Mercurial > hg > ltpda
view m-toolbox/classes/@ao/buildWhitener1D.m @ 9:fbbfcd56e449 database-connection-manager
Remove dead code
author | Daniele Nicolodi <nicolodi@science.unitn.it> |
---|---|
date | Mon, 05 Dec 2011 16:20:06 +0100 |
parents | f0afece42f48 |
children |
line wrap: on
line source
% BUILDWHITENER1D builds a whitening filter based on the input frequency-series. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % DESCRIPTION:BUILDWHITENER1D builds a whitening filter based on the input frequency-series. % The filter is built by fitting to the model provided. % If no model is provided, a fit is made to a spectral-density estimate of the % input time-series (made using psd+bin_data or lpsd). % Note: The function assumes that the input model corresponds % to the one-sided psd of the data to be whitened. % % ALGORITHM: % 1) If no model provided, make psd+bin_data or lpsd % of time-series and take it as a model % for the data power spectral density % 2) Fit a set of partial fraction z-domain filters using % utils.math.psd2wf. The fit is automatically stopped when % the accuracy tolerance is reached. % 3) Convert to array of MIIR filters % 4) Assemble into a parallel filterbank object % % % CALL: b = buildWhitener1D(a, pl) % [b1,b2,...,bn] = buildWhitener1D(a1,a2,...,an, pl); % % INPUT: % - as is a time-series analysis object or a vector of % analysis objects % - pl is a plist with the input parameters % % OUTPUT: % - b "whitening" filters, stored into a filterbank. % % <a href="matlab:utils.helper.displayMethodInfo('ao', 'buildWhitener1D')">Parameters Description</a> % % VERSION: $Id: buildWhitener1D.m,v 1.11 2011/04/18 19:46:57 mauro Exp $ % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% function varargout = buildWhitener1D(varargin) % Check if this is a call for parameters if utils.helper.isinfocall(varargin{:}) varargout{1} = getInfo(varargin{3}); return end import utils.const.* utils.helper.msg(msg.PROC3, 'running %s/%s', mfilename('class'), mfilename); % Collect input variable names in_names = cell(size(varargin)); for ii = 1:nargin,in_names{ii} = inputname(ii);end % Collect all AOs and plists [as, ao_invars] = utils.helper.collect_objects(varargin(:), 'ao', in_names); pl = utils.helper.collect_objects(varargin(:), 'plist', in_names); if nargout == 0 error('### buildWhitener cannot be used as a modifier. Please give an output variable.'); end % combine plists if isempty(pl) model = 'psd'; else model = find(pl, 'model'); if isempty(model) model = 'psd'; pl.pset('model', model); end end if ischar(model) pl = parse(pl, getDefaultPlist(model)); else pl = parse(pl, getDefaultPlist('Default')); end pl.getSetRandState(); % Collect input histories inhists = [as.hist]; % Initialize output objects bs = filterbank.initObjectWithSize(1, numel(as)); % Loop over input AOs for jj = 1:numel(as) % 1) searching for input model switch class(as(jj).data) case 'tsdata' % Build the model based on input time-series utils.helper.msg(msg.PROC1, 'user input tsdata object, estimating the model from it'); model = estimateModel(as(jj), pl); case 'fsdata' % The input data are the model utils.helper.msg(msg.PROC1, 'user input fsdata object, taking it as the model'); model = as(jj); otherwise warning('!!! %s expects ao/tsdata or ao/fsdata objects. Skipping AO %s', mfilename, ao_invars{jj}); return; end %-------------- Whiten this AO % Extract necessary parameters % Tolerance for MSE Value lrscond = find(pl, 'FITTOL'); % give an error for strange values of lrscond if lrscond < 0 error('!!! Negative values for FITTOL are not allowed !!!') end % handling data lrscond = -1 * log10(lrscond); % give a warning for strange values of lrscond if lrscond<0 warning('You are searching for a MSE lower than %s', num2str(10^(-1*lrscond))) end params.lrscond = lrscond; % Tolerance for the MSE relative variation msevar = find(pl, 'MSEVARTOL'); % handling data msevar = -1 * log10(msevar); % give a warning for strange values of msevar if msevar<0 warning('You are searching for MSE relative variation lower than %s', num2str(10^(-1*msevar))) end params.msevar = msevar; if isempty(params.msevar) params.ctp = 'chival'; else params.ctp = 'chivar'; end % Weights switch find(pl, 'Weights') case {'equal', 'flat', 1} params.weightparam = 1; case {'1/abs', '1./abs', 2} params.weightparam = 2; case {'1/abs^2', '1/abs2', '1./abs^2', '1./abs2', 3} params.weightparam = 3; otherwise warning('Unrecognized weights option %s', find(pl, 'Weights')) end % 2) Build filters % Build input structure for psd2wf params.idtp = 1; params.Nmaxiter = find(pl, 'MaxIter'); params.minorder = find(pl, 'MinOrder'); params.maxorder = find(pl, 'MaxOrder'); params.spolesopt = find(pl, 'PoleType'); params.spy = find(pl, 'Disp'); if (find(pl, 'plot')) params.plot = 1; else params.plot = 0; end fs = find(pl, 'fs'); if isempty(fs) || fs <= 0 || ~isfinite(fs) if isempty(model.fs) || model.fs <= 0 || ~isfinite(model.fs) error('### Invalid fs value %s. Please specify a meaningful fs, either via the model or in the plist', num2str(fs)); else fs = model.fs; end end params.fs = fs; params.usesym = 0; params.dterm = 0; % it is better to fit without direct term params.fullauto = 1; % call psd2wf [res, poles, dterm, mresp, rdl] = ... utils.math.psd2wf(model.y,[],[],[],model.x,params); % 3) Convert to MIIR filters % filtering with a stable model pfilts = []; for kk = 1:numel(res) ft = miir(res(kk), [ 1 -poles(kk)], fs); pfilts = [pfilts ft]; end % 4) Build the output filterbank object bs(jj) = filterbank(plist('filters', pfilts, 'type', 'parallel')); % set the input units to be the same as the model bs(jj).setIunits(sqrt(model.yunits * unit('Hz'))); % set the output units to be empty bs(jj).setOunits(unit()); % set the name for this object bs(jj).name = sprintf('buildWhitener1D(%s)', ao_invars{jj}); % add history bs(jj).addHistory(getInfo('None'), pl, ao_invars(jj), inhists(jj)); end % Set output if nargout == numel(bs) % List of outputs for ii = 1:numel(bs) varargout{ii} = bs(ii); end else % Single output varargout{1} = bs; end end %-------------------------------------------------------------------------- % Get Info Object %-------------------------------------------------------------------------- function ii = getInfo(varargin) if nargin == 1 && strcmpi(varargin{1}, 'None') sets = {}; pl = []; elseif nargin == 1 && ~isempty(varargin{1}) && ischar(varargin{1}) sets{1} = varargin{1}; pl = getDefaultPlist(sets{1}); else sets = SETS(); % get plists pl(size(sets)) = plist; for kk = 1:numel(sets) pl(kk) = getDefaultPlist(sets{kk}); end end % Build info object ii = minfo(mfilename, 'ao', 'ltpda', utils.const.categories.sigproc, '$Id: buildWhitener1D.m,v 1.11 2011/04/18 19:46:57 mauro Exp $', sets, pl); end %-------------------------------------------------------------------------- % Defintion of Sets %-------------------------------------------------------------------------- function out = SETS() out = {... 'Default', ... 'PSD', ... 'LPSD' ... }; end %-------------------------------------------------------------------------- % Get Default Plist %-------------------------------------------------------------------------- function plout = getDefaultPlist(set) persistent pl; persistent lastset; if ~exist('pl', 'var') || isempty(pl) || ~strcmp(lastset, set) pl = buildplist(set); lastset = set; end plout = pl; end function pl = buildplist(set) pl = plist(); % Model p = param({'model', ['A model estimation technique in the case of tsdata input:<br>'... '<li>PSD - using <tt>psd</tt> + <tt>bin_data</tt></li>'... '<li>LPSD - using <tt>lpsd</tt></li>']}, {1, {'PSD', 'LPSD'}, paramValue.SINGLE}); pl.append(p); % Range p = param({'range', ['The frequency range to evaluate the fitting.<br>' ... 'An empty value or [-inf inf] will include the whole range.<br>' ... 'The remaining part of the model will be completed according<br>' ... 'to the option chosen in the ''complete'' parameter.<br>' ... ]}, paramValue.EMPTY_DOUBLE); pl.append(p); % Complete p = param({'complete_hf', ['Choose how to complete the frequency range up to fs/2.<ol>' ... '<li>Assumes flat response</li>' ... '<li>Assumes 4 poles low-pass type response</li>' ... ]}, {1,{'flat', 'lowpass'}, paramValue.SINGLE}); pl.append(p); % fs p = param({'fs', ['The sampling frequency to design the output filter on.<br>' ... 'If it is not a positive number, it will be taken from the model' ... ]}, paramValue.EMPTY_DOUBLE); pl.append(p); % MaxIter p = param({'MaxIter', 'Maximum number of iterations in fit routine.'}, paramValue.DOUBLE_VALUE(30)); pl.append(p); % PoleType p = param({'PoleType', ['Choose the pole type for fitting:<ol>'... '<li>use real starting poles</li>'... '<li>generates complex conjugate poles of the<br>'... 'type <tt>a.*exp(theta*pi*j)</tt>'... 'with <tt>theta = linspace(0,pi,N/2+1)</tt></li>'... '<li>generates complex conjugate poles of the type<br>'... '<tt>a.*exp(theta*pi*j)</tt><br>'... 'with <tt>theta = linspace(0,pi,N/2+2)</tt></li></ol>']}, {1, {1, 2, 3}, paramValue.SINGLE}); pl.append(p); % MinOrder p = param({'MinOrder', 'Minimum order to fit with.'}, paramValue.DOUBLE_VALUE(2)); pl.append(p); % MaxOrder p = param({'MaxOrder', 'Maximum order to fit with.'}, paramValue.DOUBLE_VALUE(25)); pl.append(p); % Weights p = param({'Weights', ['Choose weighting method:<ol>'... '<li>equal weights for each point</li>'... '<li>weight with <tt>1/abs(model)</tt></li>'... '<li>weight with <tt>1/abs(model).^2</tt></li></ol>']}, ... {2, {'equal', '1/abs', '1/abs^2'}, paramValue.SINGLE}); pl.append(p); % Plot p = param({'Plot', 'Plot results of each fitting step.'}, paramValue.FALSE_TRUE); pl.append(p); % Disp p = param({'Disp', 'Display the progress of the fitting iteration.'}, paramValue.FALSE_TRUE); pl.append(p); % MSEVARTOL p = param({'MSEVARTOL', ['Mean Squared Error Variation - Check if the<br>'... 'relative variation of the mean squared error is<br>'... 'smaller than the value specified. This<br>'... 'option is useful for finding the minimum of Chi-squared.']}, ... paramValue.DOUBLE_VALUE(1e-1)); pl.append(p); % FITTOL p = param({'FITTOL', ['Mean Squared Error Value - Check if the mean<br>'... 'squared error value is lower than the value<br>'... 'specified.']}, paramValue.DOUBLE_VALUE(1e-2)); pl.append(p); % Append sets of parameters according to the chosen spectral estimator if ~utils.helper.ismember(lower(SETS), lower(set)) error('### Unknown set [%s]', set); end switch lower(set) case 'default' pl.remove('model'); case 'psd' pl = combine(pl, ao.getInfo('psd').plists); pl.pset(... 'model', 'PSD', ... 'Navs', 16, ... 'order', 1, ... 'olap', 50 ... ); pl = combine(pl, ao.getInfo('bin_data').plists); pl.pset(... 'method', 'MEAN', ... 'resolution', 50 ... ); case 'lpsd' pl = combine(pl, ao.getInfo('lpsd').plists); pl.pset(... 'model', 'LPSD' ... ); otherwise end end %-------------------------------------------------------------------------- % Estimate a model from the data or from user input %-------------------------------------------------------------------------- function model = estimateModel(b, pl) import utils.const.* % Estimate a model for the PSD model_all = find(pl, 'model'); if ischar(model_all) switch lower(model_all) case 'psd' % Select only the parameters associated to ao/psd pls = ao.getInfo('psd').plists; % Call ao/psd sp = psd(b, pl.subset(pls.getKeys())); % Select only the parameters associated to ao/bin_data pls = ao.getInfo('bin_data').plists; % Call ao/bin_data model_all = bin_data(sp, pl.subset(pls.getKeys())); case 'lpsd' % Select only the parameters associated to ao/lpsd pls = ao.getInfo('lpsd').plists; model_all = lpsd(b, pl.subset(pls.getKeys())); otherwise error('### Unknown model [%s]', model_all); end end % Select only a limited frequency range frange = find(pl, 'range'); if isempty(frange) frange = [-inf inf]; end model = model_all.split(plist('frequencies', frange)); f1 = frange(1); f2 = frange(2); if isfinite(f2) % Select a technique to complete the high frequency range complete_up_opt = find(pl, 'complete_hf'); switch complete_up_opt case {'flat', 'allpass', 'all pass', 'all-pass'} utils.helper.msg(msg.PROC1, 'Completing the frequency range from %s to %s with flat model', ... num2str(f2), num2str(b.fs/2)); % Build a flat model response r = ones(size(model_all.x)); case {'lowpass', 'low pass', 'low-pass'} utils.helper.msg(msg.PROC1, 'Completing the frequency range from %s to %s with 4 poles low-pass model', ... num2str(f2), num2str(b.fs/2)); % Build a 4 poles low pass resp r = abs(resp(pzmodel(plist('gain', 1, 'poles', {10*f2,11*f2,12*f2,13*f2})), plist('f', model_all.x))); r = r.y; otherwise error('### Unknown option [%s] for high frequency completion', complete_up_opt); end model_hf = r * model.y(end); model = join(model, ... ao(plist('type', 'fsdata', 'xvals', model_all.x, 'yvals', model_hf, 'fs', model_all.fs, 'yunits', model_all.yunits))); end end