======
1.04-1
======
	API Changes
	-----------
	1) Changes to the way clusteringCoefficient() is called
	
	Major Changes
	-------------
	1) Fixed clusteringCoefficient() so it now correctly calculates the Watts and Strogatz local
	   clustering coefficient - Thanks to Michael McAssey for pointing this out. It also now uses a
	   Fortran implementation for speed.
	
	Minor Changes/Bug Fixes
	-----------------------
	1) Minor source code formatting
	
======
1.03-1
======
	API Changes
	-----------
	1) added a new function, getEdgeList(m), which converts an adjacency matrix into an edge list 
	
	Major Changes
	-------------
	1) None
	
	Minor Changes/Bug Fixes
	-----------------------
	1) Updated checks on class of object m passed into idxInvert(m, idx) and pcitMemoryRequirement(m)
	2) Refactored .Last() which closes slave processors if Rmpi was used
	3) Refactored some code in pcit()to make handling force.serial and force.parallel options clearer
	4) Updated a warning message when NA's are found in the adjacency matrix during pcit()
	5) Updated some indentation in the source
	
======
1.02-2
======
	API Changes
	-----------
	1) none
	
	Major Changes
	-------------
	1) None
	
	Minor Changes/Bug Fixes
	-----------------------
	1) Fixed idxInvert() and pcitMemoryRequirement() so the user can now correctly pass a numeric or matrix as described in the documentation 
	2) Added a CITATION file with details to be used when citing this package. See: citation("PCIT") for details
	3) Cleaned up the demo file. In particular how to use get/use the results produced by pcit() and the use of plotCorCoeff()
	
======
1.02-1
======
	API Changes
	-----------
	1) pcit() now returns the indices of the significant/meaningful connections rather than the non-significant
	2) plotCorCoeff() now takes a list of indices to superimpose multiple distributions using the indices of the values
	3) pass.file removed in favour of pass.type which can take one of several different values, the default being "file"
	4) defineTasks() parameter nGenes changed to n to make it a bit more generic
	5) Added the following functions:
	   a) idx() - returns the indices associated with the result of pcit()
		 b) idxInvert() - Inverts a vector of linear indices from a matrix
	   c) maxMatrixSize() - Attempts to calculate the maximum correlation matrix size that can be handled with a given amount of RAM
	   d) pcitMemoryRequirements() - Attempts to calculate the memory requirement of pcit() for a given correlation matrix
	
	Major Changes
	-------------
	1) When pass.type="memory", the whole matrix is broadcasted to each slave, not just a portion. This is inefficient on memory as each slave creates a copy of the matrix. There is also an upper limit to the amount of data that can be broadcastedto a slave (due to serialisation) that may be reached. However, this change should make it easier to test the use of MP_SHARED_MEMORY=yes env variable and for calculating the significant indices rather than the insignificant ones! which should be faster - eventually.
	
	Minor Changes/Bug Fixes
	-----------------------
	1) Removed the redundant m1 and m2 example data sets and renamed m3 to m.
	2) Added example code to documentation.
	3) Changed warning message when Rmpi loads but fails to spawn slaves
	4) Clarified documentation on what values are used in the distribution plot.
	5) Modified labels on figure produced by defineTasks()
	6) Others!
