ODE solver should allow configurable number of age groups

MJomaba / flu-evidence-synthesis

Code for reconstructing influenza dynamics from surveillance data using evidence synthesis techniques

19 stars 10 forks source link

Closed BlackEdder closed 9 years ago

BlackEdder commented 9 years ago

Probably best done by writing separate SEIR model for each group and then interconnect them.

[x] Implement tests for the current ODE results
[x] Split out seir_t which holds needed state (S[7] E1[7], deltaS[7] etc etc)
[x] Use seir_t for each "sub population"
- [x] Convert with smart regexes
[x] Decide on how to build the full model, with multiple risk groups and age groups from SEIR modules

BlackEdder commented 9 years ago

[x] Fix performance problems with the new method
- Slow down looked to be caused by using vectors for S, E1 instead of c arrays, but replacing them with Eigen::VectorXd solved that problem again.
- Actually looks like slow down is caused by not knowing seir_t sizes at compile time Started new issue for this

Started new issue #25 to track this.

BlackEdder commented 9 years ago

Fixed by commit 7343cd1b2f266cc6406bbaf515b32da9e2e5c00c