//returns log10
__device__ double linearInterpEll(double* interpMat, double ell, constants theConst){
int ell1, ell2;
double ten1, ten2, interpValue;
//find raw value on index array
double ellIndex = ell*theConst.n_ell/theConst.ellMax;
ell1 = (int)(ellIndex);
ell2 = ell1+1;
//retrieve the array value for two indexes
ten1 = __powf(10,interpMat[ell1]);
ten2 = __powf(10,interpMat[ell2]);
interpValue = ten1 + (ten1 - ten2)*(ellIndex - (int)ellIndex);
return log10(interpValue);
}
/** Calculate the gas density, divided by the maximum density GAS_DENSITY
*
* @param y as distance in propagation direction (unit: meters / UNIT_LENGTH)
* @return float_X between 0.0 and 1.0
*/
DINLINE float_X calcNormedDensitiy( float3_X pos )
{
if (pos.y() < VACUUM_Y) return float_X(0.0);
const float3_X exponent = float3_X( math::abs((pos.x() - GAS_CENTER_X)/GAS_SIGMA_X),
math::abs((pos.y() - GAS_CENTER_Y)/GAS_SIGMA_Y),
math::abs((pos.z() - GAS_CENTER_Z)/GAS_SIGMA_Z) );
const float_X density = math::exp(GAS_FACTOR * __powf(exponent.x(), GAS_POWER))
* math::exp(GAS_FACTOR * __powf(exponent.y(), GAS_POWER))
* math::exp(GAS_FACTOR * __powf(exponent.z(), GAS_POWER));
return density;
}
//returns unLog10
__device__ double bilinInterpVal(double* interpMat, double y, double z, int energyBin, double* tempGrid, double* metalGrid, constants theConst){
int y1, y2, z1, z2;
double temp1, temp2, R1, R2, interpValue;
double ten11, ten12, ten21, ten22;
double y1val, y2val, z1val, z2val;
int nx = theConst.binCenterSize;
int ny = theConst.tGridSize;
int nz = theConst.mGridSize;
y1 = getLowerIndex(tempGrid, __powf(10,y), theConst.tGridSize);
y2 = y1+1;
z1 = getLowerIndex(metalGrid, __powf(10,z), theConst.mGridSize);
z2 = z1+1;
y1val = tempGrid[y1];
y2val = tempGrid[y2];
z1val = metalGrid[z1];
z2val = metalGrid[z2];
y = __powf(10,y);
z = __powf(10,z);
temp1 = (y2val-y)/(y2val-y1val);
temp2 = (y-y1val)/(y2val-y1val);
ten11 = __powf(10,tenRetrieve(interpMat, nx, ny, nz, energyBin, y1, z1));
ten21 = __powf(10,tenRetrieve(interpMat, nx, ny, nz, energyBin, y2, z1));
ten12 = __powf(10,tenRetrieve(interpMat, nx, ny, nz, energyBin, y1, z2));
ten22 = __powf(10,tenRetrieve(interpMat, nx, ny, nz, energyBin, y2, z2));
R1 = temp1*ten11 + temp2*ten21;
R2 = temp1*ten12 + temp2*ten22;
interpValue = ((z2val-z)/(z2val-z1val))*R1 + ((z-z1val)/(z2val-z1val))*R2;
return interpValue;
}
__device__ double getEll(double x, double y, double z, double* rotMat, double a, double b){
double xPrime, yPrime, zPrime;
xPrime = rotMat[0]*x + rotMat[1]*y + rotMat[2]*z;
yPrime = rotMat[3]*x + rotMat[4]*y + rotMat[5]*z;
zPrime = rotMat[6]*x + rotMat[7]*y + rotMat[8]*z;
return __powf(xPrime*xPrime + yPrime*yPrime/(a*a) + zPrime*zPrime/(b*b),0.5);
}
/** Calculate the gas density, divided by the maximum density GAS_DENSITY
*
* @param pos as 3D length vector offset to global left top front cell
* @return float_X between 0.0 and 1.0
*/
DINLINE float_X calcNormedDensitiy( floatD_X pos )
{
if (pos.y() < VACUUM_Y) return float_X(0.0);
float_X exponent = float_X(0.0);
if (pos.y() < GAS_CENTER_LEFT)
exponent = math::abs((pos.y() - GAS_CENTER_LEFT) / GAS_SIGMA_LEFT);
else if (pos.y() > GAS_CENTER_RIGHT)
exponent = math::abs((pos.y() - GAS_CENTER_RIGHT) / GAS_SIGMA_RIGHT);
const float_X density = math::exp(float_X(GAS_FACTOR) * __powf(exponent, GAS_POWER));
return density;
}
__device__ void single_precision_intrinsics()
{
float fX, fY;
__cosf(0.0f);
__exp10f(0.0f);
__expf(0.0f);
__fadd_rd(0.0f, 1.0f);
__fadd_rn(0.0f, 1.0f);
__fadd_ru(0.0f, 1.0f);
__fadd_rz(0.0f, 1.0f);
__fdiv_rd(4.0f, 2.0f);
__fdiv_rn(4.0f, 2.0f);
__fdiv_ru(4.0f, 2.0f);
__fdiv_rz(4.0f, 2.0f);
__fdividef(4.0f, 2.0f);
__fmaf_rd(1.0f, 2.0f, 3.0f);
__fmaf_rn(1.0f, 2.0f, 3.0f);
__fmaf_ru(1.0f, 2.0f, 3.0f);
__fmaf_rz(1.0f, 2.0f, 3.0f);
__fmul_rd(1.0f, 2.0f);
__fmul_rn(1.0f, 2.0f);
__fmul_ru(1.0f, 2.0f);
__fmul_rz(1.0f, 2.0f);
__frcp_rd(2.0f);
__frcp_rn(2.0f);
__frcp_ru(2.0f);
__frcp_rz(2.0f);
__frsqrt_rn(4.0f);
__fsqrt_rd(4.0f);
__fsqrt_rn(4.0f);
__fsqrt_ru(4.0f);
__fsqrt_rz(4.0f);
__fsub_rd(2.0f, 1.0f);
__fsub_rn(2.0f, 1.0f);
__fsub_ru(2.0f, 1.0f);
__fsub_rz(2.0f, 1.0f);
__log10f(1.0f);
__log2f(1.0f);
__logf(1.0f);
__powf(1.0f, 0.0f);
__saturatef(0.1f);
__sincosf(0.0f, &fX, &fY);
__sinf(0.0f);
__tanf(0.0f);
}
__device__ inline float occaCuda_fastPow(const float x, const float p){ return __powf(x,p); }
__global__ void integrate(double* rebinCool, double* tempArr, double* metalArr, double* emmArr, double* integral, double* tempGrid, double* metalGrid, double* rotMat, constants theConst, bool debugging){
int i=blockDim.x * blockIdx.x + threadIdx.x; //threadIdx.x is the channel/energy-bin
int j= blockIdx.x;
int x, y;
//integrate from depth/2 to -depth/2
x = j/theConst.nPixX-(theConst.nPixX/2);
y = j%theConst.nPixY-(theConst.nPixY/2);
double pixToMpc = 0.01;
double xMpc = pixToMpc*x;
double yMpc = pixToMpc*y;
int energyBin = threadIdx.x;
int n;
double a, b;
double step=1.0;
double tFunct, h, actErr, last, nextVal;
double T, Z, rebinA, rebinB;
double prevStep[200];
double ell;
b = theConst.nz/2;
a = -theConst.nz/2;
h = b-a;
actErr = 1.0;
last = 1.E30;
nextVal = 0.0;
n = 1;
ell = getEll(xMpc, yMpc, a, rotMat, theConst.a_ell, theConst.b_ell);
if (ell>theConst.ellMax) {
integral[i] = 1.0;
}
T = linearInterpEll(tempArr, ell, theConst);
Z = linearInterpEll(metalArr, ell, theConst);
double ell1 = getEll(xMpc, yMpc, a, rotMat, theConst.a_ell, theConst.b_ell);
double T1 = linearInterpEll(tempArr, ell, theConst);
double Z1 = linearInterpEll(metalArr, ell, theConst);
rebinA = bilinInterpVal(rebinCool, T, Z, energyBin, tempGrid, metalGrid, theConst);
tFunct = 0.5*__powf(__powf(10,linearInterpEll(emmArr, ell, theConst)), 2.0)*rebinA;
ell = getEll(xMpc, yMpc, b, rotMat, theConst.a_ell, theConst.b_ell);
if (ell>theConst.ellMax) {
integral[i] = 1.0;
}
T = linearInterpEll(tempArr, ell, theConst);
Z = linearInterpEll(metalArr, ell, theConst);
rebinB = bilinInterpVal(rebinCool, T, Z, energyBin, tempGrid, metalGrid, theConst);
tFunct += 0.5*__powf(__powf(10,linearInterpEll(emmArr, ell, theConst)), 2.0)*rebinB;
//iterate until convergence of integral
prevStep[0] = tFunct;
while (actErr>=0.1) {
step=step*2.0;
h = double(b-a)/step;
nextVal = 0.0;
for (int l=1; l<step; l=l+2) {
ell = getEll(xMpc, yMpc, l*h+a, rotMat, theConst.a_ell, theConst.b_ell);
if (ell>theConst.ellMax) {
integral[i] = 1.0;
break;
}
T = linearInterpEll(tempArr, ell, theConst);
Z = linearInterpEll(metalArr, ell, theConst);
nextVal+=bilinInterpVal(rebinCool, T, Z, energyBin, tempGrid, metalGrid, theConst)*__powf(__powf(10,linearInterpEll(emmArr, ell, theConst)), 2.0);
}
nextVal+=prevStep[n-1];
prevStep[n]=nextVal;
nextVal=h*(nextVal);
actErr=fabs(last-nextVal);
last=nextVal;
n++;
if (debugging && n>4) {
break;
}
}
if (debugging) {
integral[0]=xMpc;
integral[1]=energyBin;
integral[2]=ell1;
integral[3]=Z1;
integral[4]=linearInterpEll(tempArr, ell, theConst);
integral[5]=tFunct;
integral[6]=linearInterpEll(emmArr, ell, theConst);
integral[7] = rebinA;
integral[8] = n;
integral[9] = last;
integral[10] = j;
integral[11] = a;
integral[12] = b;
integral[13] = T1;
}else {
//place integrations into the 1D array by thread number
//if outside the grid then make minescule
if (ell>theConst.ellMax || integral[i] == 1.0) {
integral[i] = 1.0E-10;
}else{
integral[i] = last;
}
}
}