For the operator norm, submultiplicativity bounds the remainder of the matrix exponential byTherefore . This is precisely Fréchet differentiability at zero, with
The curve-based tangent space at consists of velocities of smooth curves satisfying . For an arbitrary subset these velocities need not form a vector space; for a smooth submanifold they do, and this is its usual tangent space. This definition suffices for the direct computation at the identity.
If is a smooth curve of orthogonal matrices with , differentiating gives . Thus each tangent velocity is a skew-symmetric matrix.
The matrix exponential is smooth, and part (a) identifies its derivative at zero as the identity. The inverse function theorem therefore gives a local diffeomorphism from an open neighborhood of zero onto a neighborhood of . Choose a sufficiently small operator norm ball and let . This is again a diffeomorphism onto an open set, since it is a restriction of the original one.
The ball is preserved by and . Since and , is preserved by adjoint and inversion. For its local inverse , injectivity also gives
Part (b) already shows . Conversely, if and , thenBoth and lie in , where the matrix exponential is injective. Therefore , proving the reverse inclusion. The restricted map is a bijection, indeed a diffeomorphism between these local sets.
The linear map is the projection onto the space of symmetric matrices, with kernel . Define on . Since is invertible everywhere on and is onto, is onto everywhere. By part (ii),The regular level set theorem makes this a submanifold of codimension , and henceThis neighborhood is around ; the source's final occurrence of in this step is a neighborhood-label slip. For any , left multiplication by is an invertible linear map of the ambient matrix space and carries onto . These translated charts establish the submanifold property at every group element.
Articles by others on the same topic
There are currently no matching articles.