Projective-Differential Geometric Synthesis

1. Introduction

Projective geometry emerged in the 17th century from the study of perspective in art and the need to understand how parallel lines appear to meet at the horizon. What began as a practical tool for artists evolved into one of the most elegant and fundamental areas of mathematics, with deep connections to algebra, topology, and differential geometry.

The key insight of projective geometry is the addition of "points at infinity" to complete the Euclidean plane and higher-dimensional spaces. This completion eliminates many of the special cases that complicate Euclidean geometry and reveals underlying symmetries and structures.

In this exposition, we will journey from the basic definitions of projective spaces through the sophisticated machinery of projective transformations, culminating in the natural transition to differential geometric concepts. Our approach emphasizes rigorous proofs while maintaining accessibility for readers with a solid background in linear algebra and real analysis.

2. Foundations of Projective Geometry

2.1. Projective Spaces

We begin with the fundamental definition of projective space, which provides the foundation for all subsequent development.

Definition 1. 

Let V be a vector space over a field F. The projective space $\mathbb{P}\left(V\right)$ associated to V is the set of all lines through the origin in V, i.e.,

$$\mathbb{P}\left(V\right)=\{L\subseteq V:L\mathit{is}a𝟣\mathit{-dimensional}\mathit{subspace}\mathit{of}V\}$$

Remark 1. 

We often denote $\mathbb{P}\left(V\right)$ as ${\mathbb{P}}^{n-1}$ when $dimV=n$, and write ${\mathbb{P}}^{n-1}\left(F\right)$ to emphasize the field. The most common cases are ${\mathbb{P}}^n\left(\mathbb{R}\right)={\mathbb{RP}}^n$ and ${\mathbb{P}}^n\left(\mathbb{C}\right)={\mathbb{CP}}^n$.

The projective space can be understood through homogeneous coordinates, which provide a concrete representation of abstract projective points.

Definition 2. 

Let $V=F^{n+1}$ where F is a field. A point in ${\mathbb{P}}^n\left(F\right)$ can be represented by homogeneous coordinates $[x_0:x_1:\cdots :x_n]$, where $(x_0,x_1,\dots ,x_n)\in F^{n+1}\setminus \left\{0\right\}$ and $[x_0:x_1:\cdots :x_n]=[y_0:y_1:\cdots :y_n]$ if and only if there exists $\lambda \in F^{*}$ such that $(y_0,y_1,\dots ,y_n)=\lambda (x_0,x_1,\dots ,x_n)$.

Example 1. 

In ${\mathbb{RP}}^2$, the point $[1:2:3]$ represents the same projective point as $[2:4:6]$ or $[-1:-2:-3]$, since these are all scalar multiples of each other.

2.2. The Relationship Between Affine and Projective Geometry

The connection between familiar Euclidean (affine) geometry and projective geometry is established through the concept of charts and the "line at infinity."

Theorem 1. 

Let ${\mathbb{P}}^n\left(F\right)$ be the n-dimensional projective space over field F. For each $i\in \{0,1,\dots ,n\}$, define

$$U_i=\{[x_0:\cdots :x_n]\in {\mathbb{P}}^n\left(F\right):x_i\ne 0\}$$

Then $U_i\cong F^n$ via the map

$${\varphi }_i:U_i\to F^n,[x_0:\cdots :x_n]\mapsto \left({\displaystyle \frac{x_0}{x_i}},\dots ,{\displaystyle \frac{x_{i-1}}{x_i}},{\displaystyle \frac{x_{i+1}}{x_i}},\dots ,{\displaystyle \frac{x_n}{x_i}}\right)$$

Proof. 

We prove this for $i=0$; the other cases follow by symmetry.

First, we show ${\varphi }_0$ is well-defined. If $[x_0:x_1:\cdots :x_n]=[y_0:y_1:\cdots :y_n]$, then there exists $\lambda \ne 0$ such that $(y_0,y_1,\dots ,y_n)=\lambda (x_0,x_1,\dots ,x_n)$. Since $x_0\ne 0$, we have $y_0=\lambda x_0\ne 0$. Then

$${\displaystyle \frac{y_j}{y_0}}={\displaystyle \frac{\lambda x_j}{\lambda x_0}}={\displaystyle \frac{x_j}{x_0}}$$

for all j, so ${\varphi }_0$ is well-defined.

Next, we construct the inverse map. Define ${\psi }_0:F^n\to U_0$ by

$${\psi }_0(t_1,\dots ,t_n)=[1:t_1:\cdots :t_n]$$

For any $(t_1,\dots ,t_n)\in F^n$:

$${\varphi }_0\left({\psi }_0(t_1,\dots ,t_n)\right)={\varphi }_0\left([1:t_1:\cdots :t_n]\right)=(t_1,\dots ,t_n)$$

For any $[x_0:x_1:\cdots :x_n]\in U_0$ with $x_0\ne 0$:

$${\psi }_0\left({\varphi }_0\left([x_0:x_1:\cdots :x_n]\right)\right)={\psi }_0\left({\displaystyle \frac{x_1}{x_0}},\dots ,{\displaystyle \frac{x_n}{x_0}}\right)=\left[1:{\displaystyle \frac{x_1}{x_0}}:\cdots :{\displaystyle \frac{x_n}{x_0}}\right]$$

This equals $[x_0:x_1:\cdots :x_n]$ since

$$\left[1:{\displaystyle \frac{x_1}{x_0}}:\cdots :{\displaystyle \frac{x_n}{x_0}}\right]=\left[{\displaystyle \frac{x_0}{x_0}}:{\displaystyle \frac{x_1}{x_0}}:\cdots :{\displaystyle \frac{x_n}{x_0}}\right]=[x_0:x_1:\cdots :x_n]$$

Therefore, ${\varphi }_0$ and ${\psi }_0$ are inverse bijections, establishing the isomorphism. □

Corollary 1. 

The projective space ${\mathbb{P}}^n\left(F\right)$ can be covered by $n+1$ affine charts, each isomorphic to $F^n$.

2.3. Points at Infinity

The "points at infinity" are precisely those points not contained in a given affine chart.

Definition 3. 

In ${\mathbb{P}}^n\left(F\right)$, the hyperplane at infinity with respect to the chart $U_0$ is

$$H_{\infty }=\{[x_0:x_1:\cdots :x_n]:x_0=0\}=\{[0:x_1:\cdots :x_n]:(x_1,\dots ,x_n)\ne (0,\dots ,0)\}$$

Theorem 2. 

$H_{\infty }\cong {\mathbb{P}}^{n-1}\left(F\right)$.

Proof. 

Define the map $\varphi :H_{\infty }\to {\mathbb{P}}^{n-1}\left(F\right)$ by

$$\varphi \left([0:x_1:\cdots :x_n]\right)=[x_1:\cdots :x_n]$$

This is well-defined since if $[0:x_1:\cdots :x_n]=[0:y_1:\cdots :y_n]$, then there exists $\lambda \ne 0$ such that $(0,y_1,\dots ,y_n)=\lambda (0,x_1,\dots ,x_n)=(0,\lambda x_1,\dots ,\lambda x_n)$. Thus $(y_1,\dots ,y_n)=\lambda (x_1,\dots ,x_n)$, which means $[x_1:\cdots :x_n]=[y_1:\cdots :y_n]$ in ${\mathbb{P}}^{n-1}\left(F\right)$.

The inverse map $\psi :{\mathbb{P}}^{n-1}\left(F\right)\to H_{\infty }$ is given by

$$\psi \left([y_1:\cdots :y_n]\right)=[0:y_1:\cdots :y_n]$$

It’s straightforward to verify that $\varphi$ and $\psi$ are inverse bijections, establishing the isomorphism. □

3. Projective Transformations

3.1. Linear Maps and Projective Maps

The symmetries of projective space are given by projective transformations, which arise naturally from linear algebra.

Definition 4. 

A projective transformation (or projective map ) of ${\mathbb{P}}^n\left(F\right)$ is a bijection $f:{\mathbb{P}}^n\left(F\right)\to {\mathbb{P}}^n\left(F\right)$ that preserves collinearity, i.e., if three points are collinear, then their images are also collinear.

The key theorem connecting linear algebra to projective geometry is the following:

Theorem 3. 

Every projective transformation of ${\mathbb{P}}^n\left(F\right)$ is induced by an invertible linear transformation of $F^{n+1}$.

Proof Sketch. 

The complete proof requires the fundamental theorem of projective geometry, which is quite technical. We outline the main ideas:

1. First, one shows that any projective transformation preserves harmonic division (cross-ratio). 2. Using this, one can prove that projective transformations are determined by their action on frames (sets of $n+2$ points in general position). 3. Finally, one shows that any map between frames can be realized by a linear transformation of the ambient vector space.

The full proof involves careful case analysis and is beyond the scope of this introduction, but can be found in standard references such as Hartshorne’s "Foundations of Projective Geometry." □

Definition 5. 

Let $T:F^{n+1}\to F^{n+1}$ be an invertible linear transformation. The induced projective transformation $\overline{T}:{\mathbb{P}}^n\left(F\right)\to {\mathbb{P}}^n\left(F\right)$ is defined by

$$\overline{T}\left([x_0:\cdots :x_n]\right)=\left[T(x_0,\dots ,x_n)\right]$$

where $\left[T(x_0,\dots ,x_n)\right]$ denotes the projective point determined by the vector $T(x_0,\dots ,x_n)$.

Lemma 1. 

The induced projective transformation $\overline{T}$ is well-defined.

Proof. 

Suppose $[x_0:\cdots :x_n]=[y_0:\cdots :y_n]$ in ${\mathbb{P}}^n\left(F\right)$. Then there exists $\lambda \ne 0$ such that $(y_0,\dots ,y_n)=\lambda (x_0,\dots ,x_n)$. Since T is linear:

$$T(y_0,\dots ,y_n)=T\left(\lambda (x_0,\dots ,x_n)\right)=\lambda T(x_0,\dots ,x_n)$$

Therefore, $\left[T(y_0,\dots ,y_n)\right]=\left[T(x_0,\dots ,x_n)\right]$ in ${\mathbb{P}}^n\left(F\right)$, showing that $\overline{T}$ is well-defined. □

3.2. The Projective General Linear Group

Definition 6. 

The projective general linear group ${\mathit{PGL}}_{n+1}\left(F\right)$ is the group of all projective transformations of ${\mathbb{P}}^n\left(F\right)$. It is isomorphic to ${\mathit{GL}}_{n+1}\left(F\right)/Z\left({\mathit{GL}}_{n+1}\left(F\right)\right)$, where $Z\left({\mathit{GL}}_{n+1}\left(F\right)\right)$ is the center of ${\mathit{GL}}_{n+1}\left(F\right)$ (the scalar matrices).

Theorem 4. 

${\mathit{PGL}}_{n+1}\left(F\right)\cong {\mathit{GL}}_{n+1}\left(F\right)/F^{*}$, where $F^{*}$ denotes the group of scalar matrices $\{\lambda I:\lambda \in F^{*}\}$.

Proof. 

Consider the natural map $\pi :{\mathrm{GL}}_{n+1}\left(F\right)\to {\mathrm{PGL}}_{n+1}\left(F\right)$ that sends a matrix A to the induced projective transformation $\overline{A}$.

First, we show $\pi$ is a homomorphism. For matrices $A,B\in {\mathrm{GL}}_{n+1}\left(F\right)$:

$$\overline{AB}\left(\left[x\right]\right)=\left[\left(AB\right)\left(x\right)\right]=\left[A\left(B\left(x\right)\right)\right]=\overline{A}\left(\left[B\left(x\right)\right]\right)=\overline{A}\left(\overline{B}\left(\left[x\right]\right)\right)=(\overline{A}\circ \overline{B})\left(\left[x\right]\right)$$

So $\overline{AB}=\overline{A}\circ \overline{B}$, confirming $\pi$ is a homomorphism.

Next, we determine $ker\left(\pi \right)$. We have $A\in ker\left(\pi \right)$ if and only if $\overline{A}={\mathrm{id}}_{{\mathbb{P}}^n\left(F\right)}$, which occurs if and only if $A\left(x\right)$ and x represent the same projective point for all $x\ne 0$. This happens precisely when $A=\lambda I$ for some $\lambda \in F^{*}$.

Therefore, $ker\left(\pi \right)=F^{*}=\{\lambda I:\lambda \in F^{*}\}$.

Finally, we show $\pi$ is surjective. By the fundamental theorem of projective geometry, every projective transformation is induced by some linear transformation, so $\pi$ is onto.

By the first isomorphism theorem, ${\mathrm{PGL}}_{n+1}\left(F\right)\cong {\mathrm{GL}}_{n+1}\left(F\right)/ker\left(\pi \right)={\mathrm{GL}}_{n+1}\left(F\right)/F^{*}$. □

3.3. Cross-Ratio and Projective Invariants

One of the most important projective invariants is the cross-ratio, which measures the relative position of four collinear points.

Definition 7. 

Let $A,B,C,D$ be four distinct points on a projective line ${\mathbb{P}}^1\left(F\right)$. The cross-ratio of these points is

$$(A,B;C,D)={\displaystyle \frac{AC·BD}{AD·BC}}$$

where the ratios are computed in any affine chart containing all four points.

Theorem 5. 

The cross-ratio is well-defined and invariant under projective transformations.

Proof. 

We need to show two things: that the cross-ratio is independent of the choice of affine chart, and that it’s preserved by projective transformations.

Well-definedness: Suppose we have four points $A,B,C,D$ on ${\mathbb{P}}^1\left(F\right)$ with homogeneous coordinates $[a_0:a_1],[b_0:b_1],[c_0:c_1],[d_0:d_1]$ respectively.

In the affine chart $U_0=\{[x_0:x_1]:x_0\ne 0\}$, these points correspond to $a_1/a_0,b_1/b_0,c_1/c_0,d_1/d_0$ respectively (assuming all are in this chart). The cross-ratio is:

$$(A,B;C,D)={\displaystyle \frac{(c_1/c_0-a_1/a_0)(d_1/d_0-b_1/b_0)}{(d_1/d_0-a_1/a_0)(c_1/c_0-b_1/b_0)}}$$

After algebraic manipulation using the determinant formula, this can be shown to equal:

$$(A,B;C,D)={\displaystyle \frac{det(A,C)·det(B,D)}{det(A,D)·det(B,C)}}$$

where $det(P,Q)=det\left(\begin{array}{cc}{p}_0& q_0\\ p_1& q_1\end{array}\right)$ for points $P=[p_0:p_1]$ and $Q=[q_0:q_1]$.

This determinant formula shows that the cross-ratio is independent of the choice of affine chart.

Invariance: Let $T\in {\mathrm{PGL}}_2\left(F\right)$ be represented by a matrix $M\in {\mathrm{GL}}_2\left(F\right)$. For any four points $A,B,C,D$:

$$\begin{array}{cc}\hfill \left(T\right(A),T(B);T(C),T(D\left)\right)& ={\displaystyle \frac{det\left(T\right(A),T(C\left)\right)·det\left(T\right(B),T(D\left)\right)}{det\left(T\right(A),T(D\left)\right)·det\left(T\right(B),T(C\left)\right)}}\hfill \end{array}$$

(1)

$$\begin{array}{cc}\hfill & ={\displaystyle \frac{det\left(M\right)·det(A,C)·det\left(M\right)·det(B,D)}{det\left(M\right)·det(A,D)·det\left(M\right)·det(B,C)}}\hfill \end{array}$$

(2)

$$\begin{array}{cc}\hfill & ={\displaystyle \frac{det(A,C)·det(B,D)}{det(A,D)·det(B,C)}}\hfill \end{array}$$

(3)

$$\begin{array}{cc}\hfill & =(A,B;C,D)\hfill \end{array}$$

(4)

Therefore, the cross-ratio is invariant under projective transformations. □

4. Duality in Projective Geometry

4.1. The Principle of Duality

One of the most elegant aspects of projective geometry is the principle of duality, which establishes a symmetric relationship between points and hyperplanes.

Definition 8. 

In ${\mathbb{P}}^n\left(F\right)$, a hyperplane is a set of the form

$$H=\{[x_0:\cdots :x_n]:a_0{x}_0+\cdots +a_n{x}_n=0\}$$

where $(a_0,\dots ,a_n)\ne (0,\dots ,0)$. We denote this hyperplane by ${[a_0:\cdots :a_n]}^{*}$.

Theorem 6 

(Projective Duality). There is a natural bijection between points in ${\mathbb{P}}^n\left(F\right)$ and hyperplanes in ${\mathbb{P}}^n\left(F\right)$.

Proof. 

Define the map $\delta :{\mathbb{P}}^n\left(F\right)\to \left\{\mathrm{hyperplanes}\mathrm{in}{\mathbb{P}}^n\left(F\right)\right\}$ by

$$\delta \left([x_0:\cdots :x_n]\right)=\{[y_0:\cdots :y_n]:x_0{y}_0+\cdots +x_n{y}_n=0\}$$

This is well-defined: if $[x_0:\cdots :x_n]=[x_0^{\prime }:\cdots :x_n^{\prime }]$, then $(x_0^{\prime },\dots ,x_n^{\prime })=\lambda (x_0,\dots ,x_n)$ for some $\lambda \ne 0$. The corresponding hyperplane is

$$\{[y_0:\cdots :y_n]:\lambda (x_0{y}_0+\cdots +x_n{y}_n)=0\}=\{[y_0:\cdots :y_n]:x_0{y}_0+\cdots +x_n{y}_n=0\}$$

So $\delta$ is well-defined.

The inverse map ${\delta }^{-1}$ sends a hyperplane $H=\{[y_0:\cdots :y_n]:a_0{y}_0+\cdots +a_n{y}_n=0\}$ to the point $[a_0:\cdots :a_n]$.

It’s straightforward to verify that $\delta$ and ${\delta }^{-1}$ are indeed inverse bijections. □

4.2. Incidence Relations

Theorem 7. 

A point $P=[x_0:\cdots :x_n]$ lies on a hyperplane $H={[a_0:\cdots :a_n]}^{*}$ if and only if $a_0{x}_0+\cdots +a_n{x}_n=0$.

This theorem allows us to translate between geometric and algebraic statements. For example:

Corollary 2 

(Duality Principle). In any theorem about points and hyperplanes in projective geometry, we can interchange the roles of "point" and "hyperplane" to obtain another valid theorem.

5. Conics and Quadrics

5.1. Projective Conics

Definition 9. 

A conic in ${\mathbb{P}}^2\left(F\right)$ is the zero set of a homogeneous quadratic polynomial:

$$C=\{[x_0:x_1:x_2]:a{x}_0^2+b{x}_1^2+c{x}_2^2+d{x}_0{x}_1+e{x}_0{x}_2+f{x}_1{x}_2=0\}$$

where not all coefficients are zero.

Equivalently, a conic can be defined using matrices:

Definition 10. 

A conic in ${\mathbb{P}}^2\left(F\right)$ is the zero set of a quadratic form ${\mathbf{x}}^{T}Q\mathbf{x}=0$, where Q is a $3\times 3$ symmetric matrix and $\mathbf{x}={(x_0,x_1,x_2)}^T$.

Theorem 8. 

Every non-degenerate conic in ${\mathbb{P}}^2\left(\mathbb{C}\right)$ is projectively equivalent to the conic $x_0^2+x_1^2+x_2^2=0$.

Proof. 

Let C be a conic defined by ${\mathbf{x}}^{T}Q\mathbf{x}=0$ where Q is non-degenerate (i.e., $detQ\ne 0$).

Since Q is symmetric, it can be diagonalized over $\mathbb{C}$. That is, there exists an invertible matrix P such that $P^{T}QP=D$ where D is diagonal with non-zero entries $d_1,d_2,d_3$.

The change of coordinates $\mathbf{y}=P^{-1}\mathbf{x}$ transforms the conic to ${\mathbf{y}}^{T}D\mathbf{y}=0$, or

$$d_1{y}_0^2+d_2{y}_1^2+d_3{y}_2^2=0$$

Since we’re working over $\mathbb{C}$, we can further transform coordinates by scaling: set $z_i=\sqrt{|d_i|}{y}_i$ if $d_i>0$ and $z_i=i\sqrt{|d_i|}{y}_i$ if $d_i<0$. This transforms the equation to $\pm z_0^2\pm z_1^2\pm z_2^2=0$.

Finally, by possibly changing signs, we can achieve the standard form $z_0^2+z_1^2+z_2^2=0$. □

5.2. The Dual of a Conic

Definition 11. 

Let C be a conic in ${\mathbb{P}}^2\left(F\right)$ defined by ${\mathbf{x}}^{T}Q\mathbf{x}=0$. The dual conic $C^{*}$ is the set of all lines tangent to C.

Theorem 9. 

If C is defined by ${\mathbf{x}}^{T}Q\mathbf{x}=0$ with Q non-degenerate, then the dual conic $C^{*}$ is defined by ${\mathbf{l}}^T{Q}^{-1}\mathbf{l}=0$, where $\mathbf{l}={(l_0,l_1,l_2)}^T$ represents a line $l_0{x}_0+l_1{x}_1+l_2{x}_2=0$.

Proof. 

A line ${\mathbf{l}}^T\mathbf{x}=0$ is tangent to the conic ${\mathbf{x}}^{T}Q\mathbf{x}=0$ if and only if the system

$$\begin{array}{cc}\hfill {\mathbf{x}}^{T}Q\mathbf{x}& =0\hfill \end{array}$$

(5)

$$\begin{array}{cc}\hfill {\mathbf{l}}^T\mathbf{x}& =0\hfill \end{array}$$

(6)

has exactly one solution (up to scaling).

Using Lagrange multipliers, the tangency condition is equivalent to the existence of a scalar $\lambda$ such that $2Q\mathbf{x}=\lambda \mathbf{l}$, or $\mathbf{x}={\displaystyle \frac{\lambda }{2}}{Q}^{-1}\mathbf{l}$ (assuming Q is invertible).

Substituting back into the conic equation:

$${\left({\displaystyle \frac{\lambda }{2}}{Q}^{-1}\mathbf{l}\right)}^{T}Q\left({\displaystyle \frac{\lambda }{2}}{Q}^{-1}\mathbf{l}\right)=0$$

$${\displaystyle \frac{{\lambda }^2}{4}}{\mathbf{l}}^T{Q}^{-1}\mathbf{l}=0$$

Since we need a non-trivial solution ($\lambda \ne 0$), we must have ${\mathbf{l}}^T{Q}^{-1}\mathbf{l}=0$.

Therefore, the dual conic $C^{*}$ is indeed defined by ${\mathbf{l}}^T{Q}^{-1}\mathbf{l}=0$. □

6. Transition to Differential Geometry

6.1. Projective Structures and Local Coordinates

The transition from projective to differential geometry begins with the observation that projective transformations, when restricted to affine charts, become rational functions. This provides a natural setting for introducing differential structures.

Definition 12. 

A projective atlas on a manifold M is a collection of charts $\left\{(U_i,{\varphi }_i)\right\}$ such that the transition maps ${\varphi }_j\circ {\varphi }_i^{-1}$ are restrictions of projective transformations.

Theorem 10. 

${\mathbb{P}}^n\left(\mathbb{R}\right)$ has a natural smooth manifold structure of dimension n.

Proof. 

We use the affine charts constructed earlier. Recall that $U_i=\{[x_0:\cdots :x_n]:x_i\ne 0\}$ and

$${\varphi }_i:U_i\to {\mathbb{R}}^n,[x_0:\cdots :x_n]\mapsto \left({\displaystyle \frac{x_0}{x_i}},\dots ,\widehat{{\displaystyle \frac{x_i}{x_i}}},\dots ,{\displaystyle \frac{x_n}{x_i}}\right)$$

The transition maps are given by rational functions. For example, the transition map ${\varphi }_1\circ {\varphi }_0^{-1}:{\varphi }_0(U_0\cap U_1)\to {\varphi }_1(U_0\cap U_1)$ is computed as follows:

Let $(t_1,\dots ,t_n)\in {\varphi }_0(U_0\cap U_1)$. This corresponds to the point $[1:t_1:\cdots :t_n]\in U_0\cap U_1$. For this point to be in $U_1$, we need $t_1\ne 0$.

Then ${\varphi }_1\left([1:t_1:\cdots :t_n]\right)=\left({\displaystyle \frac{1}{t_1}},{\displaystyle \frac{t_2}{t_1}},\dots ,{\displaystyle \frac{t_n}{t_1}}\right)$.

Therefore, ${\varphi }_1\circ {\varphi }_0^{-1}(t_1,\dots ,t_n)=\left({\displaystyle \frac{1}{t_1}},{\displaystyle \frac{t_2}{t_1}},\dots ,{\displaystyle \frac{t_n}{t_1}}\right)$.

This is a smooth map wherever $t_1\ne 0$, which is precisely the domain ${\varphi }_0(U_0\cap U_1)$.

Since all transition maps are smooth, ${\mathbb{P}}^n\left(\mathbb{R}\right)$ has a natural smooth manifold structure. □

6.2. Tangent Spaces and Vector Fields

The differential geometric structure of projective space becomes apparent when we examine tangent spaces and vector fields.

Definition 13. 

Let M be a smooth manifold and $p\in M$. The tangent space $T_{p}M$ is the vector space of all tangent vectors at p, which can be defined as derivations of the ring of germs of smooth functions at p.

Theorem 11. 

For ${\mathbb{P}}^n\left(\mathbb{R}\right)$, we have $dim{T}_p{\mathbb{P}}^n\left(\mathbb{R}\right)=n$ for any point p.

Proof. 

Let $p=[x_0:\cdots :x_n]\in {\mathbb{P}}^n\left(\mathbb{R}\right)$ and assume $x_0\ne 0$ so that $p\in U_0$. In the chart $(U_0,{\varphi }_0)$, the point p corresponds to $\left({\displaystyle \frac{x_1}{x_0}},\dots ,{\displaystyle \frac{x_n}{x_0}}\right)\in {\mathbb{R}}^n$.

The tangent space $T_p{\mathbb{P}}^n\left(\mathbb{R}\right)$ is isomorphic to $T_{{\varphi }_0\left(p\right)}{\mathbb{R}}^n\cong {\mathbb{R}}^n$ via the differential of the chart map.

Therefore, $dim{T}_p{\mathbb{P}}^n\left(\mathbb{R}\right)=n$. □

6.3. The Fubini-Study Metric

One of the most important connections between projective and differential geometry is the Fubini-Study metric on complex projective space.

Definition 14. 

The Fubini-Study metric on ${\mathbb{CP}}^n$ is the Riemannian metric whose restriction to each affine chart $U_j$ is given by $d{s}^2={\displaystyle \frac{{\sum }_{k\ne j}{\left|d{z}_k\right|}^2\left({\sum }_{l\ne j}{\left|z_l\right|}^2+1\right)-{\left|{\sum }_{k\ne j}\overline{z_k}d{z}_k\right|}^2}{{\left({\sum }_{l\ne j}{\left|z_l\right|}^2+1\right)}^2}}$ where ${\left(z_k\right)}_{k\ne j}$ are the affine coordinates in $U_j$.

Theorem 12. 

The Fubini-Study metric is well-defined and has constant holomorphic sectional curvature equal to 4.

Proof Outline. 

The proof involves several steps:

1. Well-definedness: One must verify that the metric definitions on overlapping charts are compatible. This follows from the invariance of the metric under the action of $U(n+1)$.

2. Curvature computation: The holomorphic sectional curvature can be computed using the standard formulas for Kähler metrics. The Fubini-Study metric arises as the quotient of the flat metric on ${\mathbb{C}}^{n+1}\setminus \left\{0\right\}$ by the ${\mathbb{C}}^{*}$ action.

3. Constant curvature: The homogeneity of the construction (invariance under $U(n+1)$) implies that the curvature is constant.

The detailed computation involves substantial calculation and is typically covered in specialized texts on Kähler geometry. □

6.4. Chern Classes and Characteristic Classes

The projective space ${\mathbb{CP}}^n$ serves as a fundamental example in the theory of characteristic classes.

Definition 15. 

Let $E\to M$ be a complex vector bundle. The Chern classes $c_i\left(E\right)\in H^{2i}(M;\mathbb{Z})$ are characteristic classes that measure the "twisting" of the bundle.

Theorem 13. 

For the tautological line bundle $\mathcal{O}(-1)\to {\mathbb{CP}}^n$, we have $c_1\left(\mathcal{O}(-1)\right)=-h$, where h is the generator of $H^2({\mathbb{CP}}^n;\mathbb{Z})\cong \mathbb{Z}$.

Proof Sketch. 

The tautological bundle $\mathcal{O}(-1)$ is defined as the subbundle of ${\mathbb{CP}}^n\times {\mathbb{C}}^{n+1}$ consisting of pairs $(L,v)$ where L is a line through the origin and $v\in L$.

The first Chern class can be computed using the connection induced from the flat connection on ${\mathbb{C}}^{n+1}$. The curvature form, when integrated over any line ${\mathbb{CP}}^1\subset {\mathbb{CP}}^n$, gives $-1$.

Since $H^2({\mathbb{CP}}^n;\mathbb{Z})\cong \mathbb{Z}$ is generated by the class of a line, we have $c_1\left(\mathcal{O}(-1)\right)=-h$. □

7. Advanced Topics

7.1. Grassmannians and Schubert Calculus

The ideas of projective geometry extend naturally to Grassmannians, which parametrize linear subspaces.

Definition 16. 

The Grassmannian $\mathrm{Gr}(k,n)$ is the set of all k-dimensional linear subspaces of ${\mathbb{C}}^n$ (or ${\mathbb{R}}^n$).

Theorem 14. 

$\mathit{Gr}(k,n)$ has a natural structure as a smooth manifold of dimension $k(n-k)$.

Proof Sketch. 

The Grassmannian can be realized as a quotient of the Stiefel manifold $V_{k,n}$ (the space of orthonormal k-frames in ${\mathbb{R}}^n$) by the orthogonal group $O\left(k\right)$.

Since $dim{V}_{k,n}=nk-{\displaystyle \frac{k(k+1)}{2}}$ and $dimO\left(k\right)={\displaystyle \frac{k(k-1)}{2}}$, we have $dim\mathrm{Gr}(k,n)=nk-{\displaystyle \frac{k(k+1)}{2}}-{\displaystyle \frac{k(k-1)}{2}}=nk-k^2=k(n-k)$

The smooth manifold structure comes from the quotient construction and local charts via Plücker coordinates. □

7.2. Algebraic Curves and Riemann Surfaces

Projective geometry provides the natural setting for studying algebraic curves, which become Riemann surfaces when considered over $\mathbb{C}$.

Definition 17. 

A smooth projective curve of genus g is a compact Riemann surface that can be embedded in some projective space ${\mathbb{CP}}^n$.

Theorem 15 

(Riemann-Roch Theorem). Let C be a smooth projective curve of genus g and let D be a divisor on C. Then $dim{H}^0(C,\mathcal{O}\left(D\right))-dim{H}^1(C,\mathcal{O}\left(D\right))=degD+1-g$

This theorem connects the geometric properties of the curve (its genus) with algebraic properties (dimensions of cohomology groups).

7.3. Moduli Spaces

The study of families of geometric objects leads naturally to moduli spaces, which parametrize geometric structures up to equivalence.

Definition 18. 

The moduli space ${\mathcal{M}}_g$ of curves of genus g is the space parametrizing isomorphism classes of smooth projective curves of genus g.

Theorem 16. 

For $g\ge 2$, the moduli space ${\mathcal{M}}_g$ has dimension $3g-3$.

Proof Outline. 

This follows from deformation theory. A curve of genus g has $3g-3$ complex moduli, which can be seen by:

1. Embedding the curve in ${\mathbb{CP}}^{5g-5}$ via the canonical map 2. Counting the dimension of the space of such embeddings 3. Subtracting the dimension of the automorphism group

The precise argument requires substantial machinery from algebraic geometry and deformation theory. □

8. Applications to Modern Geometry

8.1. Mirror Symmetry

Projective geometry plays a crucial role in mirror symmetry, a duality between Calabi-Yau manifolds that has deep implications for both mathematics and physics.

Definition 19. 

Two Calabi-Yau threefolds X and Y are called mirror partners if there exists an isomorphism between their Hodge diamonds that exchanges $h^{p,q}\left(X\right)$ and $h^{q,p}\left(Y\right)$.

The construction of mirror pairs often involves toric geometry, which builds on projective geometric ideas:

Theorem 17 

(Batyrev Construction). Let Δ and ${\Delta }^{*}$ be dual reflexive polytopes in ${\mathbb{R}}^4$. Then the corresponding toric varieties $X_{\Delta }$ and $X_{{\Delta }^{*}}$ are mirror Calabi-Yau threefolds.

8.2. Geometric Invariant Theory

The quotient constructions that appear throughout projective geometry are formalized in Geometric Invariant Theory (GIT).

Definition 20. 

Let G be a reductive group acting on a projective variety X. A point $x\in X$ is called GIT-stable if its orbit closure does not contain 0 and its stabilizer is finite.

Theorem 18 

(GIT Quotient Theorem). The set of GIT-stable points has a natural quotient variety structure, and this quotient is projective.

This theorem provides a systematic way to construct moduli spaces as quotients, generalizing classical constructions in projective geometry.

9. Conclusion

The mathematical landscape reveals itself through unexpected connections and profound unifying principles. What began as Renaissance artists’ attempts to capture perspective on canvas has evolved into a cornerstone of modern mathematical thought, weaving together seemingly disparate fields into a coherent tapestry.

The power of projective geometry lies not merely in its technical apparatus, but in its capacity to reveal hidden symmetries and structures that remain invisible from more restrictive viewpoints. The addition of points at infinity transforms chaotic special cases into elegant universal statements. Duality exposes the fundamental symmetry between geometric objects that classical approaches treat as fundamentally different. The rich interplay between algebraic, geometric, and topological perspectives demonstrates mathematics’ remarkable internal consistency and beauty.

Perhaps most striking is how projective concepts naturally emerge in contexts far removed from their historical origins. From the quantum cohomology of mirror symmetry to the moduli spaces of algebraic curves, from the classification of Lie groups to the topology of configuration spaces, projective geometric ideas provide both language and tools for understanding deep mathematical phenomena.

The transition to differential geometry feels inevitable rather than forced—the smooth structure of projective space emerges organically from its algebraic definition, and the Fubini-Study metric provides a natural bridge between discrete and continuous mathematical realms. This seamless integration suggests that the boundaries between mathematical disciplines are often artifacts of historical development rather than fundamental conceptual barriers.

Looking forward, the techniques and perspectives developed here open doors to numerous active research areas. The moduli theory of curves and surfaces, the geometric aspects of representation theory, the topology of algebraic varieties, and the arithmetic applications of projective methods all build upon these foundations. Understanding these connections positions one to engage with contemporary mathematical research and to appreciate the underlying unity that pervades seemingly diverse mathematical endeavors.

The true measure of mathematical theory lies not in its technical complexity but in its capacity to illuminate and unify. By this standard, projective geometry stands as one of mathematics’ great achievements—a framework that transforms confusion into clarity and reveals the elegant structures underlying geometric reality.

Appendix A.1. Vector Spaces and Linear Maps

For completeness, we review the essential linear algebra concepts used throughout this exposition.

Definition A1. 

A vector space over a field F is a set V with operations of addition and scalar multiplication satisfying:

1. 

$(V,+)$ is an abelian group

2. 

Scalar multiplication is associative: $\alpha \left(\beta v\right)=\left(\alpha \beta \right)v$

3. 

Distributive laws: $\alpha (u+v)=\alpha u+\alpha v$ and $(\alpha +\beta )v=\alpha v+\beta v$

4. 

Identity: $1·v=v$ for all $v\in V$

Definition A2. 

A linear map $T:V\to W$ between vector spaces satisfies:

1. 

$T(u+v)=T\left(u\right)+T\left(v\right)$ for all $u,v\in V$

2. 

$T\left(\alpha v\right)=\alpha T\left(v\right)$ for all $\alpha \in F,v\in V$

Appendix A.2. Quotient Spaces

Definition A3. 

Let V be a vector space and $W\subseteq V$ a subspace. The quotient space $V/W$ is the set of equivalence classes $\{v+W:v\in V\}$ where $v+W=\{v+w:w\in W\}$.

Theorem A1. 

$V/W$ is a vector space with operations:

$$\begin{array}{cc}\hfill (v_1+W)+(v_2+W)& =(v_1+v_2)+W\hfill \end{array}$$

(A1)

$$\begin{array}{cc}\hfill \alpha (v+W)& =\left(\alpha v\right)+W\hfill \end{array}$$

(A2)

Appendix A.3. Matrix Groups

Definition A4. 

The general linear group ${\mathit{GL}}_n\left(F\right)$ is the group of invertible $n\times n$ matrices over field F.

Definition A5. 

The special linear group ${\mathit{SL}}_n\left(F\right)=\{A\in {\mathit{GL}}_n\left(F\right):detA=1\}$.

Appendix B.1. Topological Spaces

Definition A6. 

A topological space is a pair $(X,\mathcal{T})$ where X is a set and $\mathcal{T}$ is a collection of subsets (called open sets) satisfying:

1. 

$\varnothing ,X\in \mathcal{T}$

2. 

Arbitrary unions of sets in $\mathcal{T}$ are in $\mathcal{T}$

3. 

Finite intersections of sets in $\mathcal{T}$ are in $\mathcal{T}$

Appendix B.2. Smooth Manifolds

Definition A7. 

A smooth manifold of dimension n is a topological space M that is locally homeomorphic to ${\mathbb{R}}^n$ and equipped with a smooth atlas.

Definition A8. 

An atlas for a topological space M is a collection $\left\{(U_i,{\varphi }_i)\right\}$ where:

1. 

Each $U_i\subseteq M$ is open and ${\bigcup }_i{U}_i=M$

2. 

Each ${\varphi }_i:U_i\to V_i\subseteq {\mathbb{R}}^n$ is a homeomorphism

3. 

For overlapping charts, ${\varphi }_j\circ {\varphi }_i^{-1}:{\varphi }_i(U_i\cap U_j)\to {\varphi }_j(U_i\cap U_j)$ is smooth

Appendix C.1. Affine and Projective Varieties

Definition A9. 

An affine variety over field F is the zero set of a collection of polynomials in $F[x_1,\dots ,x_n]$.

Definition A10. 

A projective variety over field F is the zero set of a collection of homogeneous polynomials in $F[x_0,x_1,\dots ,x_n]$, considered as a subset of ${\mathbb{P}}^n\left(F\right)$.

Appendix C.2. Sheaves and Cohomology

Definition A11. 

A presheaf $\mathcal{F}$ of abelian groups on a topological space X assigns to each open set $U\subseteq X$ an abelian group $\mathcal{F}\left(U\right)$ and to each inclusion $V\subseteq U$ a restriction map ${\rho }_{U,V}:\mathcal{F}\left(U\right)\to \mathcal{F}\left(V\right)$.

Definition A12. 

A presheaf $\mathcal{F}$ is a sheaf if it satisfies the gluing axiom: for any open cover $\left\{U_i\right\}$ of an open set U, sections that agree on overlaps can be uniquely glued to a global section.

Appendix D.1. Projective Geometry

• Hartshorne, R. Foundations of Projective Geometry
• Baer, R. Linear Algebra and Projective Geometry
• Coxeter, H.S.M. Projective Geometry

Appendix D.2. Differential Geometry

• Lee, J. Introduction to Smooth Manifolds
• Spivak, M. Differential Geometry
• Kobayashi, S. and Nomizu, K. Foundations of Differential Geometry

Appendix D.3. Algebraic Geometry

• Hartshorne, R. Algebraic Geometry
• Shafarevich, I.R. Basic Algebraic Geometry
• Griffiths, P. and Harris, J. Principles of Algebraic Geometry

Appendix D.4. Advanced Topics

• Mumford, D. Geometric Invariant Theory
• Hori, K. et al. Mirror Symmetry
• Fulton, W. Intersection Theory