I grew tired of always starting gdb when it would've sufficed to know the function.
This routine will demangle C++ symbols, so you can see in which function we crashed.
FATAL_LOG_LEVEL: Use, if we are going to crash.
ERROR_LOG_LEVEL: Use, if there is no reasonable way to continue.
WARN_LOG_LEVEL: Use, if we got something the average user should read.
INFO_LOG_LEVEL: Use, if this might in some cases be of interest.
DEBUG_LOG_LEVEL: Use, if this should usually not be relevant to a user.
TRACE_LOG_LEVEL: Use only during development.
There are three levels of verbosity:
- default: WARN and above
- verbose: INFO and above
- debug: DEBUG and above
FATAL_LOG_LEVEL: Use, if we are going to crash.
ERROR_LOG_LEVEL: Use, if there is no reasonable way to continue.
WARN_LOG_LEVEL: Use, if we got something the average user should read.
INFO_LOG_LEVEL: Use, if this might in some cases be of interest.
DEBUG_LOG_LEVEL: Use, if this should usually not be relevant to a user.
TRACE_LOG_LEVEL: Use only during development.
There are three levels of verbosity:
- default: WARN and above
- verbose: INFO and above
- debug: DEBUG and above
* Removed templates from various functions, as we can only produce double models anyway.
* Added methods to compute state rewards and state labelings
* Added toModel() routine
The new adapter will generate an intermediate datastructure that holds all transitions to be inserted.
This will combine the two phases (computing the state space and actually generating the matrix) and can also be used for dtmc and mdp models.
This datastructure is only a list of maps for each state. Each map represents a nondeterministic choice and maps target node ids to their probability.
Implemented multithreading using TBB inside of GMM for usage in Sparse Matrix Multiplication against Dense Vectors
Usage: #define GMM_USE_TBB to enable TBB, additionally define GMM_USE_TBB_FOR_INNER to enable multithreading for EACH row (only feasible of the number of NNZ per Row is large - as in near dense)
dehnert
gereon
PBerger