savefig and main thread is not in main loop in python3 - python-3.x

Question

After using
plt.savefig() in a function, I obtain
File "/usr/lib/python3.8/tkinter/__init__.py", line 4017, in __del__
self.tk.call('image', 'delete', self.name)
RuntimeError: main thread is not in main loop
Tcl_AsyncDelete: async handler deleted by the wrong thread
Aborted (core dumped)
I tried to add
plt.switch_backend('agg')
however, the code is not going to finish after this edit.
What to do, please?
I have downloaded https://github.com/amerand/PMOIRED
and I added
plt.savefig('bootstrap1.png')
before return in the function showBootstrap in the file oimodels.py.
def showBootstrap(b, indx=None, fig=0, figWidth=None, showRejected=False,
combParam={}, sigmaClipping=4.5, showChi2=False):
"""
you can look at combination of parameters:
combParam: {'sep':'np.sqrt($x**2 + $y**2)'} assuming 'x' and 'y' are parameters from the model
"""
global _AX, _AY
boot = copy.deepcopy(b)
# -- THIS IS NOT WORKING :(
#if showChi2:
# if not type(combParam) == dict:
# combParam = {}
# combParam.update({'_chi2': 'chi2'})
#boot['fitOnly'] = boot['fitOnly']
if len(combParam)>0:
s = '$'
for k in combParam:
if not k in boot['fitOnly']:
boot['fitOnly'].append(k)
for i,f in enumerate(boot['fitOnly']):
if 'chi2' in k:
boot['all fits'][i]['best'][k] = boot['all fits'][i]['chi2']
boot['all fits'][i]['uncer'][k] = 0.0
else:
tmp = combParam[k]+''
j = 0
while s in tmp and j<5:
for x in boot['best'].keys():
if s+x in tmp:
tmp = tmp.replace(s+x, '('+str(boot['best'][x])+')')
j+=1
boot['all fits'][i]['best'][k] = eval(tmp)
boot['all fits'][i]['uncer'][k] = 0.0
print('analyse')
boot = analyseBootstrap(boot, indx=inxd, inx=None, verbose=2, sigmaClipping=sigmaClipping)
print('done')
if figWidth is None:
figWidth = min(9.5, 1+2*len(boot['fitOnly']))
fontsize = max(min(4*figWidth/len(boot['fitOnly']), 14), 6)
plt.close(fig)
plt.figure(fig, figsize=(figWidth, figWidth))
_AX = {}
color1 = 'orange'
color2 = (0.2, 0.4, 1.0)
color3 = (0.8, 0.2, 0.4)
colorC2 = (0, 0.4, 0.2)
combi = False
# -- for each fitted parameters, show histogram
showP = sorted(boot['fitOnly'])
if showChi2:
showP.append('chi2')
offs = {}
amps = {}
for i1, k1 in enumerate(showP):
# -- create histogram plot
_AX[i1] = plt.subplot(len(showP),
len(showP),
1+i1*len(showP)+i1)
if k1=='chi2':
# -- assumes bins is already defined (chi2 cannot be first!)
# -- show histogram: line and area
h = plt.hist(boot['all chi2'], bins=bins,
color=colorC2, histtype='step', alpha=0.9)
h = plt.hist(boot['all chi2'], bins=bins,
color=colorC2, histtype='stepfilled', alpha=0.05)
plt.plot(boot['fit to all data']['chi2'],
0.4*max(h[0]), 's', markersize=fontsize/2,
color=colorC2, label='fit to all data')
plt.title(r'$\chi_{\rm red}^2$', fontsize=fontsize)
offs['chi2'] = 0
amps['chi2'] = 1
plt.xlim(min(min(h[1]),
boot['fit to all data']['chi2']-0.1*np.ptp(h[1])),
max(max(h[1]),
boot['fit to all data']['chi2']+0.1*np.ptp(h[1])))
#print('Xlim:', min(h[1]), max(h[1]), boot['fit to all data']['chi2'])
else:
# -- guess number of bins
if k1 in boot['uncer']:
bins = int(3*np.ptp(boot['all best'][k1])/boot['uncer'][k1])
else:
bins = 10
bins = min(bins, len(boot['mask'])//5)
bins = max(bins, 5)
nd = np.abs(np.mean(boot['all best'][k1])/np.ptp(boot['all best'][k1]))
if nd>0:
nd = int(np.log10(nd))
if nd>=4:
offs[k1] = np.round(np.mean(boot['all best'][k1]), nd)
amps[k1] = 10**(nd+1)
else:
offs[k1] = 0.0
amps[k1] = 1.0
#print(k1, nd, offs[k1])
# -- show histogram: line and area
h = plt.hist(amps[k1]*(boot['all best'][k1]-offs[k1]), bins=bins,
color='k', histtype='step', alpha=0.9)
h = plt.hist(amps[k1]*(boot['all best'][k1]-offs[k1]), bins=bins,
color='k', histtype='stepfilled', alpha=0.05)
# -- fitted and bootstrap values and uncertainties
if k1 in boot['fit to all data']['best']:
plt.errorbar(amps[k1]*(boot['fit to all data']['best'][k1]-offs[k1]),
0.4*max(h[0]), markersize=fontsize/2,
xerr=amps[k1]*boot['fit to all data']['uncer'][k1],
color=color1, fmt='s', capsize=fontsize/2,
label='fit to all data')
combi = False
else:
combi = True
plt.errorbar(amps[k1]*(boot['best'][k1]-offs[k1]), 0.5*max(h[0]),
xerr=amps[k1]*boot['uncer'][k1],
color=color3 if combi else color2, fmt='d',
capsize=fontsize/2, label='bootstrap', markersize=fontsize/2)
n = int(np.ceil(-np.log10(boot['uncer'][k1])+1))
fmt = '%s=\n'+'%.'+'%d'%n+'f'+'$\pm$'+'%.'+'%d'%n+'f'
plt.title(fmt%(k1, boot['best'][k1], boot['uncer'][k1]),
fontsize=fontsize)
plt.legend(fontsize=5)
# -- title
_AX[i1].yaxis.set_visible(False)
if i1!=(len(showP)-1):
_AX[i1].xaxis.set_visible(False)
else:
_AX[i1].tick_params(axis='x', labelsize=fontsize*0.8)
_AX[i1].set_xlabel(k1, fontsize=fontsize)
if offs[k1]<0:
_AX[i1].set_xlabel(k1+'\n+%f (%.0e)'%(np.abs(offs[k1]), 1/amps[k1]),
fontsize=fontsize)
elif offs[k1]>0:
_AX[i1].set_xlabel(k1+'\n-%f (%.0e)'%(np.abs(offs[k1]), 1/amps[k1]),
fontsize=fontsize)
_AX[i1].callbacks.connect('ylim_changed', _callbackAxes)
# -- end histogram
_AY = {}
# -- show density plots
for i1, k1 in enumerate(showP):
for i2 in range(i1+1, len(showP)):
k2 = showP[i2]
if i1==0:
_AY[i2] = plt.subplot(len(showP),
len(showP),
1+i2*len(showP)+i1,
sharex=_AX[i1])
ax = _AY[i2]
else:
ax = plt.subplot(len(showP),
len(showP),
1+i2*len(showP)+i1,
#sharex=_AX[i1],
sharey=_AY[i2])
#if k1=='chi2' or k2=='chi2':
# continue
#print(k1, k2)
c, m = 'k', '.'
if k1=='chi2':
X1 = boot['all chi2']
X1r = boot['all chi2 ignored']
c = colorC2
else:
X1 = boot['all best'][k1]
X1r = boot['all best ignored'][k1]
if k2=='chi2':
X2 = boot['all chi2']
X2r = boot['all chi2 ignored']
c = colorC2
else:
X2 = boot['all best'][k2]
X2r = boot['all best ignored'][k2]
plt.plot(amps[k1]*(X1-offs[k1]), amps[k2]*(X2-offs[k2]), m,
alpha=np.sqrt(2/len(boot['mask'])), color=c)
if showRejected:
plt.plot(amps[k1]*(X1r-offs[k1]), amps[k2]*(X2r-offs[k2]),
'xr', alpha=0.3)
# -- combined parameters function of the other one?
#print(combParam, k1, k2)
if (k1 in combParam and k2 in combParam[k1]) or \
(k2 in combParam and k1 in combParam[k2]):
_c = color3
else:
_c = color2
#x, y = dpfit.errorEllipse(boot, k1, k2)
if k1=='chi2':
_i1 = len(boot['fitOnly'])
_c = colorC2
else:
_i1 = boot['fitOnly'].index(k1)
if k2=='chi2':
_i2 = len(boot['fitOnly'])
_c = colorC2
else:
_i2 = boot['fitOnly'].index(k2)
t = np.linspace(0,2*np.pi,100)
sMa, sma, a = dpfit._ellParam(boot['cov'][_i1,_i1],
boot['cov'][_i2,_i2],
boot['cov'][_i1,_i2])
_X,_Y = sMa*np.cos(t), sma*np.sin(t)
_X,_Y = _X*np.cos(a)+_Y*np.sin(a),-_X*np.sin(a)+_Y*np.cos(a)
if (k1.endswith(',x') and k2.endswith(',y')) or \
(k1.endswith(',y') and k2.endswith(',x')):
print('ellipse (emin, emax, PA) for %s/%s: %.4f %.4f %.1f'%(
k1, k2, sMa, sma, a*180/np.pi))
if k1=='chi2':
x = np.mean(boot['all chi2'])+_X
else:
x = boot['best'][k1]+_X
if k2=='chi2':
y = np.mean(boot['all chi2'])+_Y
else:
y = boot['best'][k2]+_Y
plt.plot(amps[k1]*(x-offs[k1]),
amps[k2]*(y-offs[k2]), '-', color=_c,
label='c=%.2f'%boot['cord'][k1][k2])
plt.legend(fontsize=5)
if k1 in boot['fit to all data']['best'] and \
k2 in boot['fit to all data']['best']:
plt.plot(amps[k1]*(boot['fit to all data']['best'][k1]-offs[k1]),
amps[k2]*(boot['fit to all data']['best'][k2]-offs[k2]),
'x', color='0.5')
x, y = dpfit.errorEllipse(boot['fit to all data'], k1, k2)
plt.plot(amps[k1]*(x-offs[k1]), amps[k2]*(y-offs[k2]), '-',
color=color1)#, label='c=%.2f'%boot['cord'][k1][k2])
plt.plot(amps[k1]*(boot['best'][k1]-offs[k1]),
amps[k2]*(boot['best'][k2]-offs[k2]),
'+', color=_c)
if i2==(len(showP)-1):
plt.xlabel(k1, fontsize=fontsize)
if offs[k1]<0:
plt.xlabel(k1+'\n+%f (%.0e)'%(np.abs(offs[k1]), 1/amps[k1]),
fontsize=fontsize)
elif offs[k1]>0:
plt.xlabel(k1+'\n-%f (%.0e)'%(np.abs(offs[k1]), 1/amps[k1]),
fontsize=fontsize)
ax.tick_params(axis='x', labelsize=fontsize*0.8)
else:
ax.xaxis.set_visible(False)
if i1==0:
plt.ylabel(k2, fontsize=fontsize)
if offs[k2]<0:
plt.ylabel(k2+'\n+%f (%.0e)'%(np.abs(offs[k2]), 1/amps[k2]),
fontsize=fontsize)
elif offs[k2]>0:
plt.ylabel(k2+'\n-%f (%.0e)'%(np.abs(offs[k2]), 1/amps[k2]),
fontsize=fontsize)
_AY[i2].callbacks.connect('ylim_changed', _callbackAxes)
_AY[i2].tick_params(axis='y', labelsize=fontsize*0.8)
else:
ax.yaxis.set_visible(False)
plt.tight_layout()
plt.subplots_adjust(hspace = 0, # 0.65*fig.subplotpars.hspace,
wspace = 0, # 0.65*fig.subplotpars.wspace
)
plt.savefig('bootstrap1.png')
return