Fast linear interpolation of a regular grid

Computes a linear interpolation of multidimensional regularly gridded data.

interp(object, ...)

# S3 method for grid.par
interp(object, data, newx, ...)

# S3 method for data.grid
interp(object, data.ind = 1, newx, ...)

# S3 method for locpol.bin
predict(object, newx = NULL, hat.data = FALSE, ...)

# S3 method for np.den
predict(object, newx = NULL, ...)

Arguments

object

(gridded data) object used to select a method.

...

further arguments passed to or from other methods.

data

vector or array of data values.

newx

vector or matrix with the (irregular) locations to interpolate. Columns correspond with dimensions and rows with data.

data.ind

integer (or character) with the index (or name) of the data component.

hat.data

logical; if TRUE (and possible), the hat matrix corresponding to the (original) data is returned.

Value

A list with two components:

x

interpolation locations.

y

interpolated values.

If newx == NULL, predict.locpol.bin returns the estimates (and optionally the hat matrix) corresponding to the data (otherwise interp.data.grid is called).

Details

interp methods are interfaces to the fortran routine interp_data_grid (in grid_module.f90).

predict.locpol.bin is an interface to the fortran routine predict_lp (in lp_module.f90).

Note

Linear extrapolation is performed from the end nodes of the grid.

WARNING: May fail with missing values (especially if object$locpol$ncv > 0).

See also

interp.surface.