drawingBasis.m

function basis = drawingBasis(snips, kin)
basis = cell(size(kin));
for day=1:length(kin)
    basis{day} = struct();
    dim = size(kin{day}.velX);
    
    % These are the nonlinear variables we will use to construct a linear basis
    % set
    illusion = repmat(snips{day}.is_illusion,1,dim(2));
    illusion = illusion(:);
%     ramp = repmat((0:dim(2)-1)/(dim(2)-1),dim(1),1);
%     ramp = ramp(:);
    cycle = repmat([nan 1:5 nan],dim(2)/7,1);
    cycle = repmat(cycle(:)',dim(1),1);
    cycle = cycle(:);
    
    % We are going to create some basis functions that will allow us to fit
    % position, velocity, curvature, whatever
    D = [kin{day}.velX(:) kin{day}.velY(:)];
    D = bsxfun(@rdivide,D,sqrt(sum(D.^2,2)));
    data = [bsxfun(@times,D,cycle==1),...
            bsxfun(@times,D,cycle==2),...
            bsxfun(@times,D,cycle==3),...
            bsxfun(@times,D,cycle==4),...
            bsxfun(@times,D,cycle==5)];
    data = [data bsxfun(@times,data,illusion)]; %#ok<AGROW>
    
    for b=1:size(data,2)
        basis{day}.(sprintf('basis%0.2d',b)) = reshape(data(:,b),dim);
    end
end