Updates to the OptimizerFactory and the MatlabInterface. Better access to EvAClient.

resources/MatlabInterface/@JEInterface/makeOptions.m

@@ -0,0 +1,65 @@
+function opts = makeOptions(int, varargin)
+% Create a JEInterface options set from scratch. Possible fields are:
+%'Display';
+%'MaxFunEvals';'MaxIter';'TolFun';'TolFunEvals';TolX';'TolXEvals', where
+% all but 'TolFunEvals', 'TolXEvals' are used similar to the optimset. 
+% The latter two are interpreted as the numbers of evaluations required
+% to assume convergence. Default values are TolXEvals=TolFunEvals=200,
+% TolX=TolFun=1e-4, MaxFunEvals uses a default from EvA2.
+% Notice that this method creates a parameter set but does not assign it
+% to the interface instance. Use setOptions to do that.
+
+allfields = {'Display'; 'MaxFunEvals';'MaxIter';'TolFun';'TolFunEvals';...
+    'TolX';'TolXEvals'};
+
+specialfields={'TolFunEvals', 'TolXEvals'};
+
+nvararg=nargin-1;
+if rem(nvararg,2)==1
+    error('Pass options in name-value pairs!');
+end
+
+% create cell array
+structinput = cell(2,length(allfields));
+% fields go in first row
+structinput(1,:) = allfields';
+% []'s go in second row
+structinput(2,:) = {[]};
+% turn it into correctly ordered comma separated list and call struct
+opts = struct(structinput{:});
+
+stdSet=optimset();
+
+% standard options:
+opts.('MaxFunEvals') = eva2.OptimizerFactory.getDefaultFitCalls;
+opts.('TolX') = 1e-4;
+opts.('TolXEvals') = 200;
+opts.('TolFun') = 1e-4;
+opts.('TolFunEvals') = 200;
+
+for i=1:nvararg/2
+    name=varargin{2*i-1};
+    value=varargin{(2*i)};
+    % parse arguments
+    if ~ischar(name)
+        error('Expected char parameter name at index %d!', 2*i+1);
+    else
+        optIndex=strmatch(name,allfields, 'exact');
+        if isempty(optIndex)
+            error('Unknown option %s !', name);
+        else
+            if ~isempty(strmatch(name, specialfields,'exact'))
+                % test for integer
+                if (~isscalar(value) || ~isnumeric(value) || round(value)<1)
+                    error('invalid value type for %s, expecting numeric scalar > 1!', name);
+                end
+                value=round(value);
+            else
+                % test using optimset
+                optimset(stdSet, name, value);
+            end
+            % assign to struct
+            opts.(allfields{optIndex,:}) = value;
+        end
+    end
+end