Nested tiledlayout: can parent tiles have different widths depending on number of child tiles?

Hello,

I am trying to plot several vertical profiles grouped into categories in one figure. Each category contains a different number of profiles. My goal is for the width of each category block to scale with the number of profiles it contains, so that each individual profile has the same width across the whole figure.

My code:

% outer layout
out = tiledlayout(1, length(groups), 'TileSpacing', 'compact', 'Padding', 'compact');

Unrecognized function or variable 'groups'.

for i = 1:length(groups)
    idx = groups{i};
    n = length(idx);
    % inner layout for this group
    inn = tiledlayout(out, 1, n, 'TileSpacing', 'compact', 'Padding', 'compact');
    inn.Layout.Tile = i;
    title(inn, group_titles{i}, 'FontSize', 12, 'FontWeight', 'bold');
    for j = 1:n
        k = idx(j);
        ax = nexttile(inn);
        h = imagesc([0 1], depth_vec, salinity_mat(:, k));
        clim([cmin cmax])
        h.AlphaData = ~isnan(salinity_mat(:, k));
        title(ax, order{k})
        if k == 1
            ylabel('Depth (m)')
        else
            set(ax, 'YTickLabel', [])
        end
        set(ax, 'XTick', [])
    end
end

[Matt J]

imshow(imread('Figure_1.png'), Border='tight')

Answer 1

[Matt J]

It might be better to us e a grouped bar plot, e.g.,

Y = [5 3 7 0 0;
     6 4 2 1 0;
     4 8 5 1 6];

gradientGroupedBar(Y)

function gradientGroupedBar(Y)

% Y: [nGroups x nBarsPerGroup]

[nGroups, nBars] = size(Y);

% Basic bar geometry
groupWidth = min(0.8, nBars/(nBars+1.5));

figure; hold on;

for iBar = 1:nBars
    % Compute x positions for this bar within each group
    x = (1:nGroups) - groupWidth/2 + ...
        (2*iBar-1) * groupWidth / (2*nBars);

    for iGroup = 1:nGroups
        % Bar corners
        xL = x(iGroup) - groupWidth/(2*nBars);
        xR = x(iGroup) + groupWidth/(2*nBars);
        yB = 0;
        yT = Y(iGroup,iBar);

        N = 1000;  % controls smoothness
        
        yVals = linspace(yB, yT, N)';
        
        X = [xL*ones(N,1), xR*ones(N,1)];
        Ysurf = [yVals, yVals];
        
        % Smooth gradient values
        C = repmat(linspace(0,1,N)', 1, 2);

        % Draw surface
        surf(X, Ysurf, zeros(N,2), C, ...
            'EdgeColor','none', ...
            'FaceColor','interp');
    end
end

colormap(parula);   % choose your gradient
colorbar;

xlim([0.5 nGroups+0.5]);
view(2);            % 2D view
box on;

end

Answer 2

[Stephen23]

I don't think there is an automatic way to do this, but you could specify the TileSpan to achieve (approximately) equal sizes.

rd0 = {{rand(7,3),rand(7,3)},{rand(7,3),rand(7,3),rand(7,3)}}; % fake data
vd0 = cellfun(@numel,rd0); % count images
nd0 = sum(vd0); %  total number of images
cs0 = cumsum([1,vd0]); % cummulative count
tl0 = tiledlayout(1,nd0, 'TileSpacing','compact', 'Padding','compact');
for ii = 1:numel(rd0)
    rd1 = rd0{ii};
    nd1 = numel(rd1);
    tl1 = tiledlayout(tl0, 1,nd1, 'TileSpacing','compact', 'Padding','compact');
    tl1.Layout.Tile = cs0(ii); % first tile
    tl1.Layout.TileSpan = [1,vd0(ii)]; % span
    title(tl1, "Title of "+vd0(ii)+" Images", 'FontSize',12, 'FontWeight','bold');
    for jj = 1:nd1
        ax = nexttile(tl1);
        imagesc(ax,rd1{jj});
        title(ax, "T"+ii+" Im"+jj)
    end
end