scikits.cuda.cublas.cublasDrotmg

scikits.cuda.cublas.cublasDrotmg(handle, d1, d2, x1, y1)[source]

Construct a double-precision real modified Givens rotation matrix.

Constructs the double-precision real modified Givens rotation matrix h = [[h11, h12], [h21, h22]] that zeros out the second entry of the vector [[sqrt(d1)*x1], [sqrt(d2)*x2]].

Parameters:

handle : int

CUBLAS context.

d1 : numpy.float64

double-precision real value.

d2 : numpy.float64

double-precision real value.

x1 : numpy.float64

double-precision real value.

x2 : numpy.float64

double-precision real value.

Returns:

sparam : numpy.ndarray

sparam[0] contains the flag described below; sparam[1:5] contains the values [h00, h10, h01, h11] that determine the rotation matrix h.

Notes

The rotation matrix may assume the following values:

for flag == -1.0, h == [[h00, h01], [h10, h11]] for flag == 0.0, h == [[1.0, h01], [h10, 1.0]] for flag == 1.0, h == [[h00, 1.0], [-1.0, h11]] for flag == -2.0, h == [[1.0, 0.0], [0.0, 1.0]]