Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have heard that the least dimension $m$ required for $\mathbb{R}P_2$ to be embedded in the Euclidean space is 4, thus I wanted to find an explicit formulae for it. I found two possible strategies, but is not sure that they'll work.

  1. Define $\phi([x_1,x_2,x_3])=(|x_1|,|x_2|,|x_3|,x_1x_2+x_2x_3+x_3x_1)$, where $[x_1,x_2,x_3]$ is the eq.class under quotient from $S^2$. I hope that would be an embedding, since the last is a symmetric polynomial which is equal for $(x_1,x_2,x_3)$ and $-(x_1,x_2,x_3)$.

  2. My second stragy is more geometric approach. Note that the projective plane deleted a circle is a Möbius band $M$. Thus if I could "paste" the boundary of a closed disk (via the fourth dimension) onto the boundary circle of a Möbius Band, then I'm done. But since I'm no good at "visualizing" four dimensions, I don't know exactly how to proceed.

My question is:

1) is 1. an embedding or not? (Or give another more elegant imbedding)

2) is there a way to realize my second approach? or is it hopeless?

3) Is there a more systematic way of doing such embeddings? (Frankly, If our world is in $\mathbb{R}_2$, I would probably not even be able to imagine how to imbed the torus into $\mathbb{R}_3$)

share|cite|improve this question
from… $f(x,y,z)=(xy,xz,y^2-z^2,2yz)$, $f:S^2\to\mathbb{R}^4$, satisfies $f(-x,-y,-z)=f(x,y,z)$ so factors through the real projective plane. – yoyo Dec 16 '11 at 15:14
Related:… – Bruno Stonek Dec 16 '11 at 23:17

An example of an embedding (real-algebraic, hence real analytic, hence $C^\infty$, hence topological) of $\mathbb P^2(\mathbb R)$ into $\mathbb R^4$ is given by $$i: \mathbb P^2(\mathbb R)\to \mathbb R^4: [x:y:z]\mapsto [xy:xz:yz:x^2+2y^2+3z^2]$$

It is however impossible to embed $\mathbb P^2(\mathbb R)$ into $\mathbb R^3$: the simplest reason is that a closed smooth surface in $\mathbb R^3$ is orientable, whereas $\mathbb P^2(\mathbb R)$ is well-known not to be orientable.

As a consolation, it is possible to immerse non injectively $\mathbb P^2(\mathbb R)$ into $\mathbb R^3$: the image variety is called Boy's surface (no misogyny here: the name honors the mathematician Werner Boy)

As an answer to Michael's comment let me mention that according to a celebrated theorem of Whitney, every compact $n$-dimensional smoooth manifold ($n\gt1$) can be immersed into $ \mathbb R^{2n-1}$ and embedded into $ \mathbb R^{2n}$.
However for $\mathbb P^n(\mathbb R)$ you sometimes get better results and this has been the object of much research. For example $\mathbb P^{15}(\mathbb R)$ can be embedded into $ \mathbb R^{24}$.
You will find many such results in an interesting table here .

share|cite|improve this answer
Hence I guess $\mathbb{P}^n(\mathbb{R})$ could be embedded similarly? Say, $[x_1,\dots,x_n]$ to $[x_1x_2,x_1x_3,\dots,{x_1}^2+\dots+n{x_n}^2]$, a total of $2n+1$ dimensions? Is $2n+1$ minimal here? Or is there a general conclusion for the minimal dimension required for an embedding? Thanks – Michael Luo Dec 20 '11 at 9:09
Sorry, $(n-1)n/2+1$ dimensions. I am inclined to think $\mathbb{P}^n\mathbb{R}$ can be embedded into $\mathbb{R}^{n+1}$. – Michael Luo Dec 20 '11 at 9:29
Dear Michael, I have added an edit in my answer addressing your comments. – Georges Elencwajg Dec 20 '11 at 12:23

You can do #2 by what are called "Movie moves". Think of $\mathbb{R}^4 = \mathbb{R}^3 \times \mathbb{R}$. Then in level $\mathbb{R}^3\times 0$ embed the Möbius band. Smoothly propagate the boundary of the Möbius band to $\mathbb{R}^3\times \epsilon$. Now, the boundary of the Möbius band is a smooth unknot so we can insert the isotopy to the standard unknot in $\mathbb{R}^3\times [\epsilon,1-\epsilon]$. Now cap off the unknot inside of $\mathbb{R}^3\times [1-\epsilon,1]$.

share|cite|improve this answer
I think I got it. Thanks! – Michael Luo Dec 20 '11 at 9:13

With regard to your 2nd strategy, you remark that you have trouble "visualizing" four dimensions.

One way to think about your 2nd strategy is as follows: you can immerse $\mathbb RP^2$ into $\mathbb R^3$, by gluing a disk and a Möbius strip together along their boudnaries. The problem, as you know, is that you get self-intersections. But now imagine actually carrying this out (say with rubber models of a disk and a Möbius strip), but plotting not just the resulting points in $\mathbb R^3$, but the times at which you placed each of the points of your model in their final positions. (So now you are plotting points in $\mathbb R^4$, not $\mathbb R^3$.)

The points which intersect will arrive at their final positions at different times, and so you will obtain an embedding of $\mathbb R P^2$ in $\mathbb R^4$.

More generally speaking, the old cliche of time being the "fourth dimension" can actually be very helpful for visualizing certain constructions in $\mathbb R^4$: you imagine making some geometric construction in $\mathbb R^3$, but then also imagine the construction as a process taking place over time; this then lets you "see" a construction in $\mathbb R^4$.

share|cite|improve this answer
Ah. I see your point! By assigning (continously) a time to parts of the surface and seeing it as a process rather than a concrete shape, one can eliminate the "intersections". This was very helpful! – Michael Luo Dec 20 '11 at 9:03
@Michael: Dear Michael, I'm glad it helped. Regards, – Matt E Dec 20 '11 at 13:15
An equally good fourth dimensional illustrator is hue. Draw a mobius strip which is green in the center turning to yellow on the edges, then attach a disk which is yellow on the edges turning to red in the center. The intersection points occupy the same physical space but have different color coordinates. Tada! – Eric Jan 16 '12 at 8:26

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.