def plot_comp_sims(figure_path, figsize=(2.5, 2.5)):
non_complexy_results = restore_results('../data/sim_comp_cost/non_complexy')
con_complexy_results = restore_results('../data/sim_comp_cost/con_complexy')
fig, axis = plt.subplots(1, 1, figsize=figsize)
axis.plot(range(1,65), non_complexy_results.mean(axis=1), color=colors.categories[0], label='Random')
axis.plot(range(1,65), con_complexy_results.mean(axis=1), color=colors.categories[2], label='Convex')
axis.annotate ('', (16, non_complexy_results[15].mean()), (16, con_complexy_results[15].mean()), arrowprops={'arrowstyle':'<->', 'shrinkA':0, 'shrinkB':0})
axis.set_ylim(0-(700*0.025), 700+(700*0.025))
axis.set_xlim(0,65)
axis.set_xticks([1, 16, 32, 48, 64])
axis.set_xlabel('Number of categories', fontsize=8)
axis.set_yticks([0, 100, 200, 300, 400, 500, 600, 700])
axis.set_ylabel('Complexity (bits)', fontsize=8)
for tick in axis.yaxis.get_major_ticks():
tick.label.set_fontsize(7)
for tick in axis.xaxis.get_major_ticks():
tick.label.set_fontsize(7)
handles, labels = axis.get_legend_handles_labels()
fig.legend(handles, labels, loc='lower center', ncol=2, frameon=False, bbox_to_anchor=(0.5, -0.04), fontsize=8)
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5, rect=[0,0.08,1,1])
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot_all(figure_path):
cost_rand = restore_results('../data/sim_comp_cost/non_commcost')
comp_rand = restore_results('../data/sim_comp_cost/non_complexy')
cost_conv = restore_results('../data/sim_comp_cost/con_commcost')
comp_conv = restore_results('../data/sim_comp_cost/con_complexy')
fig, axis = plt.subplots(1, 1, figsize=(2.5,2.5))
axis.scatter([comp_rand[0,0]], [cost_rand[0,0]], c='#E6CFDA', s=2.0)
for i in range(1, 63):
x, y = reduce_points(comp_rand[i], cost_rand[i])
axis.scatter(x, y, c=colors.fake_alpha(colors.categories[0], (i+32)/95.), s=2.0)
for i in range(1, 63):
x, y = reduce_points(comp_conv[i], cost_conv[i])
axis.scatter(x, y, c=colors.fake_alpha(colors.categories[2], (i+32)/95.), s=2.0)
axis.scatter(comp_rand[63,0], cost_rand[63,0], c=colors.categories[0], s=2.0, label='Random')
axis.scatter(comp_conv[63,0], cost_conv[63,0], c=colors.categories[2], s=2.0, label='Convex')
axis.set_xlabel('Complexity (bits)', fontsize=8)
axis.set_ylabel('Communicative cost (bits)', fontsize=8)
for tick in axis.yaxis.get_major_ticks():
tick.label.set_fontsize(7)
for tick in axis.xaxis.get_major_ticks():
tick.label.set_fontsize(7)
handles, labels = axis.get_legend_handles_labels()
fig.legend(handles, labels, loc='lower center', ncol=2, markerscale=3, frameon=False, bbox_to_anchor=(0.5, -0.04), fontsize=8)
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5, rect=[0,0.08,1,1])
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot(datasets, shape, nsims, maxcats, figure_path, figsize=(5, 4.8)):
fig, axes = plt.subplots(len(datasets), 1, figsize=figsize, squeeze=False)
for (dataset, xlim, ylim), axis in zip(datasets, axes.flatten()):
plot_space(axis, dataset, shape, xlim, ylim, nsims, maxcats)
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot_prop_correct(production_results, comprehension_results, figure_path):
fig, axes = plt.subplots(1, 2, figsize=(5, 2.5))
plot_densities(axes[0], production_results)
plot_densities(axes[1], comprehension_results)
axes[0].set_title('Production', fontsize=10)
axes[1].set_title('Comprehension', fontsize=10)
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot_advantage(figure_path, figsize=(5, 2.5)):
non_commcost = restore_results('../data/sim_comp_cost/non_commcost')
con_commcost = restore_results('../data/sim_comp_cost/con_commcost')
non_complexy = restore_results('../data/sim_comp_cost/non_complexy')
con_complexy = restore_results('../data/sim_comp_cost/con_complexy')
complexy_adv = non_complexy.mean(1) - con_complexy.mean(1)
commcost_adv = non_commcost.mean(1) - con_commcost.mean(1)
fig, axes = plt.subplots(1, 2, figsize=figsize)
axes[0].plot(range(1,65), complexy_adv, color='k')
axes[1].plot(range(1,65), commcost_adv, color='k')
axes[0].set_ylim(0-(400*0.025), 400+(400*0.025))
axes[1].set_ylim(0-(0.7*0.025), 0.7+(0.7*0.025))
axes[0].set_xlim(0,65)
axes[1].set_xlim(0,65)
axes[0].set_xticks([1, 16, 32, 48, 64])
axes[1].set_xticks([1, 16, 32, 48, 64])
axes[0].set_xlabel('Number of categories', fontsize=8)
axes[1].set_xlabel('Number of categories', fontsize=8)
# axes[0].set_yticks([0, 1, 2, 3, 4, 5, 6])
axes[1].set_yticks([0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7])
axes[0].set_ylabel('Compactness advantage (bits)', fontsize=8)
axes[0].set_title('Complexity', fontsize=8)
axes[1].set_title('Communicative cost', fontsize=8)
for tick in axes[0].yaxis.get_major_ticks():
tick.label.set_fontsize(7)
for tick in axes[0].xaxis.get_major_ticks():
tick.label.set_fontsize(7)
for tick in axes[1].yaxis.get_major_ticks():
tick.label.set_fontsize(7)
for tick in axes[1].xaxis.get_major_ticks():
tick.label.set_fontsize(7)
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
# run_sims(100)
# plot_sims('../visuals/simulated_complexity_cost.svg')
# plot_cost_sims('../visuals/simulated_cost.svg')
# plot_comp_sims('../visuals/simulated_comp.svg')
# plot_all('../visuals/simulated_complexity_cost.svg')
# plot_advantage('../visuals/compactness_advantage.svg')
def make_gaussians_figure(figure_path):
gammas = [0.001, 0.1, 10]
metric = 'circle_euclidean'
y_max = 0.07
N = commcost.Need((64, ))
P = commcost.Partition([0] * 16 + [1] * 16 + [2] * 16 + [3] * 16)
S = P.spawn_speaker()
fig = plt.figure(figsize=(5.5, 2))
grid = gridspec.GridSpec(nrows=1, ncols=len(gammas))
for i in range(len(gammas)):
sp = fig.add_subplot(grid[0, i])
sp.set_axisbelow(True)
if gammas[i] == 'inf':
L = P.spawn_listener(999999999)
K = commcost.cost(P, N, S, L)
plt.title('γ = ∞', fontsize=10)
else:
L = P.spawn_listener(gammas[i], metric)
K = commcost.cost(P, N, S, L)
plt.title('γ = ' + str(gammas[i]), fontsize=10)
for j in range(len(P)):
category_distribution = list(L[j].probabilities.flatten())
category_distribution = [
category_distribution[-1]
] + category_distribution + [category_distribution[0]]
sp.plot(range(0, P.size + 2),
category_distribution,
linewidth=2.0,
color=colors.categories[j])
plt.xticks(fontsize=7)
plt.yticks(fontsize=7)
if i == 0:
plt.ylabel('Probability', fontsize=10)
else:
sp.set_yticklabels([])
if i == len(gammas) // 2:
plt.xlabel('Universe', fontsize=10)
plt.xlim(1, P.size)
y_max = y_max * 1.05
y_min = 0 - (y_max * 0.05)
plt.ylim(-0.005, 0.065)
ticks = [1] + [(P.size // len(P)) * j for j in range(1, len(P) + 1)]
sp.set_xticks(ticks)
grid.tight_layout(fig, pad=0.1, h_pad=0.1, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot_click_entropy_all_experiments(figure_path):
exp1 = tools.read_json_lines('../data/experiments/exp1_participants.json')
exp2 = tools.read_json_lines('../data/experiments/exp2_participants.json')
res1p = click_entropy(exp1, 'production')
res1c = click_entropy(exp1, 'comprehension')
res2 = click_entropy(exp2)
fig, axes = plt.subplots(1, 3, figsize=(5.5, 2))
y, c = zip(*res1p)
axes[0].scatter(range(len(res1p)), y, c=c, marker='x')
axes[0].set_ylim(-0.1, 2.1)
axes[0].set_xlim(0, len(res1p))
axes[0].set_xticks([])
axes[0].set_xticklabels([])
axes[0].set_title('Experiment 1 (P)', fontsize=10)
y, c = zip(*res1c)
axes[1].scatter(range(len(res1c)), y, c=c, marker='x')
axes[1].set_ylim(0, 6)
axes[1].set_xlim(0, len(res1p))
axes[1].set_xticks([])
axes[1].set_xticklabels([])
axes[1].set_title('Experiment 1 (C)', fontsize=10)
y, c = zip(*res2)
axes[2].scatter(range(len(res2)), y, c=c, marker='x')
axes[2].plot([0, len(res1p)], [1.75, 1.75], c='red', linestyle='--')
axes[2].set_ylim(-0.1, 2.1)
axes[2].set_xlim(0, len(res1p))
axes[2].set_xticks([])
axes[2].set_xticklabels([])
axes[2].set_title('Experiment 2', fontsize=10)
axes[0].set_ylabel('Button-click entropy')
axes[1].set_xlabel('Participant')
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def plot_final_gen_distributions(datasets, figure_path):
fig, axes = plt.subplots(1, len(datasets), figsize=(7.2, 1.75))
for i, (label, dataset) in enumerate(datasets):
plot_final_gen_densities(axes[i], dataset)
axes[i].set_title(label, fontsize=7)
axes[0].set_xticks([25, 75, 125])
axes[1].set_xticks([25, 75, 125])
axes[2].set_xticks([25, 75, 125])
axes[0].set_xlim(0, 150)
axes[1].set_xlim(0, 150)
axes[2].set_xlim(0, 150)
axes[1].set_xlabel('Complexity')
axes[3].set_xticks([4.25, 4.3, 4.35])
axes[4].set_xticks([4.25, 4.3, 4.35])
axes[5].set_xticks([4.25, 4.3, 4.35])
axes[3].set_xlim(4.2, 4.4)
axes[4].set_xlim(4.2, 4.4)
axes[5].set_xlim(4.2, 4.4)
axes[4].set_xlabel('Cost')
fig.tight_layout(pad=0.01, h_pad=0.01, w_pad=0.01)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def make_fig(figure_path):
comp_weights = np.linspace(0, 3, 100)
cost_weights = np.linspace(0, 100, 100)
comp_stripe = np.array([-35.90911969399966 * i for i in comp_weights])
comp_qudrnt = np.array([-39.25512476486815 * i for i in comp_weights])
comp_sum = np.logaddexp2(comp_stripe, comp_qudrnt)
p_comp_stripe = 2**(comp_stripe - comp_sum)
p_comp_qudrnt = 2**(comp_qudrnt - comp_sum)
cost_stripe = np.array([-4.294113843380405 * i for i in cost_weights])
cost_qudrnt = np.array([-4.200174400715539 * i for i in cost_weights])
cost_sum = np.logaddexp2(cost_stripe, cost_qudrnt)
p_cost_stripe = 2**(cost_stripe - cost_sum)
p_cost_qudrnt = 2**(cost_qudrnt - cost_sum)
fig, axes = plt.subplots(1, 2, figsize=(5.5, 2.1), sharey=True)
axes[0].plot(comp_weights,
p_comp_stripe,
c=colors.blue,
linewidth=2,
linestyle='-')
axes[0].plot(comp_weights,
p_comp_qudrnt,
c=colors.blue,
linewidth=2,
linestyle=':')
axes[0].set_ylabel('Prior probability')
axes[0].set_xlabel('Weight (w)')
axes[0].set_title('Simplicity prior (πsim)', fontsize=10)
axes[0].set_xlim(0, comp_weights[-1])
axes[0].text(comp_weights[-1] * 0.97,
0.9,
'stripes',
horizontalalignment='right',
verticalalignment='center')
axes[0].text(comp_weights[-1] * 0.97,
0.1,
'quadrants',
horizontalalignment='right',
verticalalignment='center')
axes[1].plot(cost_weights,
p_cost_stripe,
c=colors.red,
linewidth=2,
linestyle='-')
axes[1].plot(cost_weights,
p_cost_qudrnt,
c=colors.red,
linewidth=2,
linestyle=':')
axes[1].set_xlabel('Weight (w)')
axes[1].set_title('Informativeness prior (πinf)', fontsize=10)
axes[1].set_xlim(0, cost_weights[-1])
axes[1].text(cost_weights[-1] * 0.97,
0.9,
'quadrants',
horizontalalignment='right',
verticalalignment='center')
axes[1].text(cost_weights[-1] * 0.97,
0.1,
'stripes',
horizontalalignment='right',
verticalalignment='center')
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5)
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def make_figure(datasets,
figure_path,
title=None,
show_legend=False,
deep_legend=False,
figsize=None):
if show_legend:
if deep_legend:
if figsize is None:
figsize = (5.5, 5.0)
fig, axes = plt.subplots(len(figure_layout),
len(figure_layout[0]),
figsize=figsize,
squeeze=False,
sharex=True)
else:
if figsize is None:
figsize = (7.2, 1.75)
fig, axes = plt.subplots(len(figure_layout),
len(figure_layout[0]),
figsize=figsize,
squeeze=False)
else:
if figsize is None:
figsize = (7.2, 1.6)
fig, axes = plt.subplots(len(figure_layout),
len(figure_layout[0]),
figsize=figsize,
squeeze=False)
for (i, j), axis in np.ndenumerate(axes):
measure = figure_layout[i][j]
for k, (dataset, label, color, color_conf,
linestyle) in enumerate(datasets):
mean, conf_pos, conf_neg, start_gen = dataset[measure]
xvals = list(range(start_gen, start_gen + len(mean)))
axis.plot(xvals,
mean,
c=color,
label=label,
linestyle=linestyle,
dash_capstyle="round",
linewidth=2)
axis.fill_between(xvals,
conf_neg,
conf_pos,
facecolor=color_conf,
alpha=0.5)
handles, labels = axis.get_legend_handles_labels()
ylim = measure_bounds[measure]
ypad = (ylim[1] - ylim[0]) * 0.05
axis.set_ylim(ylim[0] - ypad, ylim[1] + ypad)
axis.set_ylabel(measures_names[measure])
x_max = len(dataset[figure_layout[i][j]][0]) - 1
if figure_layout[i][j] == 'error':
x_max += 1
axis.set_xlim(0, x_max)
axis.set_xticks(list(range(0, x_max + 1, x_max // 5)))
if i == len(axes) - 1:
axis.set_xlabel('Generation')
if title:
fig.suptitle(title)
if show_legend:
if deep_legend:
fig.legend(handles,
labels,
loc='lower center',
ncol=1,
frameon=False)
fig.tight_layout(pad=2, h_pad=0.5, w_pad=0.5, rect=(0, 0.1, 1, 1))
else:
legend_offset = 1 / (figsize[1] / 0.15)
if len(figure_layout) == 1:
fig.legend(handles,
labels,
ncol=3,
frameon=False,
bbox_to_anchor=(0.5, -0.04),
loc='lower center')
else:
fig.legend(handles,
labels,
ncol=3,
frameon=False,
bbox_to_anchor=(0.5, -0.02),
loc='lower center')
fig.tight_layout(pad=0.1,
h_pad=0.5,
w_pad=0.5,
rect=(0.01, legend_offset, 1, 1))
else:
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5, rect=(0.01, 0, 1, 1))
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def make_carstensen_figure(datasets, figure_path, figsize):
fig, axes = plt.subplots(1, 2, figsize=figsize, sharey=True)
y_bounds = (4.6, 5)
y_pad = (y_bounds[1] - y_bounds[0]) * 0.05
for k, (dataset, label, color, color_conf,
linestyle) in enumerate(datasets):
mean, conf_pos, conf_neg, start_gen = dataset['cost']
xvals = list(range(start_gen, start_gen + len(mean)))
axes[0].plot(xvals,
mean,
c=color,
label=label,
linestyle=linestyle,
dash_capstyle='round',
linewidth=2)
axes[0].fill_between(xvals,
conf_neg,
conf_pos,
facecolor=color_conf,
alpha=0.5)
handles, labels = axes[0].get_legend_handles_labels()
axes[0].set_ylim(y_bounds[0] - y_pad, y_bounds[1] + y_pad)
axes[0].set_ylabel('Communicative cost (bits)')
axes[0].set_xlim(0, 10)
axes[0].set_xticks(list(range(0, 11, 2)))
axes[0].set_xlabel('Generation')
for k, (dataset, label, color, color_conf,
linestyle) in enumerate(datasets):
mean, conf_pos, conf_neg, start_gen = dataset['cost']
xvals = list(range(start_gen, start_gen + len(mean)))
axes[1].plot(xvals,
mean,
c=color,
label=label,
linestyle=linestyle,
dash_capstyle='round',
linewidth=2)
axes[1].fill_between(xvals,
conf_neg,
conf_pos,
facecolor=color_conf,
alpha=0.5)
axes[1].set_ylim(y_bounds[0] - y_pad, y_bounds[1] + y_pad)
axes[1].set_xlim(0, 100)
axes[1].set_xticks(list(range(0, 101, 20)))
axes[1].set_xlabel('Generation')
legend_offset = 1 / (figsize[1] / 0.15)
fig.legend(handles,
labels,
ncol=3,
frameon=False,
bbox_to_anchor=(0.5, -0.04),
loc='lower center')
fig.tight_layout(pad=0.1,
h_pad=0.5,
w_pad=0.5,
rect=(0.01, legend_offset, 1, 1))
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)
def make_figure(datasets, opt, figure_path, n_levels=32):
fig = plt.figure(figsize=(5.1, 1.9))
gs = gridspec.GridSpec(2, 4, width_ratios=[8, 8, 10, 1])
exp_axis = plt.subplot(gs[0, 0])
err_axis = plt.subplot(gs[0, 1])
com_axis = plt.subplot(gs[1, 0])
cst_axis = plt.subplot(gs[1, 1])
fit_axis = plt.subplot(gs[0:2, 2])
leg_axis = plt.subplot(gs[0:2, 3])
(dataset, label, color, color_conf, linestyle) = datasets
for measure, axis in [('expressivity', exp_axis), ('error', err_axis),
('complexity', com_axis), ('cost', cst_axis)]:
mean, conf_pos, conf_neg, start_gen = dataset[measure]
xvals = list(range(start_gen, start_gen + len(mean)))
axis.plot(xvals,
mean,
c=color,
label=label,
linestyle=linestyle,
dash_capstyle="round",
linewidth=2)
axis.fill_between(xvals,
conf_neg,
conf_pos,
facecolor=color_conf,
alpha=0.5)
ylim = measure_bounds[measure]
ypad = (ylim[1] - ylim[0]) * 0.05
axis.set_ylim(ylim[0] - ypad, ylim[1] + ypad)
axis.set_ylabel(measures_names[measure], fontsize=8)
axis.set_yticklabels([])
axis.set_yticks([])
for tick in axis.xaxis.get_major_ticks():
tick.label.set_fontsize(7)
axis.set_xlim(0, 10)
if measure == 'expressivity' or measure == 'error':
axis.set_xticks([])
else:
axis.set_xticks([0, 2, 4, 6, 8, 10])
axis.set_xlabel('Generation', fontsize=8)
(w_star, e_star), neg_log_p_sim = expected_minimum(opt)
print('Simplicity:', len(opt.func_vals), 'iterations)')
print('w* =', w_star)
print('e* =', e_star)
print('log(P) =', -neg_log_p_sim)
space = opt.space
samples = np.asarray(opt.x_iters)
rvs_transformed = space.transform(space.rvs(n_samples=250))
xi, yi, zi = plots.partial_dependence(space, opt.models[-1], 1, 0,
rvs_transformed, 250)
zmin = zi.min()
zmax = zi.max()
levels = np.geomspace(zmin, zmax, n_levels + 1)
cs = fit_axis.contourf(xi, yi, zi, levels, cmap='viridis_r')
fit_axis.plot([0, w_star, w_star], [e_star, e_star, 0],
c='k',
linestyle=':')
fit_axis.scatter(w_star, e_star, c='k', s=100, lw=0, marker='*')
fit_axis.set_xlim(0, 2)
fit_axis.set_ylim(0, 1)
fit_axis.set_xticks([0, 0.5, 1.0, 1.5, 2.0])
for tick in fit_axis.xaxis.get_major_ticks():
tick.label.set_fontsize(7)
fit_axis.set_yticklabels([])
fit_axis.set_yticks([])
fit_axis.set_xlabel('Weight (w)', fontsize=8)
fit_axis.set_ylabel('Noise (ε)', fontsize=8)
cb = plt.colorbar(cs, cax=leg_axis)
cb.set_ticks([])
leg_axis.yaxis.set_label_position("left")
leg_axis.set_ylabel('log likelihood', fontsize=8)
leg_axis.invert_yaxis()
fig.tight_layout(pad=0.1, h_pad=0.5, w_pad=0.5, rect=(0.01, 0, 1, 1))
fig.savefig(figure_path, format='svg')
tools.format_svg_labels(figure_path)
if not figure_path.endswith('.svg'):
tools.convert_svg(figure_path, figure_path)