Fastnet is a Julia package that allows very fast (linear-time) simulation of discrete-state dynamical processes on networks, such as commonly studied models of epidemics
Fastnet achieves linear-time performance by using an innovative data structure. The underlying network is a potentially directed and potentially non-simple graph. The package provides a convenient syntax that allows to implement common models in a few simple lines of code. The simulations are done using an event-driven (Gillespie) algortithm offering fast performance and excellent agreement with real world continuous-time processes. Using Fastnet models with millions of nodes can be run within minutes on a standard laptop.
The following file defines and runs an epidemiological SIS model:
using Fastnet
const S=1 # Node state 1: Susceptible node
const I=2 # Node state 2: Infected node
const SI=1 # Link state 1: Susceptible-Infected link
const p=0.05 # Infection rate (per SI-link)
const r=0.1 # Recovery rate (per I-node)
SI_link=LinkType(S,I) # This describes what we mean by SI-link
# Let's make a network of 1M nodes and 4M links, 2 node states, that keeps track of SI links
net=FastNet(1000000,4000000,2,[SI_link]; nodealias=["S","I"], linkalias=["SI"])
randomgraph!(net) # Initialize as ER-random graph (all nodes will be in state 1: S)
for i=1:20 # Infect 20 susceptible nodes at random
node=randomnode(net,S)
nodestate!(net,node,I)
end
function rates!(rates,t) # This functins computes the rates of processes
infected=countnodes_f(net,I) # count the infected nodes
activelinks=countlinks_f(net,SI) # count the SI links
rates[1]=p*activelinks # compute total infection rate
rates[2]=r*infected # compute total recovery rate
nothing
end
function recovery!() # This is what we do when the recovery process is triggered
inode=randomnode_f(net,I) # Find a random infected node
nodestate_f!(net,inode,S) # Set the state of the node to susceptible
end
function infection!() # This is what we do when the infection process is triggered
alink=randomlink_f(net,SI) # Find a random SI link
nodestate_f!(net,linksrc_f(net,alink),I) # Set both endpoints of the link to infected
nodestate_f!(net,linkdst_f(net,alink),I)
end
sim=FastSim(net,rates!,[infection!,recovery!]) # initialize the simulation
@time runsim!(sim,60.0,5.0) # Run for 60 timeunits (reporting every 5)This produces the output:
Time S I SI
0.0 999980 20 146
5.0 999927 73 570
10.0 999710 290 2199
15.0 998734 1266 9499
20.0 995152 4848 35743
25.0 982053 17947 130184
30.0 936454 63546 437860
35.0 807731 192269 1133438
40.0 585665 414335 1744756
45.0 402526 597474 1708492
50.0 319610 680390 1550403
55.0 291675 708325 1482735
60.0 281941 718059 1458951
65.271643 seconds (806.64 M allocations: 12.244 GiB, 3.87% gc time, 0.04% compilation time)Install with the Julia package manager.
From the Julia REPL, type ] to enter the Pkg REPL mode and
pkg> add Fastnet
Alternatively from Julia
julia> import Pkg
julia> Pkg.add("Fastnet")Full documentation can be found here: https://bridgewalker.github.io/Fastnet.jl.
This package was tested on Julia 1.6.2 on Windows. This is still in an early version. More testing is necessary, so use with caution. I am still actively developing this package, so comments, feature requests etc. are very welcome. You can contact me via thilo2gross@gmail.com!
Many thanks to Pietro Monticone (@pitmonticone) for work on the documentation.
The original development of Fastnet was supported by the Volkswagen foundation. The current implementation in Julia was developed at HIFMB, a collaboration between the Alfred-Wegener-Institute, Helmholtz-Center for Polar and Marine Research, and the Carl-von-Ossietzky University Oldenburg, initially funded by the Ministry for Science and Culture of Lower Saxony (MWK) and the Volkswagen Foundation through the “Niedersächsisches Vorab” grant program (grant number ZN3285).