Struct klu_rs::FixedKluMatrix

source · [−]

pub struct FixedKluMatrix<I: KluIndex, D: KluData> { /* private fields */ }

Expand description

A compressed column form SparsMatrix whose shape is fixed

Implementations

source

impl<I: KluIndex, D: KluData> FixedKluMatrix<I, D>

source

pub fn into_alloc(self) -> Vec<Cell<D>>

Obtain the allocation of the matrix data This function can be used with [from_raw] and KluMatrixSpec::reinit to reuse a matrix allocation.

Safety

This function invalidates any pointers into the matrix

source

pub fn new_with_alloc(
spec: Rc<KluMatrixSpec>,
alloc: Vec<Cell<D>>
) -> Option<Self>

Constructs a new matrix for the provided KluMatrixSpec. alloc is used to store the data. If not enough space is available alloc is resized as required.

The matrix is always intially filled with zeros no matter the previous content of alloc

Returns

the constructed matrix if the matrix is not empty. If the matrix is empty None is retruned instead

source

pub fn new(spec: Rc<KluMatrixSpec>) -> Option<Self>

Constructs a new matrix for the provided KluMatrixSpec by allocating space where the data can be stored.

The matrix is intially filled with zeros.

Returns

the constructed matrix if the matrix is not empty. If the matrix is empty None is retruned instead

source

pub fn data(&self) -> &[Cell<D>]

source

pub fn data_ptr(&self) -> *mut D

Returns a pointer to the matrix data

Safety

The returned pointer must never be dereferenced. Turing this data into any reference always causes undefined behaviour

This pointer point to data inside an UnsafeCell so you should use std::ptr methods to access it or turn it into &Cell<D> or &UnsafeCell<D>

source

pub fn write_all(&self, val: D)

source

pub fn lu_factorize(&mut self, refactor_threshold: Option<f64>) -> bool

Perform lu_factorization of the matrix using KLU If the matrix was already factorized previously and refactor_threshold is given, it is first attempted to refactorize the matrix. If this fails (either due to an KLU error or because the rcond is larger than the provided threshold) full factorization is preformed.

Calling to funciton is a prerequisite to calling [solve_linear_system]

source

pub fn solve_linear_system(&self, rhs: &mut [D])

solves the linear system Ax=b. The b vector is read from rhs at the beginning of the function. After the functin completes x was written into rhs

Note: This function assumes that [lu_factorize] was called first. If this is not the case this functions panics.

source

pub fn solve_linear_tranose_system(&self, rhs: &mut [D])

solves the linear system A^T x=b The b vector is read from rhs at the beginning of the function. After the functin completes x was written into rhs

Note: This function assumes that [lu_factorize] was called first. If this is not the case this functions panics.

source