Matrix Stability
Cross-source consensus on Matrix Stability from 1 sources and 6 claims.
1 sources · 6 claims
How it works
Evidence quality
Other
Highlighted claims
- The main stability matrix is A_eta = D_mu(I - P_pi + eta ee^T). — On the Divergence of Differential Temporal Difference Learning without Local Clocks
- For L = D_mu(I - P_pi), nonzero eigenvalues have strictly positive real parts. — On the Divergence of Differential Temporal Difference Learning without Local Clocks
- A_eta is treated as a rank-one perturbation of a singular M-matrix. — On the Divergence of Differential Temporal Difference Learning without Local Clocks
- The stability region for A_eta is always a finite union of open intervals. — On the Divergence of Differential Temporal Difference Learning without Local Clocks
- Discounted TD has an M-matrix structure that preserves positive stability under positive diagonal multiplication. — On the Divergence of Differential Temporal Difference Learning without Local Clocks
- Differential TD lacks a general M-matrix guarantee for global-clock stability. — On the Divergence of Differential Temporal Difference Learning without Local Clocks