Added some print statements to more transparently report progress.

figure_3.py

@@ -66,7 +66,7 @@ def _plot_all(xs, ns=[2**12], s_gs=[1.0], d_Zs=[3], d_Xs=[3], sigma_Xs=[1.0], si
   plt.errorbar(xs, _mse(oracle_errors),     _ste(oracle_errors),              marker='s', markersize=10, lw=2, label='Oracle')
   plt.errorbar(xs, _mse(oracle_cv_errors),  _ste(oracle_cv_errors),  ls='--', marker='s', markersize=10, lw=2, label='Oracle CV')
 
-# Plot performance over n
+print('Generating Panel (a): performance over sample size n...')
 plt.subplot(2, 3, 1)
 ns = np.logspace(4, 12, 9, base=2., dtype=int)
 _plot_all(ns=ns, xs=ns)
@@ -76,21 +76,21 @@ plt.xlabel('Sample size ($n$)')
 plt.ylabel('Mean Squared Error')
 plt.legend()
 
-# Plot performance over d_Z
+print('Generating Panel (b): performance over intrinsic dimention d_Z...')
 plt.subplot(2, 3, 2)
 d_Zs = range(1, 11)
 _plot_all(d_Zs=d_Zs, xs=d_Zs)
 plt.gca().set_yscale('log')
 plt.xlabel('Intrinsic Dimension ($d_Z$)')
 
-# Plot performance over sigma_X
+print('Generating Panel (c): performance over covariate noise level sigma_X...')
 plt.subplot(2, 3, 3)
 sigma_Xs = np.logspace(-1, 1, 10)
 _plot_all(sigma_Xs=sigma_Xs, xs=sigma_Xs)
 plt.loglog()
 plt.xlabel('Covariate Noise Level ($\sigma_X$)')
 
-# Plot performance over s_g
+print('Generating Panel (d): performance over smoothness s_g of g...')
 plt.subplot(2, 3, 4)
 s_gs = np.linspace(0.1, 1.0, 10)
 _plot_all(s_gs=s_gs, xs=s_gs)
@@ -98,14 +98,14 @@ plt.gca().set_yscale('log')
 plt.xlabel('Smoothness ($s_g$) of $g$')
 plt.ylabel('Mean Squared Error')
 
-# Plot performance over d_X
+print('Generating Panel (e): performance over covariate noise dimention d_X...')
 plt.subplot(2, 3, 5)
 d_Xs = range(1, 11)
 _plot_all(d_Xs=d_Xs, xs=d_Xs)
 plt.gca().set_yscale('log')
 plt.xlabel('Covariate Noise Dimension ($d_X$)')
 
-# Plot performance over sigma_Y
+print('Generating Panel (f): performance over response noise level sigma_Y...')
 plt.subplot(2, 3, 6)
 sigma_Ys = np.logspace(-1, 1, 10)
 _plot_all(sigma_Ys=sigma_Ys, xs=sigma_Ys)