2020-02-25 Luigi Augugliaro version 2.1.6
* fixed bug in dglars.fit
* fixed bug in grcv
* fixed bug in cvdglars.fit
2019-12-03 Luigi Augugliaro version 2.1.5
* fixed bug in pc_cvdglars.f90
2019-11-21 Luigi Augugliaro version 2.1.4
* fixed bug in dglars.R
2019-06-12 Luigi Augugliaro version 2.1.3
* fixed bug in pc_cvdglars.f90
2019-06-12 Luigi Augugliaro version 2.1.2
* fixed bug in dglars.h
2019-04-06 Luigi Augugliaro version 2.1.1
* fixed bug in logLik.dglars
2018-10-09 Luigi Augugliaro version 2.1.0
* added "grcv" function to estimate the dispersion parameter
* fixed bug in cvdglars
2018-04-12 Luigi Augugliaro version 2.0.1
* cBind() is replaced by cbind()
2016-11-17 Luigi Augugliaro version 2.0.0
* the package 'Matrix' is used to handle sparce algebra in R.
* PC algorithm is extended to inverse gaussian, gamma and gaussian family.
* PC algorithm can fit dglars model to glm with standard link functions.
* if family is "binomial" then the respnce vector can be a matri with two colums,
i.e., the first "column" is the number of successes and the second column is
the number of "failures".
* added phi_hat subroutines to estimate the dispersion parameter when family is
inverse gaussian, gamma or gaussian.
* added argument 'unpenalized' to dglars/dglars.fit functions to fit dglars models
where a subset of parameters is unpenalized.
* added methods functions: coef.dglars, coef.cvdglars, predict.dglars, logLik.dglars,
AIC.dglars and BIC.dglars.
* improved the output of the method functions: print.dglars, summary.dglars and plot.dglars
* gdf function is extended to inverse gaussian, gamma and gaussian family
2014-08-09 Luigi Augugliaro version 1.0.5
* added reference to Journal of Statistical Software paper
2013-12-19 Luigi Augugliaro version 1.0.4
* output from cvdglars() function has been improved including an object of class
"formula" used to describe the model estimated by cross-validation
* output from cvdglars.fit() function has been improved including a character vector
with the name of the variables selected by cross-validation.
2013-10-08 Luigi Augugliaro version 1.0.3
* bug fixed in make_dglars
2013-08-04 Luigi Augugliaro version 1.0.2
* dglars() function has been changed to handle the formula interface; dglars.fit()
works with the design matrix X and the responce vector y.
* The control argument "method = dgLAR" has been changed in "method = dgLARS".
* The output from print.dglars() has been changed.
* The output from summary.dglars() has been changed.
* The summary.dglars() implements the a general measure of goodness-of-fit.
* The function plot.dglars() has been changed in order to evauate the fitted models
by means of a general measure of goodness-of-fit.
* gdf() is now a function of a fitted dglars object.
* cvdglars() function has been changed to handle the formula interface; cvdglars.fit()
works with the design matrix X and the responce vector y.
2013-06-14 Luigi Augugliaro version 1.0.1
* nasty bug fixed in fortran - removed array-bounds problems