Provide a helper function to scale the matrix to have a dominant eigenvalue of 1.0

[BlackEdder]

Scaling the contact matrix to have a dominant eigenvalue of 1 is a bit tricky to do by hand, but all the code to do it is already part of the package, so instead of throwing an error:

Error in finalsize::final_size(r0 = 1.5, combine_m, demo, susceptibility,  : 
  Error: contact matrix must have a maximum real eigenvalue of 1.0

it would be nice if the package could provide a helper function to do that for you. The helper function should only need the current contact matrix and the population vector.

[BlackEdder]

Just for reference, this is the code I ended up using:

cm <- m_unscaled %*% diag(demo)
Re(eigen(cm)$values[1]) -> scale
finalsize::final_size(r0 = 1.5, m_unscaled/scale, demo, ...)

[avallecam]

Hi @BlackEdder, I just synced with this feature. Wondering if this is preferred as a helper function or as an internal process, as currently available for {epidemics}

A latest feature request in {epidemics} will also include the matrix transposition (check epiverse-trace/epidemics#259 and epiverse-trace/epidemics#271). Thinking in interoperability between socialmixr and finalsize, or compatibility of contact matrix inputs for epidemics and finalsize, should we have similar input steps in finalsize as in epidemics?

Cc'ing @bahadzie @adamkucharski

[edwinvanl]

I feel helper functions are in general more flexible and better for long term support. But agree it could be added as an internal process.

A slightly different but related issue is that sometimes it is nice to be able to change the contact matrix, while assuming the same transmission rate. I wonder if it would be useful to accept an unscaled contact matrix in certain situations?