# covStruct.create

##### Spatial covariance - Class constructor

Creates a covariance structure.

##### Usage

```
covStruct.create(covtype, d, known.covparam, var.names, coef.cov = NULL, coef.var = NULL,
nugget = NULL, nugget.estim = FALSE, nugget.flag = FALSE,
iso = FALSE, scaling = FALSE, knots=NULL, kernel=NULL)
```

##### Arguments

- covtype
a character string specifying the covariance structure.

- d
an integer containing the spatial dimension.

- known.covparam
a character ("None" or "All") indicating whether covariance parameters are known or must be estimated.

- var.names
a vector of character strings containing the variable names.

- coef.cov
an optional vector containing the values for covariance parameters.

- coef.var
an optional number containing the variance value.

- nugget
an optional variance value standing for the homogenous nugget effect. Default is NULL.

- nugget.estim
is the nugget effect estimated or known?

- nugget.flag
is there a nugget effect?

- iso
an optional boolean that can be used to force a tensor-product covariance structure to have a range parameter common to all dimensions.

- scaling
an optional boolean indicating whether a scaling on the covariance structure should be used.

- knots
an optional list of knots (used if

`scaling = TRUE`

)- kernel
an optional function containing a new covariance structure

##### Value

A formal S4 class of type `covTensorProduct-class`

, `covIso-class`

(if `iso`

is `TRUE`

) (if `scaling`

is `TRUE`

),
or `covUser-class`

(if `kernel`

is `TRUE`

).

##### See Also

*Documentation reproduced from package DiceKriging, version 1.5.6, License: GPL-2 | GPL-3*