Statements
Source excerpts; proofs omitted. recall restates an existing declaration; #check records an expression, not its inferred type.
Algorithm 15.8 trajectories are the generic AD-LPMM trajectories specialized to B = -I
and c = 0.
recall IsADLPMMTrajectory
-- Proof sketch: specialize `mem_adlpmm_x_step_iff` from Algorithm 15.5 at `B = -LinearMap.id`
-- and `c = 0`, then simplify the residual `A x^k + (-id) z^k - 0`.
Source context
universe u v
variable {X : Type u} {Y : Type v}
variable [NormedAddCommGroup X] [InnerProductSpace ℝ X] [FiniteDimensional ℝ X]
variable [NormedAddCommGroup Y] [InnerProductSpace ℝ Y] [FiniteDimensional ℝ Y]In the linear-composite specialization B = -I, c = 0, membership in the AD-LPMM
x-update set is exactly the textbook proximal clause
x^(k+1) ∈ prox[((1 / α) f₁)] (x^k - (ρ / α) Aᵀ (A x^k - z^k + (1 / ρ) y^k)).
In the linear-composite specialization B = -I, c = 0, membership in the AD-LPMM
z-update set is exactly the textbook proximal clause
z^(k+1) ∈ prox[((1 / β) f₂)] (z^k + (ρ / β) (A x^(k+1) - z^k + (1 / ρ) y^k)).
In the linear-composite specialization of Algorithm 15.8, the generic AD-LPMM x-step is
exactly the textbook proximal update from part (a).
In the linear-composite specialization of Algorithm 15.8, the generic AD-LPMM z-step is
exactly the textbook proximal update from part (b).
In the linear-composite specialization of Algorithm 15.8, the generic AD-LPMM multiplier
update simplifies to y^(k+1) = y^k + ρ (A x^(k+1) - z^(k+1)).