fig, ax = plt.subplots(figsize=(colwide, 3/6*colwide),
constrained_layout=True)
# * plot reward function
R_value = data_list[1]["R_value"]
RX_value = vibly.project_Q2S(R_value, grids, proj_opt=np.mean)
RXv = RX_value.copy()
RXv[RXv==-0.4] = 1.
RXv[RXv==-1.4] = -1.
RXv[RXv< -100.] = 0.
# rmap = colors.ListedColormap(["tab:blue","tab:red", "white"])
# newcolors = [mymap(mynorm(-1.0)), (0., 0., 0., 1), (1., 1., 1., 1.)]
rmap = colors.ListedColormap(["darkgrey", "red", "white"])
# rmap = colors.ListedColormap(newcolors)
pc0 = vplot.reward_function(ax, RXv, grids, XV=XV, mymap=rmap)
ax.set_xlabel("$x_2$")
ax.set_ylabel("$x_1$")
fig.colorbar(pc0, ax=ax, orientation='horizontal', extend='both',
spacing='proportional', shrink=0.8)
plt.savefig('negproxy_R.pdf', format='pdf')
plt.close()
# * plot value functions
lvls = [-100., -50., alpha, -2., 0.]
for pdx, data in enumerate(data_list):
# data = data_list[ndx]
fig, ax = plt.subplots(figsize=(colwide, 3/6*colwide),
constrained_layout=True)
X_value = vibly.project_Q2S(data["Q_value"], grids, proj_opt=np.max)
pc1 = vplot.value_function(ax,
X_value,
# viability_threshold = X_value[XV].min()
# mynorm = colors.CenteredNorm()
# mymap = plt.get_cmap('bwr_r')
# shrunk_cmap = vplot.shiftedColorMap(mymap, start=0.2, midpoint=0.5, stop=0.8, name='shrunk')
extent = [
grids['states'][1][0], grids['states'][1][-1], grids['states'][0][0],
grids['states'][0][-1]
]
fig, axs = plt.subplots(2, 1)
# * Reward function
RX_value = vibly.project_Q2S(R_value, grids, proj_opt=np.mean)
pc0 = vplot.reward_function(axs[0], RX_value, grids, XV=XV, mymap="PRGn")
fig.colorbar(pc0, ax=axs[0])
# * Value Function
pc1 = vplot.value_function(axs[1], X_value, grids, XV=XV, mymap=mymap)
fig.colorbar(pc1, ax=axs[1])
plt.savefig(savename + str(failure_penalty) + '.pdf', format='pdf')
print("******************")
print("******************")
# plt.imshow(np.transpose(S_M), origin='lower') # visualize the S-safety measure
# plt.show()
# plt.imshow(Q_V) # visualize the viable set
# plt.show()
# these_values = [0, 3, 5, -3, -2, -1]
# these_values = range(len(data_list))
# num_plots = len(data_list)+1
# fig, axs = plt.subplots(num_plots, 1, figsize = (colwide, 16/10*colwide))
figR, axR = plt.subplots(figsize=(colwide, 3 / 6 * colwide),
constrained_layout=True)
# * plot reward function
R_value = data_list[0]["R_value"]
RX_value = vibly.project_Q2S(R_value, grids, proj_opt=np.mean)
# colors.LinearSegmentedColormap.from_list('second_map', mymap)
# newcolors = [(0., 0., 0., 1), (1., 1., 1., 1.), mymap(mynorm(1.0))]
rmap = colors.ListedColormap(["red", "white", "darkgrey"])
# rmap = colors.ListedColormap(newcolors)
pcR = vplot.reward_function(axR, RX_value, grids, XV=XV, mymap=rmap)
axR.set_xlabel("$x_2$")
axR.set_ylabel("$x_1$")
figR.colorbar(pcR,
ax=axR,
orientation='horizontal',
extend='both',
spacing='proportional',
shrink=1)
plt.savefig('posproxy_R.pdf', format='pdf')
plt.close()
# * plot value functions
lvls = [-100., -50., alpha, 1., 2.]
for pdx, data in enumerate(data_list):
fig, ax = plt.subplots(figsize=(colwide, 3 / 6 * colwide),
constrained_layout=True)
