def cm_remove_middle(cm,midpoint=0.5,cut_radius=0.2,name="my_cm"):
#removes the middle of a colormap
# so that region centered around 0 may have sharper contrast
cdict = {
'red': [],
'green': [],
'blue': [],
'alpha': []
}
donext=False
reg_index = np.linspace(0, 1, 100)
dri=reg_index[1]-reg_index[0]
slope1=1/(1-2*cut_radius)
slope2=0.5/(0.5-cut_radius)
offset=1-slope2
for ri in reg_index:
if (ri > midpoint+cut_radius) or (ri < midpoint - cut_radius):
r, g, b, a = cm(ri)
if ri < midpoint :
cdict['red'].append((ri*slope1, r, r))
cdict['green'].append((ri*slope1, g, g))
cdict['blue'].append((ri*slope1, b, b))
cdict['alpha'].append((ri*slope1, a, a))
#print str(ri) +":" + str(ri*slope1)
else:
cdict['red'].append((offset+ri*slope2, r, r))
cdict['green'].append((offset+ri*slope2, g, g))
cdict['blue'].append((offset+ri*slope2, b, b))
cdict['alpha'].append((offset+ri*slope2, a, a))
#print str(ri) +":" + str(offset+ri*slope2)
if ri < (midpoint+dri) and ri > (midpoint-dri):
r2, g2, b2, a2 = cm(midpoint)
cdict['red'].append((midpoint, r2, r2))
cdict['green'].append((midpoint, g2, g2))
cdict['blue'].append((midpoint, b2, b2))
cdict['alpha'].append((midpoint, a2, a2))
newcmap = matplotlib.colors.LinearSegmentedColormap(name, cdict,N=512)
plt.register_cmap(cmap=newcmap)
return newcmap
def plottingTracks(self, b):
self.updateModelTitle()
self.frameNumber.value = 0
self.refreshDisplay()
# Setting up color cycle for multiple tracks
colormap = plt.cm('gist_rainbow')
color_cycle = ([colormap(i) for i in np.linspace(0, 0.9, len(self.modelList))])
self.alltimes = []
# FOR look to create data for tracks and to set color for track
for i in range(len(self.modelList)):
self.trackColor = color_cycle[i]
self.manageData(self.modelList[i])
self.alltimes += self.time
# Creating an array of times from all selected tracks
self.allTimes = sorted(list(set(self.alltimes)))
self.updateUI()
def plot_robot(poses, params):
num_poses = len(poses)
cm =plt.get_cmap('gist_rainbow')
for i, pose in enumerate(poses):
col = cm(1.*i/num_poses )
r = params.sensor_range_m
# ax = plt.gca()
axes[0,0].plot([pose[0]-r*np.cos(pose[2]), pose[0]+r*np.cos(pose[2])],
[pose[1]-r*np.sin(pose[2]), pose[1]+r*np.sin(pose[2])], '--', color=col)
axes[0,0].plot([pose[0]-r*np.cos(pose[2]+np.pi/2), pose[0]+r*np.cos(pose[2]+np.pi/2)],
[pose[1]-r*np.sin(pose[2]+np.pi/2), pose[1]+r*np.sin(pose[2]+np.pi/2)], '--', color=col)
axes[0,0].arrow(pose[0], pose[1], 0.05 * np.cos(pose[2]), 0.05 * np.sin(pose[2]),
head_length=0.1, head_width=0.1)
FOV_ANGLE=math.pi/4
LENGTH = 0.8 # [m]
WIDTH = 0.5 # [m]
HALF_LENGTH = LENGTH/2.0 # [m]
SENSOR_LENGTH = 1.5 # [m]
WHEEL_LEN = 0.2 # [m]
WHEEL_WIDTH = 0.2 # [m]
sensor_outline = np.matrix([[0.0, SENSOR_LENGTH , SENSOR_LENGTH, 0.0], #sensor center
[0.0,SENSOR_LENGTH*math.tan(FOV_ANGLE), -SENSOR_LENGTH*math.tan(FOV_ANGLE), 0.0]])
outline = np.matrix([[-HALF_LENGTH, HALF_LENGTH, HALF_LENGTH, -HALF_LENGTH, -HALF_LENGTH],
[WIDTH / 2, WIDTH / 2, - WIDTH / 2, -WIDTH / 2, WIDTH / 2]])
yaw = pose[2]
Rot1 = np.matrix([[math.cos(yaw), math.sin(yaw)],
[-math.sin(yaw), math.cos(yaw)]])
outline = (outline.T * Rot1).T
outline[0, :] += pose[0]
outline[1, :] += pose[1]
sensor_outline = (sensor_outline.T * Rot1).T
sensor_outline[0, :] += pose[0]
sensor_outline[1, :] += pose[1]
#DRAW an agent_body
axes[0,0].plot(np.array(outline[0, :]).flatten(),
np.array(outline[1, :]).flatten(),color=col)
'ytick.labelsize': 17,
'text.usetex': True,
'figure.figsize': fig_size
}
plt.rcParams.update(params)
# read data
filename = 'plot.out'
data = np.loadtxt(filename)
ql = pd.DataFrame(data, columns=["x", "y"])
cm = ListedColormap([
'cornflowerblue', 'cornflowerblue', 'cornflowerblue', 'cornflowerblue',
'cornflowerblue', 'cornflowerblue', 'cornflowerblue', 'cornflowerblue',
'cornflowerblue'
])
my_cmap = cm(np.arange(cm.N))
my_cmap[:, -1] = np.linspace(0, 1, cm.N)
my_cmap = ListedColormap(my_cmap)
#plot data
g = sns.JointGrid(x="x",
y="y",
data=ql,
space=0.2,
xlim=(1.0, 1.7),
ylim=(0, 640))
g = g.plot_joint(sns.kdeplot,
cmap=my_cmap,
shade=True,
shade_lowest=False,
kernel='gau')
def perfect_2d_plot(dirlist,attribs,xattr="psi",yattr="theta",normname="norms.namelist",speciesname="species",psiN_to_psiname="psiAHat.h5",cm=cm.rainbow,lg=True,xlims=[90,100],ylims=[0,2],species=True,sort_species=True,first=["D","He"],last=["e"],generic_labels=True,label_dict={"D":"i","He":"z","N":"z","e":"e"},vlines=None,hlines=None,share_scale=[],skip_species = [],simulList=None):
#dirlist: list of simulation directories
#attribs: list of fields to plot from simulation
#speciesname: species filename in the simuldir
#normname: norm filename in the simuldir
#lg: controls whether to interpret nearby simulations as being paired
if type(attribs) is not list:
attribs=[attribs]
if simulList == None:
normlist=[x + "/" + normname for x in dirlist]
specieslist=[x + "/" + speciesname for x in dirlist]
psiN_to_psiList=[x + "/" + psiN_to_psiname for x in dirlist]
simulList=perfect_simulations_from_dirs(dirlist,normlist,specieslist,psiN_to_psiList)
else:
"p_2d_plot: simulList specified externally, ignoring dirlist, etc."
#if we translate species labels to generic ion and impurity labels
#we still need to use original species for sorting preferences
if generic_labels:
for simul in simulList:
simul.species_list=generic_species_labels(simul.species_list,label_dict)
first=generic_species_labels(first,label_dict)
last=generic_species_labels(last,label_dict)
#get the total list of species found in each simulation
species_set=set([])
for simul in simulList:
# exclude species that are in the skip list
#nonexcluded = set(simul.species).difference(skip_species)
nonexcluded = [s for s in simul.species if s not in skip_species]
simul.species_list = nonexcluded
# add nonexcluded species to total species set
species_set = species_set | set(simul.species)
# sort species according to preferences
if sort_species:
species_set=sort_species_list(list(species_set),first,last)
print species_set
for attrib in attribs:
psp_list=[] #list of perfect subplot objects
attrib_sp_dep = is_attribute_species_dependent(simulList,attrib)
#we will assign data to the following attribute related groups
attrib_groupname=attrib
if attrib_sp_dep and species:
species_attrib_groupname="species_dependent"
this_species_set=species_set
gridspec=[len(simulList),len(species_set)]
else:
species_attrib_groupname="species_independent"
this_species_set=set([''])
gridspec=[len(simulList),1]
for i,simul in enumerate(simulList):
data0=getattr(simul,attrib) #need to split this into species data
#i_s will be coordinate of plot
for i_s,s in enumerate(this_species_set):
if attrib_sp_dep:
index=[i2 for i2,s2 in enumerate(simul.species) if s2 == s]
if len(index)==0:
#no data for this species in this simulation. Set data to zeros
data=numpy.zeros(data0[:,:,0].shape)
elif len(index)==1:
data=data0[:,:,index[0]]
else:
print "perfect_2d_plot: warning: more than one of the same species in the simulation. Will add contributions, but this is untested."
data=numpy.sum(data0[:,:,index],axis=2)
else:
data=data0
subplot_coordinates=(i,i_s)
#print subplot_coordinates
if i == 0:
show_zaxis_ticklabel=True
show_xaxis_ticklabel=False
title=s
else:
show_zaxis_ticklabel=False
title=''
if i == len(simulList)-1:
show_xaxis_ticklabel=True
else:
show_xaxis_ticklabel=False
if i_s == 0:
show_yaxis_ticklabel=True
else:
show_yaxis_ticklabel=False
#print simul.local
if simul.local:
gl_grp="local"
else:
gl_grp="global"
if (yattr == "theta") or (yattr == "theta_shifted"):
y_scale = 1/numpy.pi
else:
y_scale=1
if (xattr == "theta") or (xattr == "theta_shifted"):
x_scale = 1/numpy.pi
elif (xattr == "psi") or (xattr == "actual_psiN"):
x_scale=100
else:
x_scale=1
if xattr is not "psio":
x = getattr(simul,xattr)*x_scale
else:
print "WARNING: orbit width uses hardcoded index for deuterium and pedestal start and stop values!"
vlines2 = [0.94927395957025573,0.97463697978512787]
iD = 0
ped_start_index=get_index_range(simul.actual_psiN,[vlines2[0],vlines2[0]])[1]
ped_stop_index=get_index_range(simul.actual_psiN,[vlines2[1],vlines2[1]])[1]
ow_start=simul.orbit_width[ped_start_index,iD]
ow_stop=simul.orbit_width[ped_stop_index,iD]
x = (simul.actual_psiN-vlines2[1])/simul.orbit_width[:,iD]
vlines=[(vlines2[0]-vlines2[1])/ow_start,(vlines2[1]-vlines2[1])/ow_stop]
y = getattr(simul,yattr)*y_scale
psp_list.append(perfect_subplot(data,x=x,y=y,subplot_coordinates=subplot_coordinates,show_zaxis_ticklabel=show_zaxis_ticklabel,show_yaxis_ticklabel=show_yaxis_ticklabel,show_xaxis_ticklabel=show_xaxis_ticklabel,title=title,groups=[s,"sim"+str(i),gl_grp,"pair"+str(i/2),species_attrib_groupname],dimensions=2))
#end simulList loop
for psp in psp_list:
print psp.groups
species_groups=[perfect_subplot_group(psp_list,groups=[s,"species_dependent"],logic="and") for s in species_set if len(perfect_subplot_group(psp_list,groups=[s,"species_dependent"],logic="and").p_subplot_list)>0]
nospecies_group=perfect_subplot_group(psp_list,groups=["species_independent"])
sim_groups = [perfect_subplot_group(psp_list,groups=["sim"+str(i)]) for i in range(len(simulList))]
pair_groups = [perfect_subplot_group(psp_list,groups=["pair"+str(i)]) for i in range(len(simulList)/2)]
local_group = perfect_subplot_group(psp_list,groups=["local"])
global_group = perfect_subplot_group(psp_list,groups=["global"])
all_group=perfect_subplot_group(psp_list,groups='',get_all=True)
share_scale_group=perfect_subplot_group(psp_list,groups=share_scale,logic="or")
if lg==False:
color=iter(cm(numpy.linspace(0,1,len(simulList))))
for sim_group in sim_groups:
c=next(color)
sim_group.setattrs("border_color",c)
else:
color=iter(cm(numpy.linspace(0,1,len(pair_groups))))
for pair_group in pair_groups:
c=next(color)
pair_group.setattrs("border_color",c)
local_group.setattrs("border_linestyle","dashed")
all_group.setattrs("xlims",xlims)
#all_group.setattrs("ylims",[all_group.get_min("y",xlim=False,ylim=False),all_group.get_max("y",xlim=False,ylim=False)])
all_group.setattrs("ylims",ylims)
all_group.setattrs("vlines",vlines)
all_group.setattrs("hlines",hlines)
for species_group in species_groups:
species_group.setattrs("zlims",[species_group.get_min("data"),species_group.get_max("data")])
print [species_group.get_min("data"),species_group.get_max("data")]
if len(share_scale_group.p_subplot_list)>0:
share_scale_group.setattrs("zlims",[share_scale_group.get_min("data"),share_scale_group.get_max("data")])
if len(nospecies_group.p_subplot_list)>0:
nospecies_group.setattrs("zlims",[nospecies_group.get_min("data"),nospecies_group.get_max("data")])
for i,sim_group in enumerate(sim_groups):
if i != 0:
sim_group.setattrs("show_zaxis",False)
if xattr is not "psio":
global_xlabel=r"$100\psi_N$"
else:
global_xlabel=r"$\psi^\mathrm{o}$"
perfect_visualizer(psp_list,gridspec,global_xlabel=global_xlabel,dimensions=2,global_ylabel=r"$\theta/\pi$")
plt.savefig(attrib+'.pdf')
def perfect_1d_plot(dirlist,attribs,xattr="psi",normname="norms.namelist",speciesname="species",psiN_to_psiname="psiAHat.h5",global_term_multiplier_name="globalTermMultiplier.h5",cm=cm.rainbow,lg=True,markers=None,linestyles=None,xlims=None,same_plot=False,outputname="default",ylabels=None,label_all=False,global_ylabel="",sort_species=True,first=["D","He"],last=["e"],generic_labels=True,label_dict={"D":"i","He":"z","N":"z","e":"e"},vlines=None,hlines=None,share_scale=[],interactive=False):
#dirlist: list of simulation directories
#attribs: list of fields to plot from simulation
#speciesname: species filename in the simuldir
#normname: norm filename in the simuldir
#lg: controls whether to interpret nearby simulations as being paired
if type(attribs) is not list:
attribs=[attribs]
if ylabels is not None:
if type(ylabels) is not list:
ylabels=[ylabels]*len(attribs)
else:
if len(ylabels) != len(attribs):
print "p_1d_plot: error: ylabels not the same size as attribs"
exit(1)
else:
ylabels=['']*len(attribs)
normlist=[x + "/" + normname for x in dirlist]
specieslist=[x + "/" + speciesname for x in dirlist]
psiN_to_psiList=[x + "/" + psiN_to_psiname for x in dirlist]
global_term_multiplierList=[x + "/" + global_term_multiplier_name for x in dirlist]
simulList=perfect_simulations_from_dirs(dirlist,normlist,specieslist,psiN_to_psiList,global_term_multiplierList)
if markers == None:
markers=['']*len(simulList)
#some logic to differentiate between species independent
#and species quantities further down will fail if there are simulations
#one species (which you really shouldn't do due to quasi-neutrality!!)
num_species_array=numpy.array([len(simul.species) for simul in simulList])
one_species_list=numpy.where(num_species_array<=1)
if len(num_species_array[one_species_list])>0:
print "p_1d_plot: warning: there are simulations with one (or) less species! Logic to determine whether attribute is a species property will not work."
if generic_labels:
for simul in simulList:
simul.species_list=generic_species_labels(simul.species_list,label_dict)
first=generic_species_labels(first,label_dict)
last=generic_species_labels(last,label_dict)
species_set=set([])
for simul in simulList:
species_set = species_set | set(simul.species)
#print gridspec
i=-1
psp_lists=[] #list of lists of perfect subplot object
gridspec_list=[]
colors=cm(numpy.linspace(0,1,len(simulList)))
#assign colors to simulations
all_linecolors=[]
color_index=0
local=[simul.local for simul in simulList]
noddpsi=[simul.no_ddpsi for simul in simulList]
if lg:
colors=cm(numpy.linspace(0,1,len(simulList)-sum(local)-sum(noddpsi)))
if sum(local)>0:
#if we have local simulations, increment color after them
for loc in local:
all_linecolors.append(colors[color_index])
if loc == True:
color_index=color_index+1
elif sum(noddpsi)>0:
#otherwise, increment color after noddpsi simulation
for nd in noddpsi:
all_linecolors.append(colors[color_index])
if nd == True:
color_index=color_index+1
else:
#else, always increment
all_linecolors=cm(numpy.linspace(0,1,len(simulList)))
else:
# force always increment behavior
all_linecolors=cm(numpy.linspace(0,1,len(simulList)))
if linestyles == None:
linestyles=[] #linestyles generated from local or global
for n,l in zip(noddpsi,local):
# local overrides noddpsi in code as well
if l:
linestyles=linestyles+["dashed"]
elif n:
linestyles=linestyles+["dashdot"]
else:
linestyles=linestyles+["solid"]
for i_a,attrib in enumerate(attribs):
psp_list=[]
#check whether attribute is species dependent
#it will be assumed to be if the lenght along the 1 axis
#is equal to the number of species in a simulation for all simulations
#at least as long as there is more than one species in the simulation.
attrib_sp_dep = is_attribute_species_dependent(simulList,attrib)
#we will assign data to the following attribute related groups
attrib_groupname=attrib
if attrib_sp_dep:
species_attrib_groupname="species_dependent"
else:
species_attrib_groupname="species_independent"
if same_plot:
if i_a == len(attribs)-1:
perhaps_last=True
else:
perhaps_last=False
else:
perhaps_last=True
if sort_species:
species_set=sort_species_list(list(species_set),first,last)
if attrib_sp_dep:
for i_sp,s in enumerate(species_set):
i=i+1
#data is taken for a given species for all simulations
#index of species in simulation given by index to index
index=[[ind for ind,spec in enumerate(simul.species) if spec==s] for simul in simulList]
if all(len(ind)<=1 for ind in index):
data=[getattr(simul,attrib)[:,index[i_si][0]] for i_si,simul in enumerate(simulList) if s in simul.species]
else:
print "p_1d_plot: warning: more than one of the same species in the simulation. Will add contributions."
data=[numpy.sum(getattr(simul,attrib)[:,index[i_si]],axis=1) for i_si,simul in enumerate(simulList) if s in simul.species]
if xattr=="theta":
x_scale=1/numpy.pi
else:
x_scale=1
if xattr != None:
x=[getattr(simul,xattr)*x_scale for simul in simulList if s in simul.species]
else:
# If xattrib is None, we plot against the index of the data
# This probably will not work if we are not plotting against
# the first index of data
x=[numpy.array(range(len(getattr(simul,attrib)))) for simul in simulList if s in simul.species]
if xlims == None:
# min to max among all the simulations
xlims = [numpy.min(x),numpy.max(x)]
linecolors=[all_linecolors[i_si] for i_si,simul in enumerate(simulList) if s in simul.species]
coordinates=(i,0)
if perhaps_last and (i_sp == len(species_set) - 1):
last_groupname="last"
gridspec_list.append([i+1,1])
else:
last_groupname="not_last"
psp_list.append(perfect_subplot(data,x,subplot_coordinates=coordinates,groups=[s,attrib_groupname,species_attrib_groupname,last_groupname],linestyles=linestyles,colors=linecolors,markers=markers)) #yaxis_label=ylabels[i_a]
else:
i=i+1
if perhaps_last:
last_groupname="last"
gridspec_list.append([i+1,1])
else:
last_groupname="not_last"
#species independent plot
data=[getattr(simul,attrib) for simul in simulList]
x=[getattr(simul,xattr) for simul in simulList]
linecolors=all_linecolors
coordinates=(i,0)
psp_list.append(perfect_subplot(data,x,subplot_coordinates=coordinates,groups=[attrib_groupname,species_attrib_groupname,last_groupname],linestyles=linestyles,colors=linecolors,markers=markers)) #yaxis_label=ylabels[i_a]
psp_lists.append(psp_list)
if not same_plot:
i=-1
#merge the psp_lists if everything is supposed to go in the same plot
if same_plot:
final_psp_lists=[]
for psp_list in psp_lists:
final_psp_lists = final_psp_lists + psp_list
psp_lists=[final_psp_lists]
for i_li,psp_list in enumerate(psp_lists):
for psp in psp_list:
print psp.groups
psp.xlims=xlims
psp.data=psp.data_inrange()
psp.x=psp.x_inrange()
if same_plot:
attrib_groups=[perfect_subplot_group(psp_list,groups=[a]) for a in attribs]
for ylabel,attrib_group in zip(ylabels,attrib_groups):
if label_all:
attrib_group.setattrs("yaxis_label",ylabel)
else:
attrib_group.set_middle_ylabel(ylabel)
else:
attrib_groups=[perfect_subplot_group(psp_list,groups=[a]) for a in [attribs[i_li]]]
for ylabel,attrib_group in zip([ylabels[i_li]],attrib_groups):
if label_all:
attrib_group.setattrs("yaxis_label",ylabel)
else:
attrib_group.set_middle_ylabel(ylabel)
if len(share_scale)>0:
share_scale_group=perfect_subplot_group(psp_list,groups=share_scale,logic="or")
share_scale_group.setattrs("ylims",[share_scale_group.get_min("data",margin=0.1),share_scale_group.get_max("data",margin=0.1)])
species_groups=[perfect_subplot_group(psp_list,groups=[s]) for s in species_set]
species_indep_groups=perfect_subplot_group(psp_list,groups=["species_independent"])
local_group = perfect_subplot_group(psp_list,groups=["local"])
global_group = perfect_subplot_group(psp_list,groups=["global"])
last_group = perfect_subplot_group(psp_list,groups=["last"])
all_group=perfect_subplot_group(psp_list,groups='',get_all=True)
for species_group,s in zip(species_groups,species_set):
species_group.setattrs("title",s)
for attrib_group in attrib_groups:
this_species_groups=[perfect_subplot_group(attrib_group.p_subplot_list,groups=[s]) for s in species_set]
for this_species_group in this_species_groups:
if len(this_species_group.p_subplot_list)>0:
this_species_group.setattrs("ylims",[this_species_group.get_min("data",margin=0.1),this_species_group.get_max("data",margin=0.1)])
#print [this_species_group.get_min("data",margin=0.1),this_species_group.get_max("data",margin=0.1)]
this_species_indep_group=perfect_subplot_group(attrib_group.p_subplot_list,groups=["species_independent"])
if len(this_species_indep_group.p_subplot_list)>0:
this_species_indep_group.setattrs("ylims",[this_species_indep_group.get_min("data",margin=0.1),this_species_indep_group.get_max("data",margin=0.1)])
this_share_scale_group=perfect_subplot_group(attrib_group.p_subplot_list,groups=share_scale,logic="or")
if len(this_share_scale_group.p_subplot_list)>0:
this_share_scale_group.setattrs("ylims",[this_share_scale_group.get_min("data",margin=0.1),this_share_scale_group.get_max("data",margin=0.1)])
all_group.setattrs("show_yaxis_ticklabel",True)
all_group.setattrs("vlines",vlines)
all_group.setattrs("hlines",hlines)
last_group.setattrs("show_xaxis_ticklabel",True)
#print gridspec_list[i_li]
if xattr=="psi":
global_xlabel=r"$\psi_N$"
elif xattr=="theta":
global_xlabel=r"$\theta/\pi$"
elif xattr=="sqrtpsi":
global_xlabel=r"$\sqrt{\psi_N}$"
elif xattr=="psiOverOrbitWidth":
global_xlabel=r"$\sqrt{\psi_N}$"
elif xattr=="actual_psi":
global_xlabel=r"$\hat{\psi}$"
elif xattr=="actual_psiN":
global_xlabel=r"$\psi_N$"
elif xattr=="psi_index":
global_xlabel=r"$i_\psi$"
elif xattr==None:
global_xlabel=r"$i$"
perfect_visualizer(psp_list,gridspec_list[i_li],global_xlabel=global_xlabel,dimensions=1,global_ylabel=global_ylabel)
if same_plot:
plt.savefig(outputname+".pdf")
else:
plt.savefig(attribs[i_li]+".pdf")
if interactive:
#dangerous, since it will (for some reason) be executed after all 1d_plot calls and show everything plotted in the given script.
datacursor(display='multiple', draggable=True)
plt.show()
def perfect_0d_plot(dirlist,yattribs,xattribs,normname="norms.namelist",speciesname="species",psiN_to_psiname="psiAHat.h5",cm=cm.rainbow,lg=True,markers=None,linestyles=None,same_plot=False,outputname="default",xlabels=None,ylabels=None,label_all=False,global_xlabel="",global_ylabel="",sort_species=True,first=["D","He"],last=["e"],generic_labels=True,label_dict={"D":"i","He":"z","N":"z","e":"e"},vlines=None,hlines=None,share_scale=[],interactive=False):
#dirlist: list of simulation directories
#attribs: list of fields to plot from simulation
#speciesname: species filename in the simuldir
#normname: norm filename in the simuldir
#lg: controls whether to interpret nearby simulations as being paired
if type(yattribs) is not list:
yattribs=[yattribs]
if type(xattribs) is not list:
xattribs=[xattribs]
if ylabels is not None:
if type(ylabels) is not list:
ylabels=[ylabels]*len(yattribs)
else:
if len(ylabels) != len(yattribs):
print "p_0d_plot: error: ylabels not the same size as attribs"
exit(1)
else:
ylabels=['']*len(yattribs)
if xlabels is not None:
if type(xlabels) is not list:
xlabels=[xlabels]*len(xattribs)
else:
if len(xlabels) != len(xattribs):
print "p_0d_plot: error: ylabels not the same size as attribs"
exit(1)
else:
xlabels=['']*len(xattribs)
normlist=[x + "/" + normname for x in dirlist]
specieslist=[x + "/" + speciesname for x in dirlist]
psiN_to_psiList=[x + "/" + psiN_to_psiname for x in dirlist]
simulList=perfect_simulations_from_dirs(dirlist,normlist,specieslist,psiN_to_psiList)
if markers == None:
markers=['']*len(simulList)
num_species_array=numpy.array([len(simul.species) for simul in simulList])
one_species_list=numpy.where(num_species_array<=1)
if len(num_species_array[one_species_list])>0:
print "p_0d_plot: warning: there are simulations with one (or) less species! Logic to determine whether attribute is a species property will not work."
if generic_labels:
for simul in simulList:
simul.species_list=generic_species_labels(simul.species_list,label_dict)
first=generic_species_labels(first,label_dict)
last=generic_species_labels(last,label_dict)
species_set=set([])
for simul in simulList:
species_set = species_set | set(simul.species)
#print gridspec
i=-1
psp_lists=[] #list of lists of perfect subplot object
gridspec_list=[]
colors=cm(numpy.linspace(0,1,len(simulList)))
#assign colors to simulations
all_linecolors=[]
color_index=0
local=[simul.local for simul in simulList]
noddpsi=[simul.no_ddpsi for simul in simulList]
if lg:
colors=cm(numpy.linspace(0,1,len(simulList)-sum(local)-sum(noddpsi)))
if sum(local)>0:
#if we have local simulations, increment color after them
for loc in local:
all_linecolors.append(colors[color_index])
if loc == True:
color_index=color_index+1
elif sum(noddpsi)>0:
#otherwise, increment color after noddpsi simulation
for nd in noddpsi:
all_linecolors.append(colors[color_index])
if nd == True:
color_index=color_index+1
else:
#else, always increment
all_linecolors=cm(numpy.linspace(0,1,len(simulList)))
else:
# force always increment behavior
all_linecolors=cm(numpy.linspace(0,1,len(simulList)))
if linestyles == None:
linestyles=[] #linestyles generated from local or global
for n,l in zip(noddpsi,local):
# local overrides noddpsi in code as well
if l:
linestyles=linestyles+["dashed"]
elif n:
linestyles=linestyles+["dashdot"]
else:
linestyles=linestyles+["solid"]
for i_a,(xattrib,yattrib) in enumerate(zip(xattribs,yattribs)):
psp_list=[]
yattrib_sp_dep = is_attribute_species_dependent(simulList,yattrib)
#we will assign data to the following attribute related groups
yattrib_groupname=yattrib
if yattrib_sp_dep:
species_attrib_groupname="species_dependent"
else:
species_attrib_groupname="species_independent"
if same_plot:
if i_a == len(yattribs)-1:
perhaps_last=True
else:
perhaps_last=False
else:
perhaps_last=True
if sort_species:
species_set=sort_species_list(list(species_set),first,last)
if yattrib_sp_dep:
for i_sp,s in enumerate(species_set):
i=i+1
#data is taken for a given species for all simulations
#index of species in simulation given by index to index
index=[[ind for ind,spec in enumerate(simul.species) if spec==s] for simul in simulList]
if all(len(ind)<=1 for ind in index):
ydata=[getattr(simul,yattrib)[index[i_si][0]] for i_si,simul in enumerate(simulList) if s in simul.species]
else:
print "p_0d_plot: warning: more than one of the same species in the simulation. Will add contributions."
ydata=[numpy.sum(getattr(simul,yattrib)[index[i_si]],axis=1) for i_si,simul in enumerate(simulList) if s in simul.species]
ydata = numpy.array(ydata)
xdata=[getattr(simul,xattrib) for simul in simulList if s in simul.species]
xdata = numpy.array(xdata)
print repr(xdata)
print repr(ydata)
linecolors=[all_linecolors[i_si] for i_si,simul in enumerate(simulList) if s in simul.species]
coordinates=(i,0)
if perhaps_last and (i_sp == len(species_set) - 1):
last_groupname="last"
gridspec_list.append([i+1,1])
else:
last_groupname="not_last"
psp_list.append(perfect_subplot(ydata,xdata,subplot_coordinates=coordinates,groups=[s,yattrib_groupname,species_attrib_groupname,last_groupname],linestyles=linestyles,colors=linecolors,markers=markers))
else:
i=i+1
if perhaps_last:
last_groupname="last"
gridspec_list.append([i+1,1])
else:
last_groupname="not_last"
#species independent plot
ydata=numpy.array([getattr(simul,yattrib) for simul in simulList])
xdata=numpy.array([getattr(simul,xattrib) for simul in simulList])
linecolors=all_linecolors
coordinates=(i,0)
psp_list.append(perfect_subplot(ydata,xdata,subplot_coordinates=coordinates,groups=[yattrib_groupname,species_attrib_groupname,last_groupname],linestyles=linestyles,colors=linecolors,markers=markers,dimensions=1))
psp_lists.append(psp_list)
if not same_plot:
i=-1
#merge the psp_lists if everything is supposed to go in the same plot
if same_plot:
final_psp_lists=[]
for psp_list in psp_lists:
final_psp_lists = final_psp_lists + psp_list
psp_lists=[final_psp_lists]
for i_li,psp_list in enumerate(psp_lists):
for psp in psp_list:
print psp.groups
xlims = [numpy.min(xdata),numpy.max(xdata)]
psp.xlims=xlims
psp.data=psp.data_inrange()
psp.x=psp.x_inrange()
if same_plot:
attrib_groups=[perfect_subplot_group(psp_list,groups=[a]) for a in yattribs]
for ylabel,attrib_group in zip(ylabels,attrib_groups):
if label_all:
attrib_group.setattrs("yaxis_label",ylabel)
else:
attrib_group.set_middle_ylabel(ylabel)
else:
attrib_groups=[perfect_subplot_group(psp_list,groups=[a]) for a in [attribs[i_li]]]
for ylabel,attrib_group in zip([ylabels[i_li]],attrib_groups):
if label_all:
attrib_group.setattrs("yaxis_label",ylabel)
else:
attrib_group.set_middle_ylabel(ylabel)
if len(share_scale)>0:
share_scale_group=perfect_subplot_group(psp_list,groups=share_scale,logic="or")
share_scale_group.setattrs("ylims",[share_scale_group.get_min("data",margin=0.1),share_scale_group.get_max("data",margin=0.1)])
species_groups=[perfect_subplot_group(psp_list,groups=[s]) for s in species_set]
species_indep_groups=perfect_subplot_group(psp_list,groups=["species_independent"])
local_group = perfect_subplot_group(psp_list,groups=["local"])
global_group = perfect_subplot_group(psp_list,groups=["global"])
last_group = perfect_subplot_group(psp_list,groups=["last"])
all_group=perfect_subplot_group(psp_list,groups='',get_all=True)
for species_group,s in zip(species_groups,species_set):
species_group.setattrs("title",s)
for attrib_group in attrib_groups:
this_species_groups=[perfect_subplot_group(attrib_group.p_subplot_list,groups=[s]) for s in species_set]
for this_species_group in this_species_groups:
if len(this_species_group.p_subplot_list)>0:
this_species_group.setattrs("ylims",[this_species_group.get_min("data",margin=0.1),this_species_group.get_max("data",margin=0.1)])
#print [this_species_group.get_min("data",margin=0.1),this_species_group.get_max("data",margin=0.1)]
this_species_indep_group=perfect_subplot_group(attrib_group.p_subplot_list,groups=["species_independent"])
if len(this_species_indep_group.p_subplot_list)>0:
this_species_indep_group.setattrs("ylims",[this_species_indep_group.get_min("data",margin=0.1),this_species_indep_group.get_max("data",margin=0.1)])
this_share_scale_group=perfect_subplot_group(attrib_group.p_subplot_list,groups=share_scale,logic="or")
if len(this_share_scale_group.p_subplot_list)>0:
this_share_scale_group.setattrs("ylims",[this_share_scale_group.get_min("data",margin=0.1),this_share_scale_group.get_max("data",margin=0.1)])
all_group.setattrs("show_yaxis_ticklabel",True)
all_group.setattrs("vlines",vlines)
all_group.setattrs("hlines",hlines)
last_group.setattrs("show_xaxis_ticklabel",True)
perfect_visualizer(psp_list,gridspec_list[i_li],global_xlabel=global_xlabel,dimensions=1,global_ylabel=global_ylabel)
if same_plot:
plt.savefig(outputname+".pdf")
else:
plt.savefig(attribs[i_li]+".pdf")
if interactive:
#dangerous, since it will (for some reason) be executed after all 0d_plot calls and show everything plotted in the given script.
datacursor(display='multiple', draggable=True)
plt.show()
from cycler import cycler
cm=plt.get_cmap('gist_rainbow')
NUM_COLORS=9
script_dir=os.path.dirname(__file__)
rel_path_o="Output_data/"
abs_path_o=os.path.join(script_dir,rel_path_o)
# plt.rc('axes', prop_cycle=(cycler('linestyle', ['-', '--', ':', '-.'])))
fig, ax = plt.subplots(1, 1,figsize=(12,9))
ax.get_xaxis().tick_bottom()
ax.get_yaxis().tick_left()
ax.set_prop_cycle(cycler('color',[cm(1.*i/9) for i in range(9)])*cycler('linestyle',['-', '--', ':']))
handle_list=[]
for item in glob.glob(abs_path_o+'Gradient_attack_SVM_PCA_*.txt'):
curr_array=np.genfromtxt(item,skip_header=1,delimiter=',')
handle_list.append(plt.plot(curr_array[:,1],curr_array[:,2],label=int(curr_array[0,0])))
curr_array=np.genfromtxt(abs_path_o+'Gradient_attack_SVM_no_PCA_.txt',
skip_header=1,delimiter=',')
plt.plot(curr_array[:,1],curr_array[:,2],label='noPCA')
plt.xlabel(r'Adversarial perturbation $\epsilon$')
plt.ylabel('Misclassification percentage')
plt.title('Adversarial example success vs. perturbation magnitude \n'
' for various reduced dimensions using the '