GUtils.h File Reference

Linear algebra functions for the CUDA version of PyPO. More...

#include <math.h>
#include <cuda.h>
#include <cuComplex.h>

Go to the source code of this file.

Functions
__device__ __inline__ void	dot (float(&v1)[3], float(&v2)[3], float &out)
__device__ __inline__ void	dot (cuFloatComplex(&cv1)[3], cuFloatComplex(&cv2)[3], cuFloatComplex &out)
__device__ __inline__ void	dot (cuFloatComplex(&cv1)[3], float(&v2)[3], cuFloatComplex &out)
__device__ __inline__ void	dot (float(&v1)[3], cuFloatComplex(&cv2)[3], cuFloatComplex &out)
__device__ __inline__ void	ext (float(&v1)[3], float(&v2)[3], float(&out)[3])
__device__ __inline__ void	ext (cuFloatComplex(&cv1)[3], cuFloatComplex(&cv2)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	ext (cuFloatComplex(&cv1)[3], float(&v2)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	ext (float(&v1)[3], cuFloatComplex(&cv2)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	diff (float(&v1)[3], float(&v2)[3], float(&out)[3])
__device__ __inline__ void	diff (cuFloatComplex(&cv1)[3], cuFloatComplex(&cv2)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	abs (float(&v)[3], float &out)
__device__ __inline__ void	conja (cuFloatComplex(&cv)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	normalize (float(&v)[3], float(&out)[3])
__device__ __inline__ void	add (float(&v1)[3], float(&v2)[3], float(&out)[3])
__device__ __inline__ void	s_mult (float(&v)[3], float &s, float(&out)[3])
__device__ __inline__ void	s_mult (cuFloatComplex(&cv)[3], cuFloatComplex &cs, cuFloatComplex(&out)[3])
__device__ __inline__ void	s_mult (float(&v)[3], cuFloatComplex &cs, cuFloatComplex(&out)[3])
__device__ __inline__ void	s_mult (cuFloatComplex(&cv)[3], const float &s, cuFloatComplex(&out)[3])
__device__ __inline__ void	snell (cuFloatComplex(&cvin)[3], float(&normal)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	snell (float(&vin)[3], float(&normal)[3], float(&out)[3])
__device__ __inline__ void	snell_t (float(&vin)[3], float(&normal)[3], float mu, float(&out)[3])
__device__ __inline__ void	dyad (float(&v1)[3], float(&v2)[3], float(&out)[3][3])
__device__ __inline__ void	matDiff (float(&m1)[3][3], float(&m2)[3][3], float(&out)[3][3])
__device__ __inline__ void	matVec (float(&m1)[3][3], float(&v1)[3], float(&out)[3])
__device__ __inline__ void	matVec (float(&m1)[3][3], cuFloatComplex(&cv1)[3], cuFloatComplex(&out)[3])
__device__ __inline__ void	matVec4 (float(&mat)[16], float(&cv1)[3], float(&out)[3], bool vec=false)
__device__ __inline__ void	invmatVec4 (float(&mat)[16], float(&cv1)[3], float(&out)[3], bool vec=false)
__device__ __inline__ cuFloatComplex	cuCexpf (cuFloatComplex z)
__device__ __inline__ cuFloatComplex	cuCaddSf (cuFloatComplex a, float b)
__device__ __inline__ cuFloatComplex	cuCaddSf (float a, cuFloatComplex b)
__device__ __inline__ cuFloatComplex	cuCsubSf (cuFloatComplex a, float b)
__device__ __inline__ cuFloatComplex	cuCsubSf (float a, cuFloatComplex b)
__device__ __inline__ cuFloatComplex	cuCmulSf (cuFloatComplex a, float b)
__device__ __inline__ cuFloatComplex	cuCmulSf (float a, cuFloatComplex b)
__device__ __inline__ cuFloatComplex	cuCdivSf (cuFloatComplex a, float b)
__device__ __inline__ cuFloatComplex	cuCdivSf (float a, cuFloatComplex b)

Detailed Description

Linear algebra functions for the CUDA version of PyPO.

Contains float overloaded functions for doing basic 3D vector operations. For the CUDA complex valued linear algebra, we employ the cuComplex.h library.

Function Documentation

abs()

__device__ __inline__ void abs	(	float(&)	v[3],
		float &	out
	)

Absolute value.

Calculate absolute value of real valued vector of size 3.

Parameters

v	Array of 3 float.
out	Scalar float.

add()

__device__ __inline__ void add	(	float(&)	v1[3],
		float(&)	v2[3],
		float(&)	out[3]
	)

Vector addition.

Add two real valued vectors of size 3 element-wise.

Parameters

v1	Array of 3 float.
v2	Array of 3 float.
out	Array of 3 complex float.

conja()

__device__ __inline__ void conja	(	cuFloatComplex(&)	cv[3],
		cuFloatComplex(&)	out[3]
	)

Conjugate.

Conjugate complex valued vector of size 3.

Parameters

cv	Array of 3 complex float.
out	Array of 3 complex float.

cuCaddSf() [1/2]

__device__ __inline__ cuFloatComplex cuCaddSf	(	cuFloatComplex	a,
		float	b
	)

Add scalar float to complex number.

Returns

res cuFloatComplex number.

cuCaddSf() [2/2]

__device__ __inline__ cuFloatComplex cuCaddSf	(	float	a,
		cuFloatComplex	b
	)

Add complex number to scalar float.

Returns

res cuFloatComplex number.

cuCdivSf() [1/2]

__device__ __inline__ cuFloatComplex cuCdivSf	(	cuFloatComplex	a,
		float	b
	)

Divide complex number by scalar float.

Returns

res cuFloatComplex number.

cuCdivSf() [2/2]

__device__ __inline__ cuFloatComplex cuCdivSf	(	float	a,
		cuFloatComplex	b
	)

Divide scalar by complex float.

Returns

res cuFloatComplex number.

cuCexpf()

__device__ __inline__ cuFloatComplex cuCexpf

(

cuFloatComplex

z

)

Take complex exponential.

Take complex exponential by decomposing into sine and cosine.

Returns

res cuFloatComplex number.

cuCmulSf() [1/2]

__device__ __inline__ cuFloatComplex cuCmulSf	(	cuFloatComplex	a,
		float	b
	)

Multiply complex number by scalar float.

Returns

res cuFloatComplex number.

cuCmulSf() [2/2]

__device__ __inline__ cuFloatComplex cuCmulSf	(	float	a,
		cuFloatComplex	b
	)

Multiply complex number by scalar float.

Returns

res cuFloatComplex number.

cuCsubSf() [1/2]

__device__ __inline__ cuFloatComplex cuCsubSf	(	cuFloatComplex	a,
		float	b
	)

Subtract scalar float from complex number.

Returns

res cuFloatComplex number.

cuCsubSf() [2/2]

__device__ __inline__ cuFloatComplex cuCsubSf	(	float	a,
		cuFloatComplex	b
	)

Add complex number to scalar float.

Returns

res cuFloatComplex number.

diff() [1/2]

__device__ __inline__ void diff	(	cuFloatComplex(&)	cv1[3],
		cuFloatComplex(&)	cv2[3],
		cuFloatComplex(&)	out[3]
	)

Component-wise vector difference.

Subtract two complex valued vectors of size 3, element-wise.

Parameters

cv1	Array of 3 complex float.
cv2	Array of 3 complex float.
out	Array of 3 complex float.

diff() [2/2]

__device__ __inline__ void diff	(	float(&)	v1[3],
		float(&)	v2[3],
		float(&)	out[3]
	)

Component-wise vector difference.

Subtract two real valued vectors of size 3, element-wise.

Parameters

v1	Array of 3 float.
v2	Array of 3 float.
out	Array of 3 float.

dot() [1/4]

__device__ __inline__ void dot	(	cuFloatComplex(&)	cv1[3],
		cuFloatComplex(&)	cv2[3],
		cuFloatComplex &	out
	)

Dot product.

Take the dot (inner) product of two complex valued float arrays of size 3.

Parameters

cv1	Array of 3 complex float.
cv2	Array of 3 complex float.
out	Scalar complex float.

dot() [2/4]

__device__ __inline__ void dot	(	cuFloatComplex(&)	cv1[3],
		float(&)	v2[3],
		cuFloatComplex &	out
	)

Dot product.

Take the dot (inner) product of one complex valued and one real valued float array of size 3.

Parameters

cv1	Array of 3 complex float.
v2	Array of 3 float.
out	Scalar complex float.

dot() [3/4]

__device__ __inline__ void dot	(	float(&)	v1[3],
		cuFloatComplex(&)	cv2[3],
		cuFloatComplex &	out
	)

Dot product.

Take the dot (inner) product of one real valued and one complex valued float array of size 3.

Parameters

v1	Array of 3 float.
cv2	Array of 3 complex float.
out	Scalar complex float.

dot() [4/4]

__device__ __inline__ void dot	(	float(&)	v1[3],
		float(&)	v2[3],
		float &	out
	)

Dot product.

Take the dot (inner) product of two real valued arrays of size 3.

Parameters

v1	Array of 3 float.
v2	Array of 3 float.
out	Scalar float.

dyad()

__device__ __inline__ void dyad	(	float(&)	v1[3],
		float(&)	v2[3],
		float(&)	out[3][3]
	)

Dyadic product.

Calculate dyadic product between two real valued float vectors of size 3.

Parameters

v1	Array of 3 float.
v2	Array of 3 float.
out	Array of 3 float, nested inside array of size 3.

ext() [1/4]

__device__ __inline__ void ext	(	cuFloatComplex(&)	cv1[3],
		cuFloatComplex(&)	cv2[3],
		cuFloatComplex(&)	out[3]
	)

Cross product.

Take the cross (outer) product of two complex valued float arrays of size 3.

Parameters

cv1	Array of 3 complex float.
cv2	Array of 3 complex float.
out	Array of 3 complex float.

ext() [2/4]

__device__ __inline__ void ext	(	cuFloatComplex(&)	cv1[3],
		float(&)	v2[3],
		cuFloatComplex(&)	out[3]
	)

Cross product.

Take the cross (outer) product of one complex valued and one real valued float array of size 3.

Parameters

cv1	Array of 3 complex float.
v2	Array of 3 float.
out	Array of 3 complex float.

ext() [3/4]

__device__ __inline__ void ext	(	float(&)	v1[3],
		cuFloatComplex(&)	cv2[3],
		cuFloatComplex(&)	out[3]
	)

Cross product.

Take the cross (outer) product of one real valued and one complex valued float array of size 3.

Parameters

v1	Array of 3 float.
cv2	Array of 3 complex float.
out	Array of 3 complex float.

ext() [4/4]

__device__ __inline__ void ext	(	float(&)	v1[3],
		float(&)	v2[3],
		float(&)	out[3]
	)

Cross product.

Take the cross (outer) product of two real valued float arrays of size 3.

Parameters

v1	Array of 3 float.
v2	Array of 3 float.
out	Array of 3 float.

invmatVec4()

__device__ __inline__ void invmatVec4	(	float(&)	mat[16],
		float(&)	cv1[3],
		float(&)	out[3],
		bool	vec = false
	)

Matrix-vector multiplication.

Multiply a vector by the inverse of a matrix.

Parameters

cv1	Array of 3 float.
out	Array of 3 float.
vec	Whether to rotate as a vector or as a point.

matDiff()

__device__ __inline__ void matDiff	(	float(&)	m1[3][3],
		float(&)	m2[3][3],
		float(&)	out[3][3]
	)

Matrix difference, element wise.

Subtract two 3x3 matrices, element wise.

Parameters

m1	Array of 3 float, nested inside array of size 3.
m2	Array of 3 float, nested inside array of size 3.
out	Array of 3 float, nested inside array of size 3.

matVec() [1/2]

__device__ __inline__ void matVec	(	float(&)	m1[3][3],
		cuFloatComplex(&)	cv1[3],
		cuFloatComplex(&)	out[3]
	)

Matrix-vector product.

Multiply a real valued 3x3 matrix and a complex valued size 3 vector to generate a new complex valued size 3 vector.

Parameters

m1	Array of 3 float, nested inside array of size 3.
cv1	Array of 3 complex float.
out	Array of 3 complex float.

matVec() [2/2]

__device__ __inline__ void matVec	(	float(&)	m1[3][3],
		float(&)	v1[3],
		float(&)	out[3]
	)

Matrix-vector product.

Multiply a real valued 3x3 matrix and a real valued size 3 vector to generate a new real valued size 3 vector.

Parameters

m1	Array of 3 float, nested inside array of size 3.
v1	Array of 3 float.
out	Array of 3 float.

matVec4()

__device__ __inline__ void matVec4	(	float(&)	mat[16],
		float(&)	cv1[3],
		float(&)	out[3],
		bool	vec = false
	)

Matrix-vector multiplication.

Uses mat from constant memory.

Parameters

cv1	Array of 3 float.
out	Array of 3 float.
vec	Whether to rotate as a vector or as a point.

normalize()

__device__ __inline__ void normalize	(	float(&)	v[3],
		float(&)	out[3]
	)

Normalize vector.

Normalize real valued vector of size 3.

Parameters

v	Array of 3 float.
out	Array of 3 float.

s_mult() [1/4]

__device__ __inline__ void s_mult	(	cuFloatComplex(&)	cv[3],
		const float &	s,
		cuFloatComplex(&)	out[3]
	)

Scalar multiplication.

Multiply complex valued vector of size 3 by real scalar, element-wise.

Parameters

cv	Array of 3 complex float.
s	Scalar float.
out	Array of 3 complex float.

s_mult() [2/4]

__device__ __inline__ void s_mult	(	cuFloatComplex(&)	cv[3],
		cuFloatComplex &	cs,
		cuFloatComplex(&)	out[3]
	)

Scalar multiplication.

Multiply complex valued vector of size 3 by complex scalar, element-wise.

Parameters

cv	Array of 3 complex float.
cs	Scalar complex float.
out	Array of 3 complex float.

s_mult() [3/4]

__device__ __inline__ void s_mult	(	float(&)	v[3],
		cuFloatComplex &	cs,
		cuFloatComplex(&)	out[3]
	)

Scalar multiplication.

Multiply real valued vector of size 3 by complex scalar, element-wise.

Parameters

v	Array of 3 float.
cs	Scalar complex float.
out	Array of 3 complex float.

s_mult() [4/4]

__device__ __inline__ void s_mult	(	float(&)	v[3],
		float &	s,
		float(&)	out[3]
	)

Scalar multiplication.

Multiply real valued vector of size 3 by real scalar, element-wise.

Parameters

v	Array of 3 float.
s	Scalar float.
out	Array of 3 float.

snell() [1/2]

__device__ __inline__ void snell	(	cuFloatComplex(&)	cvin[3],
		float(&)	normal[3],
		cuFloatComplex(&)	out[3]
	)

Snell's law reflection.

Calculate reflected direction vector from incoming direction and normal vector.

Parameters

cvin	Array of 3 complex float, incoming direction vector.
normal	Array of 3 float, normal vector of surface.
out	Array of 3 complex float.

snell() [2/2]

__device__ __inline__ void snell	(	float(&)	vin[3],
		float(&)	normal[3],
		float(&)	out[3]
	)

Snell's law reflection.

Calculate reflected direction vector from incoming direction and normal vector.

Parameters

vin	Array of 3 float, incoming direction vector.
normal	Array of 3 float, normal vector of surface.
out	Array of 3 float.

snell_t()

__device__ __inline__ void snell_t	(	float(&)	vin[3],
		float(&)	normal[3],
		float	mu,
		float(&)	out[3]
	)

Snell's law refraction.

Calculate refracted direction vector from incoming direction and normal vector.

Parameters

vin	Array of 3 double/float, incoming direction vector.
normal	Array of 3 double/float, normal vector of surface.
mu	Ratio of n1 to n2.
out	Array of 3 double/float.