startNWS {caretNWS} | R Documentation |
These functions can be used to quickly create Sleigh objects.
startNWS() makeSleighStarter(workerCount = 5, verbose = TRUE, launch = "local", timeout = 300, nodeList = rep("localhost", workerCount), ...)
workerCount |
the number of workers (i.e. compute nodes) |
verbose |
a logical: should the details of acquiring nodes be shown? |
launch |
a character string for the method of starting jobs on compute nodes. This is specific to the configuration of your grid/cluster. See sleigh for details and examples. Examples are: "local", "ssh" or "bsub". See the nws package documentation for more details. |
timeout |
The number of seconds to wait while compute nodes are acquired. See status for details and examples. |
nodeList |
a list of hosts that workers will be created. Note that this argument is ignored when launch equals "local". The default value is c('localhost', 'localhost', 'localhost') |
... |
Optional arguments to pass to sleigh . |
By default, trainNWS
uses these functions to create Sleigh objects (right now, you cannot pass Sleigh objects to the train function).
These fucntions are specified using the trainNWSControl
function.
A function is produces that will create sleigh objects. This will then be called from within trainNWS
.
Steve Weston, Nathan Coulter
trainNWS
, trainNWSControl
, sleigh
, status