How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
Graph charts and tangent directions for complex rank-r projectors
Statement
Fix , put (range as in Kernel and image of a linear map), and use the orthogonal decomposition . Write for the orthogonal projector onto an -dimensional subspace . Such a subspace for which the orthogonal projection is invertible is uniquely a graph of a linear map . Its projector is These graph coordinates form an open neighborhood of in the projector Grassmannian. In the block decomposition at , its tangent space is The projector representation also removes orbital-frame gauge: two orthonormal frames represent the same projector exactly when they differ by right multiplication by a unitary matrix.
Facts & Assumptions
Given: , the range/kernel decomposition of an orthogonal projector, and finite-dimensional complex inner-product spaces.
The domain consists of Hermitian idempotents with trace (Rank-r orthogonal projector domain).
A Hermitian idempotent has range dimension equal to its trace (A Hermitian idempotent has trace equal to its rank).
Every finite-dimensional inner-product space has an orthonormal basis (Every finite-dimensional real or complex inner product space has an orthonormal basis).
The kernel and image of a linear map are its nullspace and range (Kernel and image of a linear map).
On the invertible set, and matrix differentials obey the product rule; iterating this expression gives continuous derivatives of all finite orders (On the invertible locus, , Matrix differentials obey the sum rule, product rule, and adjoint rule).
Proof
Since , every decomposes as with the first term in and the second in by [F4]. If and , then by self-adjointness. Hence , giving the stated orthogonal block decomposition.
For any , is positive definite because for . The displayed is Hermitian, and . Its range is , of dimension ; [F2] gives .
Let , whose adjoint is . The block is , so it is invertible exactly when is invertible: is positive definite exactly when is injective, and the domain and codomain of have the same finite dimension. Thus the chart condition is the open condition that be invertible; it is open because and determinant are continuous, and at we have . In that domain, projection from to is bijective, so every has a unique vector ; this defines a unique linear . In blocks , that map is . The formula in step 1.2 is the unique orthogonal projector onto this graph. Matrix inversion is smooth on invertible matrices by [F5], so the graph map and are inverse smooth coordinates.
Let be a differentiable curve in with and derivative . Differentiating and gives and . In block form these equations force . Conversely, for any such , the graph curve from step 1.2 has derivative at , since . Thus these are exactly all tangent directions.
By [F3], each frame with and range exists. If and are two such frames, then satisfies and . Thus is unitary and . Conversely, for every unitary . Hence projectors identify exactly the unitary changes of occupied frame.
Depends on
- Rank-r orthogonal projector domain
- A Hermitian idempotent has trace equal to its rank
- Every finite-dimensional real or complex inner product space has an orthonormal basis
- Kernel and image of a linear map
- Rectangular matrix multiplication and the identity matrix $I_n$, including zero-sized shapes
- The trace $\operatorname{tr}(A)$ as the sum of the diagonal entries
- On the invertible locus, $D\operatorname{inv}(A)[H]=-A^{-1}HA^{-1}$
- Matrix differentials obey the sum rule, product rule, and adjoint rule
Used by
Dependency tree · two levels
31 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.
Sources
- P.-A. Absil, R. Mahony, and R. Sepulchre, Optimization Algorithms on Matrix Manifolds, Chapter 3, Princeton University Press (2008), §§3.4.4 and 3.5.7 (standard reference, not scraped)
- Alan Edelman, Tomás A. Arias, and Steven T. Smith, The Geometry of Algorithms with Orthogonality Constraints, SIAM Journal on Matrix Analysis and Applications 20 (1998), §§2.3.2 and 2.5 (standard reference, not scraped)