The simulation engine has been speed up by making its source code more efficient.
The dependency to the chron package of R has been removed to get rid of unhandy format definitions of dates in the input data to the simulation. Instead now a new date format has been defined that is based on numerical values. It is ‘yyyymmdd’, e.g. ‘20221017’ for October 17th, 2022. New functions have been designed and included into the package facilitating the handling of dates within the simulation.
There is now an initial check in the package to see whether the input data is in the right format and the rates required to run the simulation are available for the complete simulation horizon and along all ages considered.
A vignette has been added to show the capabilities of the package for running large-scale simulation applications. The example of the vignette has been designed by researchers of the Eureopean Commission. It is on the modelling and simulation of migration flows within Europe. They are now added as contributors to the MicSim package. They have also tested and checked the code of the package to remove bugs.
The example section of the ‘micSim’ function has been extended by another more complex example to highlight the versality of the package.
It is possible with this new version to define which parts of the initial population and/or migrant population go to which CPUs/cores when conducting parallel computing.
Fix two bugs related to the creation of newborns during simulation: (1) Before that version all transitions of the type "1+"->"1+" led to newborns; now only transitions where no other attributes of the state change cause newborns, e.g. "f/M/+1" -> "f/D/1+" caused a newborn before this version. This is not any longer the case. Now only transitions of the form "f/M/1+" -> "f/M/+1" cause a newborn. (2) There was a bug in transforming the fertility matrix to detect birth events; instead of rows, columns were taken as transitions. This has been fixed now.
Fix an inconsistency in the computation of transition ages (in years) related to leap years (now the related function is vectorized).
A further functionality was added to the package: now it is possible to link individuals born during simulation to their mothers (see argument 'reportMothers' in the micSim or micSimParallel function).
Fix bug in parallel computing function concerning case when there are migrants in the model (this bug resulted from fix to version 1.0.14).
Fix bug in parallel computing function concerning case when there are no migrants in the model.
Update of chron package caused an error due to new date format. This problem has been fixed.
Fix bug concerning the transition age of immigrants.
Up to version 1.0.10 deterministic events could only be defined to occur after whole numbered years. Now, they can also be defined to happen after whole numbered days.
Fix bug concerning the uniqueness of IDs of newborns when simulating natives and migrants on several cores, i.e., when using the function ‘micSimParallel’.
Slight changes in the description of the (life-course) model used in MicSim.
Add error message arising if negative rate values appear.
Besides not defining impossible transitions in the transition matrix, they can now also be specified by giving only zero rate values for the respective rate function/s.
Missing package dependency has been added. MicSim requires ‘rlecuyer’.
Add a new reference.
Modify function ‘micSim’ that it can also deal with a single life course.
Fix bug in function ‘convertToLongFormat’ leading to wrong ordering of the event sequences of individuals born during simulation.
Fix minor conversion problem in function ‘convertToWideFormat’, related to request for NA values.
Now function ‘buildTransitionMatrix’ can handle covariate dependent mortality rates of every kind, e.g., mortality rates depending on health status and gender.
Add function ‘convertToWideFormat’ to convert simulation output into wide format.
Slight changes in the examples to make them more feasible. Concretely, in several rates definitions I adjusted the year in which dynamics are assumed to change.
Fix a bug in the optional features of the ‘micSim’ function: Children who are part of the base population had been enrolled one year to late. Now, enrollment dates are OK.
Refine funtionality of ‘micSim’ function to handle deterministc events (marked by Inf rates).
Fix bug in function ‘buildTransitionMatrix’.
Add News file.
Extend function ‘buildTransitionMatrix’ to facilitate using transition rates which depend on covariates such as gender and marital status.
Correct typing error in Details section of help page of function ‘buildTransitionMatrix’.
Fix a problem in function ‘micSim’ occuring if the simulation is asked not to produce newborns (concerns argument ‘fertTr=c()’).