In short: mathematicians are not compilers. Written mathematics has some syntax rules, but they are not quite hard-and-fast, and need not be followed at the expense of clarity.

-
I like your point that the names of dummy variables are helpful. But I don't see what is incorrect about saying "let $f(x) = e^{-x}$". It may be improper to speak of $f(x)$ as being a function, but here one is describing the function $f$ by defining its value $f(x)$ at arbitrary $x$. –  Rahul Oct 15 '10 at 17:20
To elaborate, I feel that it is merely shorthand for "let [$f$ be the function such that] $f(x) = e^{-x}$ [for all $x$ in the appropriate domain]". –  Rahul Oct 15 '10 at 17:21
@Rahul: Agreed, defining $f$ that way is analogous to how let f x = exp (-x) is just syntactic sugar for let f = fun x -> exp (-x) in ML, for instance. In particular, it is perfectly consistent with the meanings of f and f x and does not break any syntax rules. –  Frank Jan 13 '13 at 14:02

The wording "let $f(x) \in C^0(\mathbb{R})$..." is really sloppy. Of course it doesn't lead to confusion, but there is something to be said for correctness. Ideally whenever you see $f(x)$ it should be because we care about what $f$'s values are rather than $f$ in and of itself as say a member of some space of functions.

Oh, and your example of "$x \to sin(x)$" is IMO better written as "$\mathbb{R} \owns x \mapsto \sin x \in \mathbb{R}$. Obviously not so useful for this particular function, but very smart for things like "$V \times V \owns (x,y) \mapsto \langle x,y \rangle \in \mathbb{C}$".

-
+1, Never thought about this: $\mathbb{R} \owns x \mapsto \sin x \in \mathbb{R}$ –  Kasper Feb 23 '13 at 15:34

Strictly speaking, I would say that $f(x)$ is the value of the function $f$ evaluated at $x$. However, $\frac{x^3-2}{x+1}$ might be used as a function; it is probably because $x\mapsto\frac{x^3-2}{x+1}$ is harder to write and takes up more space.

-

It all depends on what you have said $x$ is.

If $x$ is previously defined as a number (e.g. $x = 3$) then $f(x)$ is also a number, since you can evaluate $f$ at a certain point $x$ to get the number. (A more common way to notate that would be to say $x_1$ instead of $x$...)

If, on the other hand, $x$ is defined as part of some set (e.g. $x\in[0,1]$) then $f$ is a function defined for input values in that interval. $\sin x$ is defined as a function for $-\infty < x < \infty$, but as a number for $x=\pi/2$.

-