v0.4.5, 22 October 2016
New Capabilities:
	1. Implemented shortcut for convolutions of two
	random variables with [0,1] support

Fixed Bug:
	1. Removed pylab from the setup.py file. This is installed
	with matplotlib
	2. Fixed bug in posterior. Does integration directly instead
	of through Truncate when the support of an RV has oo or -oo

v0.4.4, 20 August 2016
New Capabilities:
	1. PlotLimits implemented to change the limits of a plot

Fixed Bugs:
	1. Fixed an issue in IDF where output wasn't returned
		for continuous distributions

v0.4.3, 19 July 2016
New Capabilities:
	1. X.save() method implemented to stored random variables
		as binary
	2. LoadRV() function implemented to load random variables
		from binary

v0.4.2, 12 June 2016

New Capabilities:
	1. New style plots

Fixed Bugs:
	1. Added code that allows users to specify variate 
		in X.variate for NormalRV

Other Comments:
	1. Removed the plot.py file. Plotting no longer depends on it
	2. setup.py file has been updated. pip install now downloads
		all dependencies

v0.4.1,

New Capabilities:

Fixed Bugs:
	1. Imported atan function in dist_type to allow for proper
		initialization of the ArcTanRV
	2. Fixed Transform to work with float fractions in the
		transformation
	3. Fixed bug in Product that didn't allow for the use of
		rational numbers as parameters
	4. Fixed CDF and SF so that values passed outside of the
		support are returned as 0 or 1 instead of returning
		an error
	5. If Convolution or Product cannot compute, APPLPy is now
		programmed to try to reverse the order of the
		calculation before returning and error
	6. Assumptions added to the following some distributions to 
		make computation of CDF more feasible
	7. Truncate modified to divide by the correct normalizing
		constant
	8. SF discrete was modified the handle intermediate values
	9. Added 'shortcut' for Weibull CDF

v0.4.0, 17 APR 2016

New Capabilities:
	1. Discrete Time Markov Chains module
	2. Improved method implmented for converting discrete RVs between
		different functional forms. Now runs more efficiently
	3. Discrete RVs with explicit values and discrete RVs defined by
		expressions can now be manipulated with RV algebra
	4. Sqrt(X) syntax implemented for transformation by sqrt(x).
	5. X**n syntax implemented for transformations by x**n
	6. Markov Chain class added, includes procedures for computing 
		the state of a system after n additional steps
	7. Support added for rational numbers in DiscreteUniformRV
	8. Support for discrete random variables added for the Expected
		 Value function
	9. BernoulliRV type added
	10. RangeStat procedure added for finding the range stastics for
		a random variable
	11. Maximum and Minimum IID now work with symbolic n unspecified
	12. Random Variable method added to simplify functions with
		assumptions
	13. VerifyPDF(X) added for compatiblity with APPL

Fixed Bugs:
	1. BootstrapRV procedure fixed. Duplicate elements are now
		eliminated from the support of Bootstrap RVs.
	2. Pr(X) for elements in the BootstrapRV are now described by
		rational numbers instead of floats.
	3. Bug in transform for RV's with negative support is fixed.
		Transform is now running properly for all supports.
	4. Fixed bug in convolution for lifetime distributions. When
		a floating point value was entered for a parameter,
		the procedure would return 0 for the cdf of the convolution.
		The procedure now works correctly.
	5. Added assumptions to variables in product procedure. No longer
		gives complex functions as output for products that occur in
		the third quadrant.
	6. Corrections made to transform for discrete random variables. No
		longer cuts out parts of the distribution that are not in
		the support of the tranformation, but simply does not 
		transform them.
	7. Corrections made to Transform to account for equality testing with
		SymPy number types
	8. Corrections made to the discrete transform procedure to handle
		situations in which the support of the transformation does
		not cover the full support of the random variable
	9. Added simplification line to end of Truncate procedure
	10. Fixed bug in computing values for discrete pdf
	11. Fixed bug related to floating point numbers in transform
	12. Fixed bug to correct computation of IDFs for Discrete RVs
	13. OrderStat modified to allow for 'wo' for samples of size
		greater than 4
	14. Transform modified to accept symbolic discrete RVs
	15. Convolution now works with rational supports
	16. IDF for symbolic discrete random variables updated to
		start the support at 0
	17. Corrected likelihood function computation in Posterior
	18. Fixed technique for finding area in Truncate
	19. ChiRV now outputs rational numbers
	20. Transform modified to handle transformations by 1/x and log(x)


v0.3.0, 12 JAN 2015

New Capabilities:
	1. The Queue procedure computes the sojourn time distribution
		for an M/M/s Queue for customer n, given k customers
		initially in the queue.
	2. Special MLE algorithms implemented for the Normal, Exponential,
		Poisson and Weibull distributions
	3. Thorough doc strings added for each applpy module
	4. Cache option has been added for functional form conversion and
		expected value procedure. If cache=True, APPLPy stores the
		values in memory to be retrieve later if the function is
		called again. This save significant CPU time on future
		function calls
	5. Doc Strings added for each distribution type
	6. Behavior of +,-,*,/ set to allow random variables to interact
		with floats and ints
	7. == operator will now test for the equality of two random variables
	8. Maximum and Minimum will now accept a variable number of arguments
	9. Box-Muller Transformation method implemented to allow for more
		efficient computation of Normal variates
	10. Users can now specify method for variate generation (generally
		inverse cdf or a numerical method)
	11. X.latex() command added to generate the LaTeX code for a
		random variable

Fixed Bugs:
	1. Maximum and Minimum are now working properly for piecewise
		distributions. Boundary conditions caused to the procedures
		to use the incorrect segments in the earlier
		implementation
	2. The gamma function is now loaded into the namespace in the
		dist_rv module.
	3. Error message added to Transform to indicate if the user has
		specified an undefined transformation

v0.2.1, 18 NOV 2014

New Capabilities:
	1. Pre-processing code has been added to the beginning
		of several procedures. Errors will be raised
		if a procedure receives bad input.
	2. PlotDist has been improved to allow for more plotting
		options. PlotDist is now capable of using different
		colors and displaying multiple plots simultaneously
	3. Discrete and continuous plots can now be displayed in
		the same overlay
	4. Improved display implemented for continuous random variables.
		Displays large piecewise random variabls more cleanly
	5. MLE and MOM now have options for numerical solving
	6. Procedure added for computing the entropy of random variables

Fixed Bugs:
	1. Start-up script has been modified. The init_session()
		commands for SymPy are entered manually. APPLPy
		can now be imported in an interactive command
		line session
	2. Updates have been made to the way the APPLPy package is
		structured. Importing APPLPy is now less memory
		instensive (does not import all of SymPy, etc)
	3. KSRV now runs properly ... arrays are no initialized
		with 0 instead of None to avoid TypeErrors
		when variables are summed
	4. Minimum and Maximum procedures have been debugged for
		discrete random variables. Old version had issues
		recognizing floating point numbers in the support

v0.2.0, 17 SEPTEMBER 2014

New Capabilities:
	1. Bayesian statistics module has been added. Capabilities
		include:
		a. Computing posterior distributions
		b. Computing posterior predictive distributions
		c. Computing Jeffreys Priors
		d. Credible Sets
	2. Functionality added for discrete random variables in
		functional forms for the following procedures:
		a. All procedures for changing functional form
			(i.e. CDF to PDF)
		b. All expected value procedures
	3. New Random Variable types added (new distributions are
		listed in the startup menu)
	4. KS Test added to the statistics module

Fixed Bugs:
	1. PlotDist now supports univariate discrete random
		variables
	2. Random Variables no longer default to floating point
		numbers when it produces output. Computation of
		exact distributions is now possible
	3. PlotDist now appears in the iPython Notebook when the
		command %pylab inline is entered

v0.1.2, 30 JULY 2014

New Capabilities:
	1. Capability to work with RV's with symbolic supports was
	added.

v0.1.1, 29 JULY 2014

Fixed Bugs:
	1. Convolution/Product/MaximumIID procedures have been updated
	so that they now function with all distributions. In the
	initial release, they only worked with lifetime distributions.

	2. Expected value procedures were modified to output
	simplified expressions

	3. init.py file modified so that all dependencies are
	automatically imported

	4. Plotting now takes place through the sympy plotting front-
	end. The PlotDist procedure is now capable of plotting all
	functions

v0.1.0, 23 JULY 2014 -- Initial Release
