Struct BlockDiagonal 

pub struct BlockDiagonal { /* private fields */ }

The stacked sparsity pattern for n_scenarios copies of one topology’s Jacobian, plus the per-block recipe needed to refill it.

Implementations

impl BlockDiagonal

pub fn analyze(buses: &[Bus], ybus: &YBusSparse, n_scenarios: usize) -> Self

Derives the stacked pattern by replicating a single-scenario JacobianPattern n_scenarios times, offsetting block s’s rows and columns by s * block_size.

All scenarios must share buses’ bus-type assignment and ybus — that shared pattern is the entire premise. batch::Scenario cannot override bus_type precisely so this stays true.

pub fn n_unknowns(&self) -> usize

Total unknowns across the batch — n_scenarios * block_size.

pub fn block_size(&self) -> usize

Unknowns per scenario.

pub fn len(&self) -> usize

Nonzeros in the stacked matrix.

pub fn is_empty(&self) -> bool

pub fn to_triplets(&self, values: &[f64]) -> Vec<(usize, usize, f64)>

The stacked pattern as (row, col, value) triplets — the shape every LinearSolver consumes for its one-time symbolic analysis.

pub fn fill( &self, states: &[Vec<Bus>], p_calc: &[Vec<f64>], q_calc: &[Vec<f64>], active: &[bool], values: &mut Vec<f64>, )

Fills the stacked value array from each scenario’s own state.

active[s] == false writes an identity block instead of scenario s’s Jacobian (see JacobianPattern::fill_identity_into), which — paired with a zero right-hand side — makes that scenario’s update exactly zero while leaving the sparsity pattern untouched.