Simulation Coding Plan

– internal use !–

I. Main Question

The mathematical formulae found by symbolic regression in the mixing measures paper is complex. Therefore it is natural to hypothise that form of the association between age mixing summary measures and HIV may vary.

Q. What is the impact of age-mixing relationships on HIV epidemics given a specific:

• age structure
• debut age
• degree mixing level
• ** in/between individual variation in age-gap preference**

  ---

II. Process

1. (Symbolic) regression

   Find formulae:

   formation baseline = f(x,target)

   where x=(demographical pars, eagerness/age mixing related pars) so that the baseline factor of formation is adjusted to maintain a reasonable mean degree.
2. Parameterisation Use ABC method.
   • Input variables: degree mixing related pars, duration related pars.
   • Targets: combinations of (1) degree mixing levels, (2) duration (fraction of short-term duration), (3) average relationships, (4) debut age
3. Simulation
4. Output analysis
   • Network statistics ~ HIV
   • Network motif
   • Local network statistics and risk of HIV

     Node removal

     Edge removal

III. To-Do list

not done = in block

1. Random simulation and symbolic regression
   • intitial input pars generator
   • run
   • output collection
2. ABC
   • ABC model => refine
   • target function
   • testing
   • run and restore ABC
3. main function
   • pars generator
   • main process => refine
4. output files manipulation
   • output collection
   • output manipulation
   • statistical analysis
5. Others
   • maintain all R files to be consistent with the RSimpactCyan package
6. Communication

   Roxy