How spectral geometry affects decentralized gradient tracking
The steady-state mean-square disagreement scales as when the nontrivial eigenvalue approaches $-1$.
The study examines how the spectral geometry of the mixing matrix affects the convergence of decentralized stochastic gradient tracking (GT). This process is used for smooth strongly convex problems.
The research shows that for two agents with identical quadratic objectives, the exact steady-state mean-square disagreement scales as . This occurs as the nontrivial eigenvalue approaches $-1$, provided the stepsize is sufficiently small relative to the gap.
The absolute spectral gap is denoted as . The gradient-noise variance is denoted as .
For diagonalizable mixing matrices, a stochastic bound averages all nontrivial eigenvalue contributions up to an eigenvector conditioning factor. As a real or complex eigenvalue approaches the unit circle away from , its contribution grows as the inverse cube of its distance to the circle.
For symmetric mixing with gaps at and at $-1$, the weighted sum replaces the prior factor . For agents, and count the nonnegative and negative nontrivial eigenvalues, respectively.
A finite-horizon lower bound for the average iterate matches the worst-case factor up to universal constants. Numerical experiments confirm this predicted scaling.
Key facts
- The steady-state mean-square disagreement scales as as the nontrivial eigenvalue approaches $-1$.
- The contribution of an eigenvalue grows as the inverse cube of its distance to the unit circle away from .
- Numerical experiments confirm the predicted scaling.