A rank-two relaxed parallel splitting version of the augmented Lagrangian method with step size in (0,2) for separable convex programming.
Bingsheng HeFeng MaShengjie XuXiaoming YuanPublished in: Math. Comput. (2023)
Keyphrases
- step size
- convex programming
- convex optimization
- convergence rate
- interior point methods
- primal dual
- linear programming
- convergence speed
- cost function
- total variation
- global convergence
- global optimum
- semidefinite programming
- wavelet coefficients
- optimal solution
- natural images
- image restoration
- wavelet transform
- low rank
- linear program