Control Options for SAEM
saemControl(
seed = 99,
nBurn = 200,
nEm = 300,
nmc = 3,
nu = c(2, 2, 2),
atol = 1e-06,
rtol = 1e-04,
method = "liblsoda",
transitAbs = FALSE,
print = 1,
trace = 0,
covMethod = c("linFim", "fim", "r,s", "r", "s", ""),
calcTables = TRUE,
logLik = FALSE,
nnodes.gq = 3,
nsd.gq = 1.6,
optExpression = FALSE,
maxsteps = 100000L,
adjObf = TRUE,
sum.prod = FALSE,
addProp = c("combined2", "combined1"),
singleOde = TRUE,
tol = 1e-06,
itmax = 30,
type = c("nelder-mead", "newuoa"),
powRange = 10,
lambdaRange = 3,
loadSymengine = FALSE,
odeRecalcFactor = 10^(0.5),
maxOdeRecalc = 5L,
...
)Random Seed for SAEM step. (Needs to be set for reproducibility.) By default this is 99.
Number of iterations in the Stochastic Approximation
(SA), or burn-in step. This is equivalent to Monolix's K_0 or K_b.
Number of iterations in the Expectation-Maximization
(EM) Step. This is equivalent to Monolix's K_1.
Number of Markov Chains. By default this is 3. When
you increase the number of chains the numerical integration by
MC method will be more accurate at the cost of more
computation. In Monolix this is equivalent to L
This is a vector of 3 integers. They represent the
numbers of transitions of the three different kernels used in
the Hasting-Metropolis algorithm. The default value is c(2,2,2),
representing 40 for each transition initially (each value is
multiplied by 20).
The first value represents the initial number of multi-variate Gibbs samples are taken from a normal distribution.
The second value represents the number of uni-variate, or multi- dimensional random walk Gibbs samples are taken.
The third value represents the number of bootstrap/reshuffling or uni-dimensional random samples are taken.
a numeric absolute tolerance (1e-8 by default) used by the ODE solver to determine if a good solution has been achieved; This is also used in the solved linear model to check if prior doses do not add anything to the solution.
a numeric relative tolerance (1e-6 by default) used
by the ODE solver to determine if a good solution has been
achieved. This is also used in the solved linear model to check
if prior doses do not add anything to the solution.
The method for solving ODEs. Currently this supports:
"liblsoda" thread safe lsoda. This supports parallel
thread-based solving, and ignores user Jacobian specification.
"lsoda" -- LSODA solver. Does not support parallel thread-based
solving, but allows user Jacobian specification.
"dop853" -- DOP853 solver. Does not support parallel thread-based
solving nor user Jacobain specification
"indLin" -- Solving through inductive linearization. The RxODE dll
must be setup specially to use this solving routine.
boolean indicating if this is a transit compartment absorption
The number it iterations that are completed before anything is printed to the console. By default, this is 1.
An integer indicating if you want to trace(1) the SAEM algorithm process. Useful for debugging, but not for typical fitting.
Method for calculating covariance. In this discussion, R is the Hessian matrix of the objective function. The S matrix is the sum of each individual's gradient cross-product (evaluated at the individual empirical Bayes estimates).
"linFim" Use the Linearized Fisher Information Matrix to calculate the covariance.
"fim" Use the SAEM-calculated Fisher Information Matrix to calculate the covariance.
"r,s" Uses the sandwich matrix to calculate the covariance, that is: \(R^-1 \times S \times R^-1\)
"r" Uses the Hessian matrix to calculate the covariance as \(2\times R^-1\)
"s" Uses the crossproduct matrix to calculate the covariance as \(4\times S^-1\)
"" Does not calculate the covariance step.
This boolean is to determine if the foceiFit
will calculate tables. By default this is TRUE
boolean indicating that log-likelihood should be calculate by Gaussian quadrature.
number of nodes to use for the Gaussian quadrature when computing the likelihood with this method (defaults to 1, equivalent to the Laplaclian likelihood)
span (in SD) over which to integrate when computing the likelihood by Gaussian quadrature. Defaults to 3 (eg 3 times the SD)
Optimize the RxODE expression to speed up calculation. By default this is turned on.
maximum number of (internally defined) steps allowed during one call to the solver. (5000 by default)
is a boolean to indicate if the objective function
should be adjusted to be closer to NONMEM's default objective
function. By default this is TRUE
Take the RxODE model and use more precise products/sums. Increases solving accuracy and solving time.
one of "combined1" and "combined2"; These are the two forms of additive+proportional errors supported by monolix/nonmem:
combined1: transform(y)=transform(f)+(a+b*f^c)*eps
combined2: transform(y)=transform(f)+(a^2+b^2*f^(2c))*eps
This option allows a single ode model to include the PK parameter information instead of splitting it into a function and a RxODE model
This is the tolerance for the regression models used for complex residual errors (ie add+prop etc)
This is the maximum number of iterations for the regression models used for complex residual errors. The number of iterations is itmax*number of parameters
indicates the type of optimization for the residuals; Can be one of c("nelder-mead", "newuoa")
This indicates the range that powers can take for residual errors; By default this is 10 indicating the range is c(1/10, 10) or c(0.1,10)
This indicates the range that Box-Cox and Yeo-Johnson parameters are constrained to be; The default is 3 indicating the range (-3,3)
Boolean indicating if the model should be
loaded into symengine. This cause all the ODEs to be collapsed
into one expression that is eventually optimized if
optExpression is TRUE.
The factor to increase the rtol/atol with bad ODE solving.
Maximum number of times to reduce the ODE tolerances and try to resolve the system if there was a bad ODE solve.
Other arguments to control SAEM.
List of options to be used in nlmixr fit for
SAEM.