In differential geometry, a projective connection is a geometric structure on a differentiable manifold that specifies a distinguished class of curves, called geodesics, up to projective reparametrization. Equivalently, in one common formulation, it is given by an equivalence class of torsion-free affine connections having the same unparametrized geodesics. Projective connections are modeled on the geometry of projective space. In modern terms, they may be described as Cartan connections modeled on projective space; in the normal torsion-free case, this Cartan-geometric description is equivalent to the classical description by projectively equivalent affine connections. Unlike a Riemannian or pseudo-Riemannian connection, a projective connection does not determine a notion of length, angle, or distance. Its basic geometric datum is instead a notion of straightness: it determines which curves are to be regarded as geodesics, while forgetting the affine parametrization of those curves.
Projective space as the model geometry The first step in defining any Cartan connection is to consider the flat case: in which the connection corresponds to the Maurer-Cartan form on a homogeneous space. In the projective setting, the underlying manifold M {\displaystyle M} of the homogeneous space is the projective space RPn which we shall represent by homogeneous coordinates [ x 0 , … , x n ] {\displaystyle [x_{0},\dots ,x_{n}]} . The symmetry group of M {\displaystyle M} is G = PSL(n+1,R). Let H be the isotropy group of the point [ 1 , 0 , 0 , … , 0 ] {\displaystyle [1,0,0,\ldots ,0]} . Thus, M = G/H presents M {\displaystyle M} as a homogeneous space. Let g {\displaystyle {\mathfrak {g}}} be the Lie algebra of G, and h {\displaystyle {\mathfrak {h}}} that of H. Note that g = s l ( n + 1 , R ) {\displaystyle {\mathfrak {g}}={\mathfrak {s}}{\mathfrak {l}}(n+1,{\mathbb {R} })} . As matrices relative to the homogeneous basis, g {\displaystyle {\mathfrak {g}}} consists of trace-free ( n + 1 ) × ( n + 1 ) {\displaystyle (n+1)\times (n+1)} matrices:
( λ v i w j a j i ) , ( v i ) ∈ R 1 × n , ( w j ) ∈ R n × 1 , ( a j i ) ∈ R n × n , λ = − ∑ i a i i {\displaystyle \left({\begin{matrix}\lambda &v^{i}\\w_{j}&a_{j}^{i}\end{matrix}}\right),\quad (v^{i})\in {\mathbb {R} }^{1\times n},(w_{j})\in {\mathbb {R} }^{n\times 1},(a_{j}^{i})\in {\mathbb {R} }^{n\times n},\lambda =-\sum _{i}a_{i}^{i}} . And h {\displaystyle {\mathfrak {h}}} consists of all these matrices with ( w j ) = 0 {\displaystyle (w_{j})=0} . Relative to the matrix representation above, the Maurer-Cartan form of G is a system of 1-forms ( ξ , α j , α j i , α i ) {\displaystyle (\xi ,\alpha _{j},\alpha _{j}^{i},\alpha ^{i})} satisfying the structural equations (written using the Einstein summation convention):
d ξ + α i ∧ α i = 0 {\displaystyle d\xi +\alpha ^{i}\wedge \alpha _{i}=0}
… excerpt ends here. Continue reading the full article.
