ax.set_yticks([])
ax.set_title('Karlsson & Frank, Nat. Neurosci. (2009)', pad=8)
sb.despine(ax=ax, left=True, bottom=True)
ax.set_aspect('auto')
ax = axes[2][1]
ax.imshow(plt.imread(os.path.join(figdir, fname_Suh2013)))
ax.set_xticklabels([])
ax.set_yticklabels([])
ax.set_xticks([])
ax.set_yticks([])
ax.set_title('Suh et al., Neuron (2013)', pad=10)
sb.despine(ax=ax, left=True, bottom=True)
x = -0.25
y = 1.2
label_panel(ax=axes[0][0], label='A', x=x, y=y + 0.1)
label_panel(ax=axes[0][1], label='B', x=x - 0.12, y=y + 0.1)
label_panel(ax=axes[1][0], label='C', x=x, y=y)
label_panel(ax=axes[1][1], label='D', x=x + 0.1, y=y)
label_panel(ax=axes[2][0], label='E', x=x, y=y + .07)
label_panel(ax=axes[2][1], label='F', x=x + 0.1, y=y + .1)
fig.subplots_adjust(left=.01,
bottom=.1,
right=.99,
top=.9,
wspace=0.5,
hspace=0.6)
fig.set_size_inches(width, height)
fname_base = 'FIGURE_8'
PROPd.etO[0, :] = prop_density_da # hack
PROPd.plot_prop_kernels(first_state=0,
cmap=cmap_prop,
autoprop_off=False,
n=1,
vmin=0,
vmax=vmax_prop)
for start in states_away:
EXPdh.plot_trajectory(state_seq=[start],
plot_env=False,
state_func_env=True)
ax15.set_title('diffusive propagator [away]')
x = -0.2
y = 1.2
label_panel(ax0, 'A', x, y)
label_panel(ax1, 'B', x, y)
label_panel(ax2, 'C', x, y)
label_panel(ax3, 'D', x, y)
label_panel(ax4, 'E', x, y)
label_panel(ax5, 'F', x, y)
label_panel(ax6, 'G', x, y)
label_panel(ax7, 'H', x, y)
label_panel(ax8, 'I', x, y)
label_panel(ax9, 'J', x, y)
label_panel(ax10, 'K', x, y)
label_panel(ax11, 'L', x, y)
label_panel(ax12, 'M', x, y)
labelbottom=False,
labeltop=False,
labelleft=False,
labelright=False)
sb.despine(ax=ax2, top=True, right=True, left=True, bottom=True)
sb.despine(ax=ax3, top=True, right=True, left=True, bottom=True)
sb.despine(ax=ax4, top=True, right=True, left=True, bottom=True)
sb.despine(ax=ax5, top=True, right=True, left=True, bottom=True)
ax2.set_title('true SR')
ax3.set_title('learned SR (superdiffusion)')
ax4.set_title('learned SR (diffusion)')
ax5.set_title('learned SR (min-autocorr)')
x = -0.25
y = 1.3
label_panel(ax0, label='A', x=x + 0.25, y=y)
label_panel(ax1, label='B', x=x, y=y)
label_panel(ax2, label='C', x=x, y=y)
label_panel(ax3, label='D', x=x + 0.14, y=y)
label_panel(ax4, label='E', x=x - 0.14, y=y)
label_panel(ax5, label='F', x=x - 0.09, y=y)
fig.subplots_adjust(left=.01,
bottom=.1,
right=.99,
top=.9,
wspace=0.6,
hspace=0.8)
fig.set_size_inches(width, height)
if save_output:
ax11.set_title('Pfeiffer & Foster, Science (2015)', pad=9)
ax11.text(x=90, y=460, s='empirical data', color='black', fontsize=10)
ax11.text(x=35, y=500, s='even-step prediction', color='red', fontsize=10)
ax11.add_patch(
patches.Rectangle((5, 420),
width=400,
height=100,
fill=False,
ec='black',
clip_on=False))
# tweak panels
sb.despine(fig, top=True, right=True)
x = -0.3
y = 1.2
label_panel(ax0, 'A', x, 1.2)
label_panel(ax1, 'B', x, 1.2 + 0.2)
label_panel(ax2, 'C', x, 1.2)
label_panel(ax3, 'D', x, 1.2)
label_panel(ax4, 'E', x - 0.1, 1.3)
label_panel(ax5, 'F', x - 0.1, 1.3)
label_panel(ax6, 'G', x - 0.1, 1.3)
label_panel(ax7, 'H', x - 0.1, 1.3)
label_panel(ax8, 'I', x - .05, 1.3)
label_panel(ax9, 'J', x, 1.3)
label_panel(ax10, 'K', x, 1.3)
label_panel(ax11, 'L', x, 1.3)
fig.subplots_adjust(left=.01,
bottom=.1,
right=.99,
across_samp=True)
EXP_sdiff_base.set_target_axis(ax=ax4)
EXP_sdiff_base.plot_coverage(color=color_superdiff,
func_of_time=False,
across_samp=True)
ax4.set_ylabel('fraction of env. visited', labelpad=5)
ax4.set_title('exploration efficiency (single)', pad=10)
ax4.set_xlabel('avg. distance traversed')
ax4.set_xlim([0, 300])
ax4.set_ylim([0, 0.12])
ax4.text(x=20, y=0.1, s='diffusion', color=color_diff)
ax4.text(x=20, y=0.09, s='superdiffusion', color=color_superdiff)
x = -0.35
y = 1.2
label_panel(ax0, 'A', x, y)
label_panel(ax1, 'B', x, y)
label_panel(ax2, 'C', x, y)
label_panel(ax3, 'D', x, y)
label_panel(ax4, 'E', x, y)
label_panel(ax5, 'F', x, y)
fig.subplots_adjust(left=.01,
bottom=.1,
right=.99,
top=.9,
wspace=0.7,
hspace=0.6)
fig.set_size_inches(width, height)
if save_output:
save_figure(fig=fig, figdir=figdir, fname_base=fname_base, file_ext='.png')
pos_initial[1],
'S',
ha="center",
va="center",
color="k")
# dominant eigenvector
ax2 = axes[0][2]
GEN.set_target_axis(ax2)
GEN.plot_real_eigenvectors(start=1, n=1)
ax2.set_title('dominant spectral \n component (DSC)', pad=1)
# LABELS, FINESSE AND SAVE
x = -0.25
y = 1.3
label_panel(axes[0][0], 'A', x=x + 0.1, y=y)
label_panel(axes[0][1], 'B', x=x - 0.1, y=y)
label_panel(axes[0][2], 'C', x=x, y=y)
label_panel(axes[0][3], 'D', x=x, y=y)
label_panel(axes[1][0], 'E', x=x + 0.1, y=y - 0.1)
label_panel(axes[1][1], 'F', x=x - 0.1, y=y - 0.1)
label_panel(axes[1][2], 'G', x=x, y=y - 0.1)
label_panel(axes[1][3], 'H', x=x, y=y - 0.1)
fig.subplots_adjust(left=.01,
bottom=.1,
right=.99,
top=.9,
wspace=0.6,
hspace=0.6)
fig.set_size_inches(width, height)
sb.lineplot(data=df_spec_rel[(df_spec.alpha==alpha_base)&(df_spec.tau==tau_shift)], x=x_index, y='gain', hue='alpha', size='tau', sizes=sizes, palette=cmap_spec, ax=ax5, legend=legend, marker='o', markersize=sizes[tau_shift]*2, markeredgewidth=0., markerfacecolor=color_diff, markeredgecolor=color_diff, clip_on=False, zorder=100)
sb.lineplot(data=df_spec_rel[(df_spec.alpha==alpha_shift)&(df_spec.tau==tau_shift)], x=x_index, y='gain', hue='alpha', size='tau', sizes=sizes, palette=cmap_spec, ax=ax5, legend=legend, marker='o', markersize=sizes[tau_shift]*2, markeredgewidth=0., markerfacecolor=color_superdiff, markeredgecolor=color_superdiff, clip_on=False, zorder=100)
sb.lineplot(data=df_spec_rel[(df_spec.alpha==alpha_shift)&(df_spec.tau==tau_base)], x=x_index, y='gain', hue='alpha', size='tau', sizes=sizes, palette=cmap_spec, ax=ax5, legend=legend, marker='o', markersize=sizes[tau_base]*2, markeredgewidth=0., markerfacecolor=color_superdiff, markeredgecolor=color_superdiff, clip_on=False, zorder=100)
ax5.set_ylim([0,0.3])
if x_index == 'eval':
ax5.set_xlabel('eigenvalue $ \lambda$')
ax5.set_xlim([-4,0])
else:
ax5.set_xlabel('spatial wavelength')
ax5.set_ylabel(r'$s_{\alpha,\tau}(\lambda) / s_{1,1}(\lambda)$ [normalized]')
ax5.set_title('relative power spectrum')
sb.despine(fig, top=True, right=True)
# label_panels(axes)
x = -0.3
y = 1.2
label_panel(axes[0][0], 'A', x, y)
label_panel(axes[0][1], 'B', x, y)
label_panel(axes[0][2], 'C', x, y)
label_panel(axes[1][0], 'D', x, y)
label_panel(axes[1][1], 'E', x, y)
label_panel(axes[1][2], 'F', x, y)
fig.subplots_adjust(left=.01, bottom=.1, right=.99, top=.9, wspace=0.6, hspace=0.6)
fig.set_size_inches(width, height)
if save_output:
save_figure(fig=fig, figdir=figdir, fname_base=fname_base, file_ext='.png')
save_figure(fig=fig, figdir=figdir, fname_base=fname_base, file_ext='.pdf')
# SPECTRAL EMBEDDING ax4 = axes[2][0] ax5 = axes[2][1] ax6 = axes[2][2] ax7 = axes[2][3] LRNd.mds_dyn_mat(ax=ax4, type="SR", learned=False) LRNd.mds_dyn_mat(ax=ax6, type="SR") LRNs.mds_dyn_mat(ax=ax5, type="SR") LRNo.mds_dyn_mat(ax=ax7, type="SR") x = -0.25 y = 1.3 label_panel(axes[0][0], label="A", x=x + 0.25, y=y) label_panel(axes[0][1], label="B", x=x, y=y) label_panel(axes[0][2], label="C", x=x, y=y) label_panel(axes[0][3], label="D", x=x, y=y) label_panel(axes[1][0], label="E", x=x + 0.05, y=y) label_panel(axes[1][1], label="F", x=x - 0.38, y=y) label_panel(axes[1][2], label="G", x=x - 0.39, y=y) label_panel(axes[1][3], label="H", x=x - 0.4, y=y) label_panel(axes[2][0], label="I", x=x + 0.23, y=y) label_panel(axes[2][1], label="J", x=x, y=y) label_panel(axes[2][2], label="K", x=x, y=y) label_panel(axes[2][3], label="L", x=x, y=y) fig.subplots_adjust(left=0.01, bottom=0.1, right=0.99, top=0.9, wspace=0.6, hspace=0.8)