Talk:Function (mathematics)/Archive 5

Archive 1

←

Archive 3

Archive 4

Archive 5

Archive 6

Archive 7

→

Archive 10

Types of functions section

I find this section redundant, high-school level mathematics at best, and without references. —Preceding unsigned comment added by VasileGaburici (talk • contribs) 13:06, 4 September 2008 (UTC)

Most of the mathematic topics about functions, equation, matrix, logarithmic and just about every single math article, so far majority of the article seems to have only include the technical details of the presence of article and they rarely talk about how it is being used in other sciences. This in terms unconditionally make people who aren't mathematician oblivious to the article, just because they don't have the knowledge. Not every single math article must to be written to the point that only the technical detail can exist and I think you guys should consider that when writing article, making the paragraph easy to understand instead of focusing on the technical details, which some call it as a geek. If you can't let majority of the society understand the article, then this article isn't even a topic in Wikipedia. Wikipedia isn't a technical documentation datasheet about the properities of every type of math there is, though I don't oppose it the presence. Just make some effort. --Ramu50 (talk) 02:07, 5 September 2008 (UTC)

(I deleted the section again while you were writing this, not seeing any explanation.) It does look like the article could explain better what functions are actually used for. However, I think the new section does not do a good job at this. Sorry, but I don't see how you can complain that the article is too technical and then add a section which is so hard to understand. It talks about a lot of things that are not explained in the article. Functions are very basic and the examples also have to be very basic. -- Jitse Niesen (talk) 02:31, 5 September 2008 (UTC)

I am going to revert it nevertheless. First you guys didn't even ask yourself, is Wikipedia a database for Technical Documentation Datasheet? The answers is definately not, therefore new idea can be added and removed. In this case, it happen to be making the article more easily readable to people who aren't mathematician.

I am positive to any criticism on my works, however, just because my works suck doesn't mean you delete my work with a simple complain and be gone with it forever. You don't even make an effort in trying to add a new section, and yet you complain the quality of my work when you could of improve it, if you agree with my plans. If you don't, as previously mention, do it your own and make other look up to you. So stop being ignorant and being immature, the fact that you made claim to be a so-called well known mathematician. So far, as of know, you seem not show any of the skills you said at all. --Ramu50 (talk) 18:31, 5 September 2008 (UTC)

I agree that the deleted section was unlikely to be helpful to anyone. It was a list of terms presented with no explanation at all. I do not see how it makes the article more easily readable to non-mathematicians. But I am afraid I don't understand most of your points. (What is "Technical Documentation Datasheet"? What does it have to do with your deleted section?) Perhaps we might come to some agreement if you could explain what goal you were trying to accomplish with your addition, and explain how you believe your addition would accomplish that goal. -- Dominus (talk) 18:36, 5 September 2008 (UTC)

I also find the material on GPGPU and CUDA to be too tangential to the topic of this article to be included. — Carl (CBM · talk) 21:18, 8 September 2008 (UTC)

I agree with Carl. If this section isn't gone VERY soon, I will remove it. Bill Wvbailey (talk) 23:23, 8 September 2008 (UTC)

You guys are being self-contradictory, first you said it is too highschool and when I made the contributions of a university level of work you guys complained it is too technical. What the hell are you trying to do, taking over the article so only mathematician can edit? By the way the technologies of CUDA and Physics Processing Units have been explained in many of the top reviews, such as CNET, ZDNet, Softpedia, TigerDirectBlog (YouTube)...etc. So why do I need to explain something that is already understand by the a greater half of the society. --Ramu50 (talk) 21:35, 8 September 2008 (UTC)

My goal is to have an article that covers the significant aspects of mathematical functions. I don't believe that GPGPU or CUDA are important to the mathematical theory, and I don't think they are particularly notable examples of functions. There are lots and lots of tangential applications of functions; why are those two more important? — Carl (CBM · talk) 21:41, 8 September 2008 (UTC)

I apologize if I came down too harsh on your initial contribution; I was a bit pressed for time then (the robot had to sign me). My point is that listing a bunch of functions on this page would make the article unnecessarily long and won't improve it any way I can think of. VasileGaburici (talk) 21:46, 8 September 2008 (UTC)

This article already contains an elementary introduction with plenty of examples, as well as rigorous definitions. Listing "more of the same" examples, which is what you did with you first edit, won't make the article anymore insightful. Your second attempt added some examples that far too application specific (GPGPU macros etc.) Believe it or not, most non-geeks have no clue how a GPGPU works, and this is not the article to touch that subject. So, I dissagree with your statement that those examples are "already understand by the a greater half of the society". VasileGaburici (talk) 22:09, 8 September 2008 (UTC)

The list I include, I intended including the highschool level of function only. I did't want to include any more functions list, because template already does that quite well. Presenting the basis of the function in a structured way I find it quite useful since it allow viewers to see how each subtopic in the field of mathematicians is being branched out. But I don't think you guys are thinking straight. Examples such as function z is a popular discussion, but not easily understood by the greater majority, since highschool don't teach it, 3D modeling doesn't teach it that in depth and many of people still don't know how many ways are there to graph the z function.

I can tell you that most of the math taught in highschool such as sequences, patterns, deviations, probability...etc. Most people don't even know how they are actually used, they just know it is probably use in DNA genetics pattern and statistics.

And putting the popular linear algebra and yet easily understood, since many people already know CUDA is wrong? As far as I know it is equally or more frequently discussed than identitiy function or inverse function. Also the easy understand of the topic which has attracted Softpedia to creating a News Tag on CUDA, which they use it for followups to attract enthusiats gamer (non-professional). By the way I can tell you guys didn't even read the stuff I improved after I reverted. [1]

I did this
Linear Algebra is used in one science and thus how it is use in other sciences (at the same level).

• Function
  • Linear Algebra (explained)

It can be used in

• Graphics
  • XGI Server

• Military

In Summary: I explained 2 different field of science.

You guys did this
Function Z, inverse function and identity explained, somewhat confusing and easily understood.

• Function
  • 3D Function (explained)

Results: Question raised (is it applicaple in reality or in 3D graphics?)

• Function
  • Inverse Function (understood), but useless and walk away.

If a guy who is doing a research on function, which information do you think is more useful. Moreso, if the guy have a job in civil engineering and he wanted to see if certain math topics can assist him in 3D Modelling for construction design, do you think function Z would give him the answer if he doesn't understand it.

I have asked numerous teacher about how reciprocal and inverse function is being used and not a single Math 12 teacher have given me an answer before. One of the math teacher degree I asked had a Ph.D in Linear Algebra. So here is the summary, do you want a dead article or do you want an article that is useful. --Ramu50 (talk) 22:30, 8 September 2008 (UTC)

Ramu, your addition is poorly written, hard to understand, hard to put into context with the rest of the article, and no-one is supporting you. In addition to that, it seems you selectively delete content from your talk pages. None of this suggests to me that your contribution is worth maintaining in the article. Carl.bunderson (talk) 01:00, 9 September 2008 (UTC)

I don't care if you support me or not, the fact is WP:CON policy allow me post and doesn't require me remove it regardless of how many people supported me. None of you have been able to give me a more legitimate reason for me to remove. Every single one of have so far have only noted the this sucks or judging from the quality of my work. First of all, the word I use aren't technical language, if they are I use bracket and denote a similar clarificaion such as interpreter (translator). The choice of the word, I am not require to change, I can choose whatever synonomous word I wish to use, however, it can't be vulgar, abusive, slang...etc. I will revert it infinite times if I have to. For many success in history, many do have oppose until the person is peace, then realization do play in role. If that may be so, then so be it.

But as of my presence I will faithly defend it against all of your immature actions of criticism which give no insights to any effort of seemingly worth being note as to expansion and passion for the function of mathematician which I believe is a disgrace that many of you call or claim yourself to be a mathetician or as others claim a so-called well known hither. --Ramu50 (talk) 02:24, 9 September 2008 (UTC)

Can some mathematicians help me very on this

I wrote a mathematical hypothesis about 4Dimensional objects, do you guys agree? Reply it on the other article talk page if interested. Location of interest --Ramu50 (talk) 19:40, 29 September 2008 (UTC)

Derive while-

h(x)=f(g(x))

Greetings. I hear you like functions, so I have placed a function inside your function so you may derive while you derive.

— Preceding unsigned comment added by 76.77.226.5 (talk) 17:22, 1 December 2008 (UTC)

Partiality and Totality

A partial function is no more a kind of function than near beer is a kind of beer. The modifiers in these two noun phrases (partial and near) are not meant to restrict classes of referents (functions and beers, respectively). Instead, their semantic function is more like that of the prefix pseudo- or the suffix -oid.

It may be quite reasonable for authors of works dealing with partial functions to abuse terminology by adopting the shorthand function in the context of those works and provided that the abuse is either clear from context or explicitly announced to the reader. But it would be confusing practice for the definition in a Wikipedia article on beer to expressly include such other beverages as near beer or root beer or ...

Unless people squawk here in a hurry, I'm going to reshape the definition to clarify that the concept is that of the [total] functions. I'm perfectly comfortable with this article's discussing partial functions as a related concept (in fact, a generalization).

Reactions?—PaulTanenbaum (talk) 06:53, 4 December 2008 (UTC)

If you think of a function as merely a set of ordered pairs such that no two ordered pairs in the set have the same first coordinate, as is quite common in set theory, then the difference between partial functions and total functions essentially disappears. It's only if you begin to attach other information to a function, such as the intended domain, that it becomes possible to ask whether a function is intrinsically a partial or total function. This is entirely parallel to the issue with surjective functions, where it is necessary to specify an intended codomain before you can determine whether the function is surjective.

However, the only mention of partial functions in this entire article is this one sentence: "In some parts of mathematics, including recursion theory and functional analysis, it is convenient to allow values of x with no association (in this case, the term partial function is often used)." The rest of the article makes no mention at all of partial functions, and this is the correct approach in my opinion.

How exactly would you reshape the definition (and which definition would you reshape?) to talk about totality? It looks to me like the present definitions do already require totality. — Carl (CBM · talk) 13:19, 4 December 2008 (UTC)

Actually, I did find a place where I think this language fits:

The terminology total function is sometimes used to indicate that every element of X does appear as the first element of an ordered pair in F; see partial function. In most contexts in mathematics, "function" is used as a synonym for "total function".

I think that might clarify something for the reader. I'm concerned about completely rewriting the definitions sections though, and since the topic of partial functions is already such a small issue i mathematics as a whole. — Carl (CBM · talk) 13:33, 4 December 2008 (UTC)

Yes, the crux of the issue is akin to whether a function is (1) merely a set of ordered pairs, no two of which share a first element, or is (2) that together with some set containing all of the second elements (i.e. its codomain), or what. While you are right that partiality does not rate many explicit mentions in the current definition, it does show up in the distinction between the source set and the domain.—PaulTanenbaum (talk) 19:43, 5 December 2008 (UTC)

Range vs. Codomain vs.

The article makes use the term range to be the same as the image. This disagrees with some references, and I would like to think of a clean way to point out that the term range can refer to either the codomain or the image, depending on the author. Does anyone have any suggestions before I start editing the article? Thenub314 (talk) 13:24, 5 January 2009 (UTC)

When I was in college, in the 1960s, the usage was "image" and "range". Today, I think the usage is "range" and "codomain". Any rewrite should begin with the modern usage, and only later mention the older usage. It would be interesting to know when and under whose influence the change came about.Rick Norwood (talk) 15:23, 5 January 2009 (UTC)

Alot of modern books use both conventions, in another discussion at the talk page range article I mentioned a few (see here). I don't think either definition has a real grasp on being the "modern usage". That being said we should probably get away with using mostly image and codomain, as I don't know of any ambiguity in these terms. And somewhere we point out that range could be used for either of these terms. Does that sound reasonable? Thenub314 (talk) 16:03, 5 January 2009 (UTC)

Agreeing with this, I've seen both uses of range in relatively new sources; we should point out this confusion but it's better to avoid it ourselves. Another common representation for the image is f(D), where D is whatever the domain is and the function is understood to be applied pointwise. Dcoetzee 07:14, 6 January 2009 (UTC)

Agree. Rick Norwood (talk) 14:36, 6 January 2009 (UTC)

---

We've been over this topic before -- see Archive 3. After doing a "survey of the literature" this is what I determined to be the usage today:

The symbol "Ø" stands for "empty"; [X]=Ø stands for "the place named "X" is empty of content." When restricted to a proper subset of the unrestricted universe of discourse = {Ø, 1, 2, 3, 4, 5} the function has the domain of definition {2, 3, 4} and is "effective at" (i.e. does a good job of) putting an output y ="o" or y="e" into the subset range={o,e}. This "effective" range is a place₁ inside the place₂ inside the place called "the codomain Y". The "computable range" {{o,e},u} (place₂) includes an output y="u" produced when the input(s) is(are) not in the defined domain D(f). Thus, because "u" is not an element of the "effective" range, D(f) is a proper subset of X: f(D) ⊂ X. If the function fails to HALT (for whatever reason) it apparently fails to put anything into the "computable range". But at the start, the counter-machine model "clears out" this place Y; a mathematician would start with a blank sheet of paper, or erase an area to work in. In this sense (due to the clearing, erasing, emptying) the function has put "nothingness" into the codomain Y, i.e. Ø → [Y]. Thus, because Ø is not an element of the "computable range", the "computable range" is a "proper subset" of the "semi-computable range". The drawing shows that this happens when the function is given no input i.e. [X]=Ø or if it is given the input "5". This example also works when symbol "5" represents any of the positive integers.

While this illustration may be based more on computability and set theory (i.e. discrete mathematics) as opposed to analysis it would be nice if all the terminology agreed. Or if it could be clarified. wvbaileyWvbailey (talk) 15:25, 6 January 2009 (UTC)

I strongly disagree that the word "range" should be given a "standard" definition. Traditionally, the word "range" has meant one and only one thing: what the word "codomain" means unambiguously today.

Unfortunately, many high school teachers in the '70s through the '90s -- thanks to errors in textbooks -- began to use the word "range" to mean the image of a function. Ignorant textbook writers and publishers only propagated this error.

Now the word "range" unfortunately means different things to different people -- and the split is about 50-50 among mathematical people.

For this reason, I strongly oppose further use of a word that has, unfortunately landed squarely in the middle of a schism of its meaning into two different things.

Rather, it makes complete sense to emphasize the remaining words that are not currently ambiguous -- "codomain" and "image". These are the words that ought to be defined, so as to avoid propagating more confusion.

As a side note, it is of course appropriate to mention the distinct meanings that have been assigned to the word "range". (I believe this has already been done somewhere in the article.)Daqu (talk) 19:42, 28 June 2009 (UTC)

Terminology for functions and types of functions

The presentation sometimes confuses variations in terminology and variations in kinds of functions. I have changed one paragraph that talked about both by splitting it in two. In that paragraph I also changed the remark about "mapping" -- some authors restrict the meaning of that word compared to "function".

I didn't say anything about "map" but in this paragraph the word "mapping" is linked to the Wikipedia entry for "map".SixWingedSeraph (talk) 18:52, 15 April 2009 (UTC)

Simplifying the Lead Section

As author of my lead text that was deleted, I wish to say that I object to the deletion. It made the concept of 'function' more understandable to readers with no particular training in mathematics. If my contribution had been inaccurate, I wouldn't complain. But it was accurate. David spector (talk) 17:47, 4 May 2009 (UTC)

Re David Spector: There were many minor issues with the text that was removed, but the major ones were that it made the lede too long and that it assumed that the reader was familiar with both electronics (voltage) and Boolean functions, which can hardly be assumed of a naive reader. The lede is not the place for long examples. The second paragraph that was added really did not belong in the lede, and it does not even make sense to me to say "There are many ways to simplify, expand, and define functions". Finally, as Wvbailey alluded, in a long and established article it is necessary to make new text fit smoothly into the old text; one cannot simply add new things to the top while ignoring the previous structure. After looking through what you added and looking at the rest of the article, the main thing that I see lacking is a mention of the black box model. So I proposed a way to remedy that. — Carl (CBM · talk) 23:57, 4 May 2009 (UTC)

Thanks for the feedback, which should have come along with the deletion, not later when I objected. Adding a simple description of the black box model of functions is certainly an improvement. I don't believe that mathematics articles should only be understandable by mathematicians, physics articles by physicists, medical articles by medical researchers or doctors, etc. Rigor can be maintained without lots of jargon by using simple language. There is nothing inherently difficult about mathematics; any particular part of it just depends on specialized definitions, lemmas, notation, etc., all of which can be learned in simple steps. At least that's been my experience having taken many mathematics and physics classes in college and graduate school. I just wish there weren't so many opaque articles here. Maybe highly technical Wikipedia articles could be paired with articles that explain the same concepts using simpler language. Then Wikipedia would be equally useful for someone outside of the field of study as well as inside. David spector (talk) 23:22, 5 May 2009 (UTC)

Note that the effort at this article is exactly to remove unnecessary jargon and try to achieve a high level of accessibility. On the other hand, the lede section is necessarily brief and terse. And I started this very section when I undid your edit... — Carl (CBM · talk) 00:20, 6 May 2009 (UTC)

The article has improved much. I thank each editor who was involved. One small remaining problem in the lead section is the use of the technical "codomain". In some contexts the 'natural' codomain (say, real numbers) is important (say, as a datatype), but in other contexts the 'actual' target set (the image) is what is important (for example, when the actual output of the function is what is relevant to a real-world problem, say the set of results of a perfect hash function for its set of keys).

Since the lead is not the place to explain lots of technical details (such as the distinction between codomain and image), we need a word that is ambiguous, that can represent either the codomain or the image. Wait! That's just what range has always meant! Why not use 'range' in the lead (with a deliberately ambiguous Venn diagram), and explain the difference between codomains and images later (as is currently done)? David spector (talk) 19:40, 5 December 2009 (UTC)

Black box / machine image needed

I undid an edit to the lede, but there was one thing in it that would be worth including in the "introduction" section: a picture of the "black box" or "machine" model of a function. This is such a common idiom for functions that we really should mention it, and I think an image with caption would be nice. But I cannot find a picture like I want (e.g. [2] would work if it were not copyrighted). Could someone who is better at drawing make one? It would be nice if the input were labeled x and the output were labeled f(x). — Carl (CBM · talk) 12:45, 3 May 2009 (UTC)

I'm not sure if that's in keeping the the definition of a function in mathematics. Further discussion is welcome, though. — Arthur Rubin (talk) 14:59, 3 May 2009 (UTC)

My experience is that the "machine" picture is used very extensively (in the U.S. at least) in grade school mathematics texts. You can search for "function machine" on google books to find lots of examples of what I mean. I don't think we should discuss it in prose, but using it as an image and caption in the "intro" section seems like a nice addition to me. — Carl (CBM · talk) 15:12, 3 May 2009 (UTC)

You mean something like this? I didn't understand the horizontal line drawn across the "rule box" and I prefer distinct funnels and the words "input" and "output" because these indicate "things" (e.g. numbers) as opposed to places IN and OUT". A question: can a function not be a machine? Wvbailey (talk) 16:05, 3 May 2009 (UTC) [I don't know how to place them side-by-side.]Wvbailey (talk) 16:18, 3 May 2009 (UTC)

A function can be visualized as a "machine" that takes an input x and returns an output f(x).

More accurate: there's a place called (labelled) "x" and another place called "f(x)"; the student's responsibility is to put a value (a number) in place "x" and then start the gizmo called "rule f". "Rule f" begins by searching for a number at place "x"; finding one that it can accept, gizmo "rule f" [attempts to] crunch it and then place the result at the place called "f(x)" [if it fails to produce a number it goes on crunching ad infinitum]. Wvbailey (talk) 16:27, 3 May 2009 (UTC)

Yes, that's the type of thing I meant. For the caption, I was thinking of something like: "A function can be visualized as a "machine" that takes an input x and returns an output f(x)." I am sure that other people will have suggestions for how to tweak the image to make it perfect. — Carl (CBM · talk) 17:02, 3 May 2009 (UTC)

I've noticed that as an article matures (as this one has been doing) there's so much time invested by so many folks that the work has to become more consensual, with alternatives etc being proposed and thrashed around here first. Carl's proposed substitute seems a decent compromise; a picture would accomplish the same thing as your wording and would dress up the lead too. (I'm a bit concerned, though, that there may be functions that can be "arrived at" by something other than mechanisms; if so then "mechanism" would be sufficient but not strictly necessary.)

A question for Carl: Can we speak in general terms about the nature of "input x" and "output f(x)"? They don't have to be numbers, correct? Bill Wvbailey (talk) 21:50, 4 May 2009 (UTC)

No, they do not need to be numbers. — Carl (CBM · talk) 23:57, 4 May 2009 (UTC)

I think the following drawing of a composite function executing substitution (cf Kleene 1952:78ff) is interesting. It can easily be changed to illustrate other functions:

A composite function can be visualized as two "machines". The first takes its input x and returns output f(x). The second takes its input as f(x) and returns output g(INPUT) = g(f(x)). If the order of functions f and g are reversed the output f(g(x))=16.

I want to show that a function need not return a "number" (it can return the words TRUE or FALSE, for instance), and the intuitive notion that "x" is really a place where function f goes to get its "input", f(x) is a place where the function f outputs its result etc (these places can be a region or regions on a piece of paper, a "register" or registers in a machine, etc.) Thoughts? (It was fun to draw, so even if nobody likes it, no animals will be hurt in this experiment.) Wvbailey (talk) 03:00, 8 May 2009 (UTC)

I think that diagram is too complex, and the "location" term will not be clear to many people (it isn't clear to me, in fact). The "if/then" is a programmer idiom that adds unnecessary complexity to this article, which we should try to make very broadly accessible. The basic function machine is a nice idiom and very common, but we have to keep our examples very minimalistic.

By the way, the article explicitly says, "Such functions need not involve numbers. For example, a function might associate each member of a set of words with its own first letter." This is in the first paragraph of the "introduction". — Carl (CBM · talk) 12:55, 8 May 2009 (UTC)

Could you take File:Function_machine2.jpg and replace the word "RULE" with the word "FUNCTION"? That would be great. — Carl (CBM · talk) 12:58, 8 May 2009 (UTC)

Done. If it doesn't appear above with the word "FUNCTION f:" then you'll need to empty the old drawing out of your cache. BTW I'm gonna modify the "composition" drawing to agree with the text and eliminate the "location" business and just leave it here. (The notion of a variable as a "location" is no different than naming a register the "accumulator" or "A" or a memory cell or naming a cell in Excel. It'd be interesting to experiment on a class of algebra students to see if the usage creates confusion or clarity). Bill Wvbailey (talk) 13:57, 8 May 2009 (UTC) Composition-of-functions drawing's text updated. Wvbailey (talk) 14:46, 8 May 2009 (UTC)

I put the main image in the article, and the composition image in the section on function composition. — Carl (CBM · talk) 14:41, 8 May 2009 (UTC)

They look pretty good, not as crisp as I'd like, but I have a problem with artifacts introduced by the jpg compression algorithm (I saved them at max resolution). If you know of a "crisper" compression I could redo the images. Bill Wvbailey (talk) 14:51, 8 May 2009 (UTC)

Why don't you save them as PNG files? They aren't photographs (or anything similar), so JPEG isn't really an appropriate format. --Zundark (talk) 15:13, 8 May 2009 (UTC)

It does look better as a png. I'll update the article. Thanks, Bill Wvbailey (talk) 18:45, 8 May 2009 (UTC)

Introduction rearranged

I was not thrilled with the order of topics in the introduction, so I rearranged it a little while keeping all the same content. It now has this outline, which I think is nice:

• Functions are used in many areas / abstract definition
• Functions are relations between independent and dependent variables
• Functions have names like f
• Some functions get permanent names
• Functions in algebra are often defined with formulas
• Functions in analysis and other areas of mathemtics
• Functions in set theory are even more general
• Reminder: functions do not need to involve numbers

I think this is a very reasonable outline, although some of the sentences in the intro are still pretty rough. — Carl (CBM · talk) 13:11, 8 May 2009 (UTC)

Complete and utter nonsense

It is astonishing that for a mathematical concept so basic and established as "function" there is no formal definition given, at least in the current version of the article.

The problem occurs when paragraphs like the following, having zero basis in fact, are inserted into the article:

"Because functions are used in so many areas of mathematics, and in so many different ways, no single definition of function has been universally adopted. Some definitions are elementary, while others use technical language that may obscure the intuitive notion. Formal definitions are set theoretical and, though there are variations, rely on the concept of relation. Intuitively, a function is a way to assign to each element of a given set (the domain or source) exactly one element of another given set (the codomain or target)."

In fact there has been no question in mathematics about the one and only definition of a function for at least 60 years, and probably a lot longer.

Elsewhere in the article this unique definition is mentioned -- thank goodness -- but it is unfortunately not positioned as the clear definition of the subject of the article:

"A function ƒ from a set X to a set Y associates to each element x in X an element y = ƒ(x) in Y."

This is the one and only correct definition of a function in mathematics, though here it is stated somewhat informally.

There are many, many areas of mathematics where only certain kinds of functions -- subclasses of functions -- are of interest; these are typically expressed by the word "function" preceded by a (well-defined) adjective: continuous function, differentiable function, analytic function.

There are also many areas of math where certain variants on the concept of function are used. These are often also expressed by the word "function" preceded by an adjective. Unlike the above case where the objects constitute a subclass of functions, in these cases only the two-word phrase makes sense; these are not certain types of actual functions. (It is this unfortunate syntactical fact that can confuse the beginner.) A common example is "multi-valued function".

Of course, as in any other broad field like mathematics, once a smaller context is clear, the terminology often becomes more informal for the sake of convenience, and the word "function" may be used to refer only to a certain type of function. (This is uncommon, but it can happen.)Daqu (talk) 19:27, 28 June 2009 (UTC)

You raise a serious issue, and your absolutely correct. Somewhere we should put the definition as the subset of the cross product, and delete that nonsense you pointed out. Thenub314 (talk) 20:30, 29 June 2009 (UTC)

That is already in the article, in the section "set-theoretical definitions". There is another section, "intuitive definitions", that has the more informal definitions which are, for better or worse, exactly the definitions that are presented in elementary algebra and calculus texts, which do not mention relations or ordered pairs at all. We would be remiss not to include those definitions here. Before working on articles like this I would also have suspected there to be more unanimity about the definition of a function in the literature than actually exists. The issue of whether a function has a uniquely defined codomain is one place where even research literature varies the definition. — Carl (CBM · talk) 22:57, 29 June 2009 (UTC)

To add to what Carl says: we should consider who the audience is. The concept of a function is a topic that is studied both at the secondary-school level and farther on. Although it may be the case that all research mathematicians use a formal set-theoretic definition (I am not convinced: see Brouwer and/or the people who want to found mathematics on categories) it is important that the article also be accessible to readers who are not yet well-versed in set theory. On top of which, saying that a function can only be a certain kind of set-theoretic relation makes no sense in the context of articles concerning the pre-set-theoretic history of mathematics. Therefore, I think leading with the formal set theory would be a mistake. —David Eppstein (talk) 00:03, 30 June 2009 (UTC)

For my part I replied to the above comment without reading through the article. I agree it would be a bad idea to lead with the set theoretic definition. And it is appropriate being in its own section. But perhaps I would like to see added the simpler definition that it is a subset of the cross product satisfying various properties before moving to the ordered triple to allow things like partial functions. You raise an interesting places that use alternate definitions. While I think that one can in modern constructivist mathematics take the usual definition, I couldn't guarantee it. And I suspect that if the foundations are taken to be category theory, then you end up with a distinct definition. So, in my long winded way I am agreeing with you, and I stand corrected. Thenub314 (talk) 08:08, 30 June 2009 (UTC)

Reading the above statements again I was thinking it would be nice to take out the language "Because functions are used in so many areas of mathematics, and in so many different ways, ..." It doesn't really seem that it is the multitude of areas of mathematics, nor the ways in which functions are used that really account for why there is no universal definition. It would be better to simply start "No single definition of function has been universally adopted." Are there any objections to this? Thenub314 (talk) 10:42, 30 June 2009 (UTC)

It may take a little work to smooth out the text, but this seems like a reasonable initial change to me. — Carl (CBM · talk) 15:49, 30 June 2009 (UTC)

Unfortunately, Daqu is wrong. Of course, his definition is correct, and all correct definitions are equivalent to his, but, for example:

Thomas, Calculus, 2nd edition, "Suppose now that with each value of the variable x in its range there is an associated a value, or several values, of the variable y. We then say that y is a function of x."

Thomas's Calculus, 11th edition, "Since the value of y is completely determined by the value of x, we say that y is a function of x."

Hamilton, Logic for Mathematicians, defines a function as a symbol, denoted by f together with a subscript and a superscript, and defines the formal rules in which well formed formulas can be constructed using this symbol.

Halmos, Naive Set Theory, "If X and Y are sets, a function from X to Y is a relation f such that dom f = X and such that for each x in X there is a unique element y in Y with (x,y) in f."

I could go on, but you get the idea. I pick four books more or less at random off my bookshelf, and get four different definitions of "function". This variety of definitions is the reason for the paragraph to which Daqu objects. Mathematicians are used to this sort of thing, the average reader is not, and needs an explanation of why the various books in which he encounters the word give such different sounding definitions. Rick Norwood (talk) 12:51, 30 June 2009 (UTC)

Thanks for looking up the references. I don't want to pick on Daqu here; it would be perfectly reasonable for a PhD mathematician to think there is only one definition of function, since they all express the same underlying intuition. One of the things I enjoy about wikipedia is that it encourages a broader view of the literature. — Carl (CBM · talk) 15:49, 30 June 2009 (UTC)

Indeed, I certainly fell into this trap. When your a working mathematician, can learn many branches of mathematics and only need one definition. It is easy to forget there are small groups of people worrying about foundations, or specific special cases, in which you may want to alter the definition. Thenub314 (talk) 18:21, 30 June 2009 (UTC)

Also: see the archive/3 for an extensive survey on this same issue (definition of function), plus the issue of the various usages of domain and range. BillWvbailey (talk) 20:08, 30 June 2009 (UTC)

I understand people being swayed by an erroneous definition of function in a calculus text. Mathematics is an exact science, and a huge amount has been written about it, so it is inevitable that mistakes will be made.

But it does not help things in the slightest if every mistake is carved in stone in Wikipedia!

(Note: For many years -- and probably still -- many calculus texts made statements about the inverse tangent -- usually in the context of transforming between cartesian and polar coordinates -- that are out-and-out wrong. That is not a cue for us that state in Wikipedia that "According to some sources [false statement goes here].")

We have a heavy burden to bear: Like it or not, what appears in Wikipedia will be copied to many other places on the Web, and many people will take what they read here as the gospel truth. We ought to be very careful before reproducing errors on the sole excuse that they have appeared in print.

Let me emphasize that I'm *****NOT***** advocating that the article needs to lead with the technical set-theoretic definition of "function". What I am saying is that all definitions of "function" that are mentioned without qualification in the article should be either logically equivalent to this one, or an informal approximation of that definition.

I'm also not saying that multi-valued functions, or other usages in lambda-calculus et al. should be banned from the article. I'm only saying that it must be made clear that these are different from the standard definition of function.

Maybe the best way to remain accurate, yet accommodate actual usage of the word "function" is to begin the article by discussing only the standard definition of function. This could perhaps be referred to as the *standard* definition of the word.

Then towards the end of the article could be a section called something like "Other uses of the word function".

Otherwise, we are risking creating more confusion about the concept than there already is.Daqu (talk) 18:27, 4 July 2009 (UTC)

Can you give an example of a definition of function actually in the article that is not somehow equivalent to the standard definition or clearly described as intuitive? — Carl (CBM · talk) 23:30, 4 July 2009 (UTC)

The Power set is a function defined on the class of all sets so it does not map from a set to a set. The leader of the codomain article mentions this possibility. I'd certainly prefer the standard definition be clearly separated out from the 'set theoretic definitions'. I'm not sure I even like the 'set theoretic definitions' header because set theory has functions like Powerset which aren't set relations and logicians don't tend to use the standard definition. Dmcq (talk) 05:58, 5 July 2009 (UTC)

Daqu: I agree with you that there is a standard definition of a function: a set of ordered pairs in which if <a,b> and <a,c> are in the set, b = c. I strongly disagree that this definition, or anything like it, belongs in the lede. This article (and all mathematical articles) need to do two things -- three things, actually. They need to give non-mathematicians some general picture of what the mathematics is all about. They need to give a standard, correct mathematical definition. And they need to make it clear which is which. I think this article does that. Rick Norwood (talk) 11:59, 5 July 2009 (UTC)

Rick: I certainly agree with your first two desiderata. Well, not exactly. Not what the mathematics is all about, but rather just what a function is. (And how many times do I need to repeat that I do NOT believe a technical definition belongs in the lead!) But I do think an informal but correct definition belongs there.

(What you mean by "which is which" is not clear to me. Which what is which what?)

The article as it currently stands is a mish-mosh that would be confusing to beginner and seasoned mathematician alike.

Finally, the existence of relatively rare usages of the word "function" is appropriate to mention in the article, preferably toward the end. But unmistakably, the main thrust should be to simply convey the main idea of the standard meaning of function in math.Daqu (talk) 02:03, 11 July 2009 (UTC)

By "which is which" I meant that the non-technical definition should say something to the effect, "A non-technical definition of a function is..." and the mathematical definition of a function should say something to the effect, "A more precise mathematical definition of a function is..." Rick Norwood (talk) 11:56, 11 July 2009 (UTC)

I had not realized just how bad the lede had gotten. It used to be much better, but this is one of those articles that every college Sophomore feels a call to edit. I've made an attempt at a rewrite. Rick Norwood (talk) 12:14, 11 July 2009 (UTC)

revert of good faith edit

Two problems with the two paragraphs recently restored and then removed again.

In the case of the first paragraph, I don't think any mathematician in the past hundred years considers a definition of a function as a formula to be workable, and this does not seem to be the place to discuss infinite series.

In the case of the second paragraph, the phrase 'the new definition' has no antecedent that makes sense.

Rick Norwood (talk) 12:15, 2 July 2009 (UTC)

Annual rewrite

About once a year, this article needs a rewrite, as it attracts college Sophomores almost as strongly as the article on sex. I've made a first attempt at a rewrite of the first three sections, removing repetition, digressions, and general blather. The remaining sections still need a lot of work, especially to remove repetition. I would hope that nobody would edit a later section of an article without first checking to see if the material he or she wishes to add is already in an earlier section. Rick Norwood (talk) 12:51, 11 July 2009 (UTC)

Rick, you also removed certain details that do not appear in later sections, as well as any attempt to explain what a function "does" (expresses a relationship between two quantities), as opposed to what it "has" (input and output). We've been through this discussion before in spring 2007: people who think that the main value of the article is to give THE definition of function and the associated vocabulary vs others who feel that the article should address the multiple uses of functions in mathematics within their contexts. Although at the time I was unhappy with the emphasis on "inputs" and "outputs", the result was a bit more balanced than your latest version, because it tried to address the substance of functions, not just their syntax. Apropos, certain amount of repetition is actually good: for example, the lead should summarize the content of the article, the same frequently applies to the leading paragraphs of each section, it's good to establish a context for terse statements like

There are many other ways of defining functions. Examples include piecewise definitions, induction or recursion, algebraic or analytic closure, limits, analytic continuation, infinite series, and as solutions to integral and differential equations.

I agree that this article attracts some less-than-stellar editors, but in my opinion the main damage is the transformation of an encyclopedic article into a definitions/examples crib-sheet. Arcfrk (talk) 13:25, 11 July 2009 (UTC)

I hope you'll work to improve what I've written. My main goal was, first to remove what was incomprehensible, and second to stick to the point and avoid digressions.

The idea that a function expresses a relationship is, of course, the original use of the word. Mathematicians now accept the idea of purely abstract functions (such as the choice function) that do not express relationships.

The passage you quote is not one that I contributed to the article. I agree that it is too terse. Rick Norwood (talk) 13:36, 11 July 2009 (UTC)

I'm tempted to revert the changes entirely, under WP:BRD. The technical definition, although not significantly changed, is now wrong. I don't agree with the change in emphasis, but I could live with that provided the technical definition remains correct at all times. — Arthur Rubin (talk) 16:28, 11 July 2009 (UTC)

I actually quoted the prior lead-definition in some writing I've been doing:

"The mathematical concept of a function expresses dependence between two quantities, one of which is known and the other which is produced. A function associates a single output to each input element drawn from a fixed set, such as the real numbers (ℝ), although different inputs may have the same output"

Notice the emphasis on dependence, production, and association. Now I read this:

"In science and mathematics, a function has an input and an output. The defining property of a function is that, for a given input, there is one and only one output."

This is not a definition of a function, it is the definition of, among other things: A hose (it has an input and an output), a leaking bucket with a single hole in it (it has an input and an output), a doggie port (the input is the output and vice versa), a cable plugged into a wall, etc. Are these what we mean by "mathematical functions"? No -- it would seem that function is a transformational mechanism (or a description of one) that attends the input and produces the output per some algorithmic or lookup means. I know a lot of folks don't like the "mechanism" analogy, but I don't know why. Bill Wvbailey (talk) 17:20, 11 July 2009 (UTC)

If there is a mistake in the technical definition -- domain, codomain, ordered pairs -- please point it out.

The problem with the "mechanism" analogy is this: there are mathematical functions for which no mechanism exists. The classic example of this is called the choice function. Also, to a mathematician, {<17, x>, <Sam, Ralph>, <cat, 17>} is a function whose domain is {17, Sam, cat} and whose codomain is {x, Ralph, 17}. There is no "dependence", no "known", and no "unknown", only ordered pairs.

You need to read Naive Set Theory by Paul Halmos.

In bringing up physical objects with input and output, you are headed for deep waters. Do we want to say that the input and output to a function must be a) abstractions? b) symbols? c) well-formed formulas?

For this, you need to read Logic for Mathematicians by Hamilton. You should also read some of Yuri Manin's articles.

Rick Norwood (talk) 12:22, 12 July 2009 (UTC)

I don't know if you read what I wrote above before making your recent edit. Here is what you propose for the first sentence: "The mathematical concept of a function expresses dependence between two quantities, one of which is known and the other which is produced." The mathematical concept of a function has nothing to do with dependence, knowns, and production. Yes, the idea you put forth is common among my Freshman and Sophemore students. We set them right in our Junior level course, "Mathematical Reasoning".

As for why the ideas of dependence, knowns, and production are wrong, I provide examples above. Rick Norwood (talk) 12:42, 12 July 2009 (UTC)

Rick, to set the record straight: I have read both Halmos and Hamilton and Manin's text and about 20 more -- I have had both in my library for years (Halmos in particular since 1984, I stuck the price tag in the front cover). The point is -- we have to work with published sources so we need to find definitions that come from those, and -- (again) please review the 3rd archive for my survey study -- it was far more comprehensive than yours just above. [An aside: over and over I've heard the mathematicians' assertion about how there is nothing between the input and the output except "a relation", or "an association" (which I'd prefer) but I don't believe a single word of it. I assert that these things are forms of matter. As This represents a serious philosophical problem we can't address it further here, so I have to go with the published definitions even though I don't agree with them. I've been studying human consciousness for (at least) the past 5 years which means I've read all sorts of weirdness. I put the mathematicians' assertion re "function" at the same level of weirdness as panpsychism or Platonism or neutral monism (see Andrew Gluck 2007, 'Damasio's Error and Descarte's Truth' for more]. Anyway, I'm sticking with my point above. The lead etc was recently vetted by CBM and some other pros so I'm with Arthur Rubin and I'd recommend a revert. Bill Wvbailey (talk) 20:25, 12 July 2009 (UTC)

Continuing the unsourced philosophical tangent, I think the mathematical view that there is nothing between the input and output is demanded by extensionality. That is, there could be many ways to compute the same function values, but when viewed mathematically two names f and g that refer to functions with the same range and values must be the same function, regardless of "what is between" their inputs and outputs. This point of view doesn't demand that there be no structure between input and output but it does require that the structure is not used when viewing it as a function and manipulating it mathematically. —David Eppstein (talk) 20:54, 12 July 2009 (UTC)

But you've just used the words "compute" (active verb) and "manipulate" (active verb) and "structure" (noun). Your noun -- "structure", your "it" -- is either something or nothing. If it is nothing then the "function" is empty, null. Otherwise is is something, and I assert it is matter arranged in space. And you as a mathematician (aka agent) are applying your actions to your noun in order to "realize" the function. I assert that this all occurs in the physical world and nowhere else (as opposed to the neutral monist's world-view that all this happening in the mind of God -- hence the philosophical difficulties). Carl below called this the "constructive" viewpoint, but I'm going a step further to assert that for almost all readers, mathematicans included (at least the applied ones), the constructive is what they work with from day to day, and that's what they care about. So that's who this article should be addressing, primarily. Bill Wvbailey (talk) 21:37, 12 July 2009 (UTC)

Rick, you write and act in a condescending way. I assure you that several people who have worked on this article and formed a consensus on the lead that you are so much against are professional mathematicians with vast teaching experience. As I explained above, your version of the lead removed any hint of the *definition* of the function, leaving only elements of its syntax (inputs and outputs; not standard terminology in any case, if you want to be a stickler for mathematical rigor). Wvbailey elaborated on this point further. If you have an issue with the introductory clause "mathematical concept of a function" then, please, find a way to change it without removing the substance of the description. I have actually spent quite a bit of time thinking this through, and do not appreciate your wholesale revert "because you can". Whether you like or not, a wikipedia article has to address readers on many levels of sophistication, including freshmen and sophomore students whom you seem to deride. Certainly, function expresses an idea of the dependence between two quantities, at least since the time of Newton. If anyone is interested in Bourbaki's definition of a function, they can't read it in the primary source. I am going to restore my version and post a note at math project page; hopefully, we can reach a new consensus soon. Arcfrk (talk) 14:05, 12 July 2009 (UTC)

I think the "dependence" version is better. Just saying there is an input and output is not as clear to me. Even if a function is obtained in a nonconstructive way, the point of it being a function is that it establishes a correspondence, a dependency, between each element of the domain and the corresponding element of the range. — Carl (CBM · talk) 14:30, 12 July 2009 (UTC)

The following observation, from map (mathematics), may be helpful:

"Some authors such as Serge Lang use map as a general term for an association of an element in the range with every element in the domain, and function only to refer to maps in which the range is a field."

"Known" and "produced" are confusing, as no part of a function is necessarily being produced, and the property "known" would only be defined if an observer would be specified.  Cs32en  15:47, 12 July 2009 (UTC)

Yes I don't like known and produced. Functions simply are relationships, they don't do anything. The informal picture with inputs and outputs is as close as I'd go to saying anything is known or produced. Dmcq (talk) 17:42, 12 July 2009 (UTC)

They may not do anything, but then they represent specifications or "lookups" for an agent that does have to do something "to bring the function to life". The marks are matter. You the mathematician are acting as the agent (the active entity) that has the ability to interpret the symbols or marks or whatever. . ..Bill Wvbailey (talk) 20:25, 12 July 2009 (UTC)

I concur about "known" and "produced". Here is an old version that doesn't use them. One should keep in mind, however, that in most contexts functions do have (independent) variables and there is a definite asymmetry between the argument and the value of the function. Perhaps, the word "association" expresses this asymmetry better than the word "relation". Arcfrk (talk) 18:07, 12 July 2009 (UTC)

I'd say that one variable (or set of variables) is being defined in terms of the other.  Cs32en  18:15, 12 July 2009 (UTC)

This is just my 2 cents. I think the lead is essentially very good. It is clear that enormous amount of efforts must have been put to achieve the balance of accessibility and rigors, as if it were a legal document. My complain (as everyone has one) is that the lead fails to give the clear and true definition of a function in mathematics. That is, the "graph". By this I don't mean a plot in undergraduate calculus but the set of pairs satisfying a certain condition (so that the function is single-valued). Specifying a function is precisely specifying the graph. In particular, this is problematic "the modern mathematical definition of a function is discussed below." This is as if we're saying you're not sophisticated enough to understand math so we don't tell you what a function really is. Also, it is important to note in the lead that functions appear as solutions of an equation, for instance. On the other hand, I don't know if compositions are important enough; it's just repeated applications of functions. -- Taku (talk) 22:05, 12 July 2009 (UTC)

Wvbailey wrote: "I put the mathematicians' assertion re "function" at the same level of weirdness as panpsychism or Platonism or neutral monism." That being the case, he cannot also assert that his definition is what mathematicians assert. And this elementary article is not the place for discussions of the philosophy of mathematics. If we want to discuss nominalism vs. realism there is a place for that, but this article must stick to the standard mathematical definitions, as given in standard mathematical reference books.

I also find "input" and "output" ugly. I'm much more comfortable with "argument" and "value". But usage has changed over the years, and I want the article to be readable. The old meaning of "argument" has been largely replaced by "input" while the old meaning of "value" has been largely replaced by "output".

Serge Lang's definition is of interest, but clearly too specialized for this article.

Rick Norwood (talk) 15:55, 13 July 2009 (UTC)

Here we go, round again, singing a song about Molly B.

My first edit to this page was in 2005. The same discussions occur again and again. I'm sorry if that makes me sound condescending. I'll try not to.

Everyone, as best I can tell, agrees that a function is a set of ordered pairs. The problems arise when people try to put this simple idea into words that will make sense to a non-mathematician.

There seem to be two basic points of view:

1) A function is a "rule". Lebesgue would have agreed.

2) A function is a set of inputs, each paired with a unique output. Most computer scientists would agree.

On this simple point, thousands of words have been written, dozens of ledes reverted.

Rick Norwood (talk) 16:11, 13 July 2009 (UTC)

Despite the repetitiveness, I think the recent flurry of edits has improved the lead section. I'm more concerned now about the section below it, titled "Introduction". Per WP:LEAD, I'm not convinced we should have separate "introduction" sections at all, and some of the arguments in that section seem quite muddled. —David Eppstein (talk) 18:01, 13 July 2009 (UTC)

I certainly think your edits have greatly improved the lede. For the heading on the second section, how about "vocabulary" or "notation". The last paragraph in that section should probably be moved. Rick Norwood (talk) 18:10, 13 July 2009 (UTC)

I think we can all agree that defining a function as a mere arithmetic expression is unsatisfactory. The notion of function as a "rule" would probably be familiar to most 18th century and 19th century mathematicians. It does find some use to this day in constructive mathematics as well as some parts of logic and theoretical computer science, but apart from that, it has been superseded by the modern concept of algorithm. --Classicalecon (talk) 19:04, 13 July 2009 (UTC)

(An Algorithm, with all of its restrictions, is only one way to compute a function. There are many functions that cannot be computed by any algorithm, just as there are geometrical constructions that cannot be made using a (straight edge and compass) and just as there are true statements in any system of mathematical logic that cannot be proven using the axioms of that system (Gödel's Theorem).) David spector (talk) 20:06, 5 December 2009 (UTC)

Actually, not everyone agrees that a function is just a set of ordered pairs. Some authors also include information about the codomain so that a function is either intrinsically surjective or intrinsically not surjective. This is one of the reasons that it is very difficult to give a formal definition in the lede. (I have no objections to the present version [3]; this is just a reminder that not everyone uses the definition from a set theory text). — Carl (CBM · talk) 19:06, 13 July 2009 (UTC)

Good point. Since not everyone here seems aware of the recent discussions at codomain or even that codomain is often included as part of the definition of a function. --C S (talk) 21:01, 13 July 2009 (UTC)

Perhaps it's not surprising but the section "mathematical definition" is rather muddled and confuses together definitions. It starts by saying one definition of function is as a certain triple (which includes codomain). Then it appears to say (at least I think this is a straightforward reading) that this is a special case of a relation, which obviously does not include codomain info. --C S (talk) 21:08, 13 July 2009 (UTC)

Clarification: the relation article seems to have changed substantially since I saw it last, with the main definition now being of the form "relation between X and Y" which includes this info. However, the responses below to my comment indicate that the distinction that CBM and I are making is still a point of confusion here. --C S (talk) 17:13, 16 July 2009 (UTC)

Why would relation not include codomain. Thus a relation is an ordered triple consisting of a domain, a codomain, and a set of ordered pairs. Or, equivalantly, a relation is any subset of a cross product. Rick Norwood (talk) 12:22, 14 July 2009 (UTC)

No, not equivalently. Once you take a subset of a cartesian product of two sets, the info on the superset is lost. If you wish to define a relation as an ordered triple, that's different, and not at all what you states above when you wrote "a function is a set of ordered pairs." A set of ordered pairs does not contain the necessary codomain info. --C S (talk) 17:00, 16 July 2009 (UTC)

Rick Norwood is right here, see Relation (mathematics)#Is a relation more than its graph?. --Classicalecon (talk) 14:10, 14 July 2009 (UTC)

There are different definitions of relations. Rick used initially the definition with ordered pairs only and that is what I (and CBM) responded to. His recent comment continues to confuse this with another definition, which is used in the article you link. In other words, his confusion seems to be the kind discussed in the section you linked to. --C S (talk) 17:00, 16 July 2009 (UTC)

Two vocabulary sections

There are two sections entitled "vocabulary" at the moment, of which the second seems to ramble without a point. I believe I could get rid of that second section if I merge some terminology to other places and then move the rest to new better-titled sections. No content would be lost. Would anyone object to that? — Carl (CBM · talk) 19:29, 13 July 2009 (UTC)

Why not rename the first vocabulary section to "Informal overview"? --Classicalecon (talk) 20:17, 13 July 2009 (UTC)

It was titled "introduction" until this edit: [4]. But I never liked the second vocabulary section anyway, since it doesn't read well. That's why I think rearranging is better than changing the section titles. — Carl (CBM · talk) 20:31, 13 July 2009 (UTC)

There is still a great deal of repetition in the article, so I suggest that CBM carry out his plan.

Drawing

Wouldn't a drawing showing the "mapping" from the domain (set) through the function (set) to the codomain (set) (see the first drawing on this talk page -- I agree it's too complicated but maybe I or someone could redo it given some input from other editors) be of use in this article? There were some drawings in previous versions but they didn't show the function, just the domain and codomain. Bill Wvbailey (talk) 14:33, 14 July 2009 (UTC)

A function can be visualized as an "ordered triple" of sets that takes an input x from the codomain X, "passes it through "the function f and puts y = f(x) into the codomain Y.

This is just a sketch that sort of agrees with the definition from Stewart 3rd edition (page 1), but not with his drawing:

"A function f is a rule that assigns to each element x in a set A exactly one element, called f(x), in a set B."

In his "arrow drawing" Steward shows only two shapes, the domain and codomain, illustrated with a --> f(a) and x --> f(x) and null --> null.

Does the function "select" the appropriate element of the codomain or does it "put into" the codomain the appropriate element? Bill Wvbailey (talk) 14:42, 15 July 2009 (UTC)

The drawing is ugly and misleading. I'm sure we can find or create better. The arrow should not go from x to the function and then from the function to f(x). In the "arrow" analogy, the arrow is the function, and the arrow certainly does not pass anything along or put anything anywhere. (There is also the nice convention that we put a bar on the tail of the arrow if it points from x to f(x) and no bar on the tail if it points from X to Y.) Rick Norwood (talk) 20:22, 15 July 2009 (UTC)

This is from a drawing group that was in an earlier version. I added some verbage that I'm sure could be improved upon.

An arrow diagram is a way to illustrate a function. One definition of a function is an ordered triple (X, Y, f) that consists of (1) a set X that represents the domain, (2) a set Y that represents the codomain, and (3) a "graph" f, a collection (set) of arrows (see more at Graph of a function) that associates a single element of the codomain Y with each element of the domain X. Each arrow (element) of graph f represents an ordered pair (x, y) where x is an element of X and y is an element of Y, e.g. f = { (1, d), (2, d), (3, c) }. Another definition of a function does not include the domain X and Y; it is just the graph f. Note: the graph's arrow-heads form a subset of the codomain called the range, shown here as { d, c }.Wvbailey (talk) 22:55, 17 July 2009 (UTC)

These drawings are just trial balloons; what I'm after is comments re content, artistic quality comes later. I don't understand the convention re the bar at the tail of the arrow. If you could point me to a text that illustrates this that would help. Or point me to a text that well illustrates the point. If folks feel that an illustration is a not a good thing, that is okay too. Bill Wvbailey (talk) 22:10, 15 July 2009 (UTC) The arrow drawing becomes quite interesting when you try to draw a composite function. Will tend to this when I return from the pucker-brush (aka Wyoming). Bill Wvbailey (talk) 12:35, 16 July 2009 (UTC) Add the name of ordered pair as "graph" of the function to verbage under illustration.

The "bar" convention is just something I picked up from my teachers in grad school. I don't know how widespread it is. But the idea is this. If I write f:R->R, no bar, but if I write F:2|->4 then I use the bar. If anyone knows a written reference, I'd be glad to know about it. Rick Norwood (talk) 12:33, 16 July 2009 (UTC)

I'm pretty sure that's a standard convention. I'd look in one of your books...I'd be very surprised if you couldn't find any usage there. --C S (talk) 17:26, 16 July 2009 (UTC)

Re bar at head of arrow: Your example reminds me of the symbolism for entailment, or something like that, can't quite remember (this symbol taken from Microsoft Word ↦). A Question: If the 'ordered triple' is the function, then what is the name of the set of ordered pairs that constitute the collection of arrows? BillWvbailey (talk) 12:55, 16 July 2009 (UTC)

If we use the "ordered triple" definition of a function, then the set of ordered pairs is called the "graph" of the function. Rick Norwood (talk) 13:12, 16 July 2009 (UTC)

I've never seen the bar, but I've rarely seen F:2->4; normally, it's written "F such that F(2)=4", or when context is needed, "F:X->Y such that F(2)=4". — Arthur Rubin (talk) 23:13, 16 July 2009 (UTC)

I wouldn't equate sets of ordered pairs and directed graphs; most definitions of graphs that I've seen allow isolated vertices, which the set-of-pairs definition omits. Also if one views a function as being defined by a graph, the domain can be determined as the set of vertices with outgoing edges, and the codomain must include all vertices with incoming edges or with no outgoing edges, but if a vertex has outgoing edges but no incoming edges it's ambiguous whether it belongs to the codomain. So while I think directed graphs are fine for representing functions from a set to itself (see functional graph) they're a bit problematic for representing functions more generally. —David Eppstein (talk) 23:11, 17 July 2009 (UTC)

Isn't Graph of a function more what's meant rather than anything from graph theory? Dmcq (talk) 23:22, 17 July 2009 (UTC)

Maybe, but I was responding to the mention of directed graphs in Wvbailey's image caption. —David Eppstein (talk) 23:25, 17 July 2009 (UTC)

Yeah, I was wondering if "directed graph" was the appropriate link; I'm a "state diagram" kind of guy and this just popped into my head. I'll change the verbage under the diagram to Graph of a function. Thanks, Bill Wvbailey (talk) 23:44, 17 July 2009 (UTC)

Leader

I just reverted back the leader which was changed to an old version emphasising analysis in the first paragraph and which removed the explicit reference to elements of sets for the input and value. Functions may be important in analysis but so are limits and other things, and functions are important in lots of areas besides analysis, for example topology. And I do not think it is good enough to just say the mathematical definition is in terms of sets when one can give a fairly accurate description which isn't too intimidating saying the inputs and outputs are elements of sets and need not be numbers. Dmcq (talk) 17:07, 14 July 2009 (UTC)

Well, study of functions is an essence of analysis, at least, if you look at it from historical viewpoint, from time of Newton through Weierstrass. I do think that this must be stressed right from the beginning; whether or not to include the definition of a function is comparatively less important. Topology deals with shape and continuity in a very general sense, so your comparison is completely off the mark. You obviously seems to feel very strongly about this point, since you have reverted more than once: why don't you find a concise way to explain what *is* the main import of functions, then, right in the lead, instead of engaging in pointless reverts? As for the nature of the arguments and values of functions, your version says that they need not be real numbers, which is a negative statement: nowhere before do we see any mention of what they are, or could be. This is why I rearranged the material a bit. Please, keep in mind the flow of the text as you try to express your point. Ah, and "for instance, as well as" is such a grammatical atrocity that it ought to be banned and perpetrators exiled to another wiki. Arcfrk (talk) 19:16, 14 July 2009 (UTC)

The mention of analysis, and the formalistic set-theoretic description of what functions' domains and codomains are, both form advanced mathematics that will be offputting to many readers of the article. They should certainly be included in the article, but your promotion of them to the very start of the lead section seems unbalancing. —David Eppstein (talk) 20:32, 14 July 2009 (UTC)

Yes, I absolutely agree with you about domains and codomains, as well as more broadly, that the set-theoretic definition is not appropriate for the lead; I don't know where you got the idea that it was I who promoted them. On the other hand, the mention of analysis in the first paragraph is anything but formalistic: on the contrary, it forms the much-needed context for the notion of function, it is just slightly more precise than standard clause "In analysis" or "In mathematics" at the beginning of an article. I would be happy with any reasonable version of the lead, although our time might be spent more productively improving the article itself. I do hope that you keep watching over the grammar and style, some recent versions were sub-sub-par. Arcfrk (talk) 21:21, 14 July 2009 (UTC)

Maybe use the word "calculus" instead of "analysis"? "Elementary algebra"? Does anyone know when students first encounter the notion of "function"? Is it in "elementary algebra" or "calculus"? (My 7th grade nephew was introduced to algebra this year; I don't remember him fretting about functions). My son's cc of Steward (1995, Calculus: Early Transcendtals, 3rd edition, Brooks/Cole Publishing Company begins with this: "The fundamental objects that we deal with in calculus are functions" (page 1). He shows (in diagrams on page 2) all the basic ideas concerning "function" that have been batted here including both "Figure 1 Machine diagram for a function f" and "Figure 2 Arrow diagram for f". Composition appears on page 13 together with a machine diagram. The word "analysis" does not appear in some 1000+ pages (only "Analytic function" and "Analytic geometry". "Set" appears after 1000 pages in appendix A. Bill Wvbailey (talk) 22:13, 14 July 2009 (UTC)

Definitely functions are introduced at the beginning algebra level in high school...but most students in my experience do not know them when they take calculus. For example, even the "drill and kill" type books like Saxon Algebra 1 have functions. But it sounds like you're talking about pre-algebra...I don't recall if functions are covered there. --C S (talk) 17:35, 16 July 2009 (UTC)

Putting up front that their main application is in analysis or calculus just reinforces the idea people have that they are just to do with numbers. And I didn't mention the words domain or codomain, I just gave an informal version of the usual idea of a triple (X,Y,f). I know there's lots of other definitions, I put in a counterexample myself, but that's the most usual one in maths. Analysis is mentioned at the end of the leader, there just no justification for including analysis twice and not giving a halfway decent description of what a function is. Functions are important to calculus but they're important to an enormous amount besides and you don't need calculus to work with them. Dmcq (talk) 22:35, 14 July 2009 (UTC)

Constructive and Nonconstructive Definitions

Computer scientists draw a distinction between imperative programming, instantiated, for example, in procedures, and declarative programming, instantiated, for example, in functional programming. This distinction is well understood by the student of geometry, who learns early the difference between constructive proofs and proofs that involve no construction, such as derivations from axioms and proofs by contradiction (see Types of Proofs). In many areas of mathematics, it is easy to prove theorems that are true and useful, yet do not give even a remote hint as to how the conclusion can be generated from the premises. Similarly, it is possible to exhibit functions that are useful or have useful properties, yet give no clue as to how (or even if) they can be calculated.

Another way to put this is: calculation or construction is different from declaration or definition.

A function is a type of mapping, and mappings can be considered "generative" ("transformative") in the sense that they act like a machine that "converts" or "replaces" an element with another specific element, according to some simple or possibly very complex rules, including an explicit but possibly infinite list of domain/range pairs. However, just as imperative-style programs can equally well be expressed as declarative-style programs, a mapping (and hence a function) can also be viewed and manipulated in "abstract", "non-generative" ways.

The reason I bring this up is that I think only the "productive" or constructive type of function is discussed above. But functions can also be "declarative" in that the values they generate are not known, cannot be written in any closed symbolic form, etc. Furthermore, functions can be treated as abstract objects in their own right: they can be composed, and algebras can be generated from them without reference to their actual domains.

I believe the best approach to functions is to start with extremely simple, intuitive examples, such as the "black box" machines above, but stating explicitly that this is not yet the whole story. As each new example or definition is added, it should inspire the reader to see that there are many different ways to look at functions in mathematics. Eventually, by the end of the article, there should be a table giving the fundamental and rigorous definitions of "function" in various areas of mathematics. It should be clear by then that these definitions, some of which sound very different from others, are all equivalent.

I like this approach. Bill Wvbailey (talk) 00:09, 15 July 2009 (UTC)

If we don't do that, then we risk conveying only a set-theoretic view, or only an algebraic view, or only a computer science view, or only a database-theoretic view, or only a topological decomposition view, or only a view in the context of partial differential equations, etc., etc.

Just as Saul Gorn of the University of Pennsylvania (my former teacher) famously showed in his "Baker's Dozen" paper that many different implementations of an algorithm can be equivalent (including logic diagrams, programming languages, integrated circuit layout, logical rewriting systems, Turing machines, etc.), so there are many equivalent definitions of functions (and other kinds of mappings), each having its own usefulness in some area of mathematics. David spector (talk) 23:48, 14 July 2009 (UTC)

I don't think a table is the right way to go; the standard format for presenting ideas such as these is prose. Moreover, the article at present does describe the variations on the definition, and there are not really so many that a table is needed. The article also already explains, as early as the lede, that there are functions that cannot be described with a simple formula. So it seems to me that the suggested content is already in the article.

Maybe I have misunderstood your long, abstract suggestion. It would be easier to discuss if you could refer directly to the sections in which the change is proposed and tell exactly what change is desired. Can you rephrase your abstract suggestion into short concrete ones? — Carl (CBM · talk) 03:57, 15 July 2009 (UTC)

I second CBM about being more specific. Also while I'm all in favour of starting simple and inspiring readers you do have to be careful that this is an encyclopaedia and not a student textbook. It should mainly cater for people who want to look up things and haven't a great deal of patience, good websites try to make a page load in less than a second or the punters go elsewhere. So I like to see things fairly definitive from the start but phrased so people who don't already know most of the answer aren't turned away. Dmcq (talk) 07:38, 15 July 2009 (UTC)

"More specific" and "tell exactly what change" are clearly code for "please propose a rewrite for part or all of the article". I wish I had time, but even if I did I probably wouldn't waste it that way. When an article needs improving, like this one, and there are editors lurking who have strong opinions, like this one has, it is likely that any proposal will be shot down. Such is life here at Wikipedia, and I accept it as the price for such a great resource, and one that is always improving (although not necessarily monotonically--this article is a great example of ups and downs throughout its history).

I've indicated one approach to improving the article. I think that's a pretty good use of my time here. IMO, only when someone comes along who understands what I've written (a mathematician) and also understands what Dmcq wrote just above (a humanist educator), this article will have a chance to be rewritten really well. Until then, youse guys are just going to have fun opposing each other and anyone else who comes along. I was once oppositional, like many editors here, but now, at age 63, I've either grown up or gotten tired of fighting (or perhaps both). David spector (talk) 23:50, 16 July 2009 (UTC)

I'm glad you have taken the time to make a long, thoughtful suggestion, while saving the bulk of your time for more important things off Wikipedia -- a wise choice in many ways); however, if you're not interested in actually implementing what you say or engaging in discussion about it, I don't see how you expect others to magically do your bidding. That's definitely not the Wikipedia way. --C S (talk) 03:02, 17 July 2009 (UTC)

Reordering the sections

I was looking trying to reorder the sections as a first step to cleaning up the duplication and getting a good top level structure. Here's my fist thoughts on how the sections should be reordered. The idea is to firstly just move them around without changing any content and then it might become easier to concentrate on specific sections.

Leader

1. Overview
2. History
3. Mathematical definition
   1. Notation
   2. Vocabulary
   3. Identity function
   4. Restrictions and extensions
   5. Function composition
   6. Functions with multiple inputs and outputs
      1. Binary operations
   7. Inverse function
4. Specifying a function
   1. Computability
5. Function spaces
   1. Pointwise operations
6. Other properties
7. See also
8. References
   1. Notes
   2. Sources
9. External links

Thoughts anyone? Dmcq (talk) 11:52, 18 July 2009 (UTC)

I've finished the reordering of the sections and then moving bits between the sections. I guess it's open season again. Dmcq (talk) 21:25, 19 July 2009 (UTC)

I will try to give it a thorough reading tomorrow morning. — Carl (CBM · talk) 22:36, 19 July 2009 (UTC)

Merging and splitting "vocabulary"

I read through the first few sections in detail this morning and they seem nice. But then the "vocabulary" section sticks out when you reach it: the tone changes, the sentences become choppy, there is less sense of organization. Sections such as "Functions with multiple inputs and outputs" and "specifying a function" feel like they should be higher up in the article. I will try to work on these things slowly, merging a little at a time. — Carl (CBM · talk) 12:55, 24 July 2009 (UTC)

I moved bits from the original vocabulary and notation sections out to other sections because they both seemed too long. I was thinking of how to amalgamate the remainder of the vocabulary and notation sections but copped out to just putting them adjacent to each other rather than engaging in too much editing of the actual text of the sections - also I haven't a good idea how to do it nicely :) Dmcq (talk) 15:12, 24 July 2009 (UTC)

I've now reordered the vocabulary section to move bits down from the top as far as possible and make the start shorter. I moved the "Functions with multiple inputs and outputs" higher up but have done nothing to "specifying a function. Some of the content was made into two small new sections on injective and surjective functiosn and image of a set. Dmcq (talk) 21:23, 1 August 2009 (UTC)

Good work, Dmcq. Rick Norwood (talk) 13:37, 2 August 2009 (UTC)

Equivalence of definitions

I'm lost in the WY pucker-brush so I have to be brief. How are we to treat what appear to be at least 3 different versions of the definition of function (as an ordered triple, as just the graph, as a rule such as an algorithm or formula)? I'm having a hard time believing that these are equivalent. Bill Wvbailey (talk) 17:17, 21 July 2009 (UTC)

They're not the same so no need to believe three impossible things before breakfast! The definition as a function is the earliest one and the one probably used most in informal contexts. The definition as a triple is the usual one in maths except in logic and related areas. Dmcq (talk) 17:43, 21 July 2009 (UTC)

The basic idea of a function is the same throughout mathematics. A function is deterministic: the input determines the output. But this idea can be expressed in a variety of ways, increasingly abstract. Rick Norwood (talk) 12:22, 22 July 2009 (UTC)

Can someone explain succinctly how an algebraic formula such as y(x) = 3*x+1, with a domain of the real numbers (for instance), can be considered "equivalent" to an ordered triple definition. Whereas I was able to write the formula (and thereby specify the graph) with 10 symbols, the ordered triple's graph is an infinite set, and this I cannot write down. But the only way to make the two graphs equivalent is to write them down because you cannot assume, given the graph's nth element, that the n+1st element will be (n+1, 3*(n+1)+1) -- it has to be written down. Am confused, and I'm sure so are many students. Bill Wvbailey (talk) 16:52, 23 July 2009 (UTC)

The definitions are not fully equivalent but that function can be represented by the triple

$\left(\mathbb {R} ,\mathbb {R} ,\left\{\left(x,3x+1\right):x\in \mathbb {R} \right\}\right)$

Dmcq (talk) 17:37, 23 July 2009 (UTC)

This occurred to me, but I rejected the notion (in the formal sense) because it looks like a category error, i.e. mixing apples (aka ordered pairs) with oranges (formulas, algorithms). On an intuitive level, though, I understand your symbolism. Bill Wvbailey (talk) 00:56, 25 July 2009 (UTC).

I'm afraid I don't understand you. The whole expression is a set and could be expressed in a proof checker like Metamath or Mizar without any great problem. It sounds to me from what you're saying that you would have a problem with the idea of the set of natural numbers even. Dmcq (talk) 06:56, 25 July 2009 (UTC)

I'm trying to make some sense out of my concern, based as it is on an intuition at this point. I do know it has to do with mixing an intensional definition (formula, algorithm, etc) into an extensional definition (ordered triple, one element of which is a collection of ordered pairs). So I've been doing some historical research re how the set-theoretic definition of a function came to be. It turns out that Bertrand Russell took Frege seriously and, with some refinements, adopted Frege's views in his Bergriffsshrift (1879) and later writings, that "functions" were more basic than "predicates" or "relations" (cf *Russell 1903:505, and Frege 1879 in **van Heijenoort 1967:21-24). But Russell had some quibbles with Frege's notion of "function" (cf his appendix A in his 1903) and he expressed some of the same concerns I've expressed above with regards to how one goes about specifying infinite ordered pairs in a thoroughly extensional manner, in particular:

"...if we take extension pure, our class is defined by enumeration of its terms, and this method will not allow us to deal, as Symbolic Logic does, with infinite classes. Thus our classes must in general be regarded as objects denoted by concepts, and to this extent the point of view of intension is essential. It is owing to this consideration that the theory of denoting is of such importance....throughout the discussion, I must ask the reader to remember that whatever is said has to be applicable to infinite as well as to finite classes." (Russell 1903:67)

"71. Class may be defined either extensionally or intensionally. That is to say, we may define the kind of object which is a class, or the kind of concept which denotes a class: this is the precise meaning of the opposition of extensional and intension in this connection. But although the general notion can be defined in this two-fold manner, particular classes, except when they happen to be finite, can only be defined intensionally , i.e. as the objects denoted by such and such concepts. I believe this distinction to be purely psychological: logically, the extensional definition appears to be equally applicable to infinite classes, but practically, if we were to attempt it, Death would cut short our laudable endeavor before it had attained its goal. Logically, therefore, extension and intension seem to be on a par."(boldface added, Russell 1903:69)

I've found a fascinating quote re the history of the notion "ordered pair" (the problem of ordering appears tangentially in Frege as well cf van Heijenoort 1967:23):

"98. There is a temptation to regard a relation as definable in extension as a class of couples. This has the formal advantage that it avoids the necessity for the primitive proposition asserting that every couple has a relation holding between no other pair of terms. But it is necessary to give sense to the couple, to distinguish the referent from the relatum: thus a couple becomes essentially distinct from a class of two terms, and must itself be introduced as a primitive idea . . . It seems therefore more corect to take an intensional view of relations, and to identify them rather with class-concepts that with classes. This procedure is formally more convenient, and seems also nearer to the logical facts. Throughout Mathematics there is the same rather curious relation of intensional and extensional points of view: the symbols other than variable terms (i.e. the variable class-concepts and relations) stand for intensions, the the actual objects dealt with are always extensions."(Russell 1903:99)

Resolution of the ordered-pair problem (i.e. no need for an addition of Russell's "primitive idea") came only with Wiener (1914)-Hausdorff (1914)-Kuratowski (1921) (cf commentary in van Heijenoort 1967:224). But at this point I don't see the set-theoretic definition solves the "intension versus extension" issue. Maybe you or someone else reading this can help me here. (Eventually I hope to add some of this into the history section.)

*Russell, Bertrand (1903), The Principles of Mathematics: Vol. 1, Cambridge at the University Press, Cambridge UK (reprinted by googlebooks).

**van Heijenoort (1967), From Frege to Gödel: A Source Book in Mathematical Logic, 1879-1931, Harvard University Press, Cambridge MA, ISBN 0-674-32449-8 (pbk)

Bill Wvbailey (talk) 16:29, 2 August 2009 (UTC)

Actually the ZFC axioms don't have any sets defined by listing their contents, they are all defined by existence rules. Something like there exists a set such that if x is an element of the set then etc etc. I'd hate to have try and express even a straightforward set that way. Dmcq (talk) 21:45, 3 August 2009 (UTC)

As another man named Bill once said, "It depends upon what the meaning of the word 'is' is." [5].

From one perspective, the function is its graph, or is the ordered triple, (domain, codomain, graph). This is the perspective from which mathematicians usually use the word: It embodies exactly the idea of "arguments determine values" or "inputs determine outputs". It says nothing about how the values are determined from the arguments, because from this viewpoint, that's irrelevant. y(x) = 3x + 1 is a convenient way of expressing the graph and nothing more; the exact same content is expressed by y(x) = 3(x + 1) − 2 or by an infinite list of ordered pairs.

But even though this is by far the most useful perspective for abstract mathematics, it's not the only one possible. Computer scientists, as well as most students, care very much about how the values are determined from the arguments. From this perspective, the two functions above are different because their method of determining their values from their arguments are different. For example, one might note that the first function, y(x) = 3x + 1, uses one multiplication and one addition, whereas the second function, y(x) = 3(x + 1) − 2, uses one multiplication and two additions. There are situations when these sorts of considerations come up in pure mathematics (in logic, the precise sequence of symbols used to represent certain things can be important), but they're not common.

The mathematical world has, as far as I'm aware, universally adopted the first perspective: Functions are determined by their arguments and values and nothing else. That's what the usual definition of a function as its graph is meant to express. But this takes some sophistication to accept; it really didn't happen in math until the 20th century, I'd say. The second perspective seems much more transparent to most people; but it lost out in mathematics because it's just not as useful.

Does that help? Ozob (talk) 17:43, 23 July 2009 (UTC)

Yes, Someone above wrote about a function as a definition "in extension", and if I understand both of you, you are saying something similar. But I see a problem (somewhat to the point of Classicalcon below): an algorithm to create an infinite list of ordered pairs one after another in sequence, say 3*x+1 one after the other ad infinitum starting with x=0: {(0,1), (1,4), (2,7), (3,10), ..., (ω, 3*ω+1) } is more complex than an algorithm that merely inputs a specific x and spits out 3*x+1. I'm not sure what to think about Rick Norwood's statement below that mathematical and computer functions are not the same thing. Bill Wvbailey (talk) 00:56, 25 July 2009 (UTC)

I don't think that the second perspective has "lost out" in mathematics, as much as been given a different name than "function". The concept of "algorithm" in mathematics is basically equivalent to your second definition. The real problem is that outside of special cases algorithms are 1. inherently partial, i.e. not necessarily terminating, and 2. not extensional, i.e. two different algorithms may have equivalent correspondences between arguments and values, and it is in general impossible to decide whether two algorithms are equivalent in this sense. So I disagree that your second perspective is genuinely "more transparent", outside of really elementary cases. --Classicalecon (talk) 19:05, 23 July 2009 (UTC)

One reason mathematicians don't use the idea that function = formula is that, in the example above, working to one significant digit and in the case of a number ending in 5 rounding to the nearest even digit, if x = 4, then 3x + 1 = 10 + 1 = 10, while 3(x+1) - 2 = 3(5) - 2 = 20 - 2 = 20. (Similar examples can be given with any number of significant digits.) In short, a computer function is not the same as a mathematical function. Rick Norwood (talk) 19:53, 23 July 2009 (UTC)

Can you expand on this? Does everyone else agree? Thanks, BillWvbailey (talk) 00:56, 25 July 2009 (UTC)

I don't think this article does ever claim that a function is defined as a rule. Can someone point out what text in the article is under discussion here? — Carl (CBM · talk) 13:09, 25 July 2009 (UTC)

The question of round-off error is discussed in that article, and really does not belong in this one. I just brought it up to show that the mathematical idea of a function is not, at its heart, computational.

The article does not claim that a function is defined as a rule. This entire discussion, which produces in me a strong sense of deja vu, is in response to an edit that did attempt to define function as a rule or formula. The edit is gone, but the discussion goes on and on. Rick Norwood (talk) 14:07, 25 July 2009 (UTC)

I'm hammering on this becasue the article says:

"One precise, mathematical definition of a function is that it consists of an ordered triple of sets, which may be written as (X,Y,f). . . .Some authors (especially in set theory) define a function as simply its graph f . . ."

This vague wording indicates to me that there are other defintions. And it's not clear from the article how a formula (or an algorithm) turns into one of these definitions, altho Dmcq's sticking the formula into the ordered triple (if it is formally correct) helps, Ozob's "in extension" should be expanded upon. Bill Wvbailey (talk) 16:37, 25 July 2009 (UTC)

The other important definition for that section is the one below, from set theory, where the function is identified with its graph. The second paragraph of the "Mathematical definition" section describes how one obtains an actual function from the formula f(x) = x^2. — Carl (CBM · talk) 16:42, 25 July 2009 (UTC)

And as said before the definitions aren't the same, different people use the name 'function' for slightly different things. A constructivist for instance wouldn't consider something a function unless you could always work out its values as closely as desired. Dmcq (talk) 22:46, 25 July 2009 (UTC)

Words in mathematics are defined both more precisely and more flexably than in other fields. Words have a precise definition only in context: many authors begin their book with a Chapter Zero that puts forth their precise definitons. But another author's Chapter Zero may have slightly different definitions. Or even completely different definitions! See, for example ring (mathematics). This takes some getting used to but, as Lewis Caroll said, "It's all a question of who's the master, you or the word." Rick Norwood (talk) 12:41, 26 July 2009 (UTC)

What you-all have written is useful, but it's here on the discussion page. I'd work on alternate wording but I'm in the puckerbrush. I definitely think that Dmcq's example (or a variant) should be included:

$\left(\mathbb {R} ,\mathbb {R} ,\left\{\left(x,3x+1\right):x\in \mathbb {R} \right\}\right)$

Bill Wvbailey (talk) 15:26, 26 July 2009 (UTC)

Ask, and it shall be given. Rick Norwood (talk) 12:16, 27 July 2009 (UTC)