TimQu
|
87c8241ec7
|
policies for games
Former-commit-id: 8bfb325b60
|
9 years ago |
|
TimQu
|
bf450688b4
|
The variable pool of carl needs to be cleared after executing a test.
Sampling for mdps now uses the policy of the previous iteration as initial guess
Former-commit-id: 3b8b25f30f
|
9 years ago |
|
TimQu
|
358bc14ded
|
Merge branch 'future' into TimParamSysAndSMT
Former-commit-id: f54a7a3c88
|
9 years ago |
|
TimQu
|
e8304e7515
|
Merge branch 'master' into future
Former-commit-id: 1ac7368569
|
9 years ago |
|
TimQu
|
b792438d88
|
Added missing include
Former-commit-id: a073cf157d
|
9 years ago |
|
TimQu
|
0089212b7e
|
adaptation to recent changes in carl
Former-commit-id: 90535a0d66
|
9 years ago |
|
TimQu
|
0863d0e51a
|
don't store ALL the occurring policies...
Former-commit-id: f442b97b8b
|
9 years ago |
dehnert
|
1d49bc6dd0
|
extracting the bisimulation quotient for MDPs; tests for MDP bisimulation
Former-commit-id: 5613c653ba
|
9 years ago |
|
dehnert
|
b31d98909b
|
Explicit MDP bisim working but unfortunately slow :(
Former-commit-id: 6714bdbd61
|
9 years ago |
|
dehnert
|
7156a63b0f
|
tried different approach for bisim for MDPs
Former-commit-id: 92d56a4620
|
9 years ago |
|
dehnert
|
c2a0bd5ab0
|
initial outline of strong MDP bisimulation
Former-commit-id: 06452543ea
|
9 years ago |
|
dehnert
|
288f0418c2
|
Merge branch 'future' into mdpbisim
Former-commit-id: 7c75979e67
|
9 years ago |
|
dehnert
|
9bac056354
|
enabled preconditioning for value iteration in gmm++-based MinMax equation solver
Former-commit-id: 39c4efdb52
|
9 years ago |
|
TimQu
|
ca917a651c
|
MinMaxLinearEqSolvers can now use some initial policy as a first guess.
First steps to use this for region approximation
Former-commit-id: 9a8151607f
|
9 years ago |
|
TimQu
|
a723cfca14
|
Made sampling for MDPs correct again
Former-commit-id: 7917200f4e
|
9 years ago |
|
dehnert
|
7833025829
|
reenabled all bisimulation tests
Former-commit-id: 24e8629270
|
9 years ago |
|
dehnert
|
3054aa9866
|
Merge branch 'future' into mdpbisim
Former-commit-id: bfc52e25d9
|
9 years ago |
|
dehnert
|
1de6d4ee07
|
Merge branch 'master' into future
Former-commit-id: ab9d23e0e1
|
9 years ago |
|
dehnert
|
9b7d4ec57b
|
made rational functions use cln again
Former-commit-id: a4b14f9271
|
9 years ago |
|
dehnert
|
51bd689c96
|
fixed a bug in the reward model
Former-commit-id: 5302a85d6f
|
9 years ago |
|
dehnert
|
9475d29164
|
fixed weak bisim for dtmc
Former-commit-id: 05462b99ec
|
9 years ago |
|
dehnert
|
29597e014f
|
more work on reimplementation of weak bisim
Former-commit-id: 5bdd8ea139
|
9 years ago |
|
TimQu
|
e80a1081bb
|
First steps to identify the parameters for which the optimal policy always choses the same boundary
Former-commit-id: c2561730f6
|
9 years ago |
|
TimQu
|
b182e3fdcf
|
Small fix for policy extraction
Former-commit-id: fa37a1ceb2
|
9 years ago |
|
dehnert
|
91bfdba528
|
Scan for equal ranges of probabilities now uses std::equal_range and reduces the number of comparisons
Former-commit-id: 3154d77a3f
|
9 years ago |
|
dehnert
|
46fee522ff
|
made strong bisim for DTMCs work again
Former-commit-id: e42bafef4d
|
9 years ago |
|
TimQu
|
6ddddd8cfa
|
Implemented policy extraction for value iteration
Former-commit-id: 604b4667b8
|
9 years ago |
|
dehnert
|
1f5110b90c
|
work on making bisimulation fast again :(
Former-commit-id: bb89091b2d
|
9 years ago |
|
dehnert
|
2484a515a0
|
some more work on bisim
Former-commit-id: aaa8088b00
|
9 years ago |
|
dehnert
|
40a75baee7
|
using some template magic to make it compile again
Former-commit-id: 69888cac45
|
9 years ago |
|
dehnert
|
11b04c7940
|
more work towards making the new bisim class available from the cl
Former-commit-id: b177287e00
|
9 years ago |
|
dehnert
|
1428f1647b
|
commented in some more tests, however the main entry points need to be fixed because of the new templating of the bisimulation class
Former-commit-id: 7133025049
|
9 years ago |
|
dehnert
|
11c21eb338
|
on my way of making (the refactored version) bisimulation work again for deterministic models
Former-commit-id: 79c089a693
|
9 years ago |
|
dehnert
|
96954ddd15
|
refactoring of bisimulation class in the prospect of extending it to (CT)MDPs, not yet done
Former-commit-id: 09f47ad977
|
9 years ago |
|
TimQu
|
eacfe7a389
|
Merge branch 'future' into TimParamSysAndSMT
Former-commit-id: eccd4d739f
|
9 years ago |
|
TimQu
|
b09d123779
|
...
Former-commit-id: 47c139f3bf
|
9 years ago |
|
TimQu
|
1860502a3a
|
Deterministic states with only constant outgoing transitions are now eliminated
Former-commit-id: be5bf4f7cc
|
9 years ago |
|
dehnert
|
97e302a78c
|
guarded timeouts in z3 by ifdef-guards
Former-commit-id: 1bf463686a
|
9 years ago |
|
dehnert
|
6cfa6ac9c7
|
added timeout to smt solver interface
Former-commit-id: 0003c2c9cc
|
9 years ago |
|
TimQu
|
77c2f397a9
|
fix for approximation model, additional test for mdps, minor changes
Former-commit-id: cc837ddf3e
|
10 years ago |
sjunges
|
b9f8de2c01
|
...
Former-commit-id: 102cc9ea40
|
10 years ago |
|
sjunges
|
703013b97c
|
program, vector, gurobi
Former-commit-id: 6cfaf78d59
|
10 years ago |
|
TimQu
|
c53b79b9b7
|
approximation model, again
Former-commit-id: a7da8e4298
|
10 years ago |
|
dehnert
|
b3ce727f6c
|
fixed minor bug, tests for smt-based permissive schedulers (for upper-bounded properties) now passing
Former-commit-id: bf0261e981
|
10 years ago |
|
dehnert
|
59501dd347
|
removed some object files of xerces. started working on smt-based permissive schedulers
Former-commit-id: de95333225
|
10 years ago |
|
TimQu
|
67ff27954e
|
refactored approximation model (almost done)
Former-commit-id: c7f285906b
|
10 years ago |
|
sjunges
|
160f9e476f
|
test descr for milp perm sched
Former-commit-id: 54dfa2587e
|
10 years ago |
|
TimQu
|
046afd3804
|
Refactored SamplingModel
Former-commit-id: b51ed752b4
|
10 years ago |
|
TimQu
|
f167a46f44
|
Merge branch 'future' into TimParamSysAndSMT
Former-commit-id: f544271dcf
|
10 years ago |
|
sjunges
|
ee0e34146f
|
build command index to action name mapping
Former-commit-id: a9b6c19e68
|
10 years ago |
