ft_freqanalysis_mvar.m

function [freq] = ft_freqanalysis_mvar(cfg, data)

% FT_FREQANALYSIS_MVAR performs frequency analysis on
% mvar data, by fourier transformation of the coefficients. The output
% contains cross-spectral density, spectral transfer matrix, and the
% covariance of the innovation noise. The dimord = 'chan_chan(_freq)(_time)
%
% The function is stand-alone, but is typically called through
% FT_FREQANALYSIS, specifying cfg.method = 'mvar'.
%
% Use as
%   [freq] = ft_freqanalysis(cfg, data), with cfg.method = 'mvar'
%
% or
%
%   [freq] = ft_freqanalysis_mvar(cfg, data)
%
% The input data structure should be a data structure created by
% FT_MVARANALYSIS, i.e. a data-structure of type 'mvar'.
%
% The configuration can contain:
%   cfg.foi = vector with the frequencies at which the spectral quantities
%               are estimated (in Hz). Default: 0:1:Nyquist
%   cfg.feedback = 'none', or any of the methods supported by FT_PROGRESS,
%                    for providing feedback to the user in the command
%                    window.
%
% To facilitate data-handling and distributed computing you can use
%   cfg.inputfile   =  ...
%   cfg.outputfile  =  ...
% If you specify one of these (or both) the input data will be read from a *.mat
% file on disk and/or the output data will be written to a *.mat file. These mat
% files should contain only a single variable, corresponding with the
% input/output structure.
%
% See also FT_MVARANALYSIS, FT_DATATYPE_MVAR, FT_PROGRESS

% Copyright (C) 2009, Jan-Mathijs Schoffelen
%
% This file is part of FieldTrip, see http://www.fieldtriptoolbox.org
% for the documentation and details.
%
%    FieldTrip is free software: you can redistribute it and/or modify
%    it under the terms of the GNU General Public License as published by
%    the Free Software Foundation, either version 3 of the License, or
%    (at your option) any later version.
%
%    FieldTrip is distributed in the hope that it will be useful,
%    but WITHOUT ANY WARRANTY; without even the implied warranty of
%    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
%    GNU General Public License for more details.
%
%    You should have received a copy of the GNU General Public License
%    along with FieldTrip. If not, see <http://www.gnu.org/licenses/>.
%
% $Id$

% these are used by the ft_preamble/ft_postamble function and scripts
ft_revision = '$Id$';
ft_nargin   = nargin;
ft_nargout  = nargout;

% do the general setup of the function
ft_defaults
ft_preamble init
ft_preamble debug
ft_preamble loadvar data
ft_preamble provenance data

% the ft_abort variable is set to true or false in ft_preamble_init
if ft_abort
  return
end

cfg.foi        = ft_getopt(cfg, 'foi',        'all');
cfg.feedback   = ft_getopt(cfg, 'feedback',   'none');
%cfg.channel    = ft_getopt(cfg, 'channel',    'all');
%cfg.keeptrials = ft_getopt(cfg, 'keeptrials', 'no');
%cfg.jackknife  = ft_getopt(cfg, 'jackknife',  'no');
%cfg.keeptapers = ft_getopt(cfg, 'keeptapers', 'yes');

if strcmp(cfg.foi, 'all')
  cfg.foi = (0:1:data.fsampleorig/2);
end

dimtok = tokenize(data.dimord, '_');
isfull = isfield(data, 'label') && sum(strcmp(dimtok, 'chan'))==2;
isuvar = isfield(data, 'label') && sum(strcmp(dimtok, 'chan'))==1;
isbvar = isfield(data, 'labelcmb');

if (isfull||isuvar) && isbvar
  ft_error('data representation is ambiguous');
end
if ~isfull && ~isbvar && ~isuvar
  ft_error('data representation is unsupported');
end

%keeprpt  = strcmp(cfg.keeptrials, 'yes');
%keeptap  = strcmp(cfg.keeptapers, 'yes');
%dojack   = strcmp(cfg.jackknife,  'yes');
%dozscore = strcmp(cfg.zscore,     'yes');

%if ~keeptap, ft_error('not keeping tapers is not possible yet'); end
%if dojack && keeprpt, ft_error('you cannot simultaneously keep trials and do jackknifing'); end

nfoi     = length(cfg.foi);
if isfield(data, 'time')
  ntoi = numel(data.time);
else
  ntoi = 1;
end

if isfull || isuvar
  cfg.channel = ft_channelselection('all', data.label);
  %cfg.channel    = ft_channelselection(cfg.channel,      data.label);
  chanindx = match_str(data.label, cfg.channel);
  nchan    = length(chanindx);
  label    = data.label(chanindx);
  nlag     = size(data.coeffs,3); %change in due course
  
  %---allocate memory
  h         = complex(zeros(nchan, nchan,  nfoi, ntoi), zeros(nchan, nchan,  nfoi, ntoi));
  a         = complex(zeros(nchan, nchan,  nfoi, ntoi), zeros(nchan, nchan,  nfoi, ntoi));
  crsspctrm = complex(zeros(nchan, nchan,  nfoi, ntoi), zeros(nchan, nchan,  nfoi, ntoi));
elseif isbvar
  ncmb      = size(data.labelcmb,1)./4;
  nlag      = size(data.coeffs,2);
  
  %---allocate memory
  h         = complex(zeros(ncmb*4, nfoi, ntoi), zeros(ncmb*4, nfoi, ntoi));
  a         = complex(zeros(ncmb*4, nfoi, ntoi), zeros(ncmb*4, nfoi, ntoi));
  crsspctrm = complex(zeros(ncmb*4, nfoi, ntoi), zeros(ncmb*4, nfoi, ntoi));
end

%FIXME build in repetitions

%---loop over the tois
ft_progress('init', cfg.feedback, 'computing MAR-model based TFR');
for j = 1:ntoi
  ft_progress(j/ntoi, 'processing timewindow %d from %d\n', j, ntoi);
  
  if isfull
    %---compute transfer function
    ar = reshape(data.coeffs(:,:,:,j), [nchan nchan*nlag]);
    [h(:,:,:,j), a(:,:,:,j)] = ar2h(ar, cfg.foi, data.fsampleorig);
    
    %---compute cross-spectra
    nc = data.noisecov(:,:,j);
    for k = 1:nfoi
      tmph               = h(:,:,k,j);
      crsspctrm(:,:,k,j) = tmph*nc*tmph';
    end
  elseif isuvar
    %---compute transfer function
    for m = 1:nchan
      ar = reshape(data.coeffs(m,:,j), [1 nlag]);
      [h(m,m,:,j), a(m,m,:,j)] = ar2h(ar, cfg.foi, data.fsampleorig);
      
      %---compute cross-spectra
      nc = data.noisecov(m,j);
      for k = 1:nfoi
        tmph               = h(m,m,k,j);
        crsspctrm(m,m,k,j) = tmph*nc*tmph';
      end
    end
  elseif isbvar
    for kk = 1:ncmb
      %---compute transfer function
      ar = reshape(data.coeffs((kk-1)*4+(1:4),:,:,j), [2 2*nlag]);
      [tmph,tmpa] = ar2h(ar, cfg.foi, data.fsampleorig);
      h((kk-1)*4+(1:4),:,:) = reshape(tmph, [4 nfoi ntoi]);
      a((kk-1)*4+(1:4),:,:) = reshape(tmpa, [4 nfoi ntoi]);
      
      %---compute cross-spectra
      nc = reshape(data.noisecov((kk-1)*4+(1:4),j), [2 2]);
      for k = 1:nfoi
        crsspctrm((kk-1)*4+(1:4),k,j) = reshape(tmph(:,:,k)*nc*tmph(:,:,k)', [4 1]);
      end
    end
  end
end
ft_progress('close');

%---create output-structure
freq          = [];
freq.freq     = cfg.foi;
%freq.cumtapcnt= ones(ntrl, 1)*ntap;
freq.transfer  = h;
%freq.itransfer = a;
freq.noisecov  = data.noisecov;
freq.crsspctrm = crsspctrm;
if isfield(data, 'dof')
  freq.dof       = data.dof;
end
if isfull
  freq.label    = label;
  if ntoi>1
    freq.time   = data.time;
    freq.dimord = 'chan_chan_freq_time';
  else
    freq.dimord = 'chan_chan_freq';
  end
elseif isbvar
  freq.labelcmb = data.labelcmb;
  if ntoi>1
    freq.time   = data.time;
    freq.dimord = 'chancmb_freq_time';
  else
    freq.dimord = 'chancmb_freq';
  end
end

% do the general cleanup and bookkeeping at the end of the function
ft_postamble debug
ft_postamble previous   data
ft_postamble provenance freq
ft_postamble history    freq
ft_postamble savevar    freq

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% SUBFUNCTION to compute transfer-function from ar-parameters
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [h, zar] = ar2h(ar, foi, fsample)

nchan = size(ar,1);
ncmb  = nchan*nchan;
nfoi  = length(foi);

%---z-transform frequency axis
zfoi  = exp(-2.*pi.*1i.*(foi./fsample));

%---reorganize the ar-parameters
ar  = reshape(ar, [ncmb size(ar,2)./nchan]);
ar  = fliplr([reshape(eye(nchan), [ncmb 1]) -ar]);

zar = complex(zeros(ncmb, nfoi), zeros(ncmb, nfoi));
for k = 1:ncmb
  zar(k,:) = polyval(ar(k,:),zfoi);
end
zar = reshape(zar, [nchan nchan nfoi]);
h   = zeros(size(zar));
for k = 1:nfoi
  h(:,:,k) = inv(zar(:,:,k));
end
h   = sqrt(2).*h; %account for the negative frequencies, normalization necessary for
%comparison with non-parametric (fft based) results in FieldTrip
%FIXME probably the normalization for the zero Hz bin is incorrect
zar = zar./sqrt(2);