Scalar2IceCubeTray.m

function out = Scalar2IceCubeTray(scalars,traywidth,emptyvalue,fullvalue, inverseFlag)
%
% helper function to map (column of) single scalars between low and fullvalue from left to
% right with integer and decimal parts such that low = [0 0 ...] and fullvalue = [1 1 ...], 
% e.g 2.5 -> [1 1 0.5 0 0] 
% if inverseFlag is true then we go from vectors to numerical scalars.
% there's probably an easier way but i don't know it.
% note that it doesn't validate that input in correct formats.

if nargin < 5; inverseFlag = false; end;

maxthreshold = 0.7;
minthreshold = 0.1;
range = fullvalue - emptyvalue;
[Nrows Ncols] = size(scalars);
unitcount = ones(Nrows, Ncols);
idx=1:traywidth;                

if inverseFlag 
    [R C] = size(scalars);
    %int part is just sum of 1's per row
    filledcells = scalars > maxthreshold;
    intpart = sum(filledcells,2);
    for rw = 1:R
        if intpart(rw) == traywidth
            %decimal part is zero
            decpart(rw) = 0;
        else
            %decimal part found in next column along
            decpart(rw) = scalars(rw, intpart(rw)+1);
            %but round down small values
            if decpart(rw) < minthreshold
                decpart(rw) = 0;
            end
        end
    end
    out = range .* (intpart+decpart')/traywidth + emptyvalue;
else
    allidx=repmat(idx,Nrows,1);
    scaledvalues = traywidth*(scalars - emptyvalue)/range;
    intpart = floor(scaledvalues)';                  %integer part
    decpart = scaledvalues' - intpart;               %decimal part
    repscaledvalues = repmat(intpart',1,traywidth);
    out = allidx<=repscaledvalues; 
    out = 1.0 * out;                               %make sure it's array of reals
    %special cases where intpart = full array of ones to avoid subscript error
    decpart(intpart==traywidth) =1;
    intpart(intpart==traywidth) = traywidth -1;
    inds = sub2ind(size(out), (1:Nrows),intpart + 1); %find where to put dec vals
    out(inds) = decpart;
end