KerMor  0.9
Model order reduction for nonlinear dynamical systems and nonlinear approximation
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Pages
Changes in Version 0.7

Final version of KerMor at the end of my PhD time (plus some extra)

Branch name: master

Class approx.algorithms.Componentwise
(Daniel Wirtz, 2011-11-22) Moved Componentwise.guessGammas to kernels.config.ExpansionConfig
Class approx.algorithms.VKOGA
(Daniel Wirtz, 2012-11-26) Renamed to VKOGA and starting to build in IClassConfig interfaces
Class data.FileData
(Daniel Wirtz, 2013-03-27) Improved the modularity of FileData with more verbosity and implemented a basic "relocate" method.
Member data.ModelData.ModelData (varargin)
(Daniel Wirtz, 2013-04-15) Using the SaveTag + ID to determine the data storage location if a BaseFullModel is passed. Then, a second parameter may be the base directory for the model's data directory.
Class Devel
(Daniel Wirtz, 2013-01-23) Removed the overwrite flag and added option to specify a path in which the file should be created.
Class dscomponents.ParamTimeKernelCoreFun
(Daniel Wirtz, 2014-01-15) Removed the inheritance from kernels.ParamTimeKernelExpansion and included an instance instead. (Favor inclusion over inheritance, Warwick said!)
Class general.interpolation.KernelInterpol

(Daniel Wirtz, 2014-01-24) Removed the preconditioning stuff and LU factorization. Instead included the Newton basis computation method from [7] and controllable interpolation precision via RelTol

(Daniel Wirtz, 2013-01-23) Re-added the LU decomposition stuff to this class from MemoryMatrix, as the class has been removed.

Class general.regression.ScalarEpsSVR
(Daniel Wirtz, 2013-01-23) Removed QP solvers and introduced intermediate class BaseQPSVR to include qp stuff.
Class general.regression.ScalarNuSVR
(Daniel Wirtz, 2013-01-23) Removed QP solvers and introduced intermediate class BaseQPSVR to include qp stuff.
Class IParallelizable
(Daniel Wirtz, 2014-01-17) Removed the KerMor.UseMatlabParallelComputing flag and hence this class is only of minimal functionality now. However, having an interface to indicate parallel computation capability is not making anything else more complicated, which is why this remains in KerMor.
Class JaRMoSExport
(Daniel Wirtz, 2013-08-11) Moved the +export package from KerMor to an own git repository (and re-included it in KerMor in extern/matlabtojarmos)
Class KerMor

(Daniel Wirtz, 2014-01-17) Removed the Flag "UseMatlabParallelComputing" as this should/must be set up separately on each machine. Instead, manually opening the matlabpool is required. Moreover, the flags models.BaseFullModel.ComputeParallel are used to determine the way of execution.

(Daniel Wirtz, 2013-07-09) Removed the TempDirectory as customizable property and simply using "DataDirectory/tmp" now.

(Daniel Wirtz, 2013-01-23) Done some restructuring in KerMor

  • Removed thee QP-Solver packages and extern/ folders as they are effectively no longer used
  • Removed associated tests
  • Moved the ODE solvers from solvers.ode to solvers
  • Moved the IClassConfig interface to +general
  • Adopted the KerMor install script to the changes
  • Removed two no longer used approx.algorithms classes (MinMax and VectorialOMP2Step)

(Daniel Wirtz, 2012-11-16)

  • Moved some tools from the extern folder into a new separate git repo "MatlabTools", including the previous general.collections package
  • Changed the warning for missing mosek quadprog
  • Adjusted the directories that are being added to PATH at startup
Class kernels.ARBFKernel
(Daniel Wirtz, 2013-01-16) Moved the Gamma property to here as it is a common setting for all RBF kernels.
Class kernels.BaseKernel
(Daniel Wirtz, 2014-01-24) Removed getDefaultConfig. Exactly a year since added :-)
Class LogPlot
(Daniel Wirtz, 2013-03-14) postprocessing does not call axis tight if axis limits have already been set manually.
Member models.BaseModel.computeTrajectory (colvec< double > mu,integer inputidx)
(Daniel Wirtz, 2013-03-19) Fixed the MaxStep setting of the ODE solver when time-scaling is used.
Class models.pcd.PCDModel
(Daniel Wirtz, 2013-03-14) Moved constants from other pcd classes to this class.
Class MUnit
(Daniel Wirtz, 2014-04-15) Added a convenience interface for easy skipping of previously successful tests (via passing an "exclude" cell array and retrieving a "successful" cell array)
Class PlotManager
(Daniel Wirtz, 2014-01-15) Changed the property "SingleSize" to "FigureSize". The sizes are now used whenever a new figure is created, independently of being in single or subplot mode.
Class PrintTable

(Daniel Wirtz, 2013-08-10) The format cell argument can now also hold function handles for advanced formatting.

(Daniel Wirtz, 2013-04-09) Added some table cell spacing by default ( \(\LaTeX\) arraystretch)

(Daniel Wirtz, 2013-04-05)

  • Automatically stripping newline characters from any strings
  • Added some more verbose output about table creation to \(\LaTeX\) output

(Daniel Wirtz, 2013-02-19)

  • Bugfix: Accidentally also wrapped in $$ for plain text output by default.

(Daniel Wirtz, 2013-02-15)

  • Bugfix for empty cell string
  • LaTeX alignment for first column is "r" if HasRowHeader is used

(Daniel Wirtz, 2012-12-11) Improved automatic setting of TabCharLen default values.

Class sampling.GridSampler
(Daniel Wirtz, 2013-09-05) Moved the Spacing property to data.ModelParam
Class solvers.AJacobianSolver
(Daniel Wirtz, 2013-05-16) Renamed this class from AImplSolver to AJacobianSolver. Now providing the jacobian stuff via this class independently from explicit/implicit solver tactics.
Class solvers.BaseCustomSolver
(Timm Strecker, 2013-01-24) accelerated method getCompTimes
Member solvers.BaseCustomSolver.customSolve (odefun, t, x0, outputtimes)
(Daniel Wirtz, 2013-01-11) Passing the outputtimes parameter to the customSolve method in order to allow for more memory-efficient implementations.
Member solvers.SemiImplicitEuler.customSolve (unused1,rowvec< double > t,colvec< double > x0,rowvec< integer > outputtimes)
(Daniel Wirtz, 2013-01-11) Using the outputtimes parameter in order to provide a more memory-efficient implementation.