2015-06-26  Andy Leung  <andy.leung@stat.ubc.ca>
    * 3.2.2
	* Increased the number of subsamples in EMVE.
	* Fixed seed problem
	
2015-06-26  Andy Leung  <andy.leung@stat.ubc.ca>
    * 3.2
	* Implemented fast 2SGS as described in the rejoinder of the 2SGS paper
	* Implemented the partial implementation procedure as described in the rejoinder of the 2SGS paper
	* Removed the accessor -- getInitial (unecessary)
	* Added 'rrcov' package in the imports field

2015-04-08  Andy Leung  <andy.leung@stat.ubc.ca>
    * 3.1.1
	* Fixed set seed issue
	* Fixed the condition number restriction bug in the estimation of EMVE

2014-08-04  Andy Leung  <andy.leung@stat.ubc.ca>
    * 3.1
	* Convert the iteration code of EM to all C++ (or Rcpp)
	* Convert the subsampling and concentration part of emve to all C++ (or Rcpp)
	* Include the initial estimates mu0 and S0 as an attribute of the GSE class
	* Include the imputed data matrix, weights, and first deriviative of weights used in the location estimation as new attributes of the GSE class
	* Create a new accessor -- getInitial -- for the new attributes of the GSE class (see above bullet)
	* Include TSGS, a new function that computes 2SGS estimate for cell- and case-wise outliers
	* Include Geochemical data that is analyzed in the 2SGS article

2013-06-04  Andy Leung  <andy.leung@stat.ubc.ca>
    * 3.0
	* Convert to S4 object system
	* Convert the iteration code of GSE to all C++ (or Rcpp)

2013-02-13  Andy Leung  <andy.leung@stat.ubc.ca>
	* 2.4.2
	* Fix the bug in rcondCov, now calculated condition number based on eigenvalues

2012-08-04  Andy Leung  <andy.leung@stat.ubc.ca>
	* 2.3
	* Include function to plot and print the result for different estimators
	* Fix the rand seed for emve C++ code
	* Update all the manuals

2012-07-15  Andy Leung  <andy.leung@stat.ubc.ca>
	* Fix the bug of asymetric covariance matrix when iteration is large
	* Algorithm stop when negative eigenvalue appears

2012-06-12  Andy Leung  <andy.leung@stat.ubc.ca>
	* 2.2
	* Include a line for only allowing up to dimension 50
	* Include GMME function
	* Extract cov.EM and slrt functions (and their related) from robmiss packages
	* No dependency on robmiss package

2012-06-10  Andy Leung  <andy.leung@stat.ubc.ca>
	* 2.1
	* Include the function to compute Quadrant Covariance
	* Include the function to compute Composite S-estimator
	* Include the version of GSE with initial estimator as quadrant covariance


2012-05-12  Andy Leung  <andy.leung@stat.ubc.ca>
    * 2.0
    * fixed bugs in the iterative step of GSE code, i.e. computing GSE based on fixed initial estimator 
	* fixed bugs in computation of scale in GSE
	* include GSE with fixed point approach
	
2012-04-30  Andy Leung  <andy.leung@stat.ubc.ca>
	* 1.1
	* GSE is rewritten in C++ at least for the iterative step but not the loop
	* Reference manual for GSE is included
	* Faster implementation of both EMVE and GSE by performing the computation
	  on groups with same missing patterns 
	* Code for simulating contaminated data set is included
	
2012-03-17  Andy Leung  <andy.leung@stat.ubc.ca>
	* 1.0
    * Depends: R (>= 2.14.1), robmiss, Rcpp (>= 0.9.8), RcppArmadillo (>= 0.2.34)
	* Code for GSE computation is included
	* Faster implementation of both EMVE scale and partial mahalanobis distance
	* Computation of ML-estimate of location and scatter for each candidate within
	  the resampling procedure is restricted to only 5 iterations
	
2012-02-14  Andy Leung  <andy.leung@stat.ubc.ca>
	* 0.1
	* Reference manual for mve.missing.C is included
	* Computation of scale for EMVE and partial mahalanobis distance is coded in C++
	
2012-01-14  Andy Leung  <andy.leung@stat.ubc.ca>
    * 0.0
    * Depends: R (>= 2.14.1), robmiss, Rcpp (>= 0.9.8), RcppArmadillo (>= 0.2.34)
    * Only include the function to compute EMVE
	* Resampling procedure is coded in C++ 
	* No code for GSE computation is included