Projective geometry extends affine geometry by introducing points at infinity, where any two distinct lines always intersect at exactly one point (eliminating parallelism exceptions). Projective space P(V) is defined as the set of one-dimensional vector subspaces of a vector space V, with dimension equal to dim(V) - 1. Points are represented using homogeneous coordinates [x₀:x₁:...:xₙ], which are defined up to scalar multiplication, or inhomogeneous coordinates on open subsets. This framework simplifies geometric theorems by removing special cases—such as conic classification depending on intersections with the line at infinity—and enables powerful proof techniques that lift geometric problems to linear algebra in the underlying vector space.
Algebraic Curves 1: Introduction to Projective Spaces | Oxford Mathematics
Added:[Music] Hello. Uh this is the algebraic curves course. Uh which is really a first course in algebraic geometry. Um so our subjects are going to be uh curves living in projective planes. in fact two-dimensional projective spaces primarily over the complex numbers but also making sense over other fields. Um I'm going to spend the first three lectures talking just about projective geometry before we explain what algebraic curves are. Um so projective spaces are a kind of space.
Um so lectures 1 to three will be about um projective geometry. So some reading for this would be the Hitchin lecture notes uh section one. Um I want to start off by telling you what projected spaces are. So these are some kind of generalizations of of the usual spaces R to the N which you're used to doing geometry in.
Um so let's talk about these. So this is some kind of generalization of uh or way of doing geometry using kind of linear algebra broadly. So we're going to work with vector spaces over some fields. So my field is going to be called f. So f be a field. Uh so usually for us uh f would be the real numbers or the complex numbers. Uh but uh you can also take um the fieldp uh of for p prime and so on.
Um okay. Uh so let's suppose V uh is a finite dimensional vector space uh over our field um f. Normally the dimension will be n + one.
Um and we define the projective space P of V um associated to V uh to be the set of one-dimensional vector subspaces of big V. So this is the place where our geometry is going to happen and I hope that by the end of the lecture you'll have some idea of why this is a sensible idea. Okay, so let's have some more notation.
uh for each nonzero vector little v in big v uh we write square brackets little v is the set of all multiples of v. So that's all lambda* v for lambda in our field f. Uh so then that's a point in p of v.
This is V is a basis vector for that onedimensional vector space. Um and um P of V is exactly the set of all point square brackets V uh for little V a non-zero vector in big V. Um and the point does not determine the vector little V uniquely. In fact, we have brackets v is brackets w if and only if v is lambda time w for some lambda in your field but lambda has to be non zero. So lambda a non-zero element to the field. So we can also write uh P of V is well V takeway zero brackets divided by well F star uh so by FAR I mean the field take away zero as a group under multiplication and the group action on this thing is just rescaling multiplication by lambda. Um so something to note about that is let's say this is a real or complex vector space then this is a quotient of well a subset of a real complex vector space that becomes a topological space this thing has a quotient topology okay so over r or c this thing becomes a topological space it's actually a very nice topological space it is compact and housed off it's something called a real or a complex manifold uh which is part of a very important story in geometry which um you may get to study later on in your degree. Anyway, we won't talk about that now. Um so we define the dimension of P of V uh to be the dimension of V minus one.
So the reason for the minus one is that well that thing here has a dimension of v but dividing by this one dimensional group um reduces dimension by one. Uh so if dimension of P of V is zero uh well in that case P of V is a point because there's only one dimensional vector sub there is only one dimensional subspace of one dimensional vector space which is a whole thing. Um uh so dimension P of V is one.
uh then p of v is what's called a uh projective line. Um if dimension of p of v is two then p of v is uh a projective plane. So projective planes are going to be particularly important in this course because we're going to define uh algebraic curves to be special subsets of projective planes defined as the zeros of some polinomial. Okay. So for example um P of R3 uh well you can take that as R3 takeway 0 divided by non-zero reals. Another way of doing that would be to say it's uh you could just take the the unit vectors in R3 and divide by the group preserving them which is plus or minus one. This is clearly a two dimensional object. Um P of R 2 is the circle divid plus - 1 which again is onedimensional.
Um so that's why subtracting one uh is a good thing to do. Okay.
Next, let's suppose that um big U is a vector subspace of our vector space big V. Um so a common notation for this is uh u less than or equal to v. So that's slightly stronger than saying u is a subset of v. It's a subset of v which is also a vector subspace.
Um then well it's just kind of obvious that P of U is a subset of P of V because a one-dimensional vector subspace of U is also a one dimensional vector subspace of V. So a point in there is a point in there.
Um and we say that um P of U is a projective linear subspace of P of V.
So that this is the first okay so first of all I've defined projector spaces P of V as a set. Now we know that they have a bunch of distinguished subsets and projective linear subspaces. Um and they're things we can do geometry with. Um so well a linear subspace um P of view with dimension of P of view is one is called a projective line in B of Um and well also a projected linear subspace P of U with um dimension of P of U is equal to dimension of P of V minus minus one. So something with dimension one less than the whole thing. This is called a hyper plane. Okay. Um so you should be seeing this is some kind of geometry generalization of geometry in R2 and R3.
In R2 and R3 we have notions of straight line.
Um the projected lines in projected spaces are kind of direct generalizations of um lines in R2 and R3 as we'll see. Um and some common notation will be uh r p to the n in black bold bold is a projective space of r to the n + one. Uh c pn is projective space of c to the n plus one when our field is complex numbers.
um and well PN or P lower N for um n dimensional projected space if you're not specifying your field. Uh so I prefer superscripts but I think the Hitchin writes P lower N in his notes.
All right, let's now talk about different notions of coordinates. So I have a section on coordinates on projected states.
So uh let's take V to be a finite dimensional vector space over our field f.
uh let's say dimension is n plus one and let's choose a basis. So things look nicer notationally if we start numbering from zero. So the basis will be v 0 v1 up to vn for v. So then um well every little v and big v uh can be written uniquely as uh v is a sum from i is zero up to n of x i vi um for x i in your field Um so let's make a definition.
Uh well if we let x0 in xn in our field to be not all zero.
Then we're going to write square brackets x0 up to xn. Um for the point uh p is brackets x0 v 0 plus plus xn vn in p of v.
So, it's important to use square brackets here because we're going to use round brackets to mean something different. Okay. So, well, every point in P of V can be written this way.
Uh so we call square brackets x0 up to xn homogeneous coordinates for the point p. So once we choose a basis for V, we have this notion of coordinates for points in P of V. Um so one thing to note is that um brackets x0 up to xn and brackets y0 up to yn uh represent the same point p uh if and only If well the vectors x0 v 0 plus up to xn vn and y 0 v 0 plus up to yn vn if know if these are proportional. Um so that is uh if and only if yj is lambda* xj uh for some lambda in your field take away zero and all j.
Um okay.
So homogeneous coordinates of a point are not unique except over the field with two elements. Um but can always be rescaled.
by some lambda in your field take away zero. So the word homogeneous is to do with rescalings. So you might talk about a homogeneous polomial to be one which um behaves in a certain way when you multiply all the coordinates in the polomial by a constant.
Okay, so that's our first notion of um coordinates. We'll have a second notion which is called inhomogeneous coordinates.
Um okay. Well, so now let P of V be a projected space of dimension N uh with uh homogeneous coordinates uh square brackets X0 0 up to xn and uh let u0 be the subset um of points um x0 up to xn in p of v uh with x0 non zero Um and then for a point in here we can put yj is xj / x0. The j is 1 to n. And then um x0 x1 up to xn is equal to the point 1 comma y1 the yn uh just rescaling by x0 inverse. Okay.
Um so we represent the point square brackets 1 comma y1 to yn in p of v by round brackets y1 up to yn without the leading Uh and we say that these round brackets y1 up to yn are in homogeneous coordinates. So homogeneous coordinates are kind of compatible with rescalings of the vectors and or the numbers. Inhomogeneous coordinates are not.
Um the inhomogeneous coordinates on the subject the subset U0 in P of V which is not the whole of P of V. Uh so inhomogeneous coordinates are coordinates uh of the usual sort.
Um so not like uh homogeneous coordinates and by the usual sort I mean well each n tuple of numbers determines a unique point.
Um so they give a one to one correspondence uh between our subset U0. It's an open subset if you're doing topology uh and the vector space f to the n.
Um and they are defined only on a subset u0 in P of V because if X0 is zero, you can't divide by it.
[Applause] Okay. So now actually we can start to get a bit more in intuition as to what projected spaces look like. Um because we got a projector space P of V.
Inside it we got a large subset which looks like a vector space F to the N. We can say what's what's the rest of P of V look like?
Um so now P of V takeway U0 is the set of points um brackets x0 x1 to xn with x0 equals z.
Uh so this is a projected subspace P of W where W is a vector subspace of V uh is the vector subspace spanned by uh the basis vectors V1 to VN but not V zero. So that P of W is a uh pro projected space of dimension n minus one which is one less than dimension w. Um okay so P of V is F to the N disjoint union P of W um so in particular let's say complex projective space CPN uh you can identify it with a copy of C to the N and a copy of CPN minus one. So that's your subset u0 where the x0 coordinate is non zero. That's your subset where the x0 coordinate is zero. Um and similarly real projector space rpn is an r the n together with a r pn minus one. In general, p to the n is f to the n together with a um a pn minus one. And of course, you can iterate. So then by inductive you could by induction you could write cpn as a copy of c to the n together with a copy of c to n minus one together with all the way down to the c to the zero which is a point.
[Applause] Um so um we think of the CPN minus one in CPN minus CPN as the points at infinity in Z to the Okay. Um, so CPN is a kind of if we're doing things with topologies, this is actually a compactification of C to the N. So C to the N is a non-compact topological space. We're adding some points, some extra points forming a copy of CPN minus one uh on which makes CPN into a compact topological space. And to get to a point here, you can take a limit of points in C to the N which are going off to infinity in some direction.
So really CPM minus one is parameterizing the complex directions in C to the N in which a point could go to infinity.
Um, okay. So, for example, um, RP1, this is an R together with infinity. Um, I'd like you to think about this as being um, uh, well, you can think about R as being a kind of straight line. um but really gets kind of rolled up. So um if you want to kind of identify a point in so r gets kind of rolled up into the component of a point in the circle and then you glue on infinity and infinity is the same as minus infinity. Uh you can get to this point infinity by going either to plus infinity or to minus infinity in r. Uh so the topological space RP1 is a circle.
Um whereas CP1 is C union infinity. Um this is a sphere topologically a two sphere. Um, think about infinity here. Maybe zero is at the bottom. You might have one i - one minus i, something like that. Uh, this is often called the reman sphere. Uh, I don't know if you've met it in complex analysis.
Um, so we we call uh r to the n and c to the n aphine space.
uh in comparison to the projected spaces um RPN and CPN um so then well so somehow aphine geometry let's say in R to the uh it's kind of geometry uh invariant under the group gl of n r but also extended by r the n um where these are the uh linear transformations And these are the translations. So this kind of bow tie thing is something called a semi-direct product. It means a a group is built from two other groups in such a way that the right hand group is a normal subgroup uh and that one is a quotient group. And it's a kind of twisted product uh because the linear transformations are kind of acting on the translations. But the you can see that there is a group of transformations of R to the N generated by the translations and the the linear transformations and um that's the you know geometry we care about uh in here. So that's I find geometry won't come up much but that's that's what the word really means. Um and this kind of aine is contrast projective. Okay. Uh so why is projective geometry a good idea?
Well, um it's a fact that things are often simpler in um projective geometry uh than in aphine geometry.
So you already know quite a lot of alphine geometry. You know all the stuff you've been doing with you know lines in the plane and triangles and things like that and hexagons um in school that's really aphine geometry happening in R2 and then there's maybe you know more geometry in R3.
Um so um just for example um in R2 we have a good notion of line um and two distinct lines um meet in a unique point. Um well, unless they are parallel. Okay, so that's one of Uklid's favorite facts. In contrast, in RP2, um, two distinct lines meet in a unique point. That's stop. Um, there's no exception for the parallel case.
Um and as a so parallel lines um in R2 um meeting a point at infinity in RP2. So uh you can see the the fine geometry has been kind of contained inside projective geometry. Um and then you tend to have when you got a theorem in in fine geometry you have to add in some kind of exceptions depending on how whatever it is you're talking about is interacting with line to infinity. If you just forget about the line infinity and working objective space then you have fewer exceptions.
um uh as well another example uh from later in the course.
Um you could look at conics in um RP2 or R2 So um in R2 you know you can get ellipses or you can get paraboly or you could get hyperbole.
Um I'm really talking about let's say non-s singular and non- empty conx. Um so in R in RP2 there's only one kind which looks a bit like an ellipse but then um the what you get in R in R2 depends upon how this conic is intersecting with a line to infinity.
So, um, uh, so uh, for the ellipse, the line to infinity misses it. For the parabola, the line to infinity hits it at one point. For the hyperola, the line to infinity hits it at two points. And therefore there's actually two chunks of curve uh in R2 corresponding to the two pieces of your hyperola. Okay. Um so okay so then the the R2 cases depend on how the conic intersects the line at infinity. Okay, so at the the cost of making things more abstract, we actually find that we get some our theorems become kind of simpler. Okay. Uh any questions?
Nope. All right. Um in that case let's move on to a section on examples of proofs in uh projective geometry.
So today I'll give you a relative relatively simple but non still non-trivial fact about lines in RP3. Um and tomorrow I'll give you a more much more non-trivial proof. Um so let's now prove the following. Um so let L1 L2 L3 be lines meaning projective lines in the projective space R P3 which is P of R4 that um do not intersect pairwise. So no two of these lines intersect. If you're making the same statement in R3, you have to say they don't intersect and also no two of the lines are parallel. The conclusion is that then uh there are an infinite number of lines in projective lines.
L in P of R3 R4 uh that intersect all three lines L1, L2 and L3.
[Applause] So the geometric picture is this. We're working in threedimensional space. You can think of it like ordinary R3 with some points added to infinity. We've got lines L1 and L2 and L3 which are somehow skew. And actually we're going to see that you can get a whole bunch of lines um like this.
Um and actually there's through each point of each of these lines there is a unique line um intersecting all three and passing through that point.
Um that that will come out of the proof. So we're going to prove a geometric fact. It's all about lines in space and how they intersect.
But the proof is going to be using linear algebra in not in the projective space but in the overlying vector space.
And that's kind of characteristic of a lot of proofs in projective geometry which is that you um you have some geometric fact but you take it up into the vector space and you do linear algebra. So let uh LJ J is a projected space of U.J. for J is 1 2 and three.
Um so then u1 and u2 and u3 are vector subspaces of r4 and as these are projective spaces dimension one um the J's have to mention one more which is two. Um so since L1 and L2 do not intersect their intersection is empty. The intersection of U1 and U2 well it can't be empty because vector spaces always contain zero.
That's the only point it's allowed to include because if there was a non-zero vector point in here that would then it's the line generated by that point would define a point in L1 and L2. So U2 U1 and U2 intersect in the zero vector space. Um so hence R4 is a direct sum of U1 and U2 just by counting dimensions.
So u and u2 they're two dimensional vector spaces they intersect in zero.
Their span is a fourdimensional vector space which has to be the whole thing because that also has dimension four. Um so let's take v in u3 be non zero.
Um so then because u3 is contained in R4 which is this direct sum we have that v is little u1 plus little u2. Uh with u1 is in big u1 and u2 is in big u2. Uh as u3 is contained in r4 which is u1 direct sum u2.
Okay. Well, so if little u1 is zero, then uh v would be u2. So the point square brackets v would lie in both l2 and l3. Uh which is a contradiction because we've assumed that the l that the lines do not intersect.
Um and if little u2 is zero then v lies in intersection of l1 and l3 which is also a contradiction. Um so hence u1 and u2 and v are all non zero. Also u1 and u2 are linearly independent because they lie in uh vector spaces with no intersection. So uh let u4 be the span of u1 and u2. It's a vector subspace of r4.
Um so then dimension of U4 is two and those two things are linear dependent. So little L4 which is a projective space of U4 um is a line a projective line in P of R4. All right. So we found ourselves a line. Um also L4 contains the point square brackets U1 which lies in P of U1 which is L1.
um and it contains a point brackets U2 which lies in L2 and it contains a point brackets V which is U1 + U2 and that lies in our line L3. Uh that's what we chose.
Um so L4 intersects L1 and L2 and L3.
So it's not equal to the line L3 because um it also contains U1, U2 which don't line L1, L3. Um, okay. So, we found one line intersecting all three of these.
Um, so um by construction uh there exists um such a line L uh through any point uh brackets V in L3. Um so uh there are infinitely many uh such lines because the points the map from points in L3 to lines is injective um and actually it's kind of sjective of well it's a one to one correspondence um and L3 is infinite because your field R is infinite so this this theorem this proposition is also true uh over any infinite fields um and that proves the proposition. Okay.
Um any questions so far? All right. Well um so let me just say that we have proved um a uh non-trivial geometric fact in objective space P of V um by using well linear algebra in the overlying vector space big V. And this is typical um of many proofs in projective geometry.
So you can get some mileage from being able to lift up um from your projective space up into a vector space of dimension one larger. Okay. And next time uh we will talk about the proof of Dezag's theorem which involves a seemingly complicated diagram with lots and lots of lines uh in a plane.
All right. If there are no questions, we can call the day there.
Up Next

Homogeneous Coordinates in Projective Geometry: Intro
@RichardSouthwell
26.2K views•2016-07-14

JEE Maths | Circles: Pole & Polar Explained with Equations
@TargetPublications
14.4K views•2019-02-23

Complex Numbers Lecture 2: Unit Circle, de Moivre's Theorem, Roots of Unity
@OxfordMathematics
2.5M views•2018-10-25

The Mathematical Impossibility of Accurate World Maps
@Vox
23.3M views•2016-12-02
Related Study Plans & Knowledge Roadmaps
Structured learning paths in Mathematics






























![Revolutionary Math Proof No One Could Explain...Until Now [Part 1]](https://i.ytimg.com/vi/RX1tZv_Nv4Y/hqdefault.jpg)



