registerdFORTRAN: Internal FORTRAN routines for working with grids and finding...

Description Format Details References Examples

Description

These are objects of class FortranRoutine and also NativeSymbolInfo They provide information for compiled functions called with .Call, or .Fortran. Ordinarily one would not need to consult these and they are used to make the search among dynamically loaded libraries ( in particular the fields library ) have less ambiguity and also be faster. These are created when the package/library is loaded are have their definitions from the compilation of init.c in the package source (src) directory.

Format

The format is a list with components:

name

The (registration ?) name of the C function.

address

See NativeSymbolInfo.

dll

Dynamically linked library information.

numParameters

Number of calling arguments in function.

Details

Registered routines are

findnorm

Finds marginal variance quickly for the case of a rectangular lattice, a specific Wendland basis, and constant a.wght at a given level. Called from LKRectangleFastNormalization

lkdist

Euclidean distance between two sets of coordinates but restricted to a maximum distance.Called from LKDist.

lkdistcomp

Same as lkdist but returns the component distances for each coordinate.Called by LKDistComponents

lkdistgrid

Finds all distances in x1 that are within delta of a set of grid points. Results are returned in the row, column, value sparse matrix format.called from LKDistGrid

lkdistgridcomp

Same as lkdistgrid but the component distances are found for each coordinate.LKDistGridComponents

lkdiag

Fills in diagonal and off diagonal elements in sparse matrix format.

See package_native_routine_registration_skeleton for the utility used to create these data objects.

References

For background on registering C, C++ and Fortran functions see 5.4 of Writing R Extensions. See http://r.789695.n4.nabble.com/Registration-of-native-routines-td4728874.html for additional discussion of code registration.

Examples

1

LatticeKrig documentation built on Nov. 9, 2019, 5:07 p.m.