The Symmetric Space $\mathrm{GL}_2/O(2)$ and Its Complex Twin
Introduction
The quotient of the real invertible group by the orthogonal group is the positive definite cone of the symmetric half, a Riemannian symmetric space of dimension three and constant negative curvature, and its complex twin, the quotient of the complex invertible group by the unitary group, is the positive definite cone of the Hermitian half, of dimension four.
The article identifies the two quotients with the two cones, computes their dimensions from the two decompositions, describes their metrics and their curvature, compares the two spaces as the real and complex forms of one construction, and records the isometry groups. It assumes The Polar Decomposition from the Two Halves at $n=2$, The Cartan Decomposition of $\mathrm{GL}_2(\mathbb{C})$ and The Positive Cone in $\mathrm{Herm}_2$ and Its Boundary.
It follows the biquaternion article The Polar Decomposition of a Biquaternion for the same decomposition on the biquaternion side.
The Two Quotients
Both quotients are smooth manifolds whose points are the positive definite matrices of the corresponding half.
Proposition. The map sending an invertible real matrix to the class of the orthogonal group is a diffeomorphism onto the positive definite cone of the symmetric half.
Proof. The polar decomposition writes the matrix uniquely as an orthogonal matrix times a positive definite symmetric matrix, so the class depends exactly on the second factor, and the construction is smooth in both directions.
Corollary. The quotient has dimension three, the difference of the dimension four of the real group and the dimension one of the orthogonal group.
Proof. The dimensions of a quotient by a closed subgroup are the difference of the dimensions when the subgroup acts freely.
Corollary. The complex quotient is diffeomorphic to the positive definite cone of the Hermitian half and has dimension four, the difference of eight and four.
Proof. The same polar decomposition with unitary and Hermitian factors gives the identification, and the dimensions are eight minus four.
The Identification with the Cones
Theorem. The real quotient is the positive definite cone of the symmetric half, and the complex quotient is the positive definite cone of the Hermitian half.
Proof. A positive definite matrix is the square of a positive definite matrix, and the polar decomposition exhibits every coset by its positive definite factor; conversely every positive definite matrix gives a coset.
Corollary. The real quotient is a convex open cone, and the complex quotient is a convex open cone of real dimension four.
Proof. Both cones are convex by the positive definite condition, and both are open because the determinant is positive on them.
Corollary. The quotient map is a smooth fibration with fibre the compact group, and the total space is the product of the fibre with the cone.
Proof. The polar decomposition is a global diffeomorphism from the product onto the group.
The Metric and the Curvature
Proposition. The quotient inherits a metric from the trace form on the group, and with that metric the real quotient has constant negative curvature.
Proof. The trace form on the Lie algebra restricts to a positive definite form on the symmetric part, and the resulting symmetric space is the hyperbolic plane.
Corollary. The real quotient is isometric to the hyperbolic plane, and its curvature is minus one up to the normalisation of the metric.
Proof. The symmetric space of the special linear group of order two by the rotation group is the hyperbolic plane, and the real quotient differs from it by the scalars.
Corollary. The complex quotient is isometric to the product of the hyperbolic three-space with a real line, so its curvature is negative in three directions and zero in the scalar direction.
Proof. The symmetric space of the complex special linear group by the special unitary group is the hyperbolic three-space, and the complex quotient adds the one-dimensional factor of the scalars.
The Two Forms
Proposition. The two quotients are the two real forms of one complex symmetric space, obtained by the two real structures on the complex group.
Proof. The real form of the complex group given by the real matrices has fixed subgroup the orthogonal group, and the compact real form has fixed subgroup the unitary group; the two quotients are the corresponding noncompact symmetric spaces.
Corollary. The real quotient is the symmetric space of the split real form, and the complex quotient is the symmetric space of the compact real form acting on the cone of the Hermitian half.
Proof. The real matrices contain the hyperbolic elements without rotation, and the unitary group is compact.
Corollary. The compact dual of the real quotient is the two-sphere, and the compact dual of the complex quotient is the three-sphere.
Proof. The compact dual replaces the noncompact group by its compact form, so the quotient of the special unitary group by the rotation group is the two-sphere, and the quotient of the product of two special unitary groups by the diagonal is the three-sphere.
Corollary. The two quotients and their compact duals have the same dimension and the same rank, and they differ in the sign of the curvature.
Proof. The complexification of a symmetric pair gives a pair of the same dimension and rank, with the curvature signs reversed.
Isometries and Transitivity
Proposition. The group acts transitively on its quotient by isometries, and the stabiliser of a point is the compact group.
Proof. The group acts transitively on its cosets by left multiplication, and left multiplication by a group element is an isometry for the trace metric when the element is orthogonal or unitary.
Corollary. The real quotient is a homogeneous space of dimension three, and the complex quotient is a homogeneous space of dimension four.
Proof. The dimensions are computed from the dimensions of the group and of the stabiliser.
Corollary. The isometry group of the real quotient contains the real group, and the isometry group of the complex quotient contains the complex group.
Proof. The action by isometries gives an inclusion of the group into the isometry group.
The Boundary and the Compactification
Proposition. The boundary of each quotient in the ambient half is the set of positive semidefinite matrices of rank one, and it is a cone over a sphere.
Proof. The closure of the positive definite cone in the half is the positive semidefinite cone, whose boundary consists of the rank-one matrices.
Corollary. The real quotient is compactified by adding a circle of directions, and the complex quotient by adding a sphere of directions.
Proof. The rank-one positive semidefinite matrices are parametrised by a line in the real case and by a complex line in the complex case, which are a circle and a sphere.
Corollary. The compactification is a closed ball in each case, of the dimension of the space.
Proof. The cone over a sphere is homeomorphic to a ball, and the cone is the closure of the space.
Summary
The quotient of the real invertible group by the orthogonal group is the positive definite cone of the symmetric half, a symmetric space of dimension three isometric to the hyperbolic plane. Its complex twin, the quotient of the complex group by the unitary group, is the positive definite cone of the Hermitian half, of dimension four and isometric to the hyperbolic three-space up to the scalars.
Both quotients are convex open cones, homogeneous under their groups, with the compact group as the stabiliser of a point, and both are compactified by a sphere of rank-one directions into a closed ball.
Summary of Notation
| Symbol | Meaning |
|---|---|
| $\mathrm{GL}_2(\mathbb{R})/O(2)$ | the real quotient |
| $\mathrm{GL}_2(\mathbb{C})/U(2)$ | the complex twin |
| $P$ | the positive definite cone of the symmetric half |
| $P_{\mathbb{C}}$ | the positive definite cone of the Hermitian half |
| $S^{1}$, $S^{2}$ | the directions at the boundary |
Further Reading
- Sigurdur Helgason, Differential Geometry, Lie Groups, and Symmetric Spaces (American Mathematical Society, 2001), for the classification of symmetric spaces.
- Armand Borel, Semisimple Groups and Riemannian Symmetric Spaces (Hindustan Book Agency, 1998), for the real and complex forms.
- The Polar Decomposition from the Two Halves at $n=2$, for the decomposition giving the quotient.
- The Cartan Decomposition of $\mathrm{GL}_2(\mathbb{C})$, for the group decomposition behind the identification.
- The Positive Cone in $\mathrm{Herm}_2$ and Its Boundary, for the cone of the complex half.