static v3_t init(num_t x = num_t(0), num_t y = num_t(0), num_t z = num_t(0)) {
v3_t result;
result.x = x;
result.y = y;
result.z = z;
return result;
}
void init(const v2_t<num_t>& _p0,
const v2_t<num_t>& _p1,
const v2_t<num_t>& _p2) {
p0 = _p0;
p1 = _p1 - _p0;
p2 = _p2 - _p0;
A = num_t(+4) * v2_t<num_t>::dot(p1, p1) -
num_t(4) * v2_t<num_t>::dot(p1, p2) + v2_t<num_t>::dot(p2, p2);
B = num_t(-6) * v2_t<num_t>::dot(p1, p1) +
num_t(3) * v2_t<num_t>::dot(p1, p2);
C = num_t(+2) * v2_t<num_t>::dot(p1, p1);
}
bool _miller_rabin(num_t n, uint32 s, num_t d, uint32 base_no) {
// n - 1 = d * 2^s.
const num_t n_prime = montgomery::calc_n_prime(n);
const num_t r2 = montgomery::calc_r2(n);
const num_t one = montgomery::init(num_t(1), r2, n, n_prime);
const num_t minus_one = n - one;
for (uint32 bi = 0; bi < limits[base_no][1]; ++bi) {
const num_t base = montgomery::init(num_t(bases[base_no][bi]), r2, n, n_prime);
if (witness(base, s, d, n, n_prime, one, minus_one)) {
return false;
}
}
return true;
}
std::pair<v2_t<num_t>, num_t> closest_point(v2_t<num_t> pos) const {
v2_t<num_t> p = pos - p0;
num_t p1p = v2_t<num_t>::dot(p1, p);
num_t p2p = v2_t<num_t>::dot(p2, p);
std::vector<num_t> sol = solve3(A, B, C - p2p + num_t(2) * p1p, -p1p);
num_t d;
num_t min_dist = p.lengthsq();
v2_t<num_t> min_pos = v2_t<num_t>::zero();
num_t min_t = num_t(0);
d = (p - p2).lengthsq();
if (d < min_dist) {
min_dist = d;
min_pos = p2;
min_t = num_t(1);
}
for (size_t i = 0; i < sol.size(); ++i) {
if (sol[i] > num_t(0) && sol[i] < num_t(1)) {
v2_t<num_t> q = get_relpos(sol[i]);
d = (p - q).lengthsq();
if (d < min_dist) {
min_dist = d;
min_pos = q;
min_t = sol[i];
}
}
}
min_dist = ::sqrt(min_dist);
if (v2_t<num_t>::orient(min_pos, min_pos + get_tangent(min_t), p) <
num_t(0)) {
min_dist = -min_dist;
}
return std::make_pair(min_pos + p0, min_dist);
}
Image2D TestSetGenerator::MakeTestSet(int number, Mask2D& rfi, unsigned width, unsigned height, int gaussianNoise)
{
Image2D image;
switch(number)
{
case 0: // Image of all zero's
return Image2D::MakeZeroImage(width, height);
case 1: // Image of all ones
image = Image2D::MakeUnsetImage(width, height);
image.SetAll(1.0);
break;
case 2: // Noise
return MakeNoise(width, height, gaussianNoise);
case 3: { // Several broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 1.0);
} break;
case 4: { // Several broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 0.5);
} break;
case 5: { // Several broadband lines of random length
image = MakeNoise(width, height, gaussianNoise);
AddVarBroadbandToTestSet(image, rfi);
} break;
case 6: { // Different broadband lines + low freq background
image = MakeNoise(width, height, gaussianNoise);
AddVarBroadbandToTestSet(image, rfi);
for(unsigned y=0;y<image.Height();++y) {
for(unsigned x=0;x<image.Width();++x) {
image.AddValue(x, y, sinn(num_t(x)*M_PIn*5.0 / image.Width()) + 0.1);
}
}
} break;
case 7: { // Different broadband lines + high freq background
image = MakeNoise(width, height, gaussianNoise);
for(unsigned y=0;y<image.Height();++y) {
for(unsigned x=0;x<image.Width();++x) {
image.AddValue(x, y, sinn((long double) (x+y*0.1)*M_PIn*5.0L / image.Width() + 0.1));
image.AddValue(x, y, sinn((long double) (x+pown(y, 1.1))*M_PIn*50.0L / image.Width() + 0.1));
}
}
AddVarBroadbandToTestSet(image, rfi);
for(unsigned y=0;y<image.Height();++y) {
for(unsigned x=0;x<image.Width();++x) {
image.AddValue(x, y, 1.0);
}
}
} break;
case 8: { // Different droadband lines + smoothed&subtracted high freq background
image = MakeNoise(width, height, gaussianNoise);
for(unsigned y=0;y<image.Height();++y) {
for(unsigned x=0;x<image.Width();++x) {
image.AddValue(x, y, sinn((num_t) (x+y*0.1)*M_PIn*5.0 / image.Width() + 0.1));
image.AddValue(x, y, sinn((num_t) (x+pown(y, 1.1))*M_PIn*50.0 / image.Width() + 0.1));
}
}
SubtractBackground(image);
AddVarBroadbandToTestSet(image, rfi);
} break;
case 9: { //FFT of 7
image = MakeTestSet(7, rfi, width, height);
Image2D copy(image);
FFTTools::CreateHorizontalFFTImage(image, copy, false);
for(unsigned y=0;y<rfi.Height();++y) {
for(unsigned x=0;x<rfi.Width();++x) {
image.SetValue(x, y, image.Value(x, y) / sqrtn(image.Width()));
}
}
} break;
case 10: { // Identity matrix
image = Image2D::MakeZeroImage(width, height);
unsigned min = width < height ? width : height;
for(unsigned i=0;i<min;++i) {
image.SetValue(i, i, 1.0);
rfi.SetValue(i, i, true);
}
} break;
case 11: { // FFT of identity matrix
image = MakeTestSet(10, rfi, width, height);
Image2D copy(image);
FFTTools::CreateHorizontalFFTImage(image, copy, false);
for(unsigned y=0;y<rfi.Height();++y) {
for(unsigned x=0;x<rfi.Width();++x) {
image.SetValue(x, y, image.Value(x, y) / sqrtn(width));
}
}
} break;
case 12: { // Broadband contaminating all channels
image = MakeNoise(width, height, gaussianNoise);
for(unsigned y=0;y<image.Height();++y) {
for(unsigned x=0;x<image.Width();++x) {
image.AddValue(x, y, sinn((num_t) (x+y*0.1)*M_PIn*5.0 / image.Width() + 0.1));
image.AddValue(x, y, sinn((num_t) (x+powl(y, 1.1))*M_PIn*50.0 / image.Width() + 0.1));
}
}
AddBroadbandToTestSet(image, rfi, 1.0);
} break;
case 13: { // Model of three point sources with broadband RFI
SetModelData(image, rfi, 3);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
AddBroadbandToTestSet(image, rfi, 1.0);
} break;
case 14: { // Model of five point sources with broadband RFI
SetModelData(image, rfi, 5);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
AddBroadbandToTestSet(image, rfi, 1.0);
} break;
case 15: { // Model of five point sources with partial broadband RFI
SetModelData(image, rfi, 5);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
AddBroadbandToTestSet(image, rfi, 0.5);
} break;
case 16: { // Model of five point sources with random broadband RFI
SetModelData(image, rfi, 5);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
AddVarBroadbandToTestSet(image, rfi);
} break;
case 17: { // Background-fitted model of five point sources with random broadband RFI
SetModelData(image, rfi, 5);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
SubtractBackground(image);
AddVarBroadbandToTestSet(image, rfi);
} break;
case 18: { // Model of three point sources with random RFI
SetModelData(image, rfi, 3);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
AddVarBroadbandToTestSet(image, rfi);
} break;
case 19: { // Model of three point sources with noise
SetModelData(image, rfi, 3);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
} break;
case 20: { // Model of five point sources with noise
SetModelData(image, rfi, 5);
Image2D noise = MakeNoise(image.Width(), image.Height(), gaussianNoise);
image += noise;
} break;
case 21: { // Model of three point sources
SetModelData(image, rfi, 3);
} break;
case 22: { // Model of five point sources
image = Image2D::MakeZeroImage(width, height);
SetModelData(image, rfi, 5);
} break;
case 23:
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 0.5, 0.1, true);
break;
case 24:
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 0.5, 10.0, true);
break;
case 25:
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 0.5, 1.0, true);
break;
case 26: { // Several Gaussian broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 1.0, 1.0, false, GaussianShape);
} break;
case 27: { // Several Sinusoidal broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddBroadbandToTestSet(image, rfi, 1.0, 1.0, false, SinusoidalShape);
} break;
case 28: { // Several slewed Gaussian broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddSlewedBroadbandToTestSet(image, rfi, 1.0);
} break;
case 29: { // Several bursty broadband lines
image = MakeNoise(width, height, gaussianNoise);
AddBurstBroadbandToTestSet(image, rfi);
} break;
case 30: { // noise + RFI ^-2 distribution
image = sampleRFIDistribution(width, height, 1.0);
rfi.SetAll<true>();
} break;
case 31: { // noise + RFI ^-2 distribution
image = sampleRFIDistribution(width, height, 0.1);
rfi.SetAll<true>();
} break;
case 32: { // noise + RFI ^-2 distribution
image = sampleRFIDistribution(width, height, 0.01);
rfi.SetAll<true>();
} break;
}
return image;
}
v3_t normalize() const {
num_t rl = num_t(1) / length();
return init(x * rl, y * rl, z * rl);
}
static v3_t zero() { return v3_t::init(num_t(0), num_t(0), num_t(0)); }
num_t v3_t<num_t>::tetrahedron_volume(const v3_t<num_t>& a,
const v3_t<num_t>& b,
const v3_t<num_t>& c,
const v3_t<num_t>& d) {
return dotcross((a - d), (b - d), (c - d)) / num_t(6);
}
static inline v3_t<num_t> operator/(const v3_t<num_t>& a, num_t b) {
return a.scale(num_t(1) / b);
}
v2_t<num_t> get_tangent(num_t t) const {
return num_t(2) * (num_t(1) - t) * p1 + num_t(2) * t * (p2 - p1);
}
v2_t<num_t> get_relpos(num_t t) const {
return num_t(2) * t * (num_t(1) - t) * p1 + t * t * p2;
}
static const num_t limit() {
return num_t(1) << (8 * sizeof(num_t) - 1);
}