El jueves, 10 de abril de 2014 10:19:14 UTC-5, David P. Sanders escribió: > > > > El miércoles, 9 de abril de 2014 03:40:40 UTC-5, Simon Frost escribió: >> >> Dear All, >> >> I'm implementing Gillespie's direct method for stochastic simulation: >> >> http://en.wikipedia.org/wiki/Gillespie_algorithm >> >> I'm loosely following the API for the now-orphaned R package GillespieSSA: >> >> http://www.jstatsoft.org/v25/i12 >> >> http://artax.karlin.mff.cuni.cz/r-help/library/GillespieSSA/html/ssa.html >> >> In Julia, the analogous function call is something like the following. >> >> function >> ssa(x0::Vector{Int64},a::Vector{Expr},nu::Matrix{Float64,2},parms::Expr,tf::Float64,ignore_negative_state::Bool,console_interval::Int64,census_interval::Int64,verbose::Bool,max_wall_time::Float64) >> >> However, as I've never done metaprogramming in Julia before, I was >> wondering whether anyone had any input to make the model definition as >> clean as possible (i.e. without cluttering the syntax with symbols, while >> not polluting the environment)? >> >> Best >> Simon >> > > Hi Simon, > > This is a nice idea! I'm very interested. > > Unfortunately, I don't have the necessary experience either, but there are > other packages which could be useful, such as ODE solvers, which have the > same issue of how to best define the model. For example: > > https://github.com/JuliaLang/Sundials.jl > > And https://github.com/JuliaLang/ODE.jl
In fact, as a first step, I guess it would be much easier to just pass in a standard Julia function, rather than trying to parse something. > Personally, the ssa() function you are proposing seems rather complicated. > Perhaps it would be better to have some kind of new type that wraps all the > needed parameters in one place? > > Best > David. > > > > > >
