==========================
Version: 0.6-1 
Release date:  22 Apr 2010
==========================

* Fixed small bug using equidistant penalization for non-equidistant knots
  in twinSIR.  For 1st order penalty the Equations in Fahrmeir & Lang (2001)
  are used -- see p. 206 of the article.
  The fix has little influence on the results.

* Removed 'extended' argument in 'grep' calls for compatibility with R 2.11.0.

* Changed definition of S4-generic "R0":
  replaced second argument 'experiment' with '...'
  
* function 'animate' is now S3-generic


==========================
Version: 0.6-0 
Release date:  22 Jun 2009
==========================

* All changes in this version concern the spatialSIR part of the package,
  which was introduced with version 0.5-0.
  
* Renamed function spatialSIR to twinSIR and function simSpatialSIR to
  simEpidata. Also renamed all spatialSIR*.R[d] files to twinSIR*.R[d].
  
* Enhanced print.summary.twinSIR to inform about simulated penalty weights.

* Restructured Rd-files concerning twinSIR: moved methods to own
  twinSIR_methods.Rd together with AIC.
  
* Bugfixes for R < 2.8.0: simplify argument in 'by' function exists only since
  R-2.8.0, semi-transparency exists only since 2.7.0.

* Computational improvement in the fitting procedure: removed redundant
  aggregations in the log-likelihood function.

* Computational improvement in the simulation procedure: pre-allocation of big
  evHist matrix (with automatic enlargements if necessary during simulation)
  leads to about 90% time reduction!
  
* Bugfix in 'twinSIR': 'knots' must be stop times and not necessarily event
  times.
  
* New function 'intersperse', which imputes blocks for extra stop times in 
  objects inheriting from class "epidata".
  
* 'animate' and 'stateplot' now also work with "summary.epidata" objects and
  'plot.epidata' is now a wrapper for 'plot.summary.epidata'.
  
* Changed default value for 'optim.args$par' (initial parameter values) in
  'twinSIR' from the unit vector to zeros for 'beta' (endemic) coefficients
  and 1's for 'alpha' (epidemic) coefficients.
  
* Simulation now stops, when the overall infection intensity has a numerically
  infinite value as concurrent events are not allowed.  In contrast, simulation
  now does not end just because all individuals are removed.

* REML based estimation of the smoothing parameter lambda.smooth in twinSIR.
  Some arguments are currently hardcoded.


==========================
Version: 0.5-1 
Release date:  5 Jun 2009
==========================

* All changes in this version concern the spatialSIR part of the package,
  which was introduced with version 0.5-0.

* Bugfix: simulation of epidemic data did not work when all covariates
  were time constant.  Added an example for this simple case in the
  documentation.

* New plotting function 'stateplot', which illustrates the state changes for
  a selected individual during the course of an epidemic.
  
* Introduced [.epidata method, which drops class "epidata" if necessary, i.e.
  if the resulting subset would not be a valid "epidata" object.

* Generalized the simulation function, the methods for "epidata" objects and
  the function intensityPlot so that epidemics with re-susceptibility (i.e.
  SIRS and SIS) can be handled.  This also includes a more detailed summary
  method with its own print method.
  
* Improved handling of meta data in CITATION file (wish of K. Hornik).

* Changed java loading mechanism of the package (make use of .jpackage instead
  of .jinit and added java parameter -Xrs to avoid R crash when sending the
  CTRL-C signal).

* Renamed ".BLOCK." column in the "epidata" class to "BLOCK".

* Truncated lines in plot.epidata to not let them reach beyond the observation
  period of the epidemic.
  
* Simulation now also works with start time != 0. Also I0.col and coords.cols
  can now be NULL, if there are no initially infected individuals or if the is
  no spatial component in the epidemic, respectively.
  
* Major computational improvement in the fitting procedure: removed redundant
  aggregations in the score function, which leads to about 80% time reduction.
  
* Bugfix: functions AIC.epidata and extractAIC.epidata now pass argument k and
  nsim to the internal .OSAICpenalty function.


==========================
Version:	0.5-0 
Release date:	2 Feb 2009
==========================

* Package now contains functionality for fitting the spatial SIR model
  proposed in Hoehle (2008). The model consists of endemic and
  epidemic components modelled by a multivariate counting process
  having additive-multiplicative intensity function. The important new
  functions for this model are: spatialSIR and simSpatialSIR.
  Also new data classes (S3) are defined for handling the results
  from estimation and simulation.