kerdecon: Bivariate kernel deconvolution estimator
In gbasulto/mvdeconvolution: Multivariate kernel deconvolution

It computes a bivariate kernel deconvolution estimator using the Fast Fourier transform (FFT) on a grid.

1 2	kerdecon(resol, samp, error, truncate, h, EMethod, columns, Kkernel, coord1Range, coord2Range, sigE = NULL, pstve = TRUE)

`resol`	An integer specifying the grid size on each coordinate. Power of two are faster.
`samp`	2-column matrix with the contaminated sample.
`error`	Matrix with 2 columns, either with the pure sample of the error, or with the
`truncate`	NULL by default. If the approximated characteristic function of the error is smaller than truncate, it will make the integrand zero. This provides numerical stability.
`h`	Bivariate vector with the bandwidth parameters.
`EMethod`	Method used to approximate the characteristic function of the error. 1 or 2 if ecf of the char. fnc. of the kde is used, respectively. differences obtained from the panel data structure. It can also accept "ecf" or "kde" respectively.
`columns`	Number of columns of the panel data. If there is one column, it assumes that instead panel data, a pure sample of the error is provided.
`Kkernel`	Kernel to be used. See 'ker' function.
`coord1Range`	A bivariate vector with the limits of first coordinate of the grid to estimate the density.
`coord2Range`	A bivariate vector with the limits of second coordinate of the grid to estimate the density.
`sigE`	Bandwidth (2x2) matrix for the error. It is NULL by default and it is required only when EMethod = 2.
`pstve`	TRUE/FALSE determining whether returning zero for those points where the estimate in negative.
`x1`	Vector of size resol with the first coordinate grid.
`x2`	Vector of size resol with the second coordinate grid.
`z`	resol x resol grid with the estimated values.