.
.
.
.
.
.
.
Formal argument init of fastbeta,
fastbeta.bootstrap, ptpi, and seir is
repositioned.
Formal arguments sigma and m of
fastbeta, fastbeta.bootstrap, ptpi, and
seir have constant default values (sigma = 1,
m = 1L) not depending on the values of other arguments.
Entry [2, 2] of the Jacobian matrix constructed by
seir(stochastic = FALSE, useCompiled = FALSE) was wrong.
seir handles more comprehensively the situation in
which the number of rows of the matrix returned by deSolve
function lsoda is less than length.out. It now
distinguishes termination due to found roots from termination due
to an unsuccessful solver call.
Warnings caught by -Wconversion are squashed.
seir gains logical argument aggregate
indicating whether latent and infectious compartments should be
aggregated.
seir arguments nu and mu gain
default values. In both cases, it is function (t) 0,
indicating no vital dynamics.
seir.R0, seir.ee, and seir.jacobian for
calculating the basic reproduction number, endemic equilibrium,
and Jacobian matrix.
fastbeta.matrix for calculating lower triangular
coefficient matrix.
Compilation now uses -DSTRICT_R_HEADERS and -DR_NO_REMAP.
Argument constants of functions sir,
fastbeta, fastbeta.bootstrap, and ptpi is
replaced by arguments sigma, gamma, delta,
init, m, and n.
sir is now a wrapper calling new seir.
Arguments a and b of function ptpi
are replaced by arguments start and end so that
the interface is more similar to that of window.
ptpi returns list(value, diff, iter, x),
no longer list(value, delta, iter, X).
Simulations sir.e01 and sir.e02 are replaced
by seir.ts01 and seir.ts02. The replacements have
a latent compartment and a different set of attributes reflecting
the API of seir.
More simulations set tl.params = list(epsilon = .)
to mitigate noise attributable to too big leaps in the adaptive
tau-leaping algorithm.
plot method for class fastbeta.bootstrap
uses dev.hold and dev.flush from grDevices
to avoid “incremental” graphical output on some devices.
Compartmental model is generalized to support multiple
infected compartments, m latent and n infectious,
resulting in several backwards incompatible changes to the API;
see below.
adaptivetau and deSolve are now suggested, not imported, as they are needed only for simulation. Usage of simulated data sets is now everywhere conditional on successful loading of the dependency.
sir
expects constants = c(S0, I0, R0, gamma, delta),
no longer constants = c(gamma, S0, I0, R0).
fastbeta and fastbeta.bootstrap
expect constants = c(S0, I0, R0, gamma, delta),
no longer constants = c(gamma, S0, I0).
ptpi gains argument constants and loses
argument start.
It expects constants = c(Sa, Ia, Ra, gamma, delta),
no longer start = Sa.
fastbeta returns (modulo attributes)
cbind(S, I, R, beta), no longer cbind(beta, S, I).
ptpi(...)[[1L]] gives the state at time 0
or at time a conditional on new argument backcalc.
The default is a, no longer (implicitly) 0.
ptpi(...)[[4L]] has dimensions
c(b-a+1, 3, iter), no longer c(b-a+1, iter).
Simulation sir.e01 is generated by the same
parameters but is no longer stochastic.
Simulation sir.e02 gives the stochastic variant.
Compartmental model is generalized to support loss of
immunity at rate delta, resulting in several backwards
incompatible changes to the API; see below.
‘data/*.R’ are no longer copied to ‘inst/scripts’
at install time by a make rule in ‘src/Makevars’.
There is no need: the scripts are preserved due to the setting of
BuildResaveData: no. Hence ‘src/Makevars’ is removed.
Typo in ‘src/sir.c’ triggering -Wformat is fixed.
Changes are logged in ‘inst/NEWS.Rd’.
‘DESCRIPTION’ sets BuildResaveData: no,
so that R CMD build works without
--no-resave-data when fastbeta is not installed.
‘data/*.R’ are copied into ‘inst/scripts’
at install time by a make rule in ‘src/Makevars’.
Depends: R (>= 4.3.0) for amsmath in PDF
output, which was not supported until 4.2.2.