-
Notifications
You must be signed in to change notification settings - Fork 6
/
FaVeST_fwd.m
217 lines (190 loc) · 7.16 KB
/
FaVeST_fwd.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
function [F1,F2] = FaVeST_fwd(T,L,X,w)
% Forward FFTs for vector spherical harmonics
% INPUTS:
% T - scalar field or component of vector field
% L,m - degree and order for vector spherical harmonic
% X,w - quadrature rule for evaluating FFT
% OUTPUTS:
% F1,F2 - Fourier coefficients for divergence-free and curl-free vector spherical harmonics of degree L,m
% scalar Fourier coeff
That1 = fftS2_fwd(T(1,:),X,L+1,w); % of size (2L+3)*(L+2)
That2 = fftS2_fwd(T(2,:),X,L+1,w); % of size (2L+3)*(L+2)
That3 = fftS2_fwd(T(3,:),X,L+1,w); % of size (2L+3)*(L+2)
L1 = L + 2; % number of columns of F1,F2
L0 = L1; % mid line of scalar fft for That1, That2, That3
F1 = zeros(2*L+1,L); % lower triangle stands for m>0
F2 = zeros(2*L+1,L); % lower triangle stands for m>0
Lmid = L+1; % the mid line for FaVeST: F1, F2
for l=1:L
for m=0:l
if m==0
% for m=0
F1(Lmid+m,l) = 1/sqrt(2)*(xi1(l-1,m-1)*(-1*That1(L0+(m-1),(l-1)+1)+1i*That2(L0+(m-1),(l-1)+1)) ...
+ xi2(l+1,m-1)*(-1*That1(L0+(m-1),(l+1)+1)+1i*That2(L0+(m-1),(l+1)+1)) ...
+ xi3(l-1,m+1)*(-1)^(m+1)*(That1(L0+(m+1),(l-1)+1)+1i*That2(L0+(m+1),(l-1)+1)) ...
+ xi4(l+1,m+1)*(-1)^(m+1)*(That1(L0+(m+1),(l+1)+1)+1i*That2(L0+(m+1),(l+1)+1))) ...
+ xi5(l-1,m)*That3(L0+m,(l-1)+1) ...
+ xi6(l+1,m)*That3(L0+m,(l+1)+1);
F2(Lmid+m,l) = -1/sqrt(2)*1i*(mu1(l,m-1)*(-1*That1(L0+(m-1),l+1)+1i*That2(L0+(m-1),l+1)) ...
+ mu3(l,m+1)*(-1)^(m+1)*(That1(L0+(m+1),l+1)+1i*That2(L0+(m+1),l+1))) ...
- 1i*mu2(l,m)*That3(L0+m,l+1);
else
% for m>0
F1(Lmid+m,l) = 1/sqrt(2)*(xi1(l-1,m-1)*(-1)^(m-1)*(-1*That1(L0+(m-1),(l-1)+1)+1i*That2(L0+(m-1),(l-1)+1)) ...
+ xi2(l+1,m-1)*(-1)^(m-1)*(-1*That1(L0+(m-1),(l+1)+1)+1i*That2(L0+(m-1),(l+1)+1)) ...
+ xi3(l-1,m+1)*(-1)^(m+1)*(That1(L0+(m+1),(l-1)+1)+1i*That2(L0+(m+1),(l-1)+1)) ...
+ xi4(l+1,m+1)*(-1)^(m+1)*(That1(L0+(m+1),(l+1)+1)+1i*That2(L0+(m+1),(l+1)+1))) ...
+ xi5(l-1,m)*(-1)^m*That3(L0+m,(l-1)+1) ...
+ xi6(l+1,m)*(-1)^m*That3(L0+m,(l+1)+1);
F2(Lmid+m,l) = -1/sqrt(2)*1i*(mu1(l,m-1)*(-1)^(m-1)*(-1*That1(L0+(m-1),l+1)+1i*That2(L0+(m-1),l+1)) ...
+ mu3(l,m+1)*(-1)^(m+1)*(That1(L0+(m+1),l+1)+1i*That2(L0+(m+1),l+1))) ...
- 1i*mu2(l,m)*(-1)^m*That3(L0+m,l+1);
end
end
for m = -l:-1
% upper triangle of F1/F2 corresponds to m<0
F1(Lmid+m,l) = (-1)^(abs(m))*conj(F1(Lmid+(-m),l));
F2(Lmid+m,l) = (-1)^(abs(m))*conj(F2(Lmid+(-m),l));
end
F1(:,l) = F1(:,l)/sqrt(l*(l+1));
F2(:,l) = F2(:,l)/sqrt(l*(l+1));
end
end
%% subfunctions
function f_hat = fftS2_fwd(f,X,N,w)
%% evaluate scalar spherical harmonic coefficient by nfsft
%
% f_hat = fftS2(f,X,N,w)
%
% INPUT:
% X - X = (x1,...,x_M), points on the 2-sphere, where each
% x_k=(\theta,\phi) with \theta the longitudes in [0,2pi] and
% \phi the colatitudes in [0,pi].
%
% f - vector (f(x_1),...,f(x_M))
%
% N - bandwidth, see f_hat below
%
% w - weight on f(x_j), default = 1
%
% OUTPT:
% f_hat - matrix size is (2N+1) x (N+1), w.r.t.
% f_hat(l,m) = \sum_{k=1}^M f(x_k) bar{Y}_{l,m}(x_k);
% for l = 0,..,N and m=-l,..,l.
%%
if length(f) ~= size(X,2)
error('Input signal f must have the same length as X!');
end
if nargin == 3
w = ones(size(f));
end
M = size(X,2);
nfsft_precompute(N,1000); % precomputation
plan = nfsft_init_advanced(N,M,NFSFT_NORMALIZED); % Create plan.
nfsft_set_x(plan,X); % Set nodes.
nfsft_precompute_x(plan); % node-dependent precomputation
nfsft_set_f(plan,f.*w); % set f
nfsft_adjoint(plan); % adjoint transform
f_hat = nfsft_get_f_hat(plan); % get f_hat
nfsft_finalize(plan); % release memory
end
function [xi_1] = xi1(l,m)
if m>=-l&&m<=l
xi_1 = c_l(l+1)*sqrt((l+1+m)*(l+m+2)/((2*l+2)*(2*l+1))); %C_{l,m,1,1}^{l+1,m+1}
else
xi_1=0;
end
end
function [xi_2] = xi2(l,m)
if m>=-l&&m<=l
xi_2 = d_l(l-1)*sqrt((l-m-1)*(l-m)/((2*l+1)*(2*l))); %C_{l,m,1,1}^{l-1,m+1}
else
xi_2 = 0;
end
end
function [xi_3] = xi3(l,m)
if m>=-l&&m<=l
xi_3 = c_l(l+1)*sqrt((l-m+2)*(l-m+1)/((2*l+2)*(2*l+1))); %C_{l,m,1,1}^{l+1,m-1}
else
xi_3 = 0;
end
end
function [xi_4] = xi4(l,m)
if m>=-l&&m<=l
xi_4 =d_l(l-1)*sqrt((l+m)*(l+m-1)/((2*l+1)*(2*l))); %C_{l,m,1,1}^{l-1,m-1}
else
xi_4 = 0;
end
end
function [xi_5] = xi5(l,m)
if m>=-l&&m<=l
xi_5 = c_l(l+1)*sqrt((l+m+1)*(l-m+1)/((l+1)*(2*l+1))); %C_{l,m,1,0}^{l+1,m}
else
xi_5 = 0;
end
end
function [xi_6] = xi6(l,m)
if m>=-l&&m<=l
xi_6 = -d_l(l-1)*sqrt((l+m)*(l-m)/((2*l+1)*(l))); %C_{l,m,1,0}^{l-1,m}
else
xi_6 = 0;
end
end
function [mu_1] = mu1(l,m)
if m>=-l&&m<=l
lambda_l=l*(l+1);
mu_1 = - sqrt(lambda_l)*sqrt((l+m+1)*(l-m)/(l*(2*l+2))); %C_{l,m,1,1}^{l,m+1}
else
mu_1 = 0;
end
end
function [mu_2] = mu2(l,m)
if m>=-l&&m<=l
lambda_l=l*(l+1);
mu_2=sqrt(lambda_l)*m/sqrt(l*(l+1)); %C_{l,m,1,0}^{l,m}
else
mu_2 = 0;
end
end
function [mu_3] = mu3(l,m)
if m>=-l&&m<=l
lambda_l=l*(l+1);
mu_3=sqrt(lambda_l)*sqrt((l+m)*(l-m+1)/(l*(2*l+2))); %C_{l,m,1,-1}^{l,m-1}
else
mu_3 = 0;
end
end
function y = c_l(l)
y = (l+1)*sqrt(l/(2*l+1));
end
function y = d_l(l)
y = l*sqrt((l+1)/(2*l+1));
end
function [x,w] = lgwt(n,a,b)
n = n-1; n1= n + 1; n2 = n + 2;
% n1 uniform nodes in [-1,1]
xu = linspace(-1,1,n1)';
% initial guess for nodes
y=cos((2*(0:n)'+1)*pi/(2*n+2)) + (0.27/n1)*sin(pi*xu*n/n2);
% Gauss-Legendre Vandermonde matrix
L=zeros(n1,n2);
% derivative of that
Lp=zeros(n1,n2);
% We compute the zeros of the n1th Legendre polynomial using the recursion
% relation and the Newton-Raphson method.
y0=2;
% Iterate until new points are uniformly within epsilon of old points.
while (max(abs(y - y0)) > eps)
L(:,1) = 1; Lp(:,1) = 0; L(:,2) = y; Lp(:,2) = 1;
for k = 2:n1
L(:,k+1) = ((2*k-1)*y.*L(:,k)-(k-1)*L(:,k-1))/k;
end
Lp = n2 * (L(:,n1) - y.*L(:,n2))./(1-y.^2); y0 = y; y = y0 - L(:,n2)./Lp;
end
% linear map from[-1,1] to [a,b]
x = (a*(1-y)+b*(1+y))/2;
if (nargout == 2)
% weights
w = (b-a)./((1-y.^2).*Lp.^2)*(n2/n1)^2;
end
end