Added all the tests

constant_vs_k.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, eps_rsm3, m_rsm3 = 300, 0.1, 5000
+nItrs_rsm3 = 200
+
+k_bounds_rsm3 = (1, 100, 5)
+
+model_rsm3 = rl.DenseNoiseModel(-2)
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, m=m_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/constant-loss-vs-k-pkl'
+outputfilename_rsm3 = 'figs/constant-loss-vs-k.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'k', k_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, m = {m}, eps = {eps}, Noise is constant +2 biased'.format(d=d_rsm3, m=m_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'k'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'k', k_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('constant-loss-vs-k')
+

constant_vs_m.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, k_rsm3, eps_rsm3 = 300, 10, 0.1 
+nItrs_rsm3 = 200
+
+m_bounds_rsm3 = (5, 350, 25)
+
+model_rsm3 = rl.DenseNoiseModel(-2)
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, k=k_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/constant-loss-vs-m-pkl'
+outputfilename_rsm3 = 'figs/constant-loss-vs-m.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'm', m_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, k = {k}, eps = {eps}, Noise is constant +2 biased'.format(d=d_rsm3, k=k_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'm'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'm', m_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('constant-loss-vs-m')
+

gd.py

@@ -5,27 +5,27 @@
 fontname='serif'
 fsize=20
 fpad=10
-d_rsm3, k_rsm3, eps_rsm3 = 1000, 40, 0.1
+d_rsm3, eps_rsm3, m_rsm3 = 300, 0.1, 5000
 nItrs_rsm3 = 200
 
-m_bounds_rsm3 = (2000, 3000, 500)
+k_bounds_rsm3 = (1, 100, 5)
 
-model_rsm3 = rl.BimodalModel()
+model_rsm3 = rl.TailFlipModel()
 
-keys_rsm3 = [rl.Oracle, rl.RME_sp_L, rl.RME_sp, rl.NP_sp, rl.RME, rl.GDAlgs]
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
 #keys_rsm3 = [rl.GDAlgs]
-model_params_rsm3 = rl.Params(d=d_rsm3, k=k_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+model_params_rsm3 = rl.Params(d=d_rsm3, m=m_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
 ylims_rsm3 = (0, 1)
 xs = []
 
-inputfilename_rsm3 = 'data/lhiding-loss-vs-m-pkl'
-outputfilename_rsm3 = 'figs/lhiding-loss-vs-m.pdf'
+inputfilename_rsm3 = 'data/tflip-loss-vs-k-pkl'
+outputfilename_rsm3 = 'figs/tflip-loss-vs-k.pdf'
 plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
-plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'm', m_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
-title_rsm3 = 'd = {d}, k = {k}, eps = {eps}, Noise is linear hiding'.format(d=d_rsm3, k=k_rsm3, eps=eps_rsm3)
-xlabel_rsm3 = 'm'
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'k', k_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, m = {m}, eps = {eps}, Noise is tail flipping'.format(d=d_rsm3, m=m_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'k'
 ylabel_rsm3 = 'L2 loss'
 plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
-plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'm', m_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
-plt.savefig('lhiding-loss-vs-m')
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'k', k_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('tflip-loss-vs-k')
 

lhiding_vs_k.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, eps_rsm3, m_rsm3 = 300, 0.1, 5000
+nItrs_rsm3 = 200
+
+k_bounds_rsm3 = (1, 200, 20)
+
+model_rsm3 = rl.BimodalModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, m=m_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/lhiding-loss-vs-k-pkl'
+outputfilename_rsm3 = 'figs/lhiding-loss-vs-k.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'k', k_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, m = {m}, eps = {eps}, Noise is linear hiding'.format(d=d_rsm3, m=m_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'k'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'k', k_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('lhiding-loss-vs-k')
+

lhiding_vs_m.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, k_rsm3, eps_rsm3 = 1000, 40, 0.1 
+nItrs_rsm3 = 200
+
+m_bounds_rsm3 = (1000, 13000, 1000)
+
+model_rsm3 = rl.BimodalModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, k=k_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/lhiding-loss-vs-m-pkl'
+outputfilename_rsm3 = 'figs/lhiding-loss-vs-m.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'm', m_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, k = {k}, eps = {eps}, Noise is linear hiding'.format(d=d_rsm3, k=k_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'm'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'm', m_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('lhiding-loss-vs-m')
+

que_vs_k.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, eps_rsm3, m_rsm3 = 300, 0.1, 5000
+nItrs_rsm3 = 200
+
+k_bounds_rsm3 = (1, 200, 20)
+
+model_rsm3 = rl.QUESyntheticModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, m=m_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/que-loss-vs-k-pkl'
+outputfilename_rsm3 = 'figs/que-loss-vs-k.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'k', k_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, m = {m}, eps = {eps}, Noise is in multiple directions'.format(d=d_rsm3, m=m_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'k'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'k', k_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('que-loss-vs-k')
+

que_vs_m.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, k_rsm3, eps_rsm3 = 1000, 40, 0.1 
+nItrs_rsm3 = 200
+
+m_bounds_rsm3 = (1000, 13000, 1000)
+
+model_rsm3 = rl.QUESyntheticModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, k=k_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/que-loss-vs-m-pkl'
+outputfilename_rsm3 = 'figs/que-loss-vs-m.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'm', m_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, k = {k}, eps = {eps}, Noise is in multiple directions'.format(d=d_rsm3, k=k_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'm'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'm', m_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('que-loss-vs-m')
+

tflip_vs_k.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, eps_rsm3, m_rsm3 = 300, 0.1, 5000
+nItrs_rsm3 = 200
+
+k_bounds_rsm3 = (1, 100, 5)
+
+model_rsm3 = rl.TailFlipModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, m=m_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/tflip-loss-vs-k-pkl'
+outputfilename_rsm3 = 'figs/tflip-loss-vs-k.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'k', k_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, m = {m}, eps = {eps}, Noise is tail flipping'.format(d=d_rsm3, m=m_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'k'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'k', k_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('tflip-loss-vs-k')
+

tflip_vs_m.py

@@ -0,0 +1,31 @@
+import robustlib as rl
+import matplotlib.pyplot as plt
+trials_rsm3 = 10
+figsize=(8,4)
+fontname='serif'
+fsize=20
+fpad=10
+d_rsm3, k_rsm3, eps_rsm3 = 10, 1, 0.1 
+nItrs_rsm3 = 200
+
+m_bounds_rsm3 = (2, 1000, 10)
+
+model_rsm3 = rl.TailFlipModel()
+
+keys_rsm3 = [rl.Oracle, rl.RME_sp, rl.NP_sp, rl.ransacGaussianMean, rl.GDAlgs]
+#keys_rsm3 = [rl.GDAlgs]
+model_params_rsm3 = rl.Params(d=d_rsm3, k=k_rsm3, eps=eps_rsm3, nItrs=nItrs_rsm3)
+ylims_rsm3 = (0, 1)
+xs = []
+
+inputfilename_rsm3 = 'data/tailflip-loss-vs-m-pkl'
+outputfilename_rsm3 = 'figs/tailflip-loss-vs-m.pdf'
+plot_m_loss_rsm3 = rl.load_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3.setdata_tofile(inputfilename_rsm3, 'm', m_bounds_rsm3, trials_rsm3, ylims_rsm3, explicit_xs=False, xs=xs)
+title_rsm3 = 'd = {d}, k = {k}, eps = {eps}, Noise is tail flipping'.format(d=d_rsm3, k=k_rsm3, eps=eps_rsm3)
+xlabel_rsm3 = 'm'
+ylabel_rsm3 = 'L2 loss'
+plot_m_loss_rsm3_plt = rl.plot_data(model_rsm3, model_params_rsm3, rl.err, keys_rsm3)
+plot_m_loss_rsm3_plt.plotxy_fromfile(outputfilename_rsm3, inputfilename_rsm3, 'm', m_bounds_rsm3, (0, 1), title_rsm3, xlabel_rsm3, ylabel_rsm3, figsize=figsize, fsize=fsize, fpad=fpad, fontname=fontname, explicit_xs = False, xs = xs)
+plt.savefig('tailfip-loss-vs-m')
+