Pretty much just the title. Obviously a vertical line isn't a function (doesn't pass the vertical line test), but technically it passes the horizontal line test for injectivity. But, I thought that an injective map needed to have distinct outputs corresponding with distinct inputs, which a vertical line doesn't have (multiple y outputs for our x input). Or does an injective map also have to be a function by definition?
-
4An injective map is a map, and a map is a function. – player3236 Sep 10 '20 at 17:06
-
1And a line is not a function, nor a map: it is a geometrical object. – Bernard Sep 10 '20 at 17:30
-
1A vertical line does represent a relation, and the injectivity property is certainly applicable to relations. However, although some authors use the phrase "injective relation" for such cases (even when the relation is not necessarily a function), usually another term is used (such as single-rooted or left unique), possibly to avoid readers accidentally thinking that a function is being considered. – Dave L. Renfro Sep 10 '20 at 17:47
-
So is a one-to-many mapping not a type of map? I guess not, even though I always thought it was. Oh well. I rarely see the term "map" or "mapping" anyway. Everyone says "function". – Adam Rubinson Sep 10 '20 at 20:25
3 Answers
As you noticed, a vertical line is not a function and therefore it is meaningless to discuss injectivity for it, which is defined using precisely the concept of map, that is
$$f(a)=f(b) \implies a=b$$
In some sense we could also claim that vertical line is surjective but also this statement is meaningless for the same reason.
Refer also to

- 154,566
-
Okay, thanks. Followup question: I was told that a map differs from a function as each input does not need to only return one output (i.e. many-to-one mapping). Using this definition, could a map be injective without being a function? – imemyhudson Sep 10 '20 at 17:35
-
When we deal with injectivity we use the term map as function. I’m not aware of its use in different context but of course we could extend the concept to other cases. – user Sep 10 '20 at 17:42
Injectivity like surjectivity is a property of functions and as you've noted the vertical line is not the graph of a function. That being said, I can define a function which has the image of a vertical line by a function from $\mathbb{R}$ to $\mathbb{R}^2$ parametrically by $x = c, y = t$ with $c$ a real constant and $t$ a real variable. In this context the line would be the image of an injective function. We can instead use $y=kt$ for some real $k\neq 0$ and this parameterization will also be an injective function with the same image, the vertical line through $x=c$.

- 11,018
- 1
- 12
- 29
I think it depends on what you define "injective" to mean on something that is not a function.
It could be argued that "injective" talks about mappings and if something isn't a function it makes no sense to talk about it being injective. Fair enough.
But you could also say that if $F \subset X\times Y$ then if $F$ such that if $(a,w)\in F$ and $(b,w)\in F$ then that is only possible of $a=b$ is a valid definition of injective. So if vertical line is $F = \{(c,y)|y\in \mathbb R\}$ and if $(a,w)\in F, (b,w)\in F$ then $a=b=c$ regardless of the $w$ so the line is injective. Fair enought.
But one can say the latin for injective means into so that means we chart everything into. (This would be compatible with bijective= injective and surjective; it'd be hard to say a vertical line is "bijective" when for and $x \ne c$ we can't have any $(x,\sim)$.) So we need that for if there is no $(m,w)\in F$ for some $m \in X$ then $m$ isn't charted at all. so the definition for injection should include that that aspect. So as a subset $\{c\}\times \mathbb R$ the line is injective but as a subset of $\mathbb R\times \mathbb R$ it isn't. Note: for this definition of injective $F= \{(x,y)| x = \tan(y)\}$ will be injective but certainly not a function. (For any $x$ there will be a $y = \arctan x$ so $(x, y)\in F$. And if $(a, y),(b,y) \in F$ then $a = b =\tan y$. But it's certainly not a function as $(x, \arctan x)$ and $(x, \pi + \arctan x)$ are both in $F$ so we don't have the single input/single output condition.)
I'd say for all practical purposes "injection" only refers to a function and if something isn't a function forget about it. After all, why make trouble.

- 124,253