Model submission consists of two steps: generating the initial values
(METHOD=CHAIN
run) and sampling for each chain (METHOD=BAYES
or
METHOD=NUTS
runs).
# S3 method for bbi_nmbayes_model
submit_model(
.mod,
.bbi_args = NULL,
.mode = getOption("bbr.bbi_exe_mode"),
...,
.overwrite = NULL,
.config_path = NULL,
.wait = TRUE,
.dry_run = FALSE
)
The model object to submit.
A named list specifying arguments to pass to bbi
formatted like list("nm_version" = "nm74gf_nmfe", "json" = T, "threads" = 4)
. Run print_bbi_args()
to see valid arguments. Note that bbr does
not support changing the output directory (including through the model or
global YAML files).
Either "sge"
, the default, to submit model(s) to the grid or
"local"
for local execution. This can be passed directly to this argument
or set globally with options("bbr.bbi_exe_mode")
.
args passed through to bbi_exec()
Logical to specify whether or not to overwrite existing
model output from a previous run. If NULL
, the default, will defer to
setting in .bbi_args
or bbi.yaml
. If not NULL
will override any
settings in .bbi_args
or bbi.yaml
.
Path to a bbi configuration file. If NULL
, the
default, will attempt to use a bbi.yaml
in the same directory as the
model.
If TRUE
, the default, wait for the bbi process to return
before this function call returns. If FALSE
function will return while
bbi process runs in the background.
Do not submit the sampling runs; just report what command would be executed via the returned object. Note: The METHOD=CHAIN model is executed to generate the initialization values regardless of this value.
bbr_nmbayes for a high-level description of how NONMEM Bayes models are structured