se-sic / splconqueror Goto Github PK

Mir sind 2 Fehler beim Hinzufügen von Non-Boolean Constraints untergekommen.

1. Bei der Eingabe des Ausdrucks "nx - ny = 0" (nx und ny sind numerische Feature, die root untergeordnet sind) wird nach dem Hinzufügen von der Variable ny leider kein Operator, sondern eine weitere Variable/Zahl erwartet.
2. Sobald ein Non-Boolean-Constraint hinzugefügt wurde, wird dieser nicht in der XML-Datei abgespeichert.

Ich habe mir gerade ein paar Alternativen angesehen und habe bemerkt, dass leider immer noch ein Bug beim Anwenden des Mappings besteht.
Er kann mit folgendem Shellbefehl reproduziert werden:
a Dune::PDELab::QkLocalFiniteElementMap < GV, GV::ctype, Real, degree >

Ausgabe ist da mitunter:
Dune::PDELab::Pk1dLocalFiniteElementMap < class D,class R >

Im Mapping steht:
D -> GV::ctype
R -> Real

Ein Vergleich der ursprünglichen kompletten Namen ergibt:
Dune::PDELab::QkLocalFiniteElementMap<GV, D, R,std::size_t>
Dune::PDELab::Pk1dLocalFiniteElementMap<D, R>
Eigentlich müsste das Mapping 2 Matches ergeben und es entsprechend einsetzen. Allerdings ist das Ergebnis:
Dune::PDELab::Pk1dLocalFiniteElementMap < class D,class R >
Eigentlich erwartetes Ergebnis:
Dune::PDELab::Pk1dLocalFiniteElementMap < GV::ctype, Real >

In my workspace, I previously had a version of SPLC that contained a 'hard copy' of ViPe.

Now, ViPe is inserted via the git submodule mechanism.

Could you document that it is necessary to pull also the submodules?

In my case, it was enough to do the following:

git submodule update --init --recursive

Probably, people cloning a fresh copy of SPLC will not have this problem, but I wanted to remark this anyway.

After SPLC finished learning, the result only takes some of the features into account. Does this mean, that it could not find an interaction between these features and the corresponding non-functional property or is this typically rather a problem of my encoding? Or, as I limit the time for learning to one hour: Does it make sense to let the learning process simply run longer although we typically had good results with the one hour time limit?

Referring to #40, I tried both solutions described there.

I tried to compile SPLC with VS 2013, which did not work because SPLC uses C#7 features (e.g., import static).

Could you extend you instructions in the README.md file regarding VS 2017 being required?

I have the problem that I get an InvalidOperationException in line 604 of Commands.cs:
GlobalState.logInfo.logLine("Termination reason: " + learnedModel.LearningHistory.Last().terminationReason);

The reason for that is that learnedModel.LearningHistory is empty. What could be the reason for this? In general this might also be something which could be checked instead of crashing. Thanks a lot for any idea!

In a configuration file, I have an invalid "all " specification.
Instead of reporting that the configuration is not valid or cannot be read, I received an "InvalidArgumentException"

The location is here:

• File: CommandLine -> Commands.cs
• Line: 432
• Expression: ConfigurationReader.readConfigurations(task.TrimEnd(), GlobalState.varModel)
  • Returned null in my case.

It would be good to have a more expressive exception here that states that the 'task' does not represent a valid configuration.

Content was already merged into master branch and is no longer required.

Twise sampling with t=1 produces a bigger set of configurations than Featurewise sampling.