[729] | 1 | #include "machdefs.h"
|
---|
| 2 | #include <iostream.h>
|
---|
| 3 | #include <math.h>
|
---|
| 4 | #include <complex>
|
---|
| 5 | #include "sphericaltransformserver.h"
|
---|
| 6 | #include "tvector.h"
|
---|
| 7 | #include "nbrandom.h"
|
---|
| 8 | #include "nbmath.h"
|
---|
| 9 |
|
---|
[1218] | 10 | /*! \class SOPHYA::SphericalTransformServer
|
---|
[729] | 11 |
|
---|
[1218] | 12 | Class for performing analysis and synthesis of sky maps using spin-0 or spin-2 spherical harmonics.
|
---|
| 13 |
|
---|
| 14 | Maps must be SOPHYA SphericalMaps (SphereGorski or SphereThetaPhi).
|
---|
| 15 |
|
---|
| 16 | Temperature and polarization (Stokes parameters) can be developped on spherical harmonics :
|
---|
| 17 | \f[
|
---|
| 18 | \frac{\Delta T}{T}(\hat{n})=\sum_{lm}a_{lm}^TY_l^m(\hat{n})
|
---|
| 19 | \f]
|
---|
| 20 | \f[
|
---|
| 21 | Q(\hat{n})=\frac{1}{\sqrt{2}}\sum_{lm}N_l\left(a_{lm}^EW_{lm}(\hat{n})+a_{lm}^BX_{lm}(\hat{n})\right)
|
---|
| 22 | \f]
|
---|
| 23 | \f[
|
---|
| 24 | U(\hat{n})=-\frac{1}{\sqrt{2}}\sum_{lm}N_l\left(a_{lm}^EX_{lm}(\hat{n})-a_{lm}^BW_{lm}(\hat{n})\right)
|
---|
| 25 | \f]
|
---|
| 26 | \f[
|
---|
| 27 | \left(Q \pm iU\right)(\hat{n})=\sum_{lm}a_{\pm 2lm}\, _{\pm 2}Y_l^m(\hat{n})
|
---|
| 28 | \f]
|
---|
| 29 |
|
---|
| 30 | \f[
|
---|
| 31 | Y_l^m(\hat{n})=\lambda_l^m(\theta)e^{im\phi}
|
---|
| 32 | \f]
|
---|
| 33 | \f[
|
---|
| 34 | _{\pm}Y_l^m(\hat{n})=_{\pm}\lambda_l^m(\theta)e^{im\phi}
|
---|
| 35 | \f]
|
---|
| 36 | \f[
|
---|
| 37 | W_{lm}(\hat{n})=\frac{1}{N_l}\,_{w}\lambda_l^m(\theta)e^{im\phi}
|
---|
| 38 | \f]
|
---|
| 39 | \f[
|
---|
| 40 | X_{lm}(\hat{n})=\frac{-i}{N_l}\,_{x}\lambda_l^m(\theta)e^{im\phi}
|
---|
| 41 | \f]
|
---|
| 42 |
|
---|
| 43 | (see LambdaLMBuilder, LambdaPMBuilder, LambdaWXBuilder classes)
|
---|
| 44 |
|
---|
| 45 | power spectra :
|
---|
| 46 |
|
---|
| 47 | \f[
|
---|
| 48 | C_l^T=\frac{1}{2l+1}\sum_{m=0}^{+ \infty }\left|a_{lm}^T\right|^2=\langle\left|a_{lm}^T\right|^2\rangle
|
---|
| 49 | \f]
|
---|
| 50 | \f[
|
---|
| 51 | C_l^E=\frac{1}{2l+1}\sum_{m=0}^{+\infty}\left|a_{lm}^E\right|^2=\langle\left|a_{lm}^E\right|^2\rangle
|
---|
| 52 | \f]
|
---|
| 53 | \f[
|
---|
| 54 | C_l^B=\frac{1}{2l+1}\sum_{m=0}^{+\infty}\left|a_{lm}^B\right|^2=\langle\left|a_{lm}^B\right|^2\rangle
|
---|
| 55 | \f]
|
---|
| 56 |
|
---|
| 57 | \arg
|
---|
| 58 | \b Synthesis : Get temperature and polarization maps from \f$a_{lm}\f$ coefficients or from power spectra, (methods GenerateFrom...).
|
---|
| 59 |
|
---|
| 60 | \b Temperature:
|
---|
| 61 | \f[
|
---|
| 62 | \frac{\Delta T}{T}(\hat{n})=\sum_{lm}a_{lm}^TY_l^m(\hat{n}) = \sum_{-\infty}^{+\infty}b_m(\theta)e^{im\phi}
|
---|
| 63 | \f]
|
---|
| 64 |
|
---|
| 65 | with
|
---|
| 66 | \f[
|
---|
| 67 | b_m(\theta)=\sum_{l=\left|m\right|}^{+\infty}a_{lm}^T\lambda_l^m(\theta)
|
---|
| 68 | \f]
|
---|
| 69 |
|
---|
| 70 | \b Polarisation
|
---|
| 71 | \f[
|
---|
| 72 | Q \pm iU = \sum_{-\infty}^{+\infty}b_m^{\pm}(\theta)e^{im\phi}
|
---|
| 73 | \f]
|
---|
| 74 |
|
---|
| 75 | where :
|
---|
| 76 | \f[
|
---|
| 77 | b_m^{\pm}(\theta) = \sum_{l=\left|m\right|}^{+\infty}a_{\pm 2lm}\,_{\pm}\lambda_l^m(\theta)
|
---|
| 78 | \f]
|
---|
| 79 |
|
---|
| 80 | or :
|
---|
| 81 | \f[
|
---|
| 82 | Q = \sum_{-\infty}^{+\infty}b_m^{Q}(\theta)e^{im\phi}
|
---|
| 83 | \f]
|
---|
| 84 | \f[
|
---|
| 85 | U = \sum_{-\infty}^{+\infty}b_m^{U}(\theta)e^{im\phi}
|
---|
| 86 | \f]
|
---|
| 87 |
|
---|
| 88 | where:
|
---|
| 89 | \f[
|
---|
| 90 | b_m^{Q}(\theta) = \frac{1}{\sqrt{2}}\sum_{l=\left|m\right|}^{+\infty}\left(a_{lm}^E\,_{w}\lambda_l^m(\theta)-ia_{lm}^B\,_{x}\lambda_l^m(\theta)\right)
|
---|
| 91 | \f]
|
---|
| 92 | \f[
|
---|
| 93 | b_m^{U}(\theta) = \frac{1}{\sqrt{2}}\sum_{l=\left|m\right|}^{+\infty}\left(ia_{lm}^E\,_{x}\lambda_l^m(\theta)+a_{lm}^B\,_{w}\lambda_l^m(\theta)\right)
|
---|
| 94 | \f]
|
---|
| 95 |
|
---|
| 96 | Since the pixelization provides "slices" with constant \f$\theta\f$ and \f$\phi\f$ equally distributed on \f$2\pi\f$ \f$\frac{\Delta T}{T}\f$, \f$Q\f$,\f$U\f$ can be computed by FFT.
|
---|
| 97 |
|
---|
| 98 |
|
---|
| 99 | \arg
|
---|
| 100 | \b Analysis : Get \f$a_{lm}\f$ coefficients or power spectra from temperature and polarization maps (methods DecomposeTo...).
|
---|
| 101 |
|
---|
| 102 | \b Temperature:
|
---|
| 103 | \f[
|
---|
| 104 | a_{lm}^T=\int\frac{\Delta T}{T}(\hat{n})Y_l^{m*}(\hat{n})d\hat{n}
|
---|
| 105 | \f]
|
---|
| 106 |
|
---|
| 107 | approximated as :
|
---|
| 108 | \f[
|
---|
| 109 | a_{lm}^T=\sum_{\theta_k}\omega_kC_m(\theta_k)\lambda_l^m(\theta_k)
|
---|
| 110 | \f]
|
---|
| 111 | where :
|
---|
| 112 | \f[
|
---|
| 113 | C_m (\theta _k)=\sum_{\phi _{k\prime}}\frac{\Delta T}{T}(\theta _k,\phi_{k\prime})e^{-im\phi _{k\prime}}
|
---|
| 114 | \f]
|
---|
| 115 | Since the pixelization provides "slices" with constant \f$\theta\f$ and \f$\phi\f$ equally distributed on \f$2\pi\f$ (\f$\omega_k\f$ is the solid angle of each pixel of the slice \f$\theta_k\f$) \f$C_m\f$ can be computed by FFT.
|
---|
| 116 |
|
---|
| 117 | \b polarisation:
|
---|
| 118 |
|
---|
| 119 | \f[
|
---|
| 120 | a_{\pm 2lm}=\sum_{\theta_k}\omega_kC_m^{\pm}(\theta_k)\,_{\pm}\lambda_l^m(\theta_k)
|
---|
| 121 | \f]
|
---|
| 122 | where :
|
---|
| 123 | \f[
|
---|
| 124 | C_m^{\pm} (\theta _k)=\sum_{\phi _{k\prime}}\left(Q \pm iU\right)(\theta _k,\phi_{k\prime})e^{-im\phi _{k\prime}}
|
---|
| 125 | \f]
|
---|
| 126 | or :
|
---|
| 127 |
|
---|
| 128 | \f[
|
---|
| 129 | a_{lm}^E=\frac{1}{\sqrt{2}}\sum_{\theta_k}\omega_k\left(C_m^{Q}(\theta_k)\,_{w}\lambda_l^m(\theta_k)-iC_m^{U}(\theta_k)\,_{x}\lambda_l^m(\theta_k)\right)
|
---|
| 130 | \f]
|
---|
| 131 | \f[
|
---|
| 132 | a_{lm}^B=\frac{1}{\sqrt{2}}\sum_{\theta_k}\omega_k\left(iC_m^{Q}(\theta_k)\,_{x}\lambda_l^m(\theta_k)+C_m^{U}(\theta_k)\,_{w}\lambda_l^m(\theta_k)\right)
|
---|
| 133 | \f]
|
---|
| 134 |
|
---|
| 135 | where :
|
---|
| 136 | \f[
|
---|
| 137 | C_m^{Q} (\theta _k)=\sum_{\phi _{k\prime}}Q(\theta _k,\phi_{k\prime})e^{-im\phi _{k\prime}}
|
---|
| 138 | \f]
|
---|
| 139 | \f[
|
---|
| 140 | C_m^{U} (\theta _k)=\sum_{\phi _{k\prime}}U(\theta _k,\phi_{k\prime})e^{-im\phi _{k\prime}}
|
---|
| 141 | \f]
|
---|
| 142 |
|
---|
| 143 | */
|
---|
| 144 |
|
---|
| 145 | /*! \fn void SOPHYA::SphericalTransformServer::GenerateFromAlm( SphericalMap<T>& map, int_4 pixelSizeIndex, const Alm<T>& alm) const
|
---|
| 146 |
|
---|
| 147 | synthesis of a temperature map from Alm coefficients
|
---|
| 148 | */
|
---|
[729] | 149 | template<class T>
|
---|
| 150 | void SphericalTransformServer<T>::GenerateFromAlm( SphericalMap<T>& map, int_4 pixelSizeIndex, const Alm<T>& alm) const
|
---|
| 151 | {
|
---|
| 152 | /*=======================================================================
|
---|
| 153 | computes a map form its alm for the HEALPIX pixelisation
|
---|
| 154 | map(theta,phi) = sum_l_m a_lm Y_lm(theta,phi)
|
---|
| 155 | = sum_m {e^(i*m*phi) sum_l a_lm*lambda_lm(theta)}
|
---|
| 156 |
|
---|
| 157 | where Y_lm(theta,phi) = lambda(theta) * e^(i*m*phi)
|
---|
| 158 |
|
---|
| 159 | * the recurrence of Ylm is the standard one (cf Num Rec)
|
---|
| 160 | * the sum over m is done by FFT
|
---|
| 161 |
|
---|
| 162 | =======================================================================*/
|
---|
| 163 | int_4 nlmax=alm.Lmax();
|
---|
| 164 | int_4 nmmax=nlmax;
|
---|
| 165 | int_4 nsmax=0;
|
---|
| 166 | map.Resize(pixelSizeIndex);
|
---|
| 167 | char* sphere_type=map.TypeOfMap();
|
---|
| 168 | if (strncmp(sphere_type,"RING",4) == 0)
|
---|
| 169 | {
|
---|
| 170 | nsmax=map.SizeIndex();
|
---|
| 171 | }
|
---|
| 172 | else
|
---|
| 173 | // pour une sphere Gorski le nombre de pixels est 12*nsmax**2
|
---|
| 174 | // on calcule une quantite equivalente a nsmax pour la sphere-theta-phi
|
---|
| 175 | // en vue de l'application du critere Healpix : nlmax<=3*nsmax-1
|
---|
| 176 | // c'est approximatif ; a raffiner.
|
---|
| 177 | if (strncmp(sphere_type,"TETAFI",6) == 0)
|
---|
| 178 | {
|
---|
| 179 | nsmax=(int_4)sqrt(map.NbPixels()/12.);
|
---|
| 180 | }
|
---|
| 181 | else
|
---|
| 182 | {
|
---|
| 183 | cout << " unknown type of sphere : " << sphere_type << endl;
|
---|
| 184 | throw IOExc(" unknown type of sphere: " + (string)sphere_type );
|
---|
| 185 | }
|
---|
| 186 | cout << "GenerateFromAlm: the sphere is of type : " << sphere_type << endl;
|
---|
| 187 | cout << "GenerateFromAlm: size index (nside) of the sphere= " << nsmax << endl;
|
---|
| 188 | cout << "GenerateFromAlm: nlmax (from Alm) = " << nlmax << endl;
|
---|
| 189 | if (nlmax>3*nsmax-1)
|
---|
| 190 | {
|
---|
| 191 | cout << "GenerateFromAlm: nlmax should be <= 3*nside-1" << endl;
|
---|
| 192 | if (strncmp(sphere_type,"TETAFI",6) == 0)
|
---|
| 193 | {
|
---|
| 194 | cout << " (for this criterium, nsmax is computed as sqrt(nbPixels/12))" << endl;
|
---|
| 195 | }
|
---|
| 196 | }
|
---|
| 197 | Bm<complex<T> > b_m_theta(nmmax);
|
---|
| 198 |
|
---|
| 199 | // map.Resize(nsmax);
|
---|
| 200 |
|
---|
| 201 |
|
---|
| 202 | // pour chaque tranche en theta
|
---|
[746] | 203 | for (int_4 ith = 0; ith < map.NbThetaSlices();ith++)
|
---|
[729] | 204 | {
|
---|
| 205 | int_4 nph;
|
---|
| 206 | r_8 phi0;
|
---|
| 207 | r_8 theta;
|
---|
| 208 | TVector<int_4> pixNumber;
|
---|
| 209 | TVector<T> datan;
|
---|
| 210 |
|
---|
| 211 | map.GetThetaSlice(ith,theta,phi0, pixNumber,datan);
|
---|
| 212 | nph = pixNumber.NElts();
|
---|
| 213 |
|
---|
| 214 | // -----------------------------------------------------
|
---|
| 215 | // for each theta, and each m, computes
|
---|
| 216 | // b(m,theta) = sum_over_l>m (lambda_l_m(theta) * a_l_m)
|
---|
| 217 | // ------------------------------------------------------
|
---|
| 218 | LambdaLMBuilder lb(theta,nlmax,nmmax);
|
---|
| 219 | // somme sur m de 0 a l'infini
|
---|
[833] | 220 | int m;
|
---|
| 221 | for (m = 0; m <= nmmax; m++)
|
---|
[729] | 222 | {
|
---|
| 223 | // somme sur l de m a l'infini
|
---|
| 224 | b_m_theta(m) = (T)( lb.lamlm(m,m) ) * alm(m,m);
|
---|
| 225 | // if (ith==0 && m==0)
|
---|
| 226 | // {
|
---|
| 227 | // cout << " guy: lmm= " << lb.lamlm(m,m) << " alm " << alm(m,m) << "b00= " << b_m_theta(m) << endl;
|
---|
| 228 | // }
|
---|
| 229 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 230 | {
|
---|
| 231 | b_m_theta(m) += (T)( lb.lamlm(l,m) ) * alm(l,m);
|
---|
| 232 |
|
---|
| 233 |
|
---|
| 234 | // if (ith==0 && m==0)
|
---|
| 235 | // {
|
---|
| 236 | // cout << " guy:l=" << l << " m= " << m << " lmm= " << lb.lamlm(l,m) << " alm " << alm(l,m) << "b00= " << b_m_theta(m) << endl;
|
---|
| 237 |
|
---|
| 238 | // }
|
---|
| 239 |
|
---|
| 240 | }
|
---|
| 241 | }
|
---|
| 242 |
|
---|
| 243 | // obtains the negative m of b(m,theta) (= complex conjugate)
|
---|
| 244 |
|
---|
[833] | 245 | for (m=1;m<=nmmax;m++)
|
---|
[729] | 246 | {
|
---|
| 247 | //compiler doesn't have conj()
|
---|
| 248 | b_m_theta(-m) = conj(b_m_theta(m));
|
---|
| 249 | }
|
---|
| 250 | // ---------------------------------------------------------------
|
---|
| 251 | // sum_m b(m,theta)*exp(i*m*phi) -> f(phi,theta)
|
---|
| 252 | // ---------------------------------------------------------------*/
|
---|
| 253 | // TVector<complex<T> > Temp = fourierSynthesisFromB(b_m_theta,nph,phi0);
|
---|
| 254 | TVector<T> Temp = RfourierSynthesisFromB(b_m_theta,nph,phi0);
|
---|
| 255 | for (int i=0;i< nph;i++)
|
---|
| 256 | {
|
---|
| 257 | // map(pixNumber(i))=Temp(i).real();
|
---|
| 258 | map(pixNumber(i))=Temp(i);
|
---|
| 259 | }
|
---|
| 260 | }
|
---|
| 261 | }
|
---|
| 262 |
|
---|
| 263 |
|
---|
| 264 |
|
---|
[1218] | 265 | /*! \fn TVector< complex<T> > SOPHYA::SphericalTransformServer::fourierSynthesisFromB(const Bm<complex<T> >& b_m, int_4 nph, r_8 phi0) const
|
---|
| 266 |
|
---|
| 267 | \return a vector with nph elements which are sums :\f$\sum_{m=-mmax}^{mmax}b_m(\theta)e^{im\varphi}\f$ for nph values of \f$\varphi\f$ regularly distributed in \f$[0,\pi]\f$ ( calculated by FFT)
|
---|
| 268 |
|
---|
| 269 | The object b_m (\f$b_m\f$) of the class Bm is a special vector which index goes from -mmax to mmax.
|
---|
| 270 | */
|
---|
[729] | 271 | template<class T>
|
---|
| 272 | TVector< complex<T> > SphericalTransformServer<T>::fourierSynthesisFromB(const Bm<complex<T> >& b_m, int_4 nph, r_8 phi0) const
|
---|
| 273 | {
|
---|
| 274 | /*=======================================================================
|
---|
| 275 | dataout(j) = sum_m datain(m) * exp(i*m*phi(j))
|
---|
| 276 | with phi(j) = j*2pi/nph + kphi0*pi/nph and kphi0 =0 or 1
|
---|
| 277 |
|
---|
| 278 | as the set of frequencies {m} is larger than nph,
|
---|
| 279 | we wrap frequencies within {0..nph-1}
|
---|
| 280 | ie m = k*nph + m' with m' in {0..nph-1}
|
---|
| 281 | then
|
---|
| 282 | noting bw(m') = exp(i*m'*phi0)
|
---|
| 283 | * sum_k (datain(k*nph+m') exp(i*k*pi*kphi0))
|
---|
| 284 | with bw(nph-m') = CONJ(bw(m')) (if datain(-m) = CONJ(datain(m)))
|
---|
| 285 | dataout(j) = sum_m' [ bw(m') exp (i*j*m'*2pi/nph) ]
|
---|
| 286 | = Fourier Transform of bw
|
---|
| 287 | is real
|
---|
| 288 |
|
---|
| 289 | NB nph is not necessarily a power of 2
|
---|
| 290 |
|
---|
| 291 | =======================================================================*/
|
---|
| 292 | //**********************************************************************
|
---|
| 293 | // pour une valeur de phi (indexee par j) la temperature est la transformee
|
---|
| 294 | // de Fourier de bm (somme sur m de -nmax a +nmmax de bm*exp(i*m*phi)).
|
---|
| 295 | // on demande nph (nombre de pixels sur la tranche) valeurs de transformees, pour nph valeurs de phi, regulierement reparties sur 2*pi. On a:
|
---|
| 296 | // DT/T(j) = sum_m b(m) * exp(i*m*phi(j))
|
---|
| 297 | // sommation de -infini a +infini, en fait limitee a -nmamx, +nmmax
|
---|
| 298 | // On pose m=k*nph + m', avec m' compris entre 0 et nph-1. Alors :
|
---|
| 299 | // DT/T(j) = somme_k somme_m' b(k*nph + m')*exp(i*(k*nph + m')*phi(j))
|
---|
| 300 | // somme_k : de -infini a +infini
|
---|
| 301 | // somme_m' : de 0 a nph-1
|
---|
| 302 | // On echange les sommations :
|
---|
| 303 | // DT/T(j) = somme_k (exp(i*m'*phi(j)) somme_m' b(k*nph + m')*exp(i*(k*nph*phi(j))
|
---|
| 304 | // mais phi(j) est un multiple entier de 2*pi/nph, la seconde exponentielle
|
---|
| 305 | // vaut 1.
|
---|
| 306 | // Il reste a calculer les transformees de Fourier de somme_m' b(k*nph + m')
|
---|
| 307 | // si phi0 n'est pas nul, il y a juste un decalage a faire.
|
---|
| 308 | //**********************************************************************
|
---|
| 309 |
|
---|
| 310 | TVector< complex<T> > bw(nph);
|
---|
| 311 | TVector< complex<T> > dataout(nph);
|
---|
| 312 | TVector< complex<T> > data(nph);
|
---|
| 313 |
|
---|
| 314 |
|
---|
| 315 | for (int kk=0; kk<bw.NElts(); kk++) bw(kk)=(T)0.;
|
---|
[833] | 316 | int m;
|
---|
| 317 | for (m=-b_m.Mmax();m<=-1;m++)
|
---|
[729] | 318 | {
|
---|
| 319 | int maux=m;
|
---|
| 320 | while (maux<0) maux+=nph;
|
---|
| 321 | int iw=maux%nph;
|
---|
| 322 | double aux=(m-iw)*phi0;
|
---|
| 323 | bw(iw) += b_m(m) * complex<T>( (T)cos(aux),(T)sin(aux) ) ;
|
---|
| 324 | }
|
---|
[833] | 325 | for (m=0;m<=b_m.Mmax();m++)
|
---|
[729] | 326 | {
|
---|
| 327 | // int iw=((m % nph) +nph) % nph; //between 0 and nph = m'
|
---|
| 328 | int iw=m%nph;
|
---|
| 329 | double aux=(m-iw)*phi0;
|
---|
| 330 | bw(iw)+=b_m(m) * complex<T>( (T)cos(aux),(T)sin(aux) );
|
---|
| 331 | }
|
---|
| 332 |
|
---|
| 333 | // applies the shift in position <-> phase factor in Fourier space
|
---|
| 334 | for (int mprime=0; mprime < nph; mprime++)
|
---|
| 335 | {
|
---|
| 336 | complex<double> aux(cos(mprime*phi0),sin(mprime*phi0));
|
---|
| 337 | data(mprime)=bw(mprime)*
|
---|
| 338 | (complex<T>)(complex<double>(cos(mprime*phi0),sin(mprime*phi0)));
|
---|
| 339 | }
|
---|
| 340 |
|
---|
| 341 | //sortie.ReSize(nph);
|
---|
| 342 | TVector< complex<T> > sortie(nph);
|
---|
| 343 |
|
---|
| 344 | fftIntfPtr_-> FFTBackward(data, sortie);
|
---|
| 345 |
|
---|
| 346 | return sortie;
|
---|
| 347 | }
|
---|
| 348 |
|
---|
| 349 | //********************************************
|
---|
[1218] | 350 | /*! \fn TVector<T> SOPHYA::SphericalTransformServer::RfourierSynthesisFromB(const Bm<complex<T> >& b_m, int_4 nph, r_8 phi0) const
|
---|
| 351 |
|
---|
| 352 | same as fourierSynthesisFromB, but return a real vector, taking into account the fact that b(-m) is conjugate of b(m) */
|
---|
[729] | 353 | template<class T>
|
---|
| 354 | TVector<T> SphericalTransformServer<T>::RfourierSynthesisFromB(const Bm<complex<T> >& b_m, int_4 nph, r_8 phi0) const
|
---|
| 355 | {
|
---|
| 356 | /*=======================================================================
|
---|
| 357 | dataout(j) = sum_m datain(m) * exp(i*m*phi(j))
|
---|
| 358 | with phi(j) = j*2pi/nph + kphi0*pi/nph and kphi0 =0 or 1
|
---|
| 359 |
|
---|
| 360 | as the set of frequencies {m} is larger than nph,
|
---|
| 361 | we wrap frequencies within {0..nph-1}
|
---|
| 362 | ie m = k*nph + m' with m' in {0..nph-1}
|
---|
| 363 | then
|
---|
| 364 | noting bw(m') = exp(i*m'*phi0)
|
---|
| 365 | * sum_k (datain(k*nph+m') exp(i*k*pi*kphi0))
|
---|
| 366 | with bw(nph-m') = CONJ(bw(m')) (if datain(-m) = CONJ(datain(m)))
|
---|
| 367 | dataout(j) = sum_m' [ bw(m') exp (i*j*m'*2pi/nph) ]
|
---|
| 368 | = Fourier Transform of bw
|
---|
| 369 | is real
|
---|
| 370 |
|
---|
| 371 | NB nph is not necessarily a power of 2
|
---|
| 372 |
|
---|
| 373 | =======================================================================*/
|
---|
| 374 | //**********************************************************************
|
---|
| 375 | // pour une valeur de phi (indexee par j) la temperature est la transformee
|
---|
| 376 | // de Fourier de bm (somme sur m de -nmax a +nmmax de bm*exp(i*m*phi)).
|
---|
| 377 | // on demande nph (nombre de pixels sur la tranche) valeurs de transformees, pour nph valeurs de phi, regulierement reparties sur 2*pi. On a:
|
---|
| 378 | // DT/T(j) = sum_m b(m) * exp(i*m*phi(j))
|
---|
| 379 | // sommation de -infini a +infini, en fait limitee a -nmamx, +nmmax
|
---|
| 380 | // On pose m=k*nph + m', avec m' compris entre 0 et nph-1. Alors :
|
---|
| 381 | // DT/T(j) = somme_k somme_m' b(k*nph + m')*exp(i*(k*nph + m')*phi(j))
|
---|
| 382 | // somme_k : de -infini a +infini
|
---|
| 383 | // somme_m' : de 0 a nph-1
|
---|
| 384 | // On echange les sommations :
|
---|
| 385 | // DT/T(j) = somme_k (exp(i*m'*phi(j)) somme_m' b(k*nph + m')*exp(i*(k*nph*phi(j))
|
---|
| 386 | // mais phi(j) est un multiple entier de 2*pi/nph, la seconde exponentielle
|
---|
| 387 | // vaut 1.
|
---|
| 388 | // Il reste a calculer les transformees de Fourier de somme_m' b(k*nph + m')
|
---|
| 389 | // si phi0 n'est pas nul, il y a juste un decalage a faire.
|
---|
| 390 | //**********************************************************************
|
---|
| 391 |
|
---|
| 392 | TVector< complex<T> > bw(nph);
|
---|
| 393 | TVector< complex<T> > dataout(nph);
|
---|
| 394 | TVector< complex<T> > data(nph/2+1);
|
---|
| 395 |
|
---|
| 396 |
|
---|
| 397 | for (int kk=0; kk<bw.NElts(); kk++) bw(kk)=(T)0.;
|
---|
[833] | 398 | int m;
|
---|
| 399 | for (m=-b_m.Mmax();m<=-1;m++)
|
---|
[729] | 400 | {
|
---|
| 401 | int maux=m;
|
---|
| 402 | while (maux<0) maux+=nph;
|
---|
| 403 | int iw=maux%nph;
|
---|
| 404 | double aux=(m-iw)*phi0;
|
---|
| 405 | bw(iw) += b_m(m) * complex<T>( (T)cos(aux),(T)sin(aux) ) ;
|
---|
| 406 | }
|
---|
[833] | 407 | for (m=0;m<=b_m.Mmax();m++)
|
---|
[729] | 408 | {
|
---|
| 409 | // int iw=((m % nph) +nph) % nph; //between 0 and nph = m'
|
---|
| 410 | int iw=m%nph;
|
---|
| 411 | double aux=(m-iw)*phi0;
|
---|
| 412 | bw(iw)+=b_m(m) * complex<T>( (T)cos(aux),(T)sin(aux) );
|
---|
| 413 | }
|
---|
| 414 |
|
---|
| 415 | // applies the shift in position <-> phase factor in Fourier space
|
---|
| 416 | for (int mprime=0; mprime <= nph/2; mprime++)
|
---|
| 417 | {
|
---|
| 418 | complex<double> aux(cos(mprime*phi0),sin(mprime*phi0));
|
---|
| 419 | data(mprime)=bw(mprime)*
|
---|
| 420 | (complex<T>)(complex<double>(cos(mprime*phi0),sin(mprime*phi0)));
|
---|
| 421 | }
|
---|
| 422 |
|
---|
| 423 | //sortie.ReSize(nph);
|
---|
| 424 | TVector<T> sortie;
|
---|
| 425 |
|
---|
| 426 | fftIntfPtr_-> FFTBackward(data, sortie);
|
---|
| 427 |
|
---|
| 428 | return sortie;
|
---|
| 429 | }
|
---|
| 430 | //*******************************************
|
---|
| 431 |
|
---|
[1218] | 432 | /*! \fn Alm<T> SOPHYA::SphericalTransformServer::DecomposeToAlm(const SphericalMap<T>& map, int_4 nlmax, r_8 cos_theta_cut) const
|
---|
| 433 |
|
---|
| 434 | \return the Alm coefficients from analysis of a temperature map.
|
---|
| 435 |
|
---|
| 436 | \param<nlmax> : maximum value of the l index
|
---|
| 437 |
|
---|
| 438 | \param<cos_theta_cut> : cosinus of the symmetric cut EULER angle theta : cos_theta_cut=0 means no cut ; cos_theta_cut=1 all the sphere is cut.
|
---|
| 439 | */
|
---|
[729] | 440 | template<class T>
|
---|
| 441 | Alm<T> SphericalTransformServer<T>::DecomposeToAlm(const SphericalMap<T>& map, int_4 nlmax, r_8 cos_theta_cut) const
|
---|
| 442 | {
|
---|
| 443 |
|
---|
| 444 | /*-----------------------------------------------------------------------
|
---|
| 445 | computes the integral in phi : phas_m(theta)
|
---|
| 446 | for each parallele from north to south pole
|
---|
| 447 | -----------------------------------------------------------------------*/
|
---|
| 448 | TVector<T> data;
|
---|
| 449 | TVector<int_4> pixNumber;
|
---|
| 450 | int_4 nmmax = nlmax;
|
---|
| 451 | TVector< complex<T> > phase(nmmax+1);
|
---|
| 452 | Alm<T> alm;
|
---|
| 453 | alm.ReSizeToLmax(nlmax);
|
---|
[746] | 454 | for (int_4 ith = 0; ith < map.NbThetaSlices(); ith++)
|
---|
[729] | 455 | {
|
---|
| 456 | r_8 phi0;
|
---|
| 457 | r_8 theta;
|
---|
| 458 | map.GetThetaSlice(ith,theta,phi0,pixNumber ,data);
|
---|
| 459 | for (int i=0;i< nmmax+1;i++)
|
---|
| 460 | {
|
---|
| 461 | phase(i)=0;
|
---|
| 462 | }
|
---|
| 463 | double cth = cos(theta);
|
---|
| 464 |
|
---|
| 465 | //part of the sky out of the symetric cut
|
---|
[1428] | 466 | bool keep_it = (fabs(cth) >= cos_theta_cut);
|
---|
[729] | 467 |
|
---|
| 468 | if (keep_it)
|
---|
| 469 | {
|
---|
| 470 | // tableau datain a supprimer
|
---|
[1328] | 471 | // TVector<complex<T> > datain(pixNumber.NElts());
|
---|
[729] | 472 | // for(int kk=0; kk<nph; kk++) datain(kk)=complex<T>(data(kk),(T)0.);
|
---|
| 473 |
|
---|
[746] | 474 | // phase = CFromFourierAnalysis(nmmax,datain,phi0);
|
---|
| 475 | phase = CFromFourierAnalysis(nmmax,data,phi0);
|
---|
[729] | 476 |
|
---|
| 477 | }
|
---|
| 478 |
|
---|
| 479 | /*-----------------------------------------------------------------------
|
---|
| 480 | computes the a_lm by integrating over theta
|
---|
| 481 | lambda_lm(theta) * phas_m(theta)
|
---|
| 482 | for each m and l
|
---|
| 483 | -----------------------------------------------------------------------*/
|
---|
| 484 | // LambdaBuilder lb(theta,nlmax,nmmax);
|
---|
| 485 | LambdaLMBuilder lb(theta,nlmax,nmmax);
|
---|
| 486 | r_8 domega=map.PixSolAngle(map.PixIndexSph(theta,phi0));
|
---|
| 487 | for (int m = 0; m <= nmmax; m++)
|
---|
| 488 | {
|
---|
| 489 | alm(m,m) += (T)lb.lamlm(m,m) * phase(m) * (T)domega; //m,m even
|
---|
| 490 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 491 | {
|
---|
| 492 | alm(l,m) += (T)lb.lamlm(l,m) * phase(m)*(T)domega;
|
---|
| 493 | }
|
---|
| 494 | }
|
---|
| 495 | }
|
---|
| 496 | return alm;
|
---|
| 497 | }
|
---|
[1218] | 498 | /*! \fn TVector< complex<T> > SOPHYA::SphericalTransformServer::CFromFourierAnalysis(int_4 nmmax, const TVector<complex<T> >datain, r_8 phi0) const
|
---|
| 499 |
|
---|
| 500 | \return a vector with mmax elements which are sums :
|
---|
| 501 | \f$\sum_{k=0}^{nphi}datain(\theta,\varphi_k)e^{im\varphi_k}\f$ for (mmax+1) values of \f$m\f$ from 0 to mmax.
|
---|
| 502 | */
|
---|
[729] | 503 | template<class T>
|
---|
[746] | 504 | TVector< complex<T> > SphericalTransformServer<T>::CFromFourierAnalysis(int_4 nmmax, const TVector<complex<T> >datain, r_8 phi0) const
|
---|
[729] | 505 | {
|
---|
| 506 | /*=======================================================================
|
---|
| 507 | integrates (data * phi-dependence-of-Ylm) over phi
|
---|
| 508 | --> function of m can be computed by FFT
|
---|
| 509 |
|
---|
| 510 | datain est modifie
|
---|
| 511 | =======================================================================*/
|
---|
| 512 | int_4 nph=datain.NElts();
|
---|
| 513 | if (nph <= 0)
|
---|
| 514 | {
|
---|
| 515 | throw PException("bizarre : vecteur datain de longueur nulle (CFromFourierAnalysis)");
|
---|
| 516 | }
|
---|
| 517 | TVector<complex<T> > transformedData(nph);
|
---|
| 518 | fftIntfPtr_-> FFTForward(datain, transformedData);
|
---|
| 519 |
|
---|
| 520 | //dataout.ReSize(nmmax+1);
|
---|
| 521 | TVector< complex<T> > dataout(nmmax+1);
|
---|
| 522 |
|
---|
| 523 | int im_max=min(nph,nmmax+1);
|
---|
[833] | 524 | int i;
|
---|
| 525 | for (i=0;i< dataout.NElts();i++) dataout(i)=complex<T>((T)0.,(T)0.);
|
---|
| 526 | for (i=0;i<im_max;i++) dataout(i)=transformedData(i);
|
---|
[729] | 527 |
|
---|
| 528 |
|
---|
| 529 | // for (int i = 0;i <im_max;i++){
|
---|
| 530 | // dataout(i)*= (complex<T>)(complex<double>(cos(-i*phi0),sin(-i*phi0)));
|
---|
| 531 | // }
|
---|
| 532 | for (int kk=nph; kk<dataout.NElts(); kk++) dataout(kk)=dataout(kk%nph);
|
---|
[833] | 533 | for (i = 0;i <dataout.NElts();i++){
|
---|
[729] | 534 | dataout(i)*= (complex<T>)(complex<double>(cos(-i*phi0),sin(-i*phi0)));
|
---|
| 535 | }
|
---|
| 536 | return dataout;
|
---|
| 537 | }
|
---|
| 538 |
|
---|
| 539 | //&&&&&&&&& nouvelle version
|
---|
[1218] | 540 | /* \fn TVector< complex<T> > SOPHYA::SphericalTransformServer::CFromFourierAnalysis(int_4 nmmax, const TVector<T> datain, r_8 phi0) const
|
---|
| 541 |
|
---|
| 542 | same as previous one, but with a "datain" which is real (not complex) */
|
---|
[729] | 543 | template<class T>
|
---|
[746] | 544 | TVector< complex<T> > SphericalTransformServer<T>::CFromFourierAnalysis(int_4 nmmax, const TVector<T> datain, r_8 phi0) const
|
---|
[729] | 545 | {
|
---|
| 546 | //=======================================================================
|
---|
| 547 | // integrates (data * phi-dependence-of-Ylm) over phi
|
---|
| 548 | // --> function of m can be computed by FFT
|
---|
| 549 | // ! with 0<= m <= npoints/2 (: Nyquist)
|
---|
| 550 | // ! because the data is real the negative m are the conjugate of the
|
---|
| 551 | // ! positive ones
|
---|
| 552 |
|
---|
| 553 | // datain est modifie
|
---|
| 554 | //
|
---|
| 555 | // =======================================================================
|
---|
| 556 | int_4 nph=datain.NElts();
|
---|
| 557 | if (nph <= 0)
|
---|
| 558 | {
|
---|
| 559 | throw PException("bizarre : vecteur datain de longueur nulle (CFromFourierAnalysis)");
|
---|
| 560 | }
|
---|
| 561 | TVector<complex<T> > transformedData;
|
---|
| 562 | // a remodifier
|
---|
| 563 | //FFTPackServer ffts;
|
---|
| 564 | //ffts.setNormalize(false);
|
---|
| 565 | //ffts.FFTForward(datain, transformedData);
|
---|
| 566 |
|
---|
| 567 | fftIntfPtr_-> FFTForward(datain, transformedData);
|
---|
| 568 | //
|
---|
| 569 |
|
---|
| 570 | //dataout.ReSize(nmmax+1);
|
---|
| 571 | TVector< complex<T> > dataout(nmmax+1);
|
---|
| 572 |
|
---|
| 573 | // on transfere le resultat de la fft dans dataout.
|
---|
| 574 | // on s'assure que ca ne depasse pas la taille de dataout
|
---|
| 575 | int sizeOfTransformToGet = min(transformedData.NElts(),nmmax+1);
|
---|
| 576 | // int im_max=min(transformedData.NElts()-1,nmmax);
|
---|
[833] | 577 | int i;
|
---|
| 578 | for (i=0;i<sizeOfTransformToGet;i++) dataout(i)=transformedData(i);
|
---|
[729] | 579 |
|
---|
| 580 |
|
---|
| 581 | // si dataout n'est pas plein, on complete jusqu'a nph valeurs (a moins
|
---|
| 582 | // que dataout ne soit plein avant d'atteindre nph)
|
---|
| 583 | if (sizeOfTransformToGet == (transformedData.NElts()))
|
---|
| 584 | {
|
---|
[833] | 585 | for (i=transformedData.NElts(); i<min(nph,dataout.NElts()); i++)
|
---|
[729] | 586 | {
|
---|
| 587 |
|
---|
| 588 | // dataout(i) = conj(dataout(2*sizeOfTransformToGet-i-2) );
|
---|
| 589 | dataout(i) = conj(dataout(nph-i) );
|
---|
| 590 | }
|
---|
| 591 | // on conplete, si necessaire, par periodicite
|
---|
| 592 | for (int kk=nph; kk<dataout.NElts(); kk++)
|
---|
| 593 | {
|
---|
| 594 | dataout(kk)=dataout(kk%nph);
|
---|
| 595 | }
|
---|
| 596 | }
|
---|
[833] | 597 | for (i = 0;i <dataout.NElts();i++){
|
---|
[729] | 598 | dataout(i)*= (complex<T>)(complex<double>(cos(-i*phi0),sin(-i*phi0)));
|
---|
| 599 | }
|
---|
| 600 | return dataout;
|
---|
| 601 | }
|
---|
| 602 |
|
---|
[1218] | 603 | /*! \fn void SOPHYA::SphericalTransformServer::GenerateFromAlm(SphericalMap<T>& mapq,
|
---|
| 604 | SphericalMap<T>& mapu,
|
---|
| 605 | int_4 pixelSizeIndex,
|
---|
| 606 | const Alm<T>& alme,
|
---|
| 607 | const Alm<T>& almb) const
|
---|
| 608 |
|
---|
| 609 | synthesis of a polarization map from Alm coefficients. The spheres mapq and mapu contain respectively the Stokes parameters. */
|
---|
[729] | 610 | template<class T>
|
---|
| 611 | void SphericalTransformServer<T>::GenerateFromAlm(SphericalMap<T>& mapq,
|
---|
| 612 | SphericalMap<T>& mapu,
|
---|
| 613 | int_4 pixelSizeIndex,
|
---|
| 614 | const Alm<T>& alme,
|
---|
| 615 | const Alm<T>& almb) const
|
---|
| 616 | {
|
---|
| 617 | /*=======================================================================
|
---|
| 618 | computes a map form its alm for the HEALPIX pixelisation
|
---|
| 619 | map(theta,phi) = sum_l_m a_lm Y_lm(theta,phi)
|
---|
| 620 | = sum_m {e^(i*m*phi) sum_l a_lm*lambda_lm(theta)}
|
---|
| 621 |
|
---|
| 622 | where Y_lm(theta,phi) = lambda(theta) * e^(i*m*phi)
|
---|
| 623 |
|
---|
| 624 | * the recurrence of Ylm is the standard one (cf Num Rec)
|
---|
| 625 | * the sum over m is done by FFT
|
---|
| 626 |
|
---|
| 627 | =======================================================================*/
|
---|
| 628 | int_4 nlmax=alme.Lmax();
|
---|
| 629 | if (nlmax != almb.Lmax())
|
---|
| 630 | {
|
---|
| 631 | cout << " SphericalTransformServer: les deux tableaux alm n'ont pas la meme taille" << endl;
|
---|
| 632 | throw SzMismatchError("SphericalTransformServer: les deux tableaux alm n'ont pas la meme taille");
|
---|
| 633 | }
|
---|
| 634 | int_4 nmmax=nlmax;
|
---|
| 635 | int_4 nsmax=0;
|
---|
| 636 | mapq.Resize(pixelSizeIndex);
|
---|
| 637 | mapu.Resize(pixelSizeIndex);
|
---|
| 638 | char* sphere_type=mapq.TypeOfMap();
|
---|
| 639 | if (strncmp(sphere_type,mapu.TypeOfMap(),4) != 0)
|
---|
| 640 | {
|
---|
| 641 | cout << " SphericalTransformServer: les deux spheres ne sont pas de meme type" << endl;
|
---|
| 642 | cout << " type 1 " << sphere_type << endl;
|
---|
| 643 | cout << " type 2 " << mapu.TypeOfMap() << endl;
|
---|
| 644 | throw SzMismatchError("SphericalTransformServer: les deux spheres ne sont pas de meme type");
|
---|
| 645 |
|
---|
| 646 | }
|
---|
| 647 | if (strncmp(sphere_type,"RING",4) == 0)
|
---|
| 648 | {
|
---|
| 649 | nsmax=mapq.SizeIndex();
|
---|
| 650 | }
|
---|
| 651 | else
|
---|
| 652 | // pour une sphere Gorski le nombre de pixels est 12*nsmax**2
|
---|
| 653 | // on calcule une quantite equivalente a nsmax pour la sphere-theta-phi
|
---|
| 654 | // en vue de l'application du critere Healpix : nlmax<=3*nsmax-1
|
---|
| 655 | // c'est approximatif ; a raffiner.
|
---|
| 656 | if (strncmp(sphere_type,"TETAFI",6) == 0)
|
---|
| 657 | {
|
---|
| 658 | nsmax=(int_4)sqrt(mapq.NbPixels()/12.);
|
---|
| 659 | }
|
---|
| 660 | else
|
---|
| 661 | {
|
---|
| 662 | cout << " unknown type of sphere : " << sphere_type << endl;
|
---|
| 663 | throw IOExc(" unknown type of sphere ");
|
---|
| 664 | }
|
---|
| 665 | cout << "GenerateFromAlm: the spheres are of type : " << sphere_type << endl;
|
---|
| 666 | cout << "GenerateFromAlm: size indices (nside) of spheres= " << nsmax << endl;
|
---|
| 667 | cout << "GenerateFromAlm: nlmax (from Alm) = " << nlmax << endl;
|
---|
| 668 | if (nlmax>3*nsmax-1)
|
---|
| 669 | {
|
---|
| 670 | cout << "GenerateFromAlm: nlmax should be <= 3*nside-1" << endl;
|
---|
| 671 | if (strncmp(sphere_type,"TETAFI",6) == 0)
|
---|
| 672 | {
|
---|
| 673 | cout << " (for this criterium, nsmax is computed as sqrt(nbPixels/12))" << endl;
|
---|
| 674 | }
|
---|
| 675 | }
|
---|
| 676 | if (alme.Lmax()!=almb.Lmax())
|
---|
| 677 | {
|
---|
| 678 | cout << "GenerateFromAlm: arrays Alme and Almb have not the same size ? " << endl;
|
---|
| 679 | throw SzMismatchError("SphericalTransformServer: arrays Alme and Almb have not the same size ? ");
|
---|
| 680 | }
|
---|
| 681 | mapFromWX(nlmax, nmmax, mapq, mapu, alme, almb);
|
---|
| 682 | // mapFromPM(nlmax, nmmax, mapq, mapu, alme, almb);
|
---|
| 683 | }
|
---|
| 684 |
|
---|
| 685 |
|
---|
[1218] | 686 | /*! \fn void SOPHYA::SphericalTransformServer::DecomposeToAlm(const SphericalMap<T>& mapq,
|
---|
| 687 | const SphericalMap<T>& mapu,
|
---|
| 688 | Alm<T>& alme,
|
---|
| 689 | Alm<T>& almb,
|
---|
| 690 | int_4 nlmax,
|
---|
| 691 | r_8 cos_theta_cut) const
|
---|
| 692 |
|
---|
| 693 | analysis of a polarization map into Alm coefficients.
|
---|
| 694 |
|
---|
| 695 | The spheres \c mapq and \c mapu contain respectively the Stokes parameters.
|
---|
| 696 |
|
---|
| 697 | \c a2lme and \c a2lmb will receive respectively electric and magnetic Alm's
|
---|
| 698 | nlmax : maximum value of the l index
|
---|
| 699 |
|
---|
| 700 | \c cos_theta_cut : cosinus of the symmetric cut EULER angle theta : cos_theta_cut=0 means no cut ; cos_theta_cut=1 all the sphere is cut.
|
---|
| 701 | */
|
---|
[729] | 702 | template<class T>
|
---|
| 703 | void SphericalTransformServer<T>::DecomposeToAlm(const SphericalMap<T>& mapq,
|
---|
| 704 | const SphericalMap<T>& mapu,
|
---|
| 705 | Alm<T>& alme,
|
---|
| 706 | Alm<T>& almb,
|
---|
| 707 | int_4 nlmax,
|
---|
| 708 | r_8 cos_theta_cut) const
|
---|
| 709 | {
|
---|
| 710 | int_4 nmmax = nlmax;
|
---|
| 711 | // resize et remise a zero
|
---|
| 712 | alme.ReSizeToLmax(nlmax);
|
---|
| 713 | almb.ReSizeToLmax(nlmax);
|
---|
| 714 |
|
---|
| 715 |
|
---|
| 716 | TVector<T> dataq;
|
---|
| 717 | TVector<T> datau;
|
---|
| 718 | TVector<int_4> pixNumber;
|
---|
| 719 |
|
---|
| 720 | char* sphere_type=mapq.TypeOfMap();
|
---|
| 721 | if (strncmp(sphere_type,mapu.TypeOfMap(),4) != 0)
|
---|
| 722 | {
|
---|
| 723 | cout << " SphericalTransformServer: les deux spheres ne sont pas de meme type" << endl;
|
---|
| 724 | cout << " type 1 " << sphere_type << endl;
|
---|
| 725 | cout << " type 2 " << mapu.TypeOfMap() << endl;
|
---|
| 726 | throw SzMismatchError("SphericalTransformServer: les deux spheres ne sont pas de meme type");
|
---|
| 727 |
|
---|
| 728 | }
|
---|
| 729 | if (mapq.NbPixels()!=mapu.NbPixels())
|
---|
| 730 | {
|
---|
| 731 | cout << " DecomposeToAlm: map Q and map U have not same size ?" << endl;
|
---|
| 732 | throw SzMismatchError("SphericalTransformServer::DecomposeToAlm: map Q and map U have not same size ");
|
---|
| 733 | }
|
---|
[746] | 734 | for (int_4 ith = 0; ith < mapq.NbThetaSlices(); ith++)
|
---|
[729] | 735 | {
|
---|
| 736 | r_8 phi0;
|
---|
| 737 | r_8 theta;
|
---|
| 738 | mapq.GetThetaSlice(ith,theta,phi0, pixNumber,dataq);
|
---|
| 739 | mapu.GetThetaSlice(ith,theta,phi0, pixNumber,datau);
|
---|
| 740 | if (dataq.NElts() != datau.NElts() )
|
---|
| 741 | {
|
---|
| 742 | throw SzMismatchError("the spheres have not the same pixelization");
|
---|
| 743 | }
|
---|
| 744 | r_8 domega=mapq.PixSolAngle(mapq.PixIndexSph(theta,phi0));
|
---|
| 745 | double cth = cos(theta);
|
---|
| 746 | //part of the sky out of the symetric cut
|
---|
[1428] | 747 | bool keep_it = (fabs(cth) >= cos_theta_cut);
|
---|
[729] | 748 | if (keep_it)
|
---|
| 749 | {
|
---|
[1328] | 750 | // almFromPM(pixNumber.NElts(), nlmax, nmmax, phi0, domega, theta, dataq, datau, alme, almb);
|
---|
[746] | 751 | almFromWX(nlmax, nmmax, phi0, domega, theta, dataq, datau, alme, almb);
|
---|
[729] | 752 | }
|
---|
| 753 | }
|
---|
| 754 | }
|
---|
| 755 |
|
---|
| 756 |
|
---|
[1218] | 757 | /*! \fn void SOPHYA::SphericalTransformServer::almFromWX(int_4 nlmax, int_4 nmmax,
|
---|
| 758 | r_8 phi0, r_8 domega,
|
---|
| 759 | r_8 theta,
|
---|
| 760 | const TVector<T>& dataq,
|
---|
| 761 | const TVector<T>& datau,
|
---|
| 762 | Alm<T>& alme,
|
---|
| 763 | Alm<T>& almb) const
|
---|
| 764 |
|
---|
| 765 | Compute polarized Alm's as :
|
---|
| 766 | \f[
|
---|
| 767 | a_{lm}^E=\frac{1}{\sqrt{2}}\sum_{slices}{\omega_{pix}\left(\,_{w}\lambda_l^m\tilde{Q}-i\,_{x}\lambda_l^m\tilde{U}\right)}
|
---|
| 768 | \f]
|
---|
| 769 | \f[
|
---|
| 770 | a_{lm}^B=\frac{1}{\sqrt{2}}\sum_{slices}{\omega_{pix}\left(i\,_{x}\lambda_l^m\tilde{Q}+\,_{w}\lambda_l^m\tilde{U}\right)}
|
---|
| 771 | \f]
|
---|
| 772 |
|
---|
| 773 | where \f$\tilde{Q}\f$ and \f$\tilde{U}\f$ are C-coefficients computed by FFT (method CFromFourierAnalysis, called by present method) from the Stokes parameters.
|
---|
| 774 |
|
---|
| 775 | \f$\omega_{pix}\f$ are solid angle of each pixel.
|
---|
| 776 |
|
---|
| 777 | dataq, datau : Stokes parameters.
|
---|
| 778 |
|
---|
| 779 | */
|
---|
[729] | 780 | template<class T>
|
---|
[746] | 781 | void SphericalTransformServer<T>::almFromWX(int_4 nlmax, int_4 nmmax,
|
---|
[729] | 782 | r_8 phi0, r_8 domega,
|
---|
| 783 | r_8 theta,
|
---|
| 784 | const TVector<T>& dataq,
|
---|
| 785 | const TVector<T>& datau,
|
---|
| 786 | Alm<T>& alme,
|
---|
| 787 | Alm<T>& almb) const
|
---|
| 788 | {
|
---|
| 789 | TVector< complex<T> > phaseq(nmmax+1);
|
---|
| 790 | TVector< complex<T> > phaseu(nmmax+1);
|
---|
| 791 | // TVector<complex<T> > datain(nph);
|
---|
| 792 | for (int i=0;i< nmmax+1;i++)
|
---|
| 793 | {
|
---|
| 794 | phaseq(i)=0;
|
---|
| 795 | phaseu(i)=0;
|
---|
| 796 | }
|
---|
| 797 | // for(int kk=0; kk<nph; kk++) datain(kk)=complex<T>(dataq(kk),0.);
|
---|
| 798 |
|
---|
[746] | 799 | // phaseq = CFromFourierAnalysis(nmmax,datain,phi0);
|
---|
| 800 | phaseq = CFromFourierAnalysis(nmmax,dataq,phi0);
|
---|
[729] | 801 |
|
---|
| 802 | // for(int kk=0; kk<nph; kk++) datain(kk)=complex<T>(datau(kk),0.);
|
---|
| 803 |
|
---|
| 804 | // phaseu= CFromFourierAnalysis(nlmax,nmmax,datain,phi0);
|
---|
[746] | 805 | phaseu= CFromFourierAnalysis(nmmax,datau,phi0);
|
---|
[729] | 806 |
|
---|
| 807 | LambdaWXBuilder lwxb(theta,nlmax,nmmax);
|
---|
| 808 |
|
---|
| 809 | r_8 sqr2inv=1/Rac2;
|
---|
| 810 | for (int m = 0; m <= nmmax; m++)
|
---|
| 811 | {
|
---|
| 812 | r_8 lambda_w=0.;
|
---|
| 813 | r_8 lambda_x=0.;
|
---|
| 814 | lwxb.lam_wx(m, m, lambda_w, lambda_x);
|
---|
| 815 | complex<T> zi_lam_x((T)0., (T)lambda_x);
|
---|
| 816 | alme(m,m) += ( (T)(lambda_w)*phaseq(m)-zi_lam_x*phaseu(m) )*(T)(domega*sqr2inv);
|
---|
| 817 | almb(m,m) += ( (T)(lambda_w)*phaseu(m)+zi_lam_x*phaseq(m) )*(T)(domega*sqr2inv);
|
---|
| 818 |
|
---|
| 819 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 820 | {
|
---|
| 821 | lwxb.lam_wx(l, m, lambda_w, lambda_x);
|
---|
| 822 | zi_lam_x = complex<T>((T)0., (T)lambda_x);
|
---|
| 823 | alme(l,m) += ( (T)(lambda_w)*phaseq(m)-zi_lam_x*phaseu(m) )*(T)(domega*sqr2inv);
|
---|
| 824 | almb(l,m) += ( (T)(lambda_w)*phaseu(m)+zi_lam_x*phaseq(m) )*(T)(domega*sqr2inv);
|
---|
| 825 | }
|
---|
| 826 | }
|
---|
| 827 | }
|
---|
| 828 |
|
---|
| 829 |
|
---|
[1218] | 830 | /*! \fn void SOPHYA::SphericalTransformServer::almFromPM(int_4 nph, int_4 nlmax,
|
---|
| 831 | int_4 nmmax,
|
---|
| 832 | r_8 phi0, r_8 domega,
|
---|
| 833 | r_8 theta,
|
---|
| 834 | const TVector<T>& dataq,
|
---|
| 835 | const TVector<T>& datau,
|
---|
| 836 | Alm<T>& alme,
|
---|
| 837 | Alm<T>& almb) const
|
---|
| 838 |
|
---|
| 839 | Compute polarized Alm's as :
|
---|
| 840 | \f[
|
---|
| 841 | a_{lm}^E=-\frac{1}{2}\sum_{slices}{\omega_{pix}\left(\,_{+}\lambda_l^m\tilde{P^+}+\,_{-}\lambda_l^m\tilde{P^-}\right)}
|
---|
| 842 | \f]
|
---|
| 843 | \f[
|
---|
| 844 | a_{lm}^B=\frac{i}{2}\sum_{slices}{\omega_{pix}\left(\,_{+}\lambda_l^m\tilde{P^+}-\,_{-}\lambda_l^m\tilde{P^-}\right)}
|
---|
| 845 | \f]
|
---|
| 846 |
|
---|
| 847 | where \f$\tilde{P^{\pm}}=\tilde{Q}\pm\tilde{U}\f$ computed by FFT (method CFromFourierAnalysis, called by present method) from the Stokes parameters,\f$Q\f$ and \f$U\f$ .
|
---|
| 848 |
|
---|
| 849 | \f$\omega_{pix}\f$ are solid angle of each pixel.
|
---|
| 850 |
|
---|
| 851 | dataq, datau : Stokes parameters.
|
---|
| 852 |
|
---|
| 853 | */
|
---|
[729] | 854 | template<class T>
|
---|
[1218] | 855 | void SphericalTransformServer<T>::almFromPM(int_4 nph, int_4 nlmax,
|
---|
| 856 | int_4 nmmax,
|
---|
[729] | 857 | r_8 phi0, r_8 domega,
|
---|
| 858 | r_8 theta,
|
---|
| 859 | const TVector<T>& dataq,
|
---|
| 860 | const TVector<T>& datau,
|
---|
| 861 | Alm<T>& alme,
|
---|
| 862 | Alm<T>& almb) const
|
---|
| 863 | {
|
---|
| 864 | TVector< complex<T> > phasep(nmmax+1);
|
---|
| 865 | TVector< complex<T> > phasem(nmmax+1);
|
---|
| 866 | TVector<complex<T> > datain(nph);
|
---|
| 867 | for (int i=0;i< nmmax+1;i++)
|
---|
| 868 | {
|
---|
| 869 | phasep(i)=0;
|
---|
| 870 | phasem(i)=0;
|
---|
| 871 | }
|
---|
[833] | 872 | int kk;
|
---|
| 873 | for(kk=0; kk<nph; kk++) datain(kk)=complex<T>(dataq(kk),datau(kk));
|
---|
[729] | 874 |
|
---|
[746] | 875 | phasep = CFromFourierAnalysis(nmmax,datain,phi0);
|
---|
[729] | 876 |
|
---|
[833] | 877 | for(kk=0; kk<nph; kk++) datain(kk)=complex<T>(dataq(kk),-datau(kk));
|
---|
[746] | 878 | phasem = CFromFourierAnalysis(nmmax,datain,phi0);
|
---|
[729] | 879 | LambdaPMBuilder lpmb(theta,nlmax,nmmax);
|
---|
| 880 |
|
---|
| 881 | for (int m = 0; m <= nmmax; m++)
|
---|
| 882 | {
|
---|
| 883 | r_8 lambda_p=0.;
|
---|
| 884 | r_8 lambda_m=0.;
|
---|
| 885 | complex<T> im((T)0.,(T)1.);
|
---|
| 886 | lpmb.lam_pm(m, m, lambda_p, lambda_m);
|
---|
| 887 |
|
---|
| 888 | alme(m,m) += -( (T)(lambda_p)*phasep(m) + (T)(lambda_m)*phasem(m) )*(T)(domega*0.5);
|
---|
| 889 | almb(m,m) += im*( (T)(lambda_p)*phasep(m) - (T)(lambda_m)*phasem(m) )*(T)(domega*0.5);
|
---|
| 890 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 891 | {
|
---|
| 892 | lpmb.lam_pm(l, m, lambda_p, lambda_m);
|
---|
| 893 | alme(l,m) += -( (T)(lambda_p)*phasep(m) + (T)(lambda_m)*phasem(m) )*(T)(domega*0.5);
|
---|
| 894 | almb(l,m) += im* ( (T)(lambda_p)*phasep(m) - (T)(lambda_m)*phasem(m) )*(T)(domega*0.5);
|
---|
| 895 | }
|
---|
| 896 | }
|
---|
| 897 | }
|
---|
| 898 |
|
---|
| 899 |
|
---|
[1218] | 900 | /*! \fn void SOPHYA::SphericalTransformServer::mapFromWX(int_4 nlmax, int_4 nmmax,
|
---|
| 901 | SphericalMap<T>& mapq,
|
---|
| 902 | SphericalMap<T>& mapu,
|
---|
| 903 | const Alm<T>& alme,
|
---|
| 904 | const Alm<T>& almb) const
|
---|
| 905 |
|
---|
| 906 | synthesis of Stokes parameters following formulae :
|
---|
| 907 |
|
---|
| 908 | \f[
|
---|
| 909 | Q=\sum_{m=-mmax}^{mmax}b_m^qe^{im\varphi}
|
---|
| 910 | \f]
|
---|
| 911 | \f[
|
---|
| 912 | U=\sum_{m=-mmax}^{mmax}b_m^ue^{im\varphi}
|
---|
| 913 | \f]
|
---|
| 914 |
|
---|
| 915 | computed by FFT (method fourierSynthesisFromB called by the present one)
|
---|
| 916 |
|
---|
| 917 | with :
|
---|
| 918 |
|
---|
| 919 | \f[
|
---|
| 920 | b_m^q=-\frac{1}{\sqrt{2}}\sum_{l=|m|}^{lmax}{\left(\,_{w}\lambda_l^ma_{lm}^E-i\,_{x}\lambda_l^ma_{lm}^B\right) }
|
---|
| 921 | \f]
|
---|
| 922 | \f[
|
---|
| 923 | b_m^u=\frac{1}{\sqrt{2}}\sum_{l=|m|}^{lmax}{\left(i\,_{x}\lambda_l^ma_{lm}^E+\,_{w}\lambda_l^ma_{lm}^B\right) }
|
---|
| 924 | \f]
|
---|
| 925 | */
|
---|
[729] | 926 | template<class T>
|
---|
| 927 | void SphericalTransformServer<T>::mapFromWX(int_4 nlmax, int_4 nmmax,
|
---|
| 928 | SphericalMap<T>& mapq,
|
---|
| 929 | SphericalMap<T>& mapu,
|
---|
| 930 | const Alm<T>& alme,
|
---|
| 931 | const Alm<T>& almb) const
|
---|
| 932 | {
|
---|
| 933 | Bm<complex<T> > b_m_theta_q(nmmax);
|
---|
| 934 | Bm<complex<T> > b_m_theta_u(nmmax);
|
---|
| 935 |
|
---|
[746] | 936 | for (int_4 ith = 0; ith < mapq.NbThetaSlices();ith++)
|
---|
[729] | 937 | {
|
---|
| 938 | int_4 nph;
|
---|
| 939 | r_8 phi0;
|
---|
| 940 | r_8 theta;
|
---|
| 941 | TVector<int_4> pixNumber;
|
---|
| 942 | TVector<T> datan;
|
---|
| 943 |
|
---|
| 944 | mapq.GetThetaSlice(ith,theta,phi0, pixNumber,datan);
|
---|
| 945 | nph = pixNumber.NElts();
|
---|
| 946 | // -----------------------------------------------------
|
---|
| 947 | // for each theta, and each m, computes
|
---|
| 948 | // b(m,theta) = sum_over_l>m (lambda_l_m(theta) * a_l_m)
|
---|
| 949 | // ------------------------------------------------------
|
---|
| 950 | LambdaWXBuilder lwxb(theta,nlmax,nmmax);
|
---|
| 951 | // LambdaPMBuilder lpmb(theta,nlmax,nmmax);
|
---|
| 952 | r_8 sqr2inv=1/Rac2;
|
---|
[833] | 953 | int m;
|
---|
| 954 | for (m = 0; m <= nmmax; m++)
|
---|
[729] | 955 | {
|
---|
| 956 | r_8 lambda_w=0.;
|
---|
| 957 | r_8 lambda_x=0.;
|
---|
| 958 | lwxb.lam_wx(m, m, lambda_w, lambda_x);
|
---|
| 959 | complex<T> zi_lam_x((T)0., (T)lambda_x);
|
---|
| 960 |
|
---|
| 961 | b_m_theta_q(m) = ( (T)(lambda_w) * alme(m,m) - zi_lam_x * almb(m,m))*(T)sqr2inv ;
|
---|
| 962 | b_m_theta_u(m) = ( (T)(lambda_w) * almb(m,m) + zi_lam_x * alme(m,m))*(T)sqr2inv;
|
---|
| 963 |
|
---|
| 964 |
|
---|
| 965 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 966 | {
|
---|
| 967 |
|
---|
| 968 | lwxb.lam_wx(l, m, lambda_w, lambda_x);
|
---|
| 969 | zi_lam_x= complex<T>((T)0., (T)lambda_x);
|
---|
| 970 |
|
---|
| 971 | b_m_theta_q(m) += ((T)(lambda_w)*alme(l,m)-zi_lam_x *almb(l,m))*(T)sqr2inv;
|
---|
| 972 | b_m_theta_u(m) += ((T)(lambda_w)*almb(l,m)+zi_lam_x *alme(l,m))*(T)sqr2inv;
|
---|
| 973 |
|
---|
| 974 | }
|
---|
| 975 | }
|
---|
| 976 | // obtains the negative m of b(m,theta) (= complex conjugate)
|
---|
[833] | 977 | for (m=1;m<=nmmax;m++)
|
---|
[729] | 978 | {
|
---|
| 979 | b_m_theta_q(-m) = conj(b_m_theta_q(m));
|
---|
| 980 | b_m_theta_u(-m) = conj(b_m_theta_u(m));
|
---|
| 981 | }
|
---|
| 982 |
|
---|
| 983 | // TVector<complex<T> > Tempq = fourierSynthesisFromB(b_m_theta_q,nph,phi0);
|
---|
| 984 | // TVector<complex<T> > Tempu = fourierSynthesisFromB(b_m_theta_u,nph,phi0);
|
---|
| 985 | TVector<T> Tempq = RfourierSynthesisFromB(b_m_theta_q,nph,phi0);
|
---|
| 986 | TVector<T> Tempu = RfourierSynthesisFromB(b_m_theta_u,nph,phi0);
|
---|
| 987 | for (int i=0;i< nph;i++)
|
---|
| 988 | {
|
---|
| 989 | // mapq(pixNumber(i))=Tempq(i).real();
|
---|
| 990 | // mapu(pixNumber(i))=Tempu(i).real();
|
---|
| 991 | mapq(pixNumber(i))=Tempq(i);
|
---|
| 992 | mapu(pixNumber(i))=Tempu(i);
|
---|
| 993 |
|
---|
| 994 | }
|
---|
| 995 | }
|
---|
| 996 | }
|
---|
[1218] | 997 | /*! \fn void SOPHYA::SphericalTransformServer::mapFromPM(int_4 nlmax, int_4 nmmax,
|
---|
| 998 | SphericalMap<T>& mapq,
|
---|
| 999 | SphericalMap<T>& mapu,
|
---|
| 1000 | const Alm<T>& alme,
|
---|
| 1001 | const Alm<T>& almb) const
|
---|
| 1002 |
|
---|
| 1003 | synthesis of polarizations following formulae :
|
---|
| 1004 |
|
---|
| 1005 | \f[
|
---|
| 1006 | P^+ = \sum_{m=-mmax}^{mmax} {b_m^+e^{im\varphi} }
|
---|
| 1007 | \f]
|
---|
| 1008 | \f[
|
---|
| 1009 | P^- = \sum_{m=-mmax}^{mmax} {b_m^-e^{im\varphi} }
|
---|
| 1010 | \f]
|
---|
| 1011 |
|
---|
| 1012 | computed by FFT (method fourierSynthesisFromB called by the present one)
|
---|
| 1013 |
|
---|
| 1014 | with :
|
---|
| 1015 |
|
---|
| 1016 | \f[
|
---|
| 1017 | b_m^+=-\sum_{l=|m|}^{lmax}{\,_{+}\lambda_l^m \left( a_{lm}^E+ia_{lm}^B \right) }
|
---|
| 1018 | \f]
|
---|
| 1019 | \f[
|
---|
| 1020 | b_m^-=-\sum_{l=|m|}^{lmax}{\,_{+}\lambda_l^m \left( a_{lm}^E-ia_{lm}^B \right) }
|
---|
| 1021 | \f]
|
---|
| 1022 | */
|
---|
[729] | 1023 | template<class T>
|
---|
| 1024 | void SphericalTransformServer<T>::mapFromPM(int_4 nlmax, int_4 nmmax,
|
---|
| 1025 | SphericalMap<T>& mapq,
|
---|
| 1026 | SphericalMap<T>& mapu,
|
---|
| 1027 | const Alm<T>& alme,
|
---|
| 1028 | const Alm<T>& almb) const
|
---|
| 1029 | {
|
---|
| 1030 | Bm<complex<T> > b_m_theta_p(nmmax);
|
---|
| 1031 | Bm<complex<T> > b_m_theta_m(nmmax);
|
---|
[746] | 1032 | for (int_4 ith = 0; ith < mapq.NbThetaSlices();ith++)
|
---|
[729] | 1033 | {
|
---|
| 1034 | int_4 nph;
|
---|
| 1035 | r_8 phi0;
|
---|
| 1036 | r_8 theta;
|
---|
| 1037 | TVector<int_4> pixNumber;
|
---|
| 1038 | TVector<T> datan;
|
---|
| 1039 |
|
---|
| 1040 | mapq.GetThetaSlice(ith,theta,phi0, pixNumber,datan);
|
---|
| 1041 | nph = pixNumber.NElts();
|
---|
| 1042 |
|
---|
| 1043 | // -----------------------------------------------------
|
---|
| 1044 | // for each theta, and each m, computes
|
---|
| 1045 | // b(m,theta) = sum_over_l>m (lambda_l_m(theta) * a_l_m)
|
---|
| 1046 | //------------------------------------------------------
|
---|
| 1047 |
|
---|
| 1048 | LambdaPMBuilder lpmb(theta,nlmax,nmmax);
|
---|
[833] | 1049 | int m;
|
---|
| 1050 | for (m = 0; m <= nmmax; m++)
|
---|
[729] | 1051 | {
|
---|
| 1052 | r_8 lambda_p=0.;
|
---|
| 1053 | r_8 lambda_m=0.;
|
---|
| 1054 | lpmb.lam_pm(m, m, lambda_p, lambda_m);
|
---|
| 1055 | complex<T> im((T)0.,(T)1.);
|
---|
| 1056 |
|
---|
| 1057 | b_m_theta_p(m) = (T)(lambda_p )* (-alme(m,m) - im * almb(m,m));
|
---|
| 1058 | b_m_theta_m(m) = (T)(lambda_m) * (-alme(m,m) + im * almb(m,m));
|
---|
| 1059 |
|
---|
| 1060 |
|
---|
| 1061 | for (int l = m+1; l<= nlmax; l++)
|
---|
| 1062 | {
|
---|
| 1063 | lpmb.lam_pm(l, m, lambda_p, lambda_m);
|
---|
| 1064 | b_m_theta_p(m) += (T)(lambda_p)*(-alme(l,m)-im *almb(l,m));
|
---|
| 1065 | b_m_theta_m(m) += (T)(lambda_m)*(-alme(l,m)+im *almb(l,m));
|
---|
| 1066 | }
|
---|
| 1067 | }
|
---|
| 1068 |
|
---|
| 1069 | // obtains the negative m of b(m,theta) (= complex conjugate)
|
---|
[833] | 1070 | for (m=1;m<=nmmax;m++)
|
---|
[729] | 1071 | {
|
---|
| 1072 | b_m_theta_p(-m) = conj(b_m_theta_m(m));
|
---|
| 1073 | b_m_theta_m(-m) = conj(b_m_theta_p(m));
|
---|
| 1074 | }
|
---|
| 1075 |
|
---|
| 1076 | TVector<complex<T> > Tempp = fourierSynthesisFromB(b_m_theta_p,nph,phi0);
|
---|
| 1077 | TVector<complex<T> > Tempm = fourierSynthesisFromB(b_m_theta_m,nph,phi0);
|
---|
| 1078 |
|
---|
| 1079 | for (int i=0;i< nph;i++)
|
---|
| 1080 | {
|
---|
| 1081 | mapq(pixNumber(i))=0.5*(Tempp(i)+Tempm(i)).real();
|
---|
| 1082 | mapu(pixNumber(i))=0.5*(Tempp(i)-Tempm(i)).imag();
|
---|
| 1083 | }
|
---|
| 1084 | }
|
---|
| 1085 | }
|
---|
| 1086 |
|
---|
| 1087 |
|
---|
[1218] | 1088 | /*! \fn void SOPHYA::SphericalTransformServer::GenerateFromCl(SphericalMap<T>& sphq,
|
---|
| 1089 | SphericalMap<T>& sphu,
|
---|
| 1090 | int_4 pixelSizeIndex,
|
---|
| 1091 | const TVector<T>& Cle,
|
---|
| 1092 | const TVector<T>& Clb,
|
---|
| 1093 | const r_8 fwhm) const
|
---|
| 1094 |
|
---|
| 1095 | synthesis of a polarization map from power spectra electric-Cl and magnetic-Cl (Alm's are generated randomly, following a gaussian distribution).
|
---|
| 1096 | \param fwhm FWHM in arcmin for random generation of Alm's (eg. 5)
|
---|
| 1097 | */
|
---|
[729] | 1098 | template<class T>
|
---|
| 1099 | void SphericalTransformServer<T>::GenerateFromCl(SphericalMap<T>& sphq,
|
---|
| 1100 | SphericalMap<T>& sphu,
|
---|
| 1101 | int_4 pixelSizeIndex,
|
---|
| 1102 | const TVector<T>& Cle,
|
---|
| 1103 | const TVector<T>& Clb,
|
---|
| 1104 | const r_8 fwhm) const
|
---|
| 1105 | {
|
---|
| 1106 | if (Cle.NElts() != Clb.NElts())
|
---|
| 1107 | {
|
---|
| 1108 | cout << " SphericalTransformServer: les deux tableaux Cl n'ont pas la meme taille" << endl;
|
---|
| 1109 | throw SzMismatchError("SphericalTransformServer::GenerateFromCl : two Cl arrays have not same size");
|
---|
| 1110 | }
|
---|
| 1111 |
|
---|
| 1112 | // Alm<T> a2lme,a2lmb;
|
---|
| 1113 | // almFromCl(a2lme, Cle, fwhm);
|
---|
| 1114 | // almFromCl(a2lmb, Clb, fwhm);
|
---|
| 1115 | // Alm<T> a2lme = almFromCl(Cle, fwhm);
|
---|
| 1116 | // Alm<T> a2lmb = almFromCl(Clb, fwhm);
|
---|
| 1117 | Alm<T> a2lme(Cle, fwhm);
|
---|
| 1118 | Alm<T> a2lmb(Clb, fwhm);
|
---|
| 1119 |
|
---|
| 1120 | GenerateFromAlm(sphq,sphu,pixelSizeIndex,a2lme,a2lmb);
|
---|
| 1121 | }
|
---|
[1218] | 1122 | /*! \fn void SOPHYA::SphericalTransformServer::GenerateFromCl(SphericalMap<T>& sph,
|
---|
| 1123 | int_4 pixelSizeIndex,
|
---|
| 1124 | const TVector<T>& Cl,
|
---|
| 1125 | const r_8 fwhm) const
|
---|
| 1126 |
|
---|
| 1127 | synthesis of a temperature map from power spectrum Cl (Alm's are generated randomly, following a gaussian distribution). */
|
---|
[729] | 1128 | template<class T>
|
---|
| 1129 | void SphericalTransformServer<T>::GenerateFromCl(SphericalMap<T>& sph,
|
---|
| 1130 | int_4 pixelSizeIndex,
|
---|
| 1131 | const TVector<T>& Cl,
|
---|
| 1132 | const r_8 fwhm) const
|
---|
| 1133 | {
|
---|
| 1134 |
|
---|
| 1135 | Alm<T> alm(Cl, fwhm);
|
---|
| 1136 | GenerateFromAlm(sph,pixelSizeIndex, alm );
|
---|
| 1137 | }
|
---|
| 1138 |
|
---|
| 1139 |
|
---|
| 1140 |
|
---|
[1218] | 1141 | /*! \fn TVector<T> SOPHYA::SphericalTransformServer::DecomposeToCl(const SphericalMap<T>& sph, int_4 nlmax, r_8 cos_theta_cut) const
|
---|
| 1142 |
|
---|
| 1143 | \return power spectrum from analysis of a temperature map.
|
---|
| 1144 |
|
---|
| 1145 | \param<nlmax> : maximum value of the l index
|
---|
| 1146 |
|
---|
| 1147 | \param<cos_theta_cut> : cosinus of the symmetric cut EULER angle theta : cos_theta_cut=0 means no cut ; cos_theta_cut=1 all the sphere is cut.
|
---|
| 1148 | */
|
---|
[729] | 1149 | template <class T>
|
---|
| 1150 | TVector<T> SphericalTransformServer<T>::DecomposeToCl(const SphericalMap<T>& sph, int_4 nlmax, r_8 cos_theta_cut) const
|
---|
| 1151 | {
|
---|
| 1152 | Alm<T> alm=DecomposeToAlm( sph, nlmax, cos_theta_cut);
|
---|
| 1153 | // power spectrum
|
---|
| 1154 | return alm.powerSpectrum();
|
---|
| 1155 | }
|
---|
| 1156 |
|
---|
| 1157 | #ifdef __CXX_PRAGMA_TEMPLATES__
|
---|
| 1158 | #pragma define_template SphericalTransformServer<r_8>
|
---|
| 1159 | #pragma define_template SphericalTransformServer<r_4>
|
---|
| 1160 | #endif
|
---|
| 1161 | #if defined(ANSI_TEMPLATES) || defined(GNU_TEMPLATES)
|
---|
| 1162 | template class SphericalTransformServer<r_8>;
|
---|
| 1163 | template class SphericalTransformServer<r_4>;
|
---|
| 1164 | #endif
|
---|