gp.list {mlegp} | R Documentation |
Gaussian Process Lists
Description
Creates an object of type gp.list
, given a list of Gaussian processes fit to separate sets of observations, or a list of Gaussian processes fit to principle component weights to approximate output of high dimension
Usage
gp.list(..., param.names = NULL, UD = NULL, gp.names = NULL)
Arguments
... |
either a |
param.names |
optionally, the parameter names corresponding to the columns of the design matrix of all Gaussian processes. By default, this will be equal to the parameter names of the first Gaussian process in |
UD |
the UD matrix, if the Gaussian process is fit to principle component weights |
gp.names |
optionally, a vector of names for the Gaussian processes, defaulting to ‘gp \#1’, ‘gp \#2’, ... |
Value
A gp.list
object is a list
object, where the first k elements correspond to k Gaussian processes passed in as ...
. This makes it straightforward to access a single Gaussian process. In addition, gp.list
contains components:
params |
a vector of parameter names, corresponding to the columns of the design matrix |
numGPs |
the number of Gaussian processes in the list |
numDim |
the number of parameters in the design matrix |
numObs |
the number of observations |
names |
the names of the Gaussian processes |
Note
currently, we require that all Gaussian processes have the same dimension (number of columns in the design matrix) and the same number of observations
this function is called by mlegp
and should not be called by the user
Author(s)
Garrett M. Dancik dancikg@easternct.edu
References
https://github.com/gdancik/mlegp/
See Also
mlegp
, mlegp-svd-functions
for more details about the UD matrix
Examples
x = -5:5
y1 = sin(x) + rnorm(length(x), sd=.1)
y2 = sin(x) + rnorm(length(x), sd = .5)
## create the gp.list object ##
fitMulti = mlegp(x, cbind(y1,y2))
plot(fitMulti)
fitMulti ## print summary of of the fitted Gaussian process list
fitMulti[[2]] ## print summary for the 2nd Gaussian process