legend=legend,
xlimits=xlimits,
ylimits=ylimits,
alpha=1.)
# plot_scatter(ax, samps=z ,xlimits=xlimits,ylimits=ylimits)
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Blues')
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Greens')
mean, logvar = model.q_dist.get_mean_logvar(samp_torch)
func = lambda zs: lognormal4(torch.Tensor(zs),
torch.squeeze(mean.data.cpu()),
torch.squeeze(logvar.data.cpu()))
plot_isocontours(ax,
func,
cmap='Greens',
xlimits=xlimits,
ylimits=ylimits)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues', alpha=.3,xlimits=xlimits,ylimits=ylimits)
# #Plot prob
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# Ws, logpW, logqW = model.sample_W() #_ , [1], [1]
# # func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
# # plot_isocontours(ax, func, cmap='Reds')
# # func = lambda zs: log_bernoulli(model.decode(Ws, Variable(torch.unsqueeze(zs,1))), Variable(torch.unsqueeze(samp,0)))+ Variable(torch.unsqueeze(lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2)), 1))
# plot_isocontours2_exp_norm_logspace(ax, func, cmap='Greens', legend=legend)
# if samp_i==0: ax.annotate('p(z|x,W1)', xytext=(.1, 1.1), xy=(0, 1), textcoords='axes fraction')
# models = []
# model_names = []
alpha=.2
rows = len(posteriors)
columns = len(models) +1 #+1 for posteriors
fig = plt.figure(figsize=(6+columns,4+rows), facecolor='white')
for p_i in range(len(posteriors)):
print '\nPosterior', p_i, posterior_names[p_i]
posterior = ttp.posterior_class(posteriors[p_i])
ax = plt.subplot2grid((rows,columns), (p_i,0), frameon=False)#, colspan=3)
plot_isocontours(ax, posterior.run_log_post, cmap='Blues')
if p_i == 0: ax.annotate('Posterior', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
for q_i in range(len(models)):
print model_names[q_i]
ax = plt.subplot2grid((rows,columns), (p_i,q_i+1), frameon=False)#, colspan=3)
model = models[q_i](posteriors[p_i])
# model.train(10000, save_to=home+'/Documents/tmp/vars.ckpt')
model.train(9999000, save_to='')
samps = model.sample(1000)
plot_kde(ax, samps, cmap='Reds')
plot_isocontours(ax, posterior.run_log_post, cmap='Blues', alpha=alpha)
if p_i == 0: ax.annotate(model_names[q_i], xytext=(.38, 1.1), xy=(0, 1), textcoords='axes fraction')
# plt.show()
#Plot prob
row +=1
ax = plt.subplot2grid((rows,cols), (row, samp_i+1), frameon=False)
func = lambda zs: model.logposterior_func(samp_torch,zs)
# plot_isocontours2_exp_norm(ax, func, cmap='Greys', legend=legend,xlimits=xlimits,ylimits=ylimits,alpha=.2)
plot_isocontours2_exp_norm(ax, func, cmap='Blues', legend=legend,xlimits=xlimits,ylimits=ylimits,alpha=1.)
# plot_scatter(ax, samps=z ,xlimits=xlimits,ylimits=ylimits)
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Blues')
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Greens')
mean, logvar = model.q_dist.get_mean_logvar(samp_torch)
func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data.cpu()), torch.squeeze(logvar.data.cpu()))
plot_isocontours(ax, func, cmap='Greens',xlimits=xlimits,ylimits=ylimits)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues', alpha=.3,xlimits=xlimits,ylimits=ylimits)
# #Plot prob
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# Ws, logpW, logqW = model.sample_W() #_ , [1], [1]
# # func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
# # plot_isocontours(ax, func, cmap='Reds')
# #Plot prior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues')
# if samp_i==0: ax.annotate('Prior p(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
#Plot q
col +=1
val = 3
ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
mean, logvar = model.encode(Variable(torch.unsqueeze(samp,0)))
func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
plot_isocontours(ax, func, cmap='Reds', xlimits=[-val, val], ylimits=[-val, val])
if samp_i==0: ax.annotate('p(z)\nq(z|x)\np(z|x)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
plot_isocontours(ax, func, cmap='Blues', alpha=.3, xlimits=[-val, val], ylimits=[-val, val])
Ws, logpW, logqW = model.sample_W() #_ , [1], [1]
func = lambda zs: log_bernoulli(model.decode(Ws, Variable(torch.unsqueeze(zs,1))), Variable(torch.unsqueeze(samp,0)))+ Variable(torch.unsqueeze(lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2)), 1))
plot_isocontours2_exp_norm(ax, func, cmap='Greens', legend=legend, xlimits=[-val, val], ylimits=[-val, val])
# #Plot logprior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontoursNoExp(ax, func, cmap='Blues', legend=legend)
# if samp_i==0: ax.annotate('Prior\nlogp(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
xlimits=xlimits,
ylimits=ylimits) #,alpha=.2)
# plot_isocontours2_exp_norm(ax, func, cmap='Blues', legend=legend,xlimits=xlimits,ylimits=ylimits,alpha=1.)
# plot_scatter(ax, samps=z ,xlimits=xlimits,ylimits=ylimits)
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Blues')
# plot_kde(ax,samps=z,xlimits=xlimits,ylimits=ylimits,cmap='Greens')
mean, logvar = model.q_dist.get_mean_logvar(samp_torch)
func = lambda zs: lognormal4(torch.Tensor(zs),
torch.squeeze(mean.data.cpu()),
torch.squeeze(logvar.data.cpu()))
# plot_isocontours(ax, func, cmap='Greens',xlimits=xlimits,ylimits=ylimits)
plot_isocontours(ax,
func,
cmap='Blues',
xlimits=xlimits,
ylimits=ylimits)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues', alpha=.3,xlimits=xlimits,ylimits=ylimits)
# #Plot prob
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# Ws, logpW, logqW = model.sample_W() #_ , [1], [1]
# # func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
# # plot_isocontours(ax, func, cmap='Reds')
# # func = lambda zs: log_bernoulli(model.decode(Ws, Variable(torch.unsqueeze(zs,1))), Variable(torch.unsqueeze(samp,0)))+ Variable(torch.unsqueeze(lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2)), 1))
# plot_isocontours2_exp_norm_logspace(ax, func, cmap='Greens', legend=legend)
# if samp_i==0: ax.annotate('p(z|x,W1)', xytext=(.1, 1.1), xy=(0, 1), textcoords='axes fraction')
fig = plt.figure(figsize=(5 + columns, 3 + rows), facecolor='white')
# for p_i in range(len(disrtibutions)):
#######################
#ROW 1
p_i = 0
######
#COL 1
ax = plt.subplot2grid((rows, columns), (p_i, 0),
frameon=False) #, colspan=3)
#Plot distribution
plot_isocontours(ax, dist.run_log_post, cmap='Blues')
if p_i == 0:
ax.annotate('p(X)',
xytext=(.4, 1.1),
xy=(0, 1),
textcoords='axes fraction')
######
#COL 2
#Plot samples from distribution
ax = plt.subplot2grid((rows, columns), (p_i, 1),
frameon=False) #, colspan=3)
if p_i == 0:
ax.annotate('Dataset Samples',
xytext=(.2, 1.1),
if samp_i==0: ax.annotate('Sample', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
# #Plot prior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues')
# if samp_i==0: ax.annotate('Prior p(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
#Plot q
col +=1
ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
mean, logvar = model.encode(Variable(torch.unsqueeze(samp,0)))
func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
plot_isocontours(ax, func, cmap='Reds',xlimits=xlimits,ylimits=ylimits)
if samp_i==0: ax.annotate('p(z)\nq(z|x)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
plot_isocontours(ax, func, cmap='Blues', alpha=.3,xlimits=xlimits,ylimits=ylimits)
# #Plot logprior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontoursNoExp(ax, func, cmap='Blues', legend=legend)
# if samp_i==0: ax.annotate('Prior\nlogp(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
# #Plot logq
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
if samp_i==0: ax.annotate('Sample', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
# #Plot prior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontours(ax, func, cmap='Blues')
# if samp_i==0: ax.annotate('Prior p(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
#Plot q
col +=1
ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
mean, logvar = model.encode(Variable(torch.unsqueeze(samp,0)))
func = lambda zs: lognormal4(torch.Tensor(zs), torch.squeeze(mean.data), torch.squeeze(logvar.data))
plot_isocontours(ax, func, cmap='Reds')
if samp_i==0: ax.annotate('p(z)\nq(z|x)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
plot_isocontours(ax, func, cmap='Blues', alpha=.3)
# #Plot logprior
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)
# func = lambda zs: lognormal4(torch.Tensor(zs), torch.zeros(2), torch.zeros(2))
# plot_isocontoursNoExp(ax, func, cmap='Blues', legend=legend)
# if samp_i==0: ax.annotate('Prior\nlogp(z)', xytext=(.3, 1.1), xy=(0, 1), textcoords='axes fraction')
# #Plot logq
# col +=1
# ax = plt.subplot2grid((rows,cols), (samp_i,col), frameon=False)