knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "README-"
)
set.seed(0)
A highly scientific and utterly addictive bird point count simulator to test statistical assumptions and to aid survey design.
"I've yet to see any problem, however complicated, which when you looked at it the right way didn't become still more complicated." -- Poul Anderson, Call Me Joe
“Love the simulation we're dreaming in” - Dua Lipa, Physical
The goal of the package is to:
Design objectives:
See the package in action in the QPAD Book.
Check out the QPAD workshop.
Read/cite the paper Agent-based simulations improve abundance estimation (DOI 10.1007/s42977-023-00183-2).
CRAN version:
install.packages("bSims")
Development version:
remotes::install_github("psolymos/bSims")
See what is new in the NEWS file.
Please cite (see citation("bSims")
) the paper:
Solymos, P. 2023. Agent-based simulations improve abundance estimation. Biologia Futura 74, 377--392 DOI 10.1007/s42977-023-00183-2, link to PDF.
Feedback and contributions are welcome:
library(bSims)
phi <- 0.5
tau <- 1:3
dur <- 10
rbr <- c(0.5, 1, 1.5, Inf)
tbr <- c(3, 5, 10)
l <- bsims_init(10, 0.5, 1)
p <- bsims_populate(l, 1)
a <- bsims_animate(p, vocal_rate=phi, duration=dur)
o <- bsims_detect(a, tau=tau)
x <- bsims_transcribe(o, tint=tbr, rint=rbr)
get_table(x)
head(get_events(a))
head(get_detections(o))
A few Shiny apps come with the package. These can be used to interactively explore the effects of different settings.
Compare distance functions:
run_app("distfunH")
run_app("distfunHER")
Compare simulation settings for single landscape:
run_app("bsimsH")
run_app("bsimsHER")
Interactive sessions can be used to explore different settings.
Settings can be copied from the Shiny apps and replicated using the
bsims_all
function:
b <- bsims_all(extent=5, road=1, density=c(1,1,0))
b
The object has handy methods:
b$settings() # retrieve settings
b$new() # replicate once
b$replicate(10) # replicate 10x
The $replicate()
function also runs on multiple cores:
library(parallel)
b <- bsims_all(density=0.5)
B <- 4 # number of runs
nc <- 2 # number of cores
## sequential
system.time(bb <- b$replicate(B, cl=NULL))
## parallel clusters
cl <- makeCluster(nc)
## note: loading the package is optional
system.time(clusterEvalQ(cl, library(bSims)))
system.time(bb <- b$replicate(B, cl=cl))
stopCluster(cl)
## parallel forking
if (.Platform$OS.type != "windows") {
system.time(bb <- b$replicate(B, cl=nc))
}