A list of (major) changes between releases, as well as information about MATLAB versions used and toolbox dependencies. Sometimes we add also changes to come to our roadmap.
- Second order models: direct reduction
Dependencies | sss Toolbox |
Programmed with | MATLAB R2016b, R2015b, R2015a |
Tested with | MATLAB R2014a, R2015b, R2016b, R2017a |
on | Windows 7 |
- DOCUMENTATION
- added documentation files to a folder doc/ inside sssMOR. In this way, everybody can profit from the doc documentation of sssMOR, not just those who download the release version.
- added p-Functions publishDoc.p and publishFunction.p to be able to update the documentation whenever headers are changed.
- CPLXPAIRALL
- added function to sort several arrays with respect to the sorting of the first input argument as of cplxpair
- CRKSM
- added a function to solve Lyapunov equations by means of cumulative rational Krylov subspace method
- STABSEP
- added an own function to separate an ustable model into a stable and antistable part. In addition to the built-in function, it returns also the projection matrices resulting in the submodels.
- L2NORM
- added function to compute the L2 norm of a dynamical system. This is especially interesting when dealing with unstable models where the H2-norm is undefined.
- SSS
- updated to v2.00
- SSS is no more a submodule of sssMOR, hence it is not anymore in src/ folder. Instead, sss and sssMOR are now distributed independently and have to be both in the path for sssMOR to work.
- SSRED
- added load and save functions to the ssRed class
- allowing conversion from sss or ss to ssRed objectsf
- made the definition of ssRed objects easier, allowing also the definition of "userDefined" ssRed objects and the definition of arbitrary reduction parameters to be stored in the ssRed object
- IRKA
- allowing to call irka with a reduced order q instead of initial shifts and/or tangential directions
- ARNOLDI, RK, IRKA, CIRKA, ...
- added nLU as output argument to count the number of LU decompositions required by the reduction algorithms
- CIRKA
- added option to take only the stable subsystem of the model function
- RKICOP
- allowing matrix valued shifts as input (for MIMO models)
- CURE
- added as output argument a cell array of all reduced models at each step of the cumulative framework
Dependencies | sss Toolbox |
Programmed with | MATLAB R2015b |
Tested with | MATLAB R2014b, R2015b, R2016b |
on | Windows 7, Ubuntu 16.04.1 LTS |
- TBR
- changed the definition of the Cholesky (like) factors of the Gramians to unify the notation with the low-rank method. They are now lower triangular matrices (A=LL') as opposed to MATLAB's built-in notation.
- CIRKA, MODELFCT, MODELFCTMOR
- added functions for the H2-optimal reduction using the model function framework, which leads to significant speedup, especially for very large scale models.
- ADDED EXTRAS
- moments, isH2opt, getSylvester moved to extras
- shiftVec factorized as own function
- ismemberf2, getDesiredOutput added
- SSS
- updated to v1.03
- MODALMOR
- invariant subspace is computed only using eigs due to its higher efficiency
- TBR
- checking for empty E matrix if model is ssRed
- RK
- checking for empty E matrix if model is ssRed
Dependencies | sss Toolbox |
Programmed with | MATLAB R2015b |
Tested with | MATLAB R2014b, R2015b, R2016b |
on | Windows 7, Ubuntu 16.04.1 LTS |
- minor improvements in MODALMOR, TBR, CURE, SPARK
- SSS toolbox: using v1.02
- SSRED
- Adding hidden properties for system matrices to cope with changed ss-class definition after R2016a
Dependencies | sss Toolbox |
Programmed with | MATLAB R2015b |
Tested with | MATLAB R2014b, R2015b, R2016b |
on | Windows 7, Ubuntu 16.04.1 LTS |
-
APP
- new functionality: plotting of impulse- and step-response
-
SSRED
- new class for reduced state-space-models
-
SPARK
- new output (reduced model, reduction with porkV or porkW)
-
ISRK
- new function: iterative SVD-Rational Krylov Algorithm
-
RKOP
- new function: determine optimal expansion point for Laguerre series
-
RKICOP
- new function: Rational Krylov with an iteratively calculated optimal point
Dependencies | sss Toolbox |
Programmed with | MATLAB R2015b |
Tested with | MATLAB R2014b, R2015b, R2016b |
on | Windows 7, Ubuntu 16.04.1 LTS |
-
APP
-
New sssMOR app added
-
ARNOLDI
- new outputs (matrices of Sylvester equation)
- new input (Opts structure)
- Opts.orth (1xMGS,2x MGS,DGKS)
- Opts.dgksTol
- Opts.reorth (1xMGS, qr)
- Opts.real (0,'real')
- Opts.krylov (cascaded krylov basis for SISO)
- Opts.lse (sparse, full, hess) for solving linear systems of equations
-
RK
- new outputs (matrices of Sylvester equation)
- new input (Opts structure for arnoldi)
-
IRKA
- new outputs (matrices of Sylvester equation)
- new output (kIter)
- Opts.suppressverbose for speed-up
-
MODALMOR
- new output (dominance analysis values)
- Opts.dominance: dominance analysis, most dominant eigenvalue reduction
-
TBR
- Alternating Direction Implicit (ADI) functionality added using the M.E.S.S. toolbox
- plot decay of hsv and ask user (if q not defined)
Dependencies | sss Toolbox |
Programmed with | MATLAB R2015b |
Tested with | MATLAB R2014b, R2015b, R2016b |
on | Windows 7, Ubuntu 16.04.1 LTS |