Merge pull request #104 from tristangdwl/kern_shift_ops

Added shifted_eval support to kernel.plus

Author: mrachh

chunkie/@kernel/plus.m

@@ -8,12 +8,28 @@
   assert(f.opdims(1) == g.opdims(1) && f.opdims(2) == g.opdims(2), ...
       'kernel dimensions must agree to add');
   f.name = ['custom ',f.name,' ',g.name];
+
+  if(isa(f.shifted_eval, 'function_handle'))
+    if(isa(g.shifted_eval, 'function_handle'))
+        f.shifted_eval = @(varargin) f.shifted_eval(varargin{:})+g.shifted_eval(varargin{:});
+    else
+        f.shifted_eval = @(varargin) f.shifted_eval(varargin{:})+g.eval(varargin{1:2});
+    end
+  else
+    if(isa(g.shifted_eval, 'function_handle'))
+        f.shifted_eval = @(varargin) f.eval(varargin{1:2})+g.shifted_eval(varargin{:});
+    else
+        f.shifted_eval = [];
+    end
+  end
+  
   f.eval = @(varargin) g.eval(varargin{:}) + f.eval(varargin{:});
   if (isa(g.fmm,'function_handle') && isa(f.fmm,'function_handle'))
     f.fmm = @(varargin) g.fmm(varargin{:}) + f.fmm(varargin{:});
   else
     f.fmm = [];
   end
+
   if or(f.isnan,g.isnan)
       f = kernel.nans(f.opdims(1),f.opdims(2));
   end