Sebastian Junges
Sebastian Junges
For future reference: It would be very convenient if we could support instantiations checkers for non-graph preserving assignments; potentially via a flag that must be set explicitly and which leads...
What is the right way to force an MDP to being a DTMC, without much overhead? That is, if I have an MDP (e.g., after applyScheduler with a full scheduler)...
Bitvector assignments sometimes require reallocations. This is currently done very conservatively, see a discussion in https://github.com/moves-rwth/storm/pull/189. Before we invest time here, it would be good to understand to what extend...
Multiple operations on programs visit all expressions either to modify or to analyse them. This yields a series of awkward code duplications that could be reduced by adding a visitor...
Action masks allow to reduce the edges that we explore during model building, this can drastically reduce overhead in the next-state-generator during simulation.
Various method in storm-pomdp currently miss tests; In particular, the transformations, the nondeterministic belief-tracker, the observation trace unfolded. Some of the qualitative analysis methods have very limited tests.
For future reference, a bug in the JIT. > > ~/i2/storm/release-build/bin/storm --jani ~/Documents/petrinet.jani --prop "Pmax=? [F deadl]" --buildfull -e jit > Storm 1.6.1 (dev) > > Date: Thu Jul 16...
Pointed out in this PR https://github.com/moves-rwth/storm/pull/34 which we never merged. Suggestion: Disable reordering completely or print a warning.
Hi, when trying to combine policy iteration with state elimination, I run into the following bug. Is this combination of methods somehow not a good idea, or does this force...
When working on stormpy, @volkm hinted at the following discrepancy that likely is a bug. This is work in progress :-) > ~/i2/storm/release-build/bin/storm --jani ~/Documents/petrinet.jani --prop "Pmax=? [F deadl]" --buildfull...