Dear Janine,

The programme is changed as follows:

- I have moved the variable runR to the argumentlist of ecology_mcmc (and changed the R function eco(...) accordingly).

- in the calculation of s (calc_s) I use a slightly bigger neighbourhood than fbefore.

- I have finished the procedure for updating the R's. 

- in accept.out numbers of acceptances and attemps are printed for all R's

- in timeseries.out timeseries are printed for all R's (so this file quickly becomes _very_ big).

- if runR=1 (full bayesian analysis) all thetas are updated regardless of whether they are identifiable or not. 

- metropolis.c contains two procedures for checking computations in metropolis_R() 

NB: all R's for a given resprouter type are initialized with the same R value as specified in initialR. Same proposal distribution is used for all R's corresponding to a given resprouter type.

NB: right now we use normal priors for the R's. This is something we might want to change.

To be done: 

a) in the full Bayesian analysis we update all thetas regardless of whether they are identifiable or not. When we print the thetas in the timeseries file it may be a good idea to also print an indicator for identifiable. Then we can compute the posterior distribution of theta given it is identifiable, respcetively, not identifiable.

b) perhaps we should only print the mean of the R's instead of printing all the R values to timeseries.out.


