Struct ResidentArrowFrameHandle 

pub struct ResidentArrowFrameHandle { /* private fields */ }

#1017 Phase 3: a device-resident Arrow-Schur frame whose constant Hessian blocks (D = H_tt, B = H_tβ, border H_ββ) and their factors stay on the device across the inner Newton loop. Construct once per frozen gate/basis frame, then call ResidentArrowFrameHandle::solve_gradient once per iterate with the fresh residual gradient — only the O(n·d + p) gradient crosses to the device and only δ crosses back, in contrast to solve_arrow_newton_step which re-uploads and re-factors the full system every call. On a non-CUDA host construction returns ArrowSchurGpuFailure::Unavailable.

Implementations

impl ResidentArrowFrameHandle

pub fn new( sys: &ArrowSchurSystem, ridge_t: f64, ridge_beta: f64, ) -> Result<Self, ArrowSchurGpuFailure>

Upload the constant Hessian blocks and perform the one-time factor work.

pub fn solve_gradient( &self, g_t: &[f64], g_beta: &[f64], ) -> Result<ArrowSchurGpuSolution, ArrowSchurGpuFailure>

Solve H δ = −gradient for a fresh gradient reusing the resident factors.

pub fn log_det_hessian(&self) -> f64

log|H| for the frame (constant; depends only on the factored Hessian).