* Clean a bit the Triadiagonalization making sure it the inplace
function really works inplace ;), and that only the lower
triangular part of the matrix is referenced.
* Remove the Tridiagonalization member object of SelfAdjointEigenSolver
exploiting the in place capability of HouseholdeSequence.
* Update unit test to check SelfAdjointEigenSolver only consider
the lower triangular part.
* use SelfAdjointView instead of Eigen2's SelfAdjoint flag.
* add tests and documentation.
* allow eigenvalues() for non-selfadjoint matrices.
* they no longer depend only on SelfAdjointEigenSolver, so move them to
a separate file
This changes the return type of:
* eigenvectors() and eigenvalues() in ComplexEigenSolver
* eigenvalues() in EigenSolver
* eigenvectors() and eigenvalues() in SelfAdjointEigenSolver
This is to avoid dynamic memory allocations in the compute() methods of
ComplexEigenSolver, EigenSolver, and SelfAdjointEigenSolver where possible.
As a result, Tridiagonalization::decomposeInPlace() is no longer used.
Biggest remaining issue is the allocation in HouseholderSequence::evalTo().
* reduce scope of declarations
* use that low = 0 and high = size-1
* rename some variables
* rename hqr2_step2() to computeEigenvectors()
* exploit that ei_isMuchSmallerThan takes absolute value of arguments
* get rid of BlockReturnType: it was not needed, and code was not always using it consistently anyway
* add topRows(), leftCols(), bottomRows(), rightCols()
* add corners unit-test covering all of that
* adapt docs, expand "porting from eigen 2 to 3"
* adapt Eigen2Support
- Updated unit tests to check above constructor.
- In the compute() method of decompositions: Made temporary matrices/vectors class members to avoid heap allocations during compute() (when dynamic matrices are used, of course).
These changes can speed up decomposition computation time when a solver instance is used to solve multiple same-sized problems. An added benefit is that the compute() method can now be invoked in contexts were heap allocations are forbidden, such as in real-time control loops.
CAVEAT: Not all of the decompositions in the Eigenvalues module have a heap-allocation-free compute() method. A future patch may address this issue, but some required API changes need to be incorporated first.
* adapt Eigenvalues module to the new rule that the RowMajorBit must have the proper value for vectors
* Fix RowMajorBit in ei_traits<ProductBase>
* Fix vectorizability logic in CoeffBasedProduct
* Introduction of strides-at-compile-time so for example the optimized code really knows when it needs to evaluate to a temporary
* StorageKind / XprKind
* Quaternion::setFromTwoVectors: use JacobiSVD instead of SVD
* ComplexSchur: support the 1x1 case
for consistency with the RealScalar type; correct ComplexEigenSolver
docs to take non-diagonalizable matrices into account; refactor
ComplexEigenSolver::compute().
of ei_matrix_array for size 0
* adapt many xprs to have the right storage order, now that it matters
* add static assert on expressions to check that vector xprs
have the righ storage order
* adapt ei_plain_matrix_type_(column|row)_major
* implement assignment of selfadjointview to matrix
(was before failing to compile) and add nestedExpression() methods
* expand product_symm test
* in ei_gemv_selector, use the PlainObject type instead of a custom Matrix<...> type
* fix VectorBlock and Block mistakes
