-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathplot_physical.py
143 lines (125 loc) · 5.13 KB
/
plot_physical.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
from hyperdopamine.utils import plot_utils
import matplotlib as mpl
mpl.use('Agg')
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
mpl.rcParams['axes.linewidth'] = 1.
plt.rcParams.update({'font.size': 10})
plt.rcParams['font.family'] = 'Times New Roman, serif'
BASE_PATH = './data/benchmark/physical'
GAMES = [
'ReacherBulletEnv',
'HopperBulletEnv',
'HalfCheetahBulletEnv',
'Walker2DBulletEnv',
'AntBulletEnv',
'Humanoid']
SEEDS = [str(i) for i in range(0,9)]
MAX_EPOCH = 300
TITLES = {
'ReacherBulletEnv': r'Reacher',
'HopperBulletEnv': r'Hopper',
'HalfCheetahBulletEnv': r'HalfCheetah',
'Walker2DBulletEnv': r'Walker2D',
'AntBulletEnv': r'Ant',
'Humanoid': r'Humanoid'}
LABELS = {
'hgqn-r1-sum': 'HGQN ($r=1$)',
'hgqn-r2-sum': 'HGQN ($r=2$)',
'hgqn-r3-sum': 'HGQN ($r=3$)',
'bdqn': 'BDQN',
'dqn': 'DQN',
'rainbow': 'Rainbow$^\dagger$',
'ddpg': 'DDPG'}
COLORS = {
'hgqn-r1-sum': '#ff414d',
'hgqn-r2-sum': '#2ECC71',
'hgqn-r3-sum': '#000CF9',
'bdqn': '#00bcd4',
'dqn': '#888888',
'rainbow': '#ffa931',
'ddpg': '#151680'}
LINESTYLES = {
'hgqn-r1-sum': '-',
'hgqn-r2-sum': '-',
'hgqn-r3-sum': '-',
'bdqn': '-',
'dqn': '--',
'rainbow': '--',
'ddpg': '--'}
DDPG_PERFORMANCE = {
'ReacherBulletEnv': 12.,
'HopperBulletEnv': 1375.,
'HalfCheetahBulletEnv': 2687.5,
'Walker2DBulletEnv': 546.875,
'AntBulletEnv': 2310.,
'Humanoid': 102.}
width = 12.65
num_figs = 3
hunits_fig = 17
vunits_fig = int(hunits_fig / 1.13333333333)
hunits_between_figs = 5
vunits_between_figs = 5
legend_units = 8
unit_width = width / (num_figs*hunits_fig + (num_figs-1)*hunits_between_figs)
height = unit_width * (2*vunits_fig + vunits_between_figs + legend_units)
fig = plt.figure(figsize=(width, height))
grid = gridspec.GridSpec(2*vunits_fig + vunits_between_figs + legend_units,
2*hunits_between_figs + 3*hunits_fig)
ax1 = fig.add_subplot(grid[0:vunits_fig,
0*hunits_between_figs + 0*hunits_fig:0*hunits_between_figs + 1*hunits_fig])
ax2 = fig.add_subplot(grid[0:vunits_fig,
1*hunits_between_figs + 1*hunits_fig:1*hunits_between_figs + 2*hunits_fig])
ax3 = fig.add_subplot(grid[0:vunits_fig,
2*hunits_between_figs + 2*hunits_fig:2*hunits_between_figs + 3*hunits_fig])
ax4 = fig.add_subplot(grid[vunits_fig + vunits_between_figs:2*vunits_fig + vunits_between_figs,
0*hunits_between_figs + 0*hunits_fig:0*hunits_between_figs + 1*hunits_fig])
ax5 = fig.add_subplot(grid[vunits_fig + vunits_between_figs:2*vunits_fig + vunits_between_figs,
1*hunits_between_figs + 1*hunits_fig:1*hunits_between_figs + 2*hunits_fig])
ax6 = fig.add_subplot(grid[vunits_fig + vunits_between_figs:2*vunits_fig + vunits_between_figs,
2*hunits_between_figs + 2*hunits_fig:2*hunits_between_figs + 3*hunits_fig])
axl = fig.add_subplot(grid[
2*vunits_fig + vunits_between_figs:2*vunits_fig + vunits_between_figs + legend_units,
0*hunits_between_figs + 0*hunits_fig:2*hunits_between_figs + 3*hunits_fig])
axl.axis('off')
ax = [ax1, ax2, ax3, ax4, ax5, ax6]
for axi in ax: axi.set_axisbelow(True)
for game_id, game in enumerate(GAMES):
print(game)
ax[game_id].hlines(y=DDPG_PERFORMANCE[game], xmin=0, xmax=MAX_EPOCH, color=COLORS['ddpg'],
linestyle=LINESTYLES['ddpg'], linewidth=2., label=LABELS['ddpg'])
if game == 'ReacherBulletEnv':
AGENTS = ['hgqn-r1-sum', 'hgqn-r2-sum', 'bdqn', 'dqn', 'rainbow']
elif game in ['HopperBulletEnv', 'HalfCheetahBulletEnv', 'Walker2DBulletEnv']:
AGENTS = ['hgqn-r1-sum', 'hgqn-r2-sum', 'hgqn-r3-sum', 'bdqn', 'dqn', 'rainbow']
elif game in ['AntBulletEnv', 'Humanoid']:
AGENTS = ['hgqn-r1-sum', 'bdqn']
for agent in AGENTS:
agent_game_data = []
for seed in SEEDS:
log_path = BASE_PATH + '/{}/{}/{}/logs'.format(game, agent, seed)
raw_data, _ = plot_utils.load_statistics(log_path, verbose=False)
summarized_data = plot_utils.summarize_data(raw_data, ['eval_episode_returns'])
agent_game_data.append(summarized_data['eval_episode_returns'][:MAX_EPOCH])
xs, means, stds = plot_utils.smooth_results(agent_game_data, 5, 1)
ax[game_id].fill_between(
range(len(means)), means-stds, means+stds, color=COLORS[agent], alpha=0.1, lw=0.)
ax[game_id].plot(means, label=LABELS[agent], color=COLORS[agent],
linestyle=LINESTYLES[agent], linewidth=1.3)
ax[game_id].set_title(TITLES[game], fontsize=14)
if game_id in [0,3]:
ax[game_id].set_ylabel('Average Score', fontsize=12)
if game_id >= 3:
ax[game_id].set_xlabel('Environment Step', fontsize=12)
ax[game_id].set_xlim([0,MAX_EPOCH])
ax[game_id].set_xticks([0,100,200,300])
ax[game_id].set_xticklabels(['0','1M','2M','3M'])
handles, labels = ax[2].get_legend_handles_labels()
new_handles = \
[handles[0], handles[4], handles[1], handles[5], handles[2], handles[-1], handles[3]]
new_labels = \
[labels[0], labels[4], labels[1], labels[5], labels[2], labels[-1], labels[3]]
ax[4].legend(new_handles, new_labels, bbox_to_anchor=(0.5, -0.3), loc='upper center',
ncol=4, frameon=False, markerscale=2., fontsize=14)
plt.savefig('physical_results.pdf')
print('figure saved')