The most separated family of subspaces · d = 5, n = 18
Also known asGrassmannian subspace packingpacking planes in Grassmannian spacechordal distance packing
Choose n two-dimensional subspaces (planes through the origin) in Rᵈ. The squared chordal distance between two planes is the sum of the squared sines of their principal angles, or equivalently 2 − tr(PᵢPⱼ), where Pᵢ and Pⱼ are their orthogonal projectors. Maximize the minimum squared chordal distance over all pairs.
Record comparison
the smallest squared chordal distance · Higher is better| Record | Value / interval | Author / holder | Source |
|---|---|---|---|
| External best known | [1.1908530119, 1.1908530120] | See source attribution | cohn.mit.edu ↗ |
| Site record | 1.190853013 | zzzcy | Record details ↓ |
Numerically exceeds the published range · awaiting review.
Formal definition
- ContainerReal d-space Rᵈ; an answer consists of n points in the Grassmannian G(d, 2), namely n planes through the origin
- SubmissionExactly n subspaces, each 2 basis vectors; the basis vectors must be linearly independent
- ProjectorThe two columns of Aᵢ are the submitted basis vectors of subspace i; Pᵢ is the orthogonal projector and does not depend on the chosen basis
- Squared chordal distanceThe squared chordal distance between two 2-dimensional subspaces is 2 minus the trace of the product of their projectors
- Scoreδ(P) is the smallest squared chordal distance over all pairs of subspaces; the verifier constructs the projectors in exact rational arithmetic via the 2×2 adjugate
- ScoringThe record is stored as floor(min squared chordal distance · 10¹⁸); the page displays the squared chordal distance, rounded down at the ninth decimal
- ObjectiveMake δ(P) as large as possible over all legal families of subspaces P