Closed andreasvarga closed 3 years ago
Are you defining Matrix(op)
where op
is a Lyapunov operator anywhere? Or are you relying on the default implementation of Matrix()
?
I am using the default implementation.
Dominique notifications@github.com schrieb am Di., 3. Dez. 2019, 01:57:
Are you defining Matrix(op) where op is a Lyapunov operator anywhere? Or are you relying on the default implementation of Matrix()?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/andreasvarga/MatrixEquations.jl/issues/2?email_source=notifications&email_token=ALJDHEGADYAIMPIYBNZGRPDQWWVHRA5CNFSM4JOFBIY2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFXWOAY#issuecomment-560948995, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALJDHEB3CO45BSONTZ3F4PTQWWVHRANCNFSM4JOFBIYQ .
The Matrix(T)' == Matrix(T')
issue has been solved by determining Matrix(T')
as Matrix(T)'
(not a very elegant solution I admit!)
Let T be, for example, a Lyapunov operator T:X -> Y, which maps the upper triangular parts of symmetric matrices X to the upper triangular parts of the symmetric matrices Y computed as Y := AX+XA'. Similarly, the transposed operator Tt =T' maps X to Y := A'X+XA. Consider the example:
However we also have, as expected,
Fixing this inconsistent behaviour would be very desirable in a future release.