Equiangular Direction Method and SR-Decomposition

• Equiangular Direction Method is a framework for constructing unit vectors with constant pairwise inner products, enabling an SR-decomposition analogous to Gram–Schmidt.
• The method introduces a fixed nonzero angle to derive new canonical forms, explicit inversion formulas, and insights into spectral and geometric properties.
• It finds applications in signal processing, coding theory, and frame design, providing robust tools for analyzing high-dimensional matrix structures.

The Equiangular Direction Method is an algorithmic framework for constructing sets of unit vectors in real inner product spaces whose pairwise inner products attain a prescribed value. Equiangular vectors form the building blocks of equiangular matrices, which enable a matrix decomposition analogous to the classical Gram–Schmidt process but parameterized by a fixed nonzero angle. This method provides new canonical forms, explicit inversion formulas, and natural connections to equiangular frames in signal processing. The technique generalizes the QR factorization and yields insights into matrix spectral properties and geometric structures in high-dimensional spaces (Sadeghi et al., 2014, Sadeghi et al., 2016).

1. Mathematical Problem Statement and Definitions

Given a set of $m$ linearly independent vectors $\{a_1, \ldots, a_m\}$ in a real inner product space $V$ with $\dim V = n \geq m$, and a target angle $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$, define $\alpha = \cos \theta$. The goal is to construct unit vectors $s_1, \ldots, s_m$ such that:

• $\|s_i\| = 1$ for all $i$,
• $\langle s_i, s_j \rangle = \alpha$ for all $\{a_1, \ldots, a_m\}$0,
• $\{a_1, \ldots, a_m\}$1.

Stacking these vectors yields a matrix $\{a_1, \ldots, a_m\}$2 whose columns are equiangular and full-rank. An upper triangular matrix $\{a_1, \ldots, a_m\}$3 is simultaneously constructed so that $\{a_1, \ldots, a_m\}$4, generalizing the QR decomposition (Sadeghi et al., 2014, Sadeghi et al., 2016).

2. Algorithmic Construction: Equiangular Algorithm

The Equiangular Algorithm (EA) builds the equiangular matrix $\{a_1, \ldots, a_m\}$5 and $\{a_1, \ldots, a_m\}$6 recursively:

1. Initialization: $\{a_1, \ldots, a_m\}$7, $\{a_1, \ldots, a_m\}$8, $\{a_1, \ldots, a_m\}$9.
2. For $V$0:
   • Orthogonalize $V$1 against $V$2: $V$3.
   • Set $V$4.
   • Calculate coefficient $V$5.
   • Form $V$6.
   • Normalize: $V$7, $V$8.
   • Update $V$9: $\dim V = n \geq m$0 for $\dim V = n \geq m$1.

At each stage, $\dim V = n \geq m$2 is chosen so that $\dim V = n \geq m$3 for all $\dim V = n \geq m$4, guaranteeing that the resulting set is equiangular with prescribed angle $\dim V = n \geq m$5. The output matrices $\dim V = n \geq m$6 satisfy $\dim V = n \geq m$7 and $\dim V = n \geq m$8, with $\dim V = n \geq m$9 the Gram matrix (Sadeghi et al., 2016, Sadeghi et al., 2014).

3. Algebraic and Spectral Properties of Equiangular Matrices

Invertibility and Gram Structure

For $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$0, the Gram matrix $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$1 is positive definite, ensuring that any $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$2 is nonsingular. The Gram structure:

$$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$3

leads to explicit eigenvalues:

• $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$4 (multiplicity $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$5),
• $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$6 (simple).

Explicit Inversion

For $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$7 equiangular $$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$8, the inverse is given by:

$$\theta \in \left(0, \arccos\left(-\frac{1}{m-1}\right)\right)$$9

with

$\alpha = \cos \theta$0

This formula requires $\alpha = \cos \theta$1 operations and the rows of $\alpha = \cos \theta$2 are themselves equiangular with cosine $\alpha = \cos \theta$3 (Sadeghi et al., 2014, Sadeghi et al., 2016).

Canonical Matrix Forms

Any $\alpha = \cos \theta$4 admits a Schur-style factorization:

$\alpha = \cos \theta$5

where $\alpha = \cos \theta$6 and $\alpha = \cos \theta$7 is block-upper-triangular. Special symmetric and normal forms can be constructed analogously (Sadeghi et al., 2016).

4. Connections to Classical Decompositions and Extensions

The Equiangular Direction Method generalizes the classical Gram–Schmidt/QR process: setting $\alpha = \cos \theta$8 recovers QR, with orthogonal but not equiangular directions. For other choices of $\alpha = \cos \theta$9 within the permissible interval, the columns of $s_1, \ldots, s_m$0 have constant mutual angle. The $s_1, \ldots, s_m$1-decomposition thus forms a one-parameter family of matrix factorizations interpolating between orthogonal and fully equiangular geometries (Sadeghi et al., 2014).

The method applies directly to full-rank, rectangular $s_1, \ldots, s_m$2 with $s_1, \ldots, s_m$3. For $s_1, \ldots, s_m$4 and $s_1, \ldots, s_m$5, the EA recovers the simplex equiangular tight frame (ETF), significant in frame theory and coding (Sadeghi et al., 2016). The construction of doubly equiangular matrices, whose rows and columns are simultaneously equiangular, is possible via Householder transforms applied to $s_1, \ldots, s_m$6 to align row and column sums with the all-ones vector. This yields symmetric equiangular matrices with prescribed geometric properties (Sadeghi et al., 2014).

5. Eigenvalue Structure, Conditioning, and Stability

The eigenvalues of $s_1, \ldots, s_m$7 dictate the numerical and analytic behavior of $s_1, \ldots, s_m$8. The singular values of $s_1, \ldots, s_m$9 are $\|s_i\| = 1$0 (multiplicity $\|s_i\| = 1$1) and $\|s_i\| = 1$2 (multiplicity $\|s_i\| = 1$3). Thus, the spectral norm condition number is

$\|s_i\| = 1$4

The conditioning deteriorates as $\|s_i\| = 1$5 or $\|s_i\| = 1$6, where $\|s_i\| = 1$7 diverges (Sadeghi et al., 2014, Sadeghi et al., 2016).

The computational complexity for constructing $\|s_i\| = 1$8 and $\|s_i\| = 1$9 is $i$0 for $i$1 matrices, and $i$2 for $i$3 input, matching the cost of Gram–Schmidt. The explicit inversion requires $i$4. Numerical experiments on Vandermonde, Minij, and Hilbert matrices confirm that the method is stable for moderate values of $i$5, with numerical loss of equiangularity potentially mitigated by re-orthogonalization steps (Sadeghi et al., 2014, Sadeghi et al., 2016).

6. Applications and Examples

Applications include the design and analysis of Equiangular Tight Frames (ETFs), maximizing the minimal angle between frame vectors (Grassmannian packing), and providing well-controlled decompositions for symmetric positive definite matrices as a Cholesky-plus-rank-one decomposition. The method allows for new block and symmetric matrix canonical forms, with potential applications in signal processing, coding theory, and evenly distributed data sampling (Sadeghi et al., 2016).

Illustrative examples in the literature include:

• EA applied to Vandermonde matrices for various angles, demonstrating distinct equiangular factors.
• SR factorization of Minij matrices at different angles, yielding nonequivalent $i$6.
• Decomposition of the $i$7 Hilbert matrix, explicitly verifying the equiangular property.
• Identity matrix decomposed to a unique upper-triangular $i$8 with positive entries; here, $i$9.
• Construction of doubly equiangular matrices through Householder-conjugation, producing matrices with constant row and column sums aligned to $\langle s_i, s_j \rangle = \alpha$0 (Sadeghi et al., 2014).

7. Open Problems and Directions for Study

Several open questions and continuing research directions remain:

• Extension of the method to complex inner product spaces and frames with $\langle s_i, s_j \rangle = \alpha$1.
• In-depth numerical stability analysis, along with the design and analysis of EA variants.
• Characterization of doubly equiangular matrices and implications for matrix theory and applications.
• Investigation of connections to polynomial root localization via equiangular decompositions.
• Construction of larger equiangular line sets and exploitation of their combinatorial and geometric properties in coding and communications contexts (Sadeghi et al., 2016).

Further study of equiangular matrix structures and their decompositions may yield new analytical and computational tools across linear algebra, discrete geometry, and applied signal processing.