Skip to content

GitLab

toolbox
toolbox
Commit cd83129b authored by Marco Berzborn's avatar Marco Berzborn
Browse files
Options

fixed phase conventions for real valued spherical harmonic basis 

functions, added input parser to ita_sph_base
- Phase convention of the real valued SH basis now matches the AMBIX
convention (used in RAVEN)
- Deprecated SH basis functions removed
parent f30f2275
Hide whitespace changes
Inline Side-by-side
Showing with 183 additions and 336 deletions
applications/Binaural-HRTF/HRTF_class/@itaHRTF/interp.m
View file @ cd83129b
......@@ -106,7 +106,7 @@ ear_d = [-0.07 0.07];
W = sparse(diag(w)); % diagonal matrix containing weights
D = sparse(diag(dweights_rep)); % decomposition order-dependent Tikhonov regularization
Y = ita_sph_base(this.dirCoord,Nmax,'orthonormal',false); % calculate real-valued SHs using the measurement grid
Y = ita_sph_base(this.dirCoord,Nmax,'real'); % calculate real-valued SHs using the measurement grid
%% Calculate HRTF data for field points
if Nmax > 25
......@@ -131,7 +131,7 @@ for ear=1:2
% % calculate weighted SH coefficients using a decomposition order-dependent Tikhonov regularization
%
% freqData_temp = data.freqData;
% Y = ita_sph_base(data.channelCoordinates,Nmax,'orthonormal',false);
% Y = ita_sph_base(data.channelCoordinates,Nmax,'real');
a0 = (Y.'*W*Y + epsilon*D) \ Y.'*W * freqData_temp.';
%a0 = (Y.'*W*Y + epsilon*D) \ Y.' *
......@@ -142,10 +142,10 @@ for ear=1:2
% calculate range-extrapolated HRTFs
a1 = a0 .* hankel_rep.';
Yest = ita_sph_base(field,N,'orthonormal',false); % use real-valued SH's
Yest = ita_sph_base(field,N,'real'); % use real-valued SH's
hrtf_arbi(:,ear:2:end) = (Yest*a1).'; % interpolated + range-extrapolated HRTFs
else
Yest = ita_sph_base(field,Nmax,'orthonormal',false); % use real-valued SH's
Yest = ita_sph_base(field,Nmax,'real'); % use real-valued SH's
hrtf_arbi(:,ear:2:end) = (Yest*a0).'; % interpolated HRTFs
end
end
......
applications/Binaural-HRTF/HRTF_class/@itaHRTF/itaHRTF.m
View file @ cd83129b
......@@ -1110,7 +1110,7 @@ classdef itaHRTF < itaAudio
W = diag(vWeights); % diagonal matrix containing weights
D = diag(regweights_rep); % decomposition order-dependent Tikhonov regularization
Y = ita_sph_base(this.dirCoord,Nmeas,'orthonormal',false); % calculate real-valued SHs using the measurement grid (high SH-order)
Y = ita_sph_base(this.dirCoord,Nmeas,'real'); % calculate real-valued SHs using the measurement grid (high SH-order)
%% Calculate spatially smoothed HRTF data set
hrtf_smoo_wo_ITD = zeros(this.nBins,2*this.dirCoord.nPoints); % init.: columns: LRLRLR...
......@@ -1302,11 +1302,11 @@ classdef itaHRTF < itaAudio
earSidePlot = sArgs.earSide;
if numel(phiC_deg)>1,
xData = phiC_deg;
strTitle =[ earSidePlot ' ear, \theta = ' num2str(round(thetaC_deg)) ''];
strTitle =[ earSidePlot ' ear, \theta = ' num2str(round(thetaC_deg)) '�'];
strXlabel = '\phi in Degree';
else
xData = thetaC_deg;
strTitle =[earSidePlot ' ear, \phi = ' num2str(round(phiC_deg)) ''];
strTitle =[earSidePlot ' ear, \phi = ' num2str(round(phiC_deg)) '�'];
strXlabel = '\theta in Degree';
end
......@@ -1394,11 +1394,11 @@ classdef itaHRTF < itaAudio
earSidePlot = sArgs.earSide;
if numel(phiC_deg)>1,
xData = phiC_deg;
strTitle =[ earSidePlot ' ear, \theta = ' num2str(round(thetaC_deg)) ''];
strTitle =[ earSidePlot ' ear, \theta = ' num2str(round(thetaC_deg)) '�'];
strXlabel = '\phi in Degree';
else
xData = thetaC_deg;
strTitle =[earSidePlot ' ear, \phi = ' num2str(round(phiC_deg)) ''];
strTitle =[earSidePlot ' ear, \phi = ' num2str(round(phiC_deg)) '�'];
strXlabel = '\theta in Degree';
end
......
applications/SpatialAudio/SourceTranslation/m/calculateHandT_slideSource.m
View file @ cd83129b
......@@ -47,7 +47,7 @@ coordinate.cart = loc;
grid = params.display.grid + coordinate;
% Spherical Harmonics: num_angels x (Nc+1)^2.
Ynm = ita_sph_base(grid, N, 'Williams', true);
Ynm = ita_sph_base(grid, N);
% prepare hankel functions for translated source
for k_ind = 1 : K
......
applications/SpatialAudio/SourceTranslation/m/initParams.m
View file @ cd83129b
......@@ -168,8 +168,7 @@ params.display.grid.r = params.array.r;
% Spherical Harmonics: num_angels x (Nc+1)^2.
params.display.Ynm = ita_sph_base(params.display.grid,...
params.source.N,...
'Williams', true);
params.source.N);
% Build up a reverse grid.
th = params.display.grid.theta;
......@@ -207,8 +206,7 @@ params.array.mics = params.array.grid.nPoints; % Number of microphones
% Spherical Harmonics: num_angels x (Narray+1)^2.
params.array.Ynm = ita_sph_base(params.array.grid,...
params.array.N,...
'Williams', true);
params.array.N);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Parameters that are relative to the center
......
applications/SpatialAudio/SourceTranslation/m/simulateHandT_firstTranslation.m
View file @ cd83129b
......@@ -53,7 +53,7 @@ for k_ind = 1 : K
grid = params.display.grid - coordinate;
% Spherical Harmonics: num_angels x (Nc+1)^2.
Ynm = ita_sph_base(grid, N, 'Williams', true);
Ynm = ita_sph_base(grid, N);
hn_disPoints_x_N = ita_sph_besselh([0:N], 1,...
kr_disPoints_x_freqs(:,k_ind));
......
applications/SpatialAudio/ita_decodeAmbisonics.m
View file @ cd83129b
......@@ -57,7 +57,7 @@ for k=1:numel(weights)
Bformat(:,k)=weights(k).*Bformat(:,k); %Apply weighting factors
end
% SH and inversion
Y = ita_sph_base(LS,TheOrder,'williams',false); % generate basefunctions
Y = ita_sph_base(LS, TheOrder, 'real'); % generate basefunctions
Yinv=pinv(Y); % calculate Pseudoinverse
......
applications/SpatialAudio/ita_encodeAmbisonics.m
View file @ cd83129b
......@@ -17,7 +17,7 @@ opts.audioSource=''; %Signal of the source (itaAudio or filename)
opts = ita_parse_arguments(opts, varargin);
% Encode Source
Bformat = ita_sph_base(sourcePos, opts.order, 'Williams', false);
Bformat = ita_sph_base(sourcePos, opts.order, 'real');
if ~isempty(opts.audioSource)
if isa(opts.audioSource,'char')
......
applications/SphericalHarmonics/itaSamplingSphReal.m
View file @ cd83129b
......@@ -14,7 +14,7 @@ classdef itaSamplingSphReal < itaSamplingSph
end
methods(Access = protected)
function this = set_base(this, nmax)
this.Y = ita_sph_base(this,nmax,'Williams',false);
this.Y = ita_sph_base(this, nmax, 'real');
ita_verbose_info(['Real valued spherical harmonics up to order ' num2str(nmax) ' calculated.']);
end
end
......
applications/SphericalHarmonics/ita_sph_base.m
View file @ cd83129b
function Y = ita_sph_base(s, nmax, type, complex)
function Y = ita_sph_base(varargin)
%ITA_SPH_BASE - creates spherical harmonics (SH) base functions
% function Y = ita_sph_base(s, nmax, type)
% function Y = ita_sph_base(sampling, Nmax, options)
% Y is a matrix with dimensions [nr_points x nr_coefs]
% calculates matrix with spherical harmonic base functions
% calculates matrix with spherical harmonic basis functions
% for the grid given in theta and phi
% give type of normalization
%
%
% the definition was taken from:
......@@ -13,49 +12,51 @@ function Y = ita_sph_base(s, nmax, type, complex)
%
% This definition includes the Condon-Shotley phase term (-1)^m
%
% Martin Pollow (mpo@akustik.rwth-aachen.de)
% Institute of Technical Acoustics, RWTH Aachen, Germany
% 03.09.2008
% Syntax:
% Y = ita_sph_base(sampling, Nmax, options)
%
% Options (default):
% 'norm' ('orthonormal') : Normalization type
% 'real' (false) : Return real valued SH
%
% Example:
% Y = ita_sph_base(sampling, Nmax, 'norm', 'Williams')
%
% See also:
% ita_toolbox_gui, ita_read, ita_write, ita_generate
%
% Reference page in Help browser
% <a href="matlab:doc ita_sph_base">doc ita_sph_base</a>
%
% Deleted and new implementation
% (careful, history seems to be lost, but checkout of old version is possible):
% Bruno Masiero (bma@akustik.rwth-aachen.de)
% Institute of Technical Acoustics, RWTH Aachen, Germany
% 12.04.2011
% <ITA-Toolbox>
% This file is part of the application SphericalHarmonics for the ITA-Toolbox. All rights reserved.
% You can find the license for this m-file in the application folder.
% This file is part of the ITA-Toolbox. Some rights reserved.
% You can find the license for this m-file in the license.txt file in the ITA-Toolbox folder.
% </ITA-Toolbox>
%
%
% Martin Pollow (mpo@akustik.rwth-aachen.de)
% Institute of Technical Acoustics, RWTH Aachen, Germany
% 03.09.2008
% check input
if nargin < 4
% Use complex version of the spherical harmonics
complex = true;
end
if nargin < 3
type = 'Williams';
% disp(['using default normalization: ' type]);
end
sArgs = struct('pos1_sampling', 'itaCoordinates', ...
'pos2_Nmax', 'int', ...
'norm', 'orthonormal', ...
'real', false);
[sampling, Nmax, sArgs] = ita_parse_arguments(sArgs, varargin);
%check for invalid sampling
if s.nPoints < 1
if sampling.nPoints < 1
Y = [];
ita_verbose_info('The sampling needs to consist of at least one point.', 0)
return
end
% vectorize grid angles
theta = s.theta;
phi = s.phi;
theta = sampling.theta;
phi = sampling.phi;
nr_points = numel(theta);
if nr_points ~= numel(phi)
error('theta and phi must have same number of points');
end
nr_coefs = (nmax+1)^2;
nm = 1:nr_coefs;
nm = 1:(Nmax+1)^2;
[~,m] = ita_sph_linear2degreeorder(nm);
exp_term = exp(1i*phi*m);
......@@ -63,16 +64,16 @@ exp_term = exp(1i*phi*m);
% calculate a matrix containing the associated legendre functions
% function Pnm = ass_legendre_func
% calculate the matrix of associated Legrendre functions
Pnm = zeros(nr_points, nr_coefs);
Pnm = zeros(sampling.nPoints, (Nmax+1)^2);
for ind = 0:nmax
for ind = 0:Nmax
% define the linear indices for the i'th degree
index_m_neg = ita_sph_degreeorder2linear(ind,-1:-1:-ind); % count in reverse order
index_m_pos = ita_sph_degreeorder2linear(ind,1:ind);
index_m_pos0 = ita_sph_degreeorder2linear(ind,0:ind);
% define positive orders with correct normalization
switch lower(type)
switch lower(sArgs.norm)
case {'orthonormal','williams'}
% the Pnm's used here are the Pnm's from Williams multiplied with the
% orthonormality factor sqrt((2n+1)./(4*pi).*(n-m)! ./ (n+m)!)
......@@ -85,7 +86,10 @@ for ind = 0:nmax
case {'schmidt','sch','semi-normalized'}
Pnm(:,index_m_pos0) = ...
bsxfun(@times,(-1).^(0:ind)*sqrt(2),legendre(ind,cos(theta.'),'sch').');
bsxfun(@times,(-1).^(0:ind),legendre(ind,cos(theta.'),'sch').');
case {'ambix'}
Pnm(:,index_m_pos0) = ...
bsxfun(@times,(-1).^(0:ind)/sqrt(8*pi),legendre(ind,cos(theta.'),'sch').');
otherwise
error('Wow! I do not know this normalization!')
......@@ -103,22 +107,12 @@ end
% compose the spherical harmonic base functions
Y = Pnm .* exp_term;
if ~complex
% now the conversion is done outside of this function
Y = ita_sph_complex2real(Y')';
% below the old code:
% for ind = 1:nmax
% % define the linear indices for the i'th degree
% index_m_neg = ita_sph_degreeorder2linear(ind,-1:-1:-ind); % count in reverse order
% index_m_pos = ita_sph_degreeorder2linear(ind,1:ind);
%
% for m = 1:length(index_m_neg)
% C = ((-1)^m*Y(:,index_m_pos(m)) + Y(:,index_m_neg(m)))/sqrt(2);
% S = ((-1)^m*Y(:,index_m_neg(m)) - Y(:,index_m_pos(m)))/sqrt(2)/1i;
%
% Y(:,ita_sph_degreeorder2linear(ind,m)) = C;
% Y(:,ita_sph_degreeorder2linear(ind,-m)) = S;
% end
% end
if strcmp(sArgs.norm, 'ambix')
% multiply by sqrt(2-delta_m)
mask = ~(m | zeros(size(m)));
Y(:,mask) = Y(:,mask) * sqrt(2);
end
if sArgs.real
Y = ita_sph_complex2real(Y.').';
end
\ No newline at end of file
applications/SphericalHarmonics/ita_sph_complex2real.m
View file @ cd83129b
function SH = ita_sph_complex2real(SH)
% converts a complex valued base or SH vector into its real base / SH vectors
function SH = ita_sph_complex2real(varargin)
%ITA_SPH_COMPLEX2REAL - Transforms from complex to valued basis functions
% This function transfroms spherical harmonic coefficients with complex
% valued basis functions to their corresponding real valued
% representation. The sign convention of the real valued basis functions
% can be chosen. The definition in 'Williams - Fourier Acoustics' is
% assumed for the complex basis functions (including the Condon-Shotley
% phase). The phase conventions for the real valued spherical harmonics
% can be 'raven' (which equals the ambix phase convention as defined in
% "Nachbar - AMBIX: A Suggested Ambisonics Format (revised by Zotter)")
% or 'zotter' as defined in "Zotter - Analysis and synthesis of
% sound-radiation with spherical arrays"
%
% Syntax:
% SH_real = ita_sph_complex2real(SH_cplx, options)
%
% Options (default):
% 'phase' ('ambix') : Phase definitions as in the AmbiX format
%
% Example:
% SH_real = ita_sph_complex2real(SH_cplx)
%
% See also:
% ita_toolbox_gui, ita_read, ita_write, ita_generate
%
% Reference page in Help browser
% <a href="matlab:doc ita_sph_complex2real">doc ita_sph_complex2real</a>
% <ITA-Toolbox>
% This file is part of the application SphericalHarmonics for the ITA-Toolbox. All rights reserved.
% You can find the license for this m-file in the application folder.
% This file is part of the ITA-Toolbox. Some rights reserved.
% You can find the license for this m-file in the license.txt file in the ITA-Toolbox folder.
% </ITA-Toolbox>
% SH can be the vector/matrix of coefficients or the maximum order to
% calculate a matrix.
if isnumeric(SH)
if numel(SH) == 1
% return a matrix
SH = ita_sph_complex2real_mat(SH);
else
% assume this as SH coefs
SH = ita_sph_complex2real_coefs(SH);
end
else
error('please check syntax')
end
end
function SH = ita_sph_complex2real_coefs(SH)
% assume the SH dimension as first dimension
sizeY = size(SH);
% Rewrite, original author unknown
%
% Author: Marco Berzborn -- Email: marco.berzborn@akustik.rwth-aachen.de
% Created: 12-Jun-2017
% check for max 2 dimension
if numel(sizeY) > 2
reshape(SH, sizeY(1), []);
end
sArgs = struct('pos1_SH', 'double', ...
'phase', 'ambix');
[SH, sArgs] = ita_parse_arguments(sArgs, varargin);
% check for SH in 1st dimension
if sizeY(1) == 1 && sizeY(2) > 1
SH = SH.';
isTransposed = true;
else
isTransposed = false;
if ~isnatural(sqrt(size(SH, 1)) - 1)
ita_verbose_info('The dimensions of the input data do not match.', 0)
SH = [];
return
end
nmax = sqrt(size(SH,1)) - 1;
if nmax ~= round(nmax)
error('ita_sph_complex2real something wrong with the number of coefficients (they do not fill up to a full order)');
end
Nmax = floor(sqrt(size(SH,1))-1);
for ind = 1:nmax
for ind = 1:Nmax
% define the linear indices for the i'th degree
index_m_neg = ita_sph_degreeorder2linear(ind,-1:-1:-ind); % count in reverse order
index_m_pos = ita_sph_degreeorder2linear(ind,1:ind);
for m = 1:length(index_m_neg)
cPos = SH(index_m_pos(m),:);
cNeg = SH(index_m_neg(m),:);
rPos = ((-1)^m * cPos + cNeg) / sqrt(2);
rNeg = ((-1)^m * cNeg - cPos) / sqrt(2) .* 1i;
switch sArgs.phase
case {'ambix', 'raven'}
rNeg = (cNeg - (-1)^m*cPos) / sqrt(2) .* 1i;
case 'zotter'
rNeg = (-cNeg + (-1)^m * cPos) / sqrt(2) .* 1i;
otherwise
ita_verbose_info('I do not know this phase convention. Aborting...', 0);
SH = [];
return
end
SH(ita_sph_degreeorder2linear(ind,m),:) = rPos;
SH(ita_sph_degreeorder2linear(ind,-m),:) = rNeg;
end
end
if isTransposed
SH = SH.';
end
% now bring to the old shape
SH = reshape(SH,sizeY);
end
function T = ita_sph_complex2real_mat(nmax)
if numel(nmax) > 1
error;
end
nSH = (nmax+1).^2;
lin = (1:nSH).';
multiplicator = 2*mod(lin,2) - 1;
[deg,order] = ita_sph_linear2degreeorder(lin);
linNeg = ita_sph_degreeorder2linear(deg,-order);
isPos = order > 0;
isNeg = order < 0;
T = sparse(eye(nSH));
T_orig = T;
T_swap = T .* diag(multiplicator);
T(lin(isPos),:) = (T_swap(lin(isPos),:) + T_orig(linNeg(isPos),:)) / sqrt(2);
T(lin(isNeg),:) = (T_swap(lin(isNeg),:) - T_orig(linNeg(isNeg),:)) / sqrt(2) .* 1i;
end
end
\ No newline at end of file
applications/SphericalHarmonics/ita_sph_mimo_error_simulation.m
View file @ cd83129b
......@@ -426,13 +426,12 @@ sArgs = struct('pos1_receiverCoords','itaCoordinates',...
'pos5_k','double',...
'r',[],...
'r_eq',1,...
'norm',false,...
'shType','complex');
'norm',false);
[receiverCoords,receiverNmax,sourceCoords,sourceNmax,k,sArgs] = ita_parse_arguments(sArgs,varargin);
[receiverLookDirection, sourceLookDirection,r] = array_orientation(receiverCoords,sourceCoords);
yReceiver = ita_sph_base(receiverLookDirection,receiverNmax,'Williams',strcmp(sArgs.shType,'complex'))';
ySource = ita_sph_base(sourceLookDirection,sourceNmax,'Williams',strcmp(sArgs.shType,'complex'))';
yReceiver = ita_sph_base(receiverLookDirection,receiverNmax)';
ySource = ita_sph_base(sourceLookDirection,sourceNmax)';
% avoid sArgs in parfor since it is a broadcast variable
if ~isempty(sArgs.r)
......
applications/SphericalHarmonics/ita_sph_real2complex.m
View file @ cd83129b
function SH = ita_sph_real2complex(SH)
% converts a real base or SH vector into its complex base / SH vectors
function SH = ita_sph_real2complex(varargin)
%ITA_SPH_REAL2COMPLEX - Transforms from real to complex valued basis functions
% This function transfroms spherical harmonic coefficients with real
% valued basis functions to their corresponding complex valued
% representation. The sign convention of the real valued basis functions
% can be chosen. The definition in 'Williams - Fourier Acoustics' is
% specified for the complex basis functions (including the Condon-Shotley
% phase). The phase conventions for the real valued spherical harmonics
% can be 'raven' (which equals the ambix phase convention as defined in
% "Nachbar - AMBIX: A Suggested Ambisonics Format (revised by Zotter)")
% or 'zotter' as defined in "Zotter - Analysis and synthesis of
% sound-radiation with spherical arrays"
%
% Syntax:
% SH_cplx = ita_sph_real2complex(SH_real, options)
%
% Options (default):
% 'phase' ('ambix') : Phase definitions as in the AmbiX format
%
% Example:
% SH_cplx = ita_sph_real2complex(SH_real)
%
% See also:
% ita_toolbox_gui, ita_read, ita_write, ita_generate
%
% Reference page in Help browser
% <a href="matlab:doc ita_sph_real2complex">doc ita_sph_real2complex</a>
% <ITA-Toolbox>
% This file is part of the ITA-Toolbox. Some rights reserved.
% You can find the license for this m-file in the license.txt file in the ITA-Toolbox folder.
% </ITA-Toolbox>
% Rewrite, original author unknown
%
% Author: Marco Berzborn -- Email: marco.berzborn@akustik.rwth-aachen.de
% Created: 12-Jun-2017
%
% <ITA-Toolbox>
......@@ -10,44 +46,20 @@ function SH = ita_sph_real2complex(SH)
% SH can be the vector/matrix of coefficients or the maximum order to
% calculate a matrix.
if isnumeric(SH)
if numel(SH) == 1
% return a matrix
SH = ita_sph_real2complex_mat(SH);
else
% assume this as SH coefs
SH = ita_sph_real2complex_coefs(SH);
end
else
error('please check syntax')
end
sArgs = struct('pos1_SH', 'double' ,...
'phase', 'ambix');
[SH, sArgs] = ita_parse_arguments(sArgs, varargin);
if ~isnatural(sqrt(size(SH, 1)) - 1)
ita_verbose_info('The dimensions of the input data do not match.', 0)
SH = [];
return
end
function SH = ita_sph_real2complex_coefs(SH)
Nmax = floor(sqrt(size(SH,1))-1);
% assume the SH dimension as first dimension
sizeY = size(SH);
% check for max 2 dimension
if numel(sizeY) > 2
reshape(SH, sizeY(1), []);
end
% check for SH in 1st dimension
if sizeY(1) == 1 && sizeY(2) > 1
SH = SH.';
isTransposed = true;
else
isTransposed = false;
end
nmax = sqrt(size(SH,1)) - 1;
if nmax ~= round(nmax)
error('ita_sph_complex2real something wrong with the number of coefficients (they do not fill up to a full order)');
end
for ind = 1:nmax
for ind = 1:Nmax
% define the linear indices for the i'th degree
index_m_neg = ita_sph_degreeorder2linear(ind,-1:-1:-ind); % count in reverse order
index_m_pos = ita_sph_degreeorder2linear(ind,1:ind);
......@@ -56,43 +68,19 @@ for ind = 1:nmax
rPos = SH(index_m_pos(m),:);
rNeg = SH(index_m_neg(m),:);
cPos = ((-1)^m .* rPos + 1i .* rNeg) ./ sqrt(2);
cNeg = (rPos - 1i .* (-1)^m .* rNeg) ./ sqrt(2);
switch lower(sArgs.phase)
case {'ambix','raven'}
cPos = (rPos + 1i.* rNeg) * (-1)^m / sqrt(2);
cNeg = (rPos - 1i * rNeg) ./ sqrt(2);
case {'zotter'}
cPos = (rPos - 1i.* rNeg) * (-1)^m / sqrt(2);
cNeg = (rPos + 1i * rNeg) ./ sqrt(2);
otherwise
ita_verbose_info('I do not know this phase convention. Aborting...', 0);
SH = [];
return
end
SH(ita_sph_degreeorder2linear(ind,m),:) = cPos;
SH(ita_sph_degreeorder2linear(ind,-m),:) = cNeg;
end
end
if isTransposed
SH = SH.';
end
% now bring to the old shape
SH = reshape(SH,sizeY);
end
function T = ita_sph_real2complex_mat(nmax)
if numel(nmax) > 1
error;
end
nSH = (nmax+1).^2;
lin = (1:nSH).';
multiplicator = 2*mod(lin,2) - 1;
[deg,order] = ita_sph_linear2degreeorder(lin);
linNeg = ita_sph_degreeorder2linear(deg,-order);
isPos = order > 0;
isNeg = order < 0;
T = sparse(eye(nSH));
T_orig = T;
T_swap = T .* diag(multiplicator);
T(lin(isPos),:) = (T_swap(lin(isPos),:) + 1i .* T_orig(linNeg(isPos),:)) / sqrt(2);
T(lin(