def zipToAnimation(self, sourcepath, destpath, filename, framedata):
destpath = self.folder(destpath)
workpath = "temp/" + filename
self.execute("unzip -o " + sourcepath + " -d " + workpath)
if stddev(framedata) < 1.0:
# assumed stable framerate
fr = 1 / round(average(framedata)) * 1000
self.execute(
"ffmpeg -framerate %d -i %s/%%6d.jpg -c:v copy %s.mkv" %
(fr, workpath, destpath + filename))
else:
# probably needs variable framerate
print(
"\n stddev %.3f, %d frames:\n" %
(stddev(framedata), len(framedata)), framedata)
cmd = "convert "
for i, d in enumerate(framedata):
cmd += "-delay %d %s/%06d.jpg " % (d / 10, workpath, i)
cmd += destpath + filename + ".gif"
self.execute(
cmd
) # we convert to gif and copy the zip too, because gif is trash
self.execute("cp " + sourcepath + " " + destpath + filename +
".zip")
def __init__(self, xs, ys, mb_size, normalize=False):
if normalize:
self.xs = (xs - np.mean(xs, axis=0)) / np.stddev(xs, axis=0)
else:
self.xs = xs.copy()
self.ys = ys.copy()
self.mb_i = 0
self.mb_size = mb_size
self.batches_in_epoch = len(xs)//mb_size
def coupling_vs_r_pupilplane_single(tele: AOtele.AOtele, pupilfields, rcore,
ncore, nclad, wl0):
k0 = 2 * np.pi / wl0
fieldshape = pupilfields[0].shape
#need to generate the available modes
V = LPmodes.get_V(k0, rcore, ncore, nclad)
modes = LPmodes.get_modes(V)
lpfields = []
for mode in modes:
if mode[0] == 0:
lpfields.append(
normalize(
LPmodes.lpfield(xg, yg, mode[0], mode[1], rcore, wl0,
ncore, nclad)))
else:
lpfields.append(
normalize(
LPmodes.lpfield(xg, yg, mode[0], mode[1], rcore, wl0,
ncore, nclad, "cos")))
lpfields.append(
normalize(
LPmodes.lpfield(xg, yg, mode[0], mode[1], rcore, wl0,
ncore, nclad, "sin")))
lppupilfields = []
#then backpropagate
for field in lpfields:
wf = hc.Wavefront(hc.Field(field.flatten(), tele.focalgrid),
wavelength=wl0 * 1.e-6)
pupil_wf = tele.back_propagate(wf, True)
lppupilfields.append(pupil_wf.electric_field.reshape(fieldshape))
lppupilfields = np.array(lppupilfields)
pupilfields = np.array(pupilfields)
#compute total overlap
powers = np.sum(pupilfields * lppupilfields, axes=(1, 2))
return np.mean(powers), np.stddev(powers)
def nonmaxsup(H,n=100,c=.9):
mindistance = []
threshold = np.mean(H) + np.stddev(H)
maxima = getmaxima(H,threshold)
for x,y,z in enumerate(maxima):
min = np.infinity
for xx,yy,zz in enumerate(maxima):
dist = sqrt((x-xx)**2 + (y-yy)**2 )
if z < c*zz and dist > 0 and dist < min:
min = dist
xmin = xx
ymin = yy
mindistance.append((xx,yy,min))
mindistance.sort(key=lambda x:x[2])
return mindistance[:n]
num_states,
obs_dim,
input_dim,
observations="input_driven_obs",
observation_kwargs=dict(C=num_categories),
transitions="standard")
fit_ll = bandit_glmhmm.fit(training_choice,
inputs=training_inpt,
method="em",
num_iters=N_iters,
tolerance=10**-4)
i_ll.append(
bandit_glmhmm.log_likelihood(test_choices,
inputs=test_inpt))
lls.append(np.mean(i_ll))
lls_var.append(np.stddev(i_ll))
log_likelihoods[:, fold] = lls
# Plot Crossvalidation
sns.pointplot(x=np.arange(5), y=np.mean(log_likelihoods, axis=1))
plt.xticks(np.arange(5), np.arange(1, 6))
plt.ylabel('LL')
plt.xlabel('Number of latents')
# Fit model
num_states = 2
obs_dim = 1
input_dim = np.shape(meta_inpts[0])[1]
num_categories = 2
N_iters = 400 # maximum number of EM iterations. Fitting with stop earlier if increase in LL is below tolerance specified by tolerance parameter
bandit_glmhmm = ssm.HMM(num_states,
setHistTitles(hist,"#mu Reference","N_{Toys}")
iHist += 1
for mu in refMus:
hist.Fill(mu)
hist.Draw()
tlatex.SetTextAlign(12)
tlatex.DrawLatex(gStyle.GetPadLeftMargin(),0.96,PRELIMINARYSTRING)
tlatex.SetTextAlign(12)
tlatex.DrawLatex(0.02+gStyle.GetPadLeftMargin(),0.85,"Reference PDF: "+PDFTITLEMAP[refPdfName])
#tlatex.DrawLatex(0.02+gStyle.GetPadLeftMargin(),0.75,"Alternate PDF: "+PDFTITLEMAP[pdfAltName])
tlatex.DrawLatex(0.02+gStyle.GetPadLeftMargin(),0.68,"m_{H} = "+str(hmass)+" GeV/c^{2}")
tlatex.SetTextAlign(32)
tlatex.DrawLatex(0.99-gStyle.GetPadRightMargin(),0.96,caption)
tlatex.DrawLatex(0.97-gStyle.GetPadRightMargin(),0.85,"Median: {0:.2f}".format(median(refMus)))
tlatex.DrawLatex(0.97-gStyle.GetPadRightMargin(),0.75,"Mean: {0:.2f}".format(mean(refMus)))
tlatex.DrawLatex(0.97-gStyle.GetPadRightMargin(),0.65,"#sigma: {0:.2f}".format(stddev(refMus)))
line = setYMaxAndDrawVertLines(hist,None)
canvas.RedrawAxis()
saveAs(canvas,outputPrefix+"_"+str(hmass)+"_MuRef_Ref"+refPdfName)
canvas.Clear()
## Mu Alt Plot
hist = root.TH1F("hist"+str(iHist),"",60,-10,10)
setHistTitles(hist,"#mu Alternate","N_{Toys}")
iHist += 1
for mu in altMus:
hist.Fill(mu)
hist.Draw()
tlatex.SetTextAlign(12)
tlatex.DrawLatex(gStyle.GetPadLeftMargin(),0.96,PRELIMINARYSTRING)
tlatex.SetTextAlign(12)
def normalize(dcm):
pixels = (pixels - np.mean(pixels)) / np.stddev(pixels)
begin = datetime.datetime.now()
total = 0
done = False
while not done:
node = shockObjects.next()['node']
start = datetime.datetime.now()
r = requests.get(shockURL + "node/" + node + "?download", auth=(username, password))
data = r.json
downloadTime = datetime.datetime.now() - start
print "genome " + str(total) + " " + shockURL + "node/" + node + "?download" + " " + str(downloadTime)
total += 1
times.append(downloadTime.total_seconds())
except Exception, e:
print str(e)
print "DONE"
ntimes = numpy.array(times)
print "Total time: " + str(datetime.datetime.now() - begin)
print "Max: " + str(numpy.max(ntimes))
print "Min: " + str(numpy.min(ntimes))
print "Average: " + str(numpy.mean(ntimes))
print "Stddev: " + str(numpy.stddev(ntimes))
def samplerate_stddev(self):
"Get the variance of the differential source sample rate"
return np.stddev(np.reciprocate(np.diff(self.fx)))