def orderbook_better_stddev(orderbook, r):
samples = [
orderbook.price_at(timestamp)
for timestamp in r
if orderbook.offer_at(timestamp) != float('inf') and orderbook.bid_at(timestamp) != 0
]
if len(samples) == 0:
return 0
return stddev(samples)
def getPedestral(tDict, vDict, timetupel):
#
# returns dict with timber var as keys and corresponding
# averaged pedestral for time period
#
#
(dtStart, dtEnd, labText) = timetupel
tsStart = stringDateToTimeStamp(dtStart)
tsEnd = stringDateToTimeStamp(dtEnd)
pedList = []
if debug:
print "Starting time", dtStart
print "Ending time", dtEnd
vars = vDict.keys()
for det in vDict.keys():
xarray, yarray = [], []
if debug: print "timber var ", det
detData = tDict[det]
for ts, dt, val in detData:
if ts > tsEnd or ts <= tsStart:
continue
yarray += [val]
meanPedestral = mean(yarray)
stddevPed = stddev(yarray)
pedList += [(det, [meanPedestral, stddevPed])]
pedDict = dict(pedList)
if debug: print "pedDict", pedDict
return pedDict
###################
# Operating point #
###################
delta_V_4 = hp.fetch2('data/operating_point.xlsx', 'err V_4 [mV]')
V_4 = hp.fetch2('data/operating_point.xlsx', 'V_4 [mV]', delta_V_4)/1000
delta_I = hp.fetch2('data/operating_point.xlsx', 'err I [mu A]')
I = hp.fetch2('data/operating_point.xlsx', 'I [mu A]', delta_I)/10**6
T = hp.fetch2('data/operating_point.xlsx', 'T [°C]', 0.1)
R_S = V_4/I
fig, ax1 = plt.subplots()
ax1.errorbar(hp.nominal(I)*10**3, hp.nominal(T), fmt='b.', xerr=hp.stddev(I)*10**3, yerr=hp.stddev(T))
ax1.set_xlabel('Current $I \, [mA]$')
ax1.set_ylabel('Temperature $T \, [^{\circ}C]$', color='b')
ax1.tick_params('y', colors='b')
ax2 = ax1.twinx()
ax2.errorbar([100000], [1000000], fmt='b.', xerr=0, yerr=0, label="measured temperature") # dummy for the label
ax2.errorbar(hp.nominal(I)*10**3, hp.nominal(R_S)/1000, fmt='r.', xerr=hp.stddev(I)*10**3, yerr=hp.stddev(R_S)/1000, label="measured resistance")
ax2.set_ylabel('Resistance $R_S \, [k\Omega]$', color='r')
ax2.tick_params('y', colors='r')
I_Op = 0.4007
V_2_Op = 3.250
hp.replace("I_Op", I_Op)
hp.replace("V_2_Op", V_2_Op)
coeffs3 = hp.phpolyfit(I_mean, deltaTplus + deltaTminus, 2)
p3 = lambda x: np.polyval(coeffs3, x)
x3 = np.linspace(0, 18, 100)
coeffs4 = hp.phpolyfit(I, deltaT, 2)
p4 = lambda x: np.polyval(coeffs4, x)
x4 = np.linspace(-18, 18, 100)
### PLOTS ###
# I - Pi12V
plt.figure(0)
plt.errorbar(hp.nominal(I_mean),
hp.nominal(Pi12V) * 10**3,
xerr=hp.stddev(I_mean),
yerr=hp.stddev(Pi12V) * 10**3,
label=r'' + Temp + '°C')
plt.ylabel(r'$\Pi_{12}^{(V)}$ [mV]')
# I - Pi12I
plt.figure(1)
plt.errorbar(hp.nominal(I_mean),
hp.nominal(Pi12I) * 10**3,
xerr=hp.stddev(I_mean),
yerr=hp.stddev(Pi12I) * 10**3,
label=r'' + Temp + '°C')
plt.ylabel(r'$\Pi_{12}^{(I)}$ [mV]')
# I - dT
plt.figure(2)
hp.replace("delta_turns", hp.physical(delta_turns, 0, 1))
# erf fit
def erf_param(x, a, b, c, d):
return (np.vectorize(lambda x, a, b, c, d: a * erf(b * x + c) + d)(x, a, b,
c, d))
guess_a, guess_b, guess_c, guess_d = -20.0, 1.2, -3.5, 20.0 # erf()
guesses = np.array([guess_a, guess_b, guess_c, guess_d])
popt, pcov = curve_fit(erf_param,
hp.nominal(turns),
hp.nominal(ctps),
p0=guesses,
sigma=hp.stddev(ctps),
absolute_sigma=True)
aerr, berr, cerr, derr = list(np.sqrt(np.diag(pcov)))
a, b, c, d = list(popt)
erf_fit = lambda x: np.vectorize(lambda x: a * erf(b * x + c) + d)(x)
# parameters of the erf fit
hp.replace("fit:erf_a", hp.physical(a, aerr, 4))
hp.replace("fit:erf_b", hp.physical(b, berr, 4))
hp.replace("fit:erf_c", hp.physical(c, cerr, 4))
hp.replace("fit:erf_d", hp.physical(d, derr, 4))
X = np.linspace(0, 6, 500)
gauss = lambda x: 2 * a * b * np.exp(-(b * x + c)**2) / np.sqrt(np.pi)
def orderbook_stddev(orderbook, r):
samples = [orderbook.price_at(timestamp) for timestamp in r]
return stddev(samples)
x = hp.fetch2('data/doserate.xlsx', 'x [cm]', 1) / 100
A = hp.fetch2('data/doserate.xlsx', 'A [muSv/hr]', 0.2)
A[10].sf = 1
def invsq_param(x, a):
return (np.vectorize(lambda x, a: a / (x**2))(x, a))
fit_x = hp.nominal(x)
fit_x[0] = 10**-10 # "very" small => "zero"
guesses = np.array([0.0207])
popt, pcov = curve_fit(invsq_param,
fit_x,
hp.nominal(A),
p0=guesses,
sigma=hp.stddev(A),
absolute_sigma=True)
aerr = list(np.sqrt(np.diag(pcov)))
#a = list(popt)
a = popt[0]
invsq_fit = lambda x: np.vectorize(lambda x: 0.02 / (x**2))(x)
X = np.linspace(0.001, 2.2, 500)
plt.figure(figNr)
plt.errorbar(hp.nominal(x * 100),
hp.nominal(A),
fmt='x',
xerr=hp.stddev(x * 100),
yerr=hp.stddev(A),
label=r'dose rate data')
