topic Hello Tianxiong Lu, in Intel® oneAPI Math Kernel Library

Question with df?EditPPSpline1D in Data Fitting Functions

Tianxiong_Lu — Fri, 02 Oct 2015 02:00:03 GMT

Hi all,

I have a question about parameter bc of function df?EditPPSpline1D. In the documantation ( https://software.intel.com/en-us/node/522222#D69BA4E7-E8BD-4413-A2A9-769D5002F832 ) says :

const float* for dfsEditPPSpline1D

const double* for dfdEditPPSpline1D

Pointer to boundary conditions. The size of the array is defined by the value of parameter bc_type:

If you set free-end or not-a-knot boundary conditions, pass the NULL pointer to this parameter.
If you combine boundary conditions at the endpoints of the interpolation interval, pass an array of two elements.
If you set a boundary condition for the default quadratic spline or a periodic condition for Hermite or the default cubic spline, pass an array of one element.

But if y is vector value function with dimension ny(>1), on the last case, it only need one element array? or ny elements array instead?

the owner of this code says -

Gennady_F_Intel — Fri, 02 Oct 2015 08:25:30 GMT

the owner of this code says - - it only need one element array

Thank you. But what value

Tianxiong_Lu — Fri, 02 Oct 2015 12:05:09 GMT

Thank you. But what value should be provided while y is vector value function with dimension ny(>1) ?

For example, the boundary condition type DF_BC_Q_VAL for default quadratic spline, the description in the documentation is "function value at point (x0+x1)/2", but function value is a vector with dimension ny.

Hello Tianxiong Lu,

VictoriyaS_F_Intel — Mon, 05 Oct 2015 15:45:55 GMT

Hello Tianxiong Lu,

Intel® MKL Data Fitting routines restricts the length of the array bc to 1 for case ny > 1 and the boundary condition type is DF_BC_Q_VAL. If you need a different boundary condition for each coordinate of the vector function, please, construct spline for each coordinate separately.

Best regards,

Victoriya