dehnert
|
df0b5fbfa5
|
fixed multiply-reduce operations in the presence of empty row groups
|
8 years ago |
TimQu
|
17a45a43eb
|
better transformation of memory states
|
8 years ago |
|
dehnert
|
d25cc4b05f
|
first version of sound value iteration
|
8 years ago |
|
TimQu
|
7d4a438e82
|
Fixes for lower bounds
|
8 years ago |
|
dehnert
|
ec61e110f2
|
introducing solver formats to enable linear equation solvers to take the fixed point rather than the equation system formulation
|
8 years ago |
|
TimQu
|
b054b67312
|
first version for lower bounded properties
|
8 years ago |
|
TimQu
|
4ba20d11d4
|
more functionality for epoch manager
|
8 years ago |
|
TimQu
|
d129683c61
|
gathered dimension related data into a struct. Also started with lower reward bounds
|
8 years ago |
|
TimQu
|
41cf4e76db
|
The solutions are now stored epoch-wise and will be erased as soon as all predecessor epochs are computed
|
8 years ago |
|
TimQu
|
3044aaa3f5
|
The product model is now handled in a separate class
|
8 years ago |
|
TimQu
|
1ccc241462
|
computations on epochs are now handled in a separate class
|
8 years ago |
|
TimQu
|
d3e50b8769
|
optimized productInState Computation
|
8 years ago |
|
TimQu
|
47ab74a16b
|
implemented single objective queries
|
8 years ago |
|
TimQu
|
8b466f1fa7
|
extended multidimensional bounded until formulas to have different subformulas in each dimension
|
8 years ago |
|
dehnert
|
00f88ed452
|
gauss-seidel-style value iteration
|
8 years ago |
|
TimQu
|
cbe73081f8
|
zig zag epoch computation order
|
8 years ago |
|
TimQu
|
6ae05560fc
|
setting solutions in place
|
8 years ago |
|
dehnert
|
9d95d2adcf
|
first version of multiply-and-reduce (only for native)
|
8 years ago |
|
TimQu
|
488479b5a7
|
better objectiveRewardFilter
|
8 years ago |
|
TimQu
|
cd6a79de23
|
used cache memory when checking each epoch
|
8 years ago |
|
TimQu
|
13ade58fa0
|
using pair<uint, uint> in the map instead of vectors of size 2
|
8 years ago |
|
dehnert
|
3c844a487f
|
some more optimizations
|
8 years ago |
|
TimQu
|
3a309caafc
|
epochs are now uints instead of vector<int>s
|
8 years ago |
|
dehnert
|
5fafe835cb
|
started on some optimizations for conditionals in MDPs
|
8 years ago |
|
TimQu
|
27ee299f63
|
more efficient comparison of epoch classes
|
8 years ago |
|
TimQu
|
cf91a34970
|
use cached memory for the epoch results
|
8 years ago |
|
TimQu
|
9735ff98d0
|
SolutionType is now a single vector instead of a struct
|
8 years ago |
|
TimQu
|
d6099a91a7
|
Further optimizations for setEpoch + started to work on single objective mode
|
8 years ago |
|
TimQu
|
57e604a815
|
optimized setCurrentEpoch a little
|
8 years ago |
|
TimQu
|
e4ac32a6fc
|
fixed selection of considered state in reduced model and optimized setting the stepsolutions
|
8 years ago |
|
TimQu
|
11fc1d29eb
|
more output for statistics
|
8 years ago |
|
dehnert
|
9bda631795
|
symbolic MDP helper respecting solver requirements
|
8 years ago |
|
dehnert
|
7c24607427
|
started on symbolic solver requirements
|
8 years ago |
|
dehnert
|
e81d979d56
|
hybrid MDP helper respecting solver requirements
|
8 years ago |
|
TimQu
|
1bb8b3b497
|
only store the results that will be needed eventually
|
8 years ago |
|
TimQu
|
62e4558af1
|
reduced number of states considered in the epoch model
|
8 years ago |
|
TimQu
|
8e56efed3a
|
reduced the number of states in the product model
|
8 years ago |
|
TimQu
|
999fd0752c
|
The model memory product can now retrieve the reachable states without actually building the product
|
8 years ago |
|
TimQu
|
11a4f8d016
|
restructured initialization of modelmemory product
|
8 years ago |
|
dehnert
|
a3cbaedcc1
|
intermediate commit to switch workplace
|
8 years ago |
|
TimQu
|
90a7ea4907
|
printing timing information
|
8 years ago |
|
dehnert
|
12b10af672
|
started on hybrid MDP helper respecting solver requirements
|
8 years ago |
|
TimQu
|
61a6b178b7
|
restricted the number of reachable product states
|
8 years ago |
|
dehnert
|
4c5cdfeafc
|
Sparse MDP helper now also respects solver requirements for reachability rewards
|
8 years ago |
|
dehnert
|
74eeaa7f81
|
computing unbounded until on MDPs with the sparse helper now respects solver requirements
|
8 years ago |
|
TimQu
|
40edc0ca4d
|
assert that there are no endcomponents with positive rewards
|
8 years ago |
|
TimQu
|
529526593b
|
improved structure in rewardunfolding
|
8 years ago |
|
TimQu
|
de1e9967a3
|
added some necessary conditions for the rewardbounded weight vector checker
|
8 years ago |
|
TimQu
|
0cca4a51d0
|
added weight vector checker for reward bounded objectives
|
8 years ago |
|
TimQu
|
721dd37a62
|
Moved reduction of the preprocessed model into the weightvectorchecker
|
8 years ago |
