Various experimental edits, removal of parallelism

robustCAR4_winsz.m

@@ -1,57 +1,68 @@
 function [data_out, ref_est, nn_ref_est] = robustCAR4_winsz(data_in, dt, n_min_contributing_channels, n_cores)
+
+% New internal functions for robust CAR
 %
 % Modded by MDT -- temporary. Need to fix all of the parallel processing!
 %
 % IN PROGRESS!
 %
 % NOT FOR USE!
 
+    %win_sz    = 10*dt; %.2        % seconds.  % daFuq?
+    n_win_sz  = 1;                    % floor( win_sz / dt );
+    n         = size( data_in, 2 );
+    n_win     = floor( n / n_win_sz );
 
-  %win_sz    = 10*dt; %.2        % seconds.
-  n_win_sz  = 1; %floor( win_sz / dt );
-  n         = size( data_in, 2 );
-  n_win     = floor( n / n_win_sz );
-  if( n_win == 0 )
-    fprintf( '\n\nShould not happen.\n\n' );
-    keyboard
-    [ data_out, ref_est, nn_ref_est ] = robustCAR_helper( data_in, dt, n_min_contributing_channels, nm, taxis );
-  else
-    data_out    = zeros( size( data_in, 1 ), n );
-    ref_est     = zeros( 1, n );
-    nn_ref_est  = zeros( 1, n );
-    fprintf( '\n\nrobustCAR4: Entering for jj = 1 : n ...\n\n' )
-    tic
-    if matlabpool( 'size' ) == 0
-      matlabpool( n_cores );
-    end
-    parfor jj = 1 : n
-      if mod( jj, 100 ) == 0 
-        fprintf( 'Sample No: %d\n', jj )
-      end
-      i_start = jj; i_stop  = i_start + n_win_sz - 1;
-      i_stop  = min( [ n i_stop ] );
-      inds    = i_start:i_stop;
-      if std( data_in(:,inds)) > 1e-20
-        [ tmp_out, ref_out, nn_out ]  = robustCAR_helper( data_in(:,inds), dt, n_min_contributing_channels );
-
-try
-        data_out(:,jj)              = tmp_out(1,:)';   
-        ref_est(jj)                 = ref_out(1); 
-        nn_ref_est(jj)              = nn_out(1);
-catch me
-        me.message
-        dbstack
+    if( n_win == 0 )
+        fprintf( '\n\nShould not happen.\n\n' );
         keyboard
-end
+        [ data_out, ref_est, nn_ref_est ] = robustCAR_helper( data_in, dt, n_min_contributing_channels, nm, taxis );
+    else
+        data_out    = zeros( size( data_in, 1 ), n );
+        ref_est     = zeros( 1, n );
+        nn_ref_est  = zeros( 1, n );
+        fprintf( '\n\nrobustCAR4_winsz: Entering for jj = 1 : n ...\n\n' )
+        tic
+        %     if isempty(gcp('nocreate'))     % matlabpool( 'size' ) == 0
+        %       matlabpool( n_cores );
+        %     end
+
+        %% here is where things are hanging!!! 
+        %parfor jj = 1 : n
+        for jj = 1:n
+
+            if mod( jj, 1 ) == 0   % orig mod was to 100
+              fprintf( 'Sample No: %d\n', jj )
+            end
+
+            i_start = jj; i_stop  = i_start + n_win_sz - 1
+            i_stop  = min( [ n i_stop ] )
+            inds    = i_start:i_stop
+
+            if std( data_in(:,inds)) > 1e-20
+                [ tmp_out, ref_out, nn_out ]  = robustCAR_helper( data_in(:,inds), dt, n_min_contributing_channels );
+
+                try
+                    data_out(:,jj)              = tmp_out(1,:)';
+                    ref_est(jj)                 = ref_out(1);
+                    nn_ref_est(jj)              = nn_out(1);
+                catch me
+                    me.message
+                    dbstack
+                    keyboard
+                end
 
-      else
-         data_out(:,jj)              = 0; ref_est(jj) = mean(data_in(:,jj)); nn_ref_est(jj) = size( data_in, 1 );
-      end
-    end % for jj 
-    matlabpool close
-    fprintf( '\n\nrobustCAR4_winsz_1b: Exiting for jj = 1 : n ...\n\n' )
-    toc
-    data_out = data_out';
-  end
+            else
+                data_out(:,jj)              = 0; ref_est(jj) = mean(data_in(:,jj)); nn_ref_est(jj) = size( data_in, 1 );
+            end
+        end % for jj
+
+        %% end of the parfor loop!
+
+        %delete(gcp('nocreate'))      %matlabpool close
+        fprintf( '\n\nrobustCAR4_winsz: Exiting for jj = 1 : n ...\n\n' )
+        toc
+        data_out = data_out';
+    end
 end
 

robustCAR_helper.m

@@ -345,7 +345,8 @@
 
   try
     n_candidates      = length( candidate_deltas );
-    new_mus           = mu + make_col( candidate_deltas )';
+    %new_mus           = mu + make_col( candidate_deltas )';
+    new_mus           = mu + candidate_deltas';
     x                 = ( d*ones(1,n_candidates) - ones(size(d,1),1)*new_mus ) ./ ( scales * ones(1,n_candidates));
     newe_s            = sum( bisquarePsiFunction( x,c ));
     [ tmp i_mu ]      = min( abs( newe_s ));