oft_nksolver Type Reference

Detailed Description

Native Newton solver.

Non-linear system solver based on Newton iteration for the system, \( M_j(x_i) = y_j \), where the Jacobian is defined as \( J_{ij} = \frac{\partial M_j}{\partial x_i} \). The non-linear update then takes the form \( x^{n+1} = x^n - J^{-1}(x^n) \cdot M(x^n) \).

Public Member Functions
procedure	apply (self, u, g)
	Solve non-linear system.
procedure	delete (self)
	Clean-up internal storage.

Public Attributes
class(oft_matrix), pointer	a => NULL()
	Metric matrix.
real(r8)	atol = 1.d-14
	Absolute convergence tolerance \( |res| < atol \).
logical	backtrack = .TRUE.
	Perform backtracking?
class(oft_solver_bc), pointer	bc => NULL()
	Boundary condition.
integer(i4)	cits = 0
	Number of iteractions to convergence.
class(oft_vector), pointer	du => NULL()
	Storage for non-linear corrections.
integer(i4)	its = -1
	Maximum iteration count.
class(oft_solver), pointer	j_inv => NULL()
	Jacobian inverse solver.
integer(i4)	lits = 0
	Number of linear solver iteractions.
integer(i4)	nlits = 0
	Number of non-linear steps to convergence.
real(r8)	rtol = 1.d-14
	Relative convergence tolerance \( |res|/|res_0| < rtol \).
integer(i4)	up_freq = 1
	Frequency of jacobian updates.

Static Public Attributes
procedure(oft_update_jacobian), pointer, nopass	j_update => NULL()
	Jacobian update subroutine.

Member Function/Subroutine Documentation

apply()

procedure apply	(	class(oft_nksolver), intent(inout)	self,
		class(oft_vector), intent(inout)	u,
		class(oft_vector), intent(inout)	g
	)

Solve non-linear system.

Parameters

[in,out]	self	Solver object
[in,out]	u	Guess (input), Solution (output)
[in,out]	g	RHS (input), Residual (output)

delete()

procedure delete

(

class(oft_nksolver), intent(inout)

self

)

Clean-up internal storage.

Parameters

[in,out]

self

Solver object

Member Data Documentation

a

class(oft_matrix), pointer a => NULL()

Metric matrix.

atol

real(r8) atol = 1.d-14

Absolute convergence tolerance \( |res| < atol \).

backtrack

logical backtrack = .TRUE.

Perform backtracking?

bc

class(oft_solver_bc), pointer bc => NULL()

Boundary condition.

cits

integer(i4) cits = 0

Number of iteractions to convergence.

du

class(oft_vector), pointer du => NULL()

Storage for non-linear corrections.

its

integer(i4) its = -1

Maximum iteration count.

j_inv

class(oft_solver), pointer j_inv => NULL()

Jacobian inverse solver.

j_update

procedure(oft_update_jacobian), pointer, nopass j_update => NULL()

static

Jacobian update subroutine.

lits

integer(i4) lits = 0

Number of linear solver iteractions.

nlits

integer(i4) nlits = 0

Number of non-linear steps to convergence.

rtol

real(r8) rtol = 1.d-14

Relative convergence tolerance \( |res|/|res_0| < rtol \).

up_freq

integer(i4) up_freq = 1

Frequency of jacobian updates.

---

The documentation for this type was generated from the following file:

• /home/runner/work/OpenFUSIONToolkit/OpenFUSIONToolkit/src/lin_alg/native_solvers.F90