0.1.0 - 2010-06-15 - First released version (TIES2010)


0.1.1 - 2010-06-28 - Minor fixes of functions
Updates:
function: default.LUR.list
fix: Changed default behavious so that it now picks all variabels 
 and not a number equal to the temporal trends

function: cond.expectation.params, predictCV
fix: crossvalidation now handels groups with only one site (a number 
 of drop=FALSE added)

function: cond.expectation
fix: Added drop=FALSE to handle no temporal trend

function: cond.expectation.params
fix: uses only y, x, and ID when combining locations. This allows for 
 pass through of other components in mesa.create.model without causing 
 follow on problems

function: create.data.model
fix: allows pass through of all elements in the location vector


0.1.2 - 2010-06-28 - Minor bug fixes of functions
functions: create.data.model
fix: Now drops times not in the range of observed times both for the trend
 and for the SpatioTemporal covariate (latter is the fix)

function: cond.expectation
fix: Corrected behaviour when predicting all points with
  spatiotemporal covariates
fix: Removed creation of X, since it can be found as covar.aux$X.


0.2.0 - 2010-06-28 - Major rewrite of conditional expectaions
function: cond.expectation, cond.expectation.params
fix: 'cond.expectation.params' removed, usage transfered to 'cond.expectation'

function: loglike
fix: Removed option, 'cond.exp=FALSE', functionallity transfered to 
 'cond.expectation'

function: loglike, loglike.naive, cond.expectation
fix: Added check of parameter size.

function: Several entries
fix: Replaced calls to loglike(...,cond.exp=FALSE), or cond.expectation.params
 with calls to cond.expectation.

manual: Several entries
fix: Minor changes to refelct the changes in loglike, cond.expectation, and 
 cond.expectation.params

manual: block_mult
fix: Added generic help for all c-functions.


0.2.1 - 2010-06-28 - Fixed documentation
manual: Several entries
fix: Replacing several TODO comments, mainly in seealso and examples.

function: cond.expectation
fix: Minor bug fix, used variable before it was declared.


0.2.2 - 2010-06-29 - Minor bug fix
function: createCV
fix: Now re-sets the random seed after using a zero seed. Grouping 
 always uses the same seed to ensure that groups stay the same.


0.2.5 - 2010-06-30 - Added functions that analys cv-results
function: predictNaive
add: Added function that does the naive predictions based on a few sites.
 No help text yet.

function: summaryStatsCV
add: Added function that computes summary statistics for the cross 
 validation. Computes stats are RMSE, R2 and coverate of 95% CI:s.
 No help text yet.

function: compute.ltaCV
add: Added function that computes the long term average for each of the
  sites in the cross-validated predictions.


0.2.6 - 2010-07-05 - Added manual entries for the plotting functions
manual: plotMesaData, plotCV, plotPrediction, compute.ltaCV
fix: Added manual entry


0.2.7 - 2010-07-05 - Added manual entries and additional plotting functions
function: CVresiduals.scatter CVresiduals.qqnorm
add: Added functions that do diagnostics plots for observations and for 
 residuals. Most of the help text has been written.

manual: summaryStatsCV
fix: Added manual entry

manual: predictNaive
fix: Added skeleton, manual entry. Needs to be expanded.


0.2.8 - 2010-07-09 - Added more plotting functions
function: printMesaDataNbrObs
add: Added function that prints summary of number of observations, possably
 broken down by type

function: plotMesaData
fix: Function now takes either mesa.data or mesa.data.model objects

function: plotMonitoringLoc
add: Added function that plots the monitoring space-time locations
 of the monitoring, possibly colour coded by type.


0.2.9 - 2010-07-21 - Bug fixes and added features to combine data
function: calc.smooth.trends
fix: When subsetting to a few sites, the data matrix will now drop rows 
  with no observations (and added warning text)

function: create.data.model
fix: Function now handles the no-data case. Also added warnings for missing
   names in LUR covariates, and spatio-temporal covariates.

function: invCholBlock, makeCholBlock, and solveTriBlock
fix: New handeling of special case n.blocks=1, that should speed up 
  computations

function: combineMesaData
add: Added function that combines a mesa.data.model and a mesa.data object
  taking obs and trend from mesa.data.model, and additional locations, 
  LUR-covariates, and ST-covariates from mesa.data

function: cond.expectation, simulateMesaData
add: Added an option combine.data that will combine the mesa.data.model
  and mesa.data, before doing predictions or simulation.


0.2.10 - 2010-07-27 - Bug fixes
function: simulateMesaData
fix: Corrected behaviour when combining datasets.

function: makeCholBlock
fix: Changed error behaviour, now fails for non-pos-def matrix

function: loglike, loglike.naive
fix: Now does try-catch on the Choleskey


0.3.0 - 2010-07-28 - Handeling of collocated points
function: simulateMesaData
fix: Added mvrnorm if choleskey of Sigma fails to guard against collocated points.

function: cond.expectation
fix: Added more help text and tests for the combine data case.

function: create.data.model
fix: Now subsets the ST.all object to only include relevant dates.
fix: Added check for observation dates outside the range of the trend function.
fix: Add a strip time argument that drops the trends outside of observation range.
fix: Added check for collocated sites with observations.
fix: Added check for collocated observations (same location and time-point)


0.3.1 - 2010-07-28 - Improved simulation speed
function: simulateMesaData
fix: Improved speed when running many replicates, mainly for collocated cases
  where mvrnorm is used.

function: combineMesaData
fix: Minor changes to combining data objects, mainly to fix behaviour for 
  the spatio-temporal covariate.


0.3.2 - 2010-07-28 - More effective conditional expectation
function: cond.expectation
fix: Changed computations to a more efficient implementation, resulting in a 
  speed gain by a factor of 4.
fix: Added a pred.covar=TRUE to compute the entire covariance matrix for the
  each of the predicted time series.
issue: Changes not yet implemented for variance computations when type="r".


0.3.3 -  2010-10-?? - Correction of Rd help files


0.3.4 -  2010-10-13 - Random cross-validation sets.
function: createCV
added: An option random=FALSE was added allowing for different 
  CV-groupings for each call.


0.3.5 -  2010-11-08 - Random cross-validation sets.
function: createCV
added: Implemented the random cross-validation added in 0.3.4.
       Also added an option that gives CV for home, fixed or comco 
       sites. For internal use by the MESA Air project.


0.3.6 -  2010-11-14 - Minor bug fixes
manual: mesa.data
changes: Altered trend help to indicate that date does not have to 
   be the last column

manual: mesa.data.model
changes: Clarified warning

function: create.data.model
changes: moved computations of nt to after observations are dropped.

manul: create.data.model
changes: Added comment that NULL gives only spatial intercept.

function: cond.expectation
changes: mesa.data will now use the trend in mesa.data.model if trend is missing.


0.4.0 -  2010-11-14 - Changed and added existing data
data: mesa.data and mesa.data.model
added: A Spatiotemporal covariate (1500 m buffer Caline)

data: mesa.data.raw
added: The entire raw dataset, along with instuctions for how to create the 
  mesa.data from raw data.


0.4.1 -  2010-11-15 - Minor tweak to plotting functions
function: plotPrediction
added: Now allows mesa.data=NA, plotting only CI:s but no observations

0.4.2 -  2010-11-?? - ???
Manual: loglike
added: See also for loglike.var.names and loglike.grad

Manual: loglike.grad, fit.mesa.model, run.MCMC
added: See also for loglike.var.names

Manual: calc.tF.times.mat
added: An elaborate example

Manual: drop.observations
added: An example

Manual: Several functions
added: minor updates, references etc.

function: default.LUR.list, default.ST.list
fixed: Ensure no duplicate covariates


0.4.3 -  2010-12-12 - Updates to several help files
Updated help files (cross-references etc.):
   plotMonitoringLoc.Rd
   printMesaDataNbrObs.Rd
   create.data.matrix.Rd
   compute.ltaCV.Rd
   sumLogDiag.Rd
   make.sigma.B.Rd
   mesa.data.Rd
   default.LUR.list.Rd
   CVresiduals.qqnorm.Rd
   loglike.grad.Rd
   estimateCV.Rd
   calc.tF.times.mat.Rd
   block.mult.Rd
   predictNaive.Rd
   plotMesaData.Rd
   loglike.var.names.Rd
   run.MCMC.Rd
   calc.iS.X.Rd
   mesa.data.raw.Rd
   construct.LUR.basis.Rd
   loglike.Rd
   dot.prod.Rd
   createCV.Rd
   mesa.data.model.Rd
   make.sigma.nu.Rd
   make.sigma.B.full.Rd
   makeCholBlock.Rd
   loglike.dim.Rd
   fit.mesa.model.Rd
   cond.expectation.Rd
   summaryStatsCV.Rd
   plotCV.Rd
   create.data.model.Rd


0.4.4 -  2010-12-12 - Accounting for the ST covariate in the help-files/examples
Files with fixed examples:
   create.data.model.Rd


0.4.5 - 2011-01-04 - Minor changes to cross-validation
function: createCV, predictCV
fix: Now has an option for using only a subset of sites in the cross-validation,
     remaining sites are keept for all CV-groups.

function: predictCV, estimateCV
fix: Added option of using either a vector or a logical-matrix to define the 
     cross-validation sets.

data: mesa.data.model
fix: Updated data to match new create.data.model function, i.e. $obs now 
     includes an ID object.


0.5.0 - 2011-01-05 - Added beta-field estimation
function: cond.exp
fix: Now also estimates the latent beta-fields, still needs some testing.


0.5.1 - 2011-01-11
function: cond.exp
fix: Bug fixes of the beta-field estimation

function: printMesaDataNbrObs
fix: Now handels structures without trend component correctly

function: plotMesaData
fix: Added plots of pacf and a lag.max pass through for acf and pacf


0.5.2 - 2011-01-25
function: create.data.model
fix: Replaced approx with spline for interpolation the temporal trends
fix: Changed behavious of "strip.time", now drops all time-points that 
     lack observations. This shouldn't break any existing code.

function: plotMesaData
fix: Replaced approx with spline for interpolation the temporal trends

function: combineMesaData
fix: Minor bug-fix of temporal trends, this should ensure correct 
     handeling of spatio-temporal covariates.

function: simulateMesaData
fix: Minor update, taking the new behaviour of create.data.model into
     account (i.e. that trend is computed for all dates in trend+obs)

data: mesa.data.res
add: Added a mesa.data.res that will contain results of some costly 
     computations, currently only contains optimisation results. CV 
     results will be added.

manual: createCV
add: Added a details section describing how to handel uneven number 
     of observations in the different CV-groups.

0.5.3 - 2011-02-12
function: create.data.model
fix: Minor bug fix, different behaviour for spline and approx when 
     calling with empty matrix...

function: plotCV & plotPrediction
add: Added options for plotting allowing for both different points 
     and line types

0.5.4 - 2011-02-18
function: estimateCV
add: Added a hessian.all option, to reduce computational time.

0.6.0 - 2011-02-21
Added a number of precomputed results into the mesa.data.res data structures.

function: create.data.model
Add: warning if temporal trends have to be interpolated to observation dates.

0.7.0 - 2011-03-01
Major overhaul of the help files. Added examples and more complete 
descreptions to all files.

0.7.1 - 2011-03-01
Added test cases to the help files as \dontshow{} fields
