Muscat.LinAlg.ConstraintsHolder module

class Ainsworth

Bases: object

Method from https://doi.org/10.1016/S0045-7825(01)00236-5 Because the CleanConstraint(C) matrix is orthonormal then C.dot(C.T) = I this make the P Q R expression simpler

GetCOp(op)

GetCRhs(op, rhs)

GetNumberOfDofs()

RestoreSolution(op, rhs, arg)

RestrictSolution(op, rhs, arg)

UpdateSystem(CH)

matvec(op, arg)

CheckIntegrity(GUI: bool = False)

CheckIntegrityMasterSlaveDetection(GUI: bool = False)

CheckIntegrityTTC(ttc, GUI=False)

class ConstraintsHolder

Bases: object

Constrained linear problem: Class to store and to enforce constraints for a quadratic system of the form:

minimize 1/2 u.K.u - u.f subject to: A.u = b

This class can store, and manipulate the system Au=b and enforce the constraints by different methods (penalization, substitution, lagrange multiplier)

the first set of function is used to fill this class with the system Au=b:

NextEquation must be called after the call of a set of : AddFactor, AddConstant One per constraint

AddEquationSparse, AddEquation, AddEquationsFromIJV to add a full constraint at once

Compact will clean (zeros, and duplicated) entries

AddConstant(constant)

AddConstraint(constraints)

AddEquation(vals, const=0)

AddEquationSparse(index, vals, const)

AddEquationsFromIJV(ei, ej, ev)

AddEquationsFromOperatorAAndb(A, b=None)

AddFactor(ddl, factor)

CleanEmptyColumns(mat)

Function to clean empty columns for a matrix,

this function returns:

res : a sparse matrix with only non zero columns of mat (the last column is always present, event if it is full of zeros) usedDofs : the indices of the columns prensent in res

Compact()

ComputeConstraintsEquations(mesh, unknownFields, solutionFields=None, reactionFields=None)

GetCOp()

GetCRhs(rhs=None)

GetCleanConstraintBase(purePython=False)

This function compute the QR decomposition of the augmented system [A|b] Il will use the self.tol value to detect redundant constraints (rank revealing)

this functions returns:

res : a parse matrix q’ with only the relevant part of the QR decomposition. Q’ constraint only the non zero columns of [A|b] and always the last column (the b) usedDofs : the indices of the columns present in Q

option:

purePython: to force scipy for computing the QR decomposition

GetNumberOfConstraints()

GetNumberOfDofsOnCSystem()

GetNumberOfDofsOnOriginalSystem()

NextEquation()

Reset()

Residual(u: _SupportsArray[dtype[Any]] | _NestedSequence[_SupportsArray[dtype[Any]]] | bool | int | float | complex | str | bytes | _NestedSequence[bool | int | float | complex | str | bytes]) → Tuple[ndarray, ndarray]

Recover A.u and b

Parameters:

u (ArrayLike) – Vector of unknowns

Returns:

se result of

Return type:

Tuple[np.ndarray,np.ndarray]

RestoreSolution(arg)

RestrictSolution(arg)

SetConstraintsMethod(method: str)

SetNumberOfDofs(nbdof)

SetOp(op, rhs=None)

ToSparse()

Function to convert to the constraints system Ax=B to a sparse matrix

this function returns:

• res : a sparse matrix of [A’|B] , A’ has only the non zero columns of A

• usedDofs : the indices of the columns present in A’

ToSparseFull()

Function to convert to the constraints system Ax=B to a sparse matrix

this function returns:

res : a sparse matrix of [A|B] ,

UpdateCSystem()

matvec(arg)

rmatvec(arg)

ExpandMatrix(op, mattoglobal, nbdofs, treatrows=True, treatcols=True)

class Lagrange

Bases: object

GetCOp(op)

GetCRhs(op, rhs)

GetNumberOfDofs()

RestoreSolution(op, rhs, arg)

RestrictSolution(op, rhs, arg)

UpdateSystem(CH)

matvec(op, arg)

class Penalisation

Bases: object

GetCOp(op)

GetCRhs(op, rhs)

GetNumberOfDofs()

RestoreSolution(op, rhs, arg)

RestrictSolution(op, rhs, arg)

UpdateSystem(CH)

matvec(op, arg)

class Projection

Bases: object

GetCOp(op)

GetCRhs(op, rhs)

GetNumberOfDofs()

RestoreSolution(op, rhs, arg)

RestrictSolution(op, rhs, arg)

UpdateSystem(CH)

matvec(op, arg)