Let the integral approximation of a scalar function in space:図 1.
with the so-called smoothing length and a kernel approximation such that:図 2.
and in a suitable sense図 3.
denotes the Dirac function.
Let a set of particles =1, n at positions (=1,n) with mass and density . The smoothed approximation of the function is (summation over neighboring particles and the
particle itself):図 4.
The derivatives of the smoothed approximation are obtained
by ordinary differentiation.図 5.
The following kernel 1 which is an approximation of Gaussian
kernel by cubic splines was chosen (図 1):図 6.
図 7.
and 図 8.
図 9. Kernel Based on Spline Functions
This kernel has compact support, so that for each particle , only the closest particles contribute to
approximations at (this feature is computationally efficient). The
accuracy of approximating 式 1 by 式 4 depends on
the order of the particles.