#################################################################
 	Fourier-Motzkin elimination for information theory
#################################################################

Version: 0.6.7
Kernel: 0.6.7

STI kernel updates:
===================
- Tolerance bug fix
- MATLAB solver algorithm changed from 'simplex' to 'dual-simplex'.

#################################################################

Version: 0.6.6
Kernel: 0.6.6

STI kernel updates:
===================
- Memory comsumption reduced by sparse matrices.

FME-IT updates:
===============
- Random variables are now defined from both equations and the PMF.

#################################################################

Version: 0.6.5
Kernel: 0.6.5

STI kernel updates:
===================
- Fixed bug using Gurobi solver. Non-negativity inequalities won't be removed during the process.
- Fixed bug in sti_pmf2consts.m.

FME-IT updates:
===============
- NOTE: To handle a pmf with variables that are not included in the inequalities, instert dummy inequalities such as H(X)>0. 
  See LiangBCrelay.m for example.

#################################################################

Version: 0.6.4
Kernel: 0.6.4

STI kernel updates:
===================
- Fixed bug in sti_pmf2consts.m.
- sti_genelem.m and sti_pmf2consts.m provides strings with the information measures of the constraints.
- MATLAB solver is available.
- Zhang-Berger inner bound (example) - correct pmf is now provided.

FME-IT updates:
===============
-   To handle variables that are described in PMF but not given in inequalities, enter dummy constraints such as H(X)>0.
    See ZhengBergerInnerBound.mat for example.
