Following code reproduces the Figure 1 (EM and MM minorizes the log-likelihood) in manuscript.
versioninfo()
Read in the data points:
L = readdlm("logL.txt", Float64)
Qem = readdlm("g_EM.txt", Float64)
Gmm = readdlm("g_MM.txt", Float64);
using PyPlot, LaTeXStrings
σ1grid = 17.7:0.01:19
σ2grid = 0.65:0.01:0.8
plot_surface(σ1grid, σ2grid, L, color=:black, alpha=0.75)
plot_surface(σ1grid, σ2grid, Gmm, color=:black, alpha=0.50)
plot_surface(σ1grid, σ2grid, Qem, color=:gray, alpha=0.25)
xlabel(L"\sigma_1^2", fontsize=12)
ylabel(L"\sigma_2^2", fontsize=12)
xticks([18, 18.5, 19])
yticks([0.7, 0.75, 0.8])
zticks([-522, -521.5, -521, -520.5])
annotate("logL", xy=[0.08; 0.47], xycoords="axes fraction", fontsize=15)
annotate("MM", xy=[0.09; 0.39], xycoords="axes fraction", fontsize=15)
annotate("EM", xy=[0.11; 0.23], xycoords="axes fraction", fontsize=15)
savefig("MMvsEM.svg")
savefig("MMvsEM.png")