PReMo version 1.3.1:

    * bug fixes

PReMo version 1.3:

    * support for specifying discrete-time Quasi-Birth-Death (QBD) 
processes via a probabilistic 1-counter automata (p1CA)
    * support for analyzing termination probabilities, expected reward, 
second moment of the reward and stationary distribution of a QBD (p1CA)
    * steady state probability analysis gives meaningful result only if 
the QBD is irreducible and positive recurrent
    * meaning of the variables: g_i_j - (i,j)-th entry of matrix G; 
r_i_j - (i,j)-th entry of matrix R; rw_i_j - expected reward starting at 
(i,1) and if terminating at (j,0); q_i_j - second moment of that reward; 
u0_i - steady state probability of being at (i,0); u_i - steady state 
prob. of being at (i,1); for level k it can be computed as uk_i = u_i * 
R^k
    * small example QBDs can be downloaded from the homepage
    * C++ style // comments can be used in a source code of any model
    * changes in fonts and syntax coloring
    * equations can be edited/added after begin generated 

PReMo version 1.2:

    * strategy improvement finally accessible from the GUI
    * generating expected termination time equations for Stochastic 
Context Free Grammars
    * new universal solvers: Bad Broyden and a modified version of 
Sparse Newton that caches the Jacobian matrix so that it can be modified 
later
    * new efficient Dense and Sparse Linear Solvers for linear equations 
systems
    * cached Sparse Newton is designed to be used with the strategy 
improvement for non-linear equations systems arising in termination 
1-RSSGs whilst dense and sparse linear solvers should be used for 
equations systems arising in the context of reward 1-RSSGs
    * Dense Linear Solver is robust in a sense that it returns the 
correct solution even if some of the values of the least fixed point 
(LFP) are infinite and on top of that it generates warnings if the 
solution might be significantly influenced by the arithmetic errors.
    * strategy improvement available for the maximizer and as well as 
for the minimizer (rather experimental feature, since using it the found 
solution might not be the LFP solution)
    * three variants of strategy improvement for each of the players 
resulting in few hundreds ways of solving a given min-max equation 
system
    * PReMo restarts itself with 256MB heap memory available to it. It 
is also possible to run it with any given amount of memory by issuing 
eg: java -jar PReMo-1.2-win.jar 512 to allocate 512MB to PReMo. If your 
system does not have 256MB available then you have to run it from the 
command line as desribed since otherwise it will just generate an 
exception and exit.