README_TODO

check for bugs with:
valgrind --leak_check=yes discotress


To do:
-deadts member of Edge class only gets used in un-eliminating node in kPS (indicates that an edge no longer exists in the restored network) and
 in subsequent calculation of updated numbers of i<-j transitions. The h member of the Node and Edge classes only get used in kPS if not
 statereduction, and the dt member of the Edge class only gets used in kPS with certain state reduction methods.
 Could try and optimize memory usage and performance in these regards.


Other comments:
-discretetime is a member of Traj_Method but branchprobs is a member of Network; I think this makes sense though


Known bugs:
-STEADYSTATE (WRAPPER FIXEDT) simulations do not compute committor probabilities (NB committor probs will be same as for noneq result)
-in FIXEDT, if STEADYSTATE, I think hits to A (for calculating SS MFPT) are counted twice if no BKL steps are taken
-committor probabilities for initial state wrong if there is a direct connection between nodes of the initial and absorbing sets