andreasvarga / matrixpencils.jl Goto Github PK

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

If you'd like for me to do this for you, comment TagBot fix on this issue.
I'll open a PR within a few hours, please be patient!

Hi, I started sketching out a package for Rational Functions based on Polynomials. The code is here https://github.com/jverzani/PolynomialRatios.jl . I don't know if it is helpful to you, but I could replicate the RationalFunctions code you have and include the transfer function type as a subtype of AbtractRationalFunction. Were I to do so, would I just include the methods you've defined in https://github.com/andreasvarga/DescriptorSystems.jl/blob/main/src/types/RationalFunction.jl or are there more elsewhere in the code base?

The subfunction tgsyl! throw an exception while applying the function

fiblkdiag(A, E, missing, missing; fast=false, finite_infinite = true)

to the regular pencil

A = [-8.828446764485715e7 0.0 0.0 0.0 1.0 0.0; 0.0 -599202.6439090811 0.0 0.0 1.0 0.0; 0.0 0.0 -44599.09517911089 3.1081161629028176e6 2.0 0.0; 0.0 0.0 -3.1081161629028176e6 -44599.09517911089 0.0 0.0; 0.0 0.0 0.0 0.0 -1.0 0.0; 1.4376687293449677e10 33249.50029419384 29200.436106193243 22920.479599602455 -161.90134290419934 -1.0]

E = [1.0 0.0 0.0 0.0 0.0 0.0; 0.0 1.0 0.0 0.0 0.0 0.0; 0.0 0.0 1.0 0.0 0.0 0.0; 0.0 0.0 0.0 1.0 0.0 0.0; 0.0 0.0 0.0 0.0 0.0 0.0; 0.0 0.0 0.0 0.0 -1.8422498299893581e-6 0.0]

The same happens in the sylvsys function (package MatrixEquations) when computing the Weierstrass canonical form as in

http://www.control.isy.liu.se/research/reports/2004/2602.pdf

while using the fischur function for diagonalisation.

Do you have any suggestions for solving this problem?

Polynomials.jl realeased a new major version a while back, and I'm currently facing some compatibility problems in our stack due to MatrixPencils not supporting this new version. Would it be possible to add compat for Polynomials v4?

Today, minreal works on Windows 11/ Julia v1.9. THANKS!

The result from minreal is, however, "wrong" -- or perhaps extremely sensitive to tolerances...

Here is my system eigenvalues -- I insert copyable matrices for the system below:

From physics, there should be 4 states (independent differential equations) and 2 algebraic equations.

• If I apply the minreal function to the system with default tolerances, the system is reduced to a 3 state ODE.
• If I specify the relative tolerance rtol, the system minimal realization has 3 states all the way to rtol=1e-16.
• If I specify relative tolerance at rtol=1e-17, the minimal realization gives 1 state
• If I specify relative tolerance rtol at 1e-18 and smaller, the minimal realization gives 6 states

A = [-6.537773175952662 0.0 0.0 0.0 -9.892378564622923e-9 0.0; 0.0 -6.537773175952662 0.0 0.0 0.0 -9.892378564622923e-9; 2.0163803998106024e8 2.0163803998106024e8 -0.006223894167415392 -1.551620418759878e8 0.002358202548321148 0.002358202548321148; 0.0 0.0 5.063545034365582e-9 -0.4479539754649166 0.0 0.0; -2.824060629317756e8 2.0198389074625736e8 -0.006234569427701143 -1.5542817673286995e8 -0.7305736722226711 0.0023622473513548576; 2.0198389074625736e8 -2.824060629317756e8 -0.006234569427701143 -1.5542817673286995e8 0.0023622473513548576 -0.7305736722226711]

B = [0.004019511633336128; 0.004019511633336128; 0.0; 0.0; 297809.51426114445; 297809.51426114445]

C = [0.0 0.0 0.0 1.0 0.0 0.0]

D = [0.0]

linsys = ss(A,B,C,D)

Hello Andreas, hope you're doing well!

A user of ControlSystems.jl recently reported an error calling lsminreal in Julia v1.9 running on Windows.
The error looks like this

(minreal is a thin wrapper around MatrixPencils.lsminreal)

I suspect the error stems from some manual management of the lapack installation in this package, a search came up with this line mentioning libblastrampoline explicitly