aphin.operators package
Submodules
aphin.operators.linear_operator_skew_sym module
TensorFlow linear operators that represent [batch] matrices with special properties e.g. symmetric, skew-symmetric, symmetric positive semi-definite matrices based on their degrees of freedom (DOFs)
- class aphin.operators.linear_operator_skew_sym.LinearOperatorSkewSym(dof, is_non_singular=None, is_self_adjoint=None, is_positive_definite=None, is_square=None, name='LinearOperatorSkewSym')[source]
Bases:
LinearOperatorFullMatrixLinearOperator acting like a [batch] square skew-symmetric matrix. Each square skew-symmetric matrix with shape (N, N) has a number of n_dof = N * (N - 1) / 2 degrees of freedom (DOF).
This operator acts like a [batch] square skew-symmetric matrix A with shape [B1,…,Bb, N, N] for some b >= 0. The first b indices index a batch member. For every batch index (i1,…,ib), A[i1,…,ib, : :] is an square symmetric N x N matrix. This matrix A is not materialized but only the degrees of freedom (DOF) of each square skew-symmetric matrix is stored. Only for purposes of broadcasting this shape will be relevant.
LinearOperatorSkewSym is initialized with a (batch) vector that contains the DOF of each square skew-symmetric matrix.
aphin.operators.linear_operator_sym module
TensorFlow linear operators that represent [batch] matrices with special properties e.g. symmetric, skew-symmetric, symmetric positive semi-definite matrices based on their degrees of freedom (DOFs)
- class aphin.operators.linear_operator_sym.LinearOperatorSym(dof, is_non_singular=None, is_self_adjoint=None, is_positive_definite=None, is_square=None, name='LinearOperatorSym')[source]
Bases:
LinearOperatorFullMatrixLinearOperator acting like a [batch] square symmetric matrix. Each square symmetric matrix with shape (N, N) has a number of n_dof = N * (N + 1) / 2 degrees of freedom (DOF).
This operator acts like a [batch] square symmetric matrix A with shape [B1,…,Bb, N, N] for some b >= 0. The first b indices index a batch member. For every batch index (i1,…,ib), A[i1,…,ib, : :] is an square symmetric N x N matrix. This matrix A is not materialized but only the degrees of freedom (DOF) of each square symmetric matrix are stored. Only for purposes of broadcasting this shape will be relevant.
LinearOperatorSym is initialized with a (batch) vector that contains the DOF of each square symmetric matrix.
aphin.operators.linear_operator_sym_pos_def module
TensorFlow linear operators that represent [batch] matrices with special properties e.g. symmetric, skew-symmetric, symmetric positive semi-definite matrices based on their degrees of freedom (DOFs)
- class aphin.operators.linear_operator_sym_pos_def.LinearOperatorSymPosDef(dof, is_non_singular=None, is_self_adjoint=None, is_positive_definite=None, is_square=None, name='LinearOperatorSymPosDef', epsilon=1e-12)[source]
Bases:
LinearOperatorFullMatrixLinearOperator acting like a [batch] square symmetric positive definite matrix. Each square symmetric positive definite matrix with shape (N, N) has a number of n_dof = N * (N + 1) / 2 degrees of freedom (DOF).
This operator acts like a [batch] square symmetric positive definite matrix A with shape [B1,…,Bb, N, N] for some b >= 0. The first b indices index a batch member. For every batch index (i1,…,ib), A[i1,…,ib, : :] is an square symmetric N x N matrix. This matrix A is not materialized but only the degrees of freedom (DOF) of each square symmetric positive definite matrix is stored. Only for purposes of broadcasting this shape will be relevant.
LinearOperatorSymPosDef is initialized with a (batch) vector that contains the DOF of each square symmetric positive definite matrix.