def fit_rabi_simple(g_f, g_A, g_a, g_phi, *arg):
"""
fits a cosine,
y(x) = a + A * cos(2pi*f*x)
Initial guesses, in this order:
g_f : frequency
g_A : initial amplitude of the oscillation
g_a : offset
"""
fitfunc_str = "a + A * cos(2pi*(f*x + phi/360))"
f = fit.Parameter(g_f, 'f')
A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
phi = fit.Parameter(g_phi, 'phi')
# tau = fit.Parameter(g_tau, 'tau')
p0 = [f, A, a, phi]
def fitfunc(x):
return a() + A() * cos(2 * pi * (f() * x + phi() / 360.))
return p0, fitfunc, fitfunc_str
def fit_calibration(V, freq):
guess_b = (freq[-1] - freq[0]) / (V[-1] - V[0])
guess_a = freq[0]
print 'Initial guess: ', guess_a, guess_b
a = fit.Parameter(guess_a, 'a')
b = fit.Parameter(guess_b, 'b')
p0 = [a, b]
fitfunc_str = ''
def fitfunc(x):
return a() + b() * x
fit_result = fit.fit1d(V,
freq,
None,
p0=p0,
fitfunc=fitfunc,
fixed=[],
do_print=False,
ret=True)
a_fit = fit_result['params_dict']['a']
b_fit = fit_result['params_dict']['b']
print 'a= ', a_fit
print 'b=', b_fit
return a_fit, b_fit
def fit_sweep_pump_power(x, y, L=0.042):
guess_b = 1
guess_a = max(y)
a = fit.Parameter(guess_a, 'a')
b = fit.Parameter(guess_b, 'b')
p0 = [a, b]
fitfunc_str = ''
def fitfunc(x):
return a() * np.sin(L * (b() * x)**0.5)**2
fit_result = fit.fit1d(x,
y,
None,
p0=p0,
fitfunc=fitfunc,
fixed=[],
do_print=True,
ret=True)
a_fit = fit_result['params_dict']['a']
b_fit = fit_result['params_dict']['b']
print 'a= ', a_fit
print 'b=', b_fit
return a_fit, b_fit
def epulse_fidelity(folder, ax, *args):
a = sequence.SequenceAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='ssro')
a.get_electron_ROC()
a.plot_result_vs_sweepparam(ax=ax, name='ssro')
x = a.sweep_pts.reshape(-1)[:]
y = a.p0.reshape(-1)[:]
of = fit.Parameter(args[0], 'of')
of2 = fit.Parameter(0, 'of2')
of3 = fit.Parameter(0, 'of3')
fitfunc_str = '(1-of)'
def fitfunc_fid(x):
return (1. - of())
fit_result = fit.fit1d(x,
y,
None,
p0=[of],
fixed=[],
fitfunc=fitfunc_fid,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
#plot.plot_fit1d(fit_result, np.linspace(x[0],x[-1],201), ax=ax,
# plot_data=False, print_info=False)
return fit_result
def calibrate_epulse_amplitude(folder, ax, *args):
a = sequence.SequenceAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='ssro')
a.get_electron_ROC()
a.plot_result_vs_sweepparam(ax=ax, name='ssro')
x = a.sweep_pts.reshape(-1)[:]
y = a.p0.reshape(-1)[:]
ax.set_title(a.timestamp)
x0 = fit.Parameter(args[0], 'x0')
of = fit.Parameter(args[1], 'of')
a = fit.Parameter(args[2], 'a')
fitfunc_str = '(1-of) + a (x-x0)**2'
def fitfunc_parabolic(x):
return (1. - of()) + a() * (x - x0())**2
fit_result = fit.fit1d(x,
y,
None,
p0=[of, a, x0],
fitfunc=fitfunc_parabolic,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
plot.plot_fit1d(fit_result,
np.linspace(x[0], x[-1], 201),
ax=ax,
plot_data=False,
print_info=True)
return fit_result
def fit_linear(folder, ax, value, *args):
a = sequence.SequenceAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='ssro')
a.get_electron_ROC()
a.plot_result_vs_sweepparam(ax=ax, name='ssro')
x = a.sweep_pts.reshape(-1)[:]
y = a.p0.reshape(-1)[:]
ax.set_title(a.timestamp)
a = fit.Parameter(args[0], 'a')
b = fit.Parameter(args[0], 'b')
fitfunc_str = 'a x + b'
def fitfunc(x):
return a() * x + b()
fit_result = fit.fit1d(x,
y,
None,
p0=[a, b],
fitfunc=fitfunc,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
plot.plot_fit1d(fit_result,
np.linspace(x[0], x[-1], 201),
ax=ax,
plot_data=False,
print_info=False)
return fit_result
def fit_rabi_on_linslope(g_f, g_A, g_a, g_b, *arg):
"""
fits (and plots) a cosine on a slope, with offset,
y(x) = a + bx + Acos(f*x + phi)
Initial guesses (1st args, in this order):
g_f : float
guess for the frequency
g_A : float
guess for amplitude
g_a : float
guess for offset
g_b = 0. : float
guess for slope
"""
fitfunc_str = "a + b*x + A*cos(2pi*f*x + phi)"
f = fit.Parameter(g_f, 'frequency')
A = fit.Parameter(g_A, 'amplitude')
a = fit.Parameter(g_a, 'offset')
b = fit.Parameter(g_b, 'slope')
p0 = [f, A, a, b]
def fitfunc(x):
return a() + b() * x + A() * cos(2 * pi * f() * x)
return p0, fitfunc, fitfunc_str
def fit_correlation_parabolic(folder, ax, *args, **kw):
which_correlation = kw.pop('which_correlation', 2)
a = mbi.MBIAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='adwindata')
a.get_correlations(name='adwindata')
a.plot_results_vs_sweepparam(ax=ax, mode='correlations')
x = a.sweep_pts.reshape(-1)[:]
y = a.normalized_correlations[:, which_correlation]
x0 = fit.Parameter(args[0], 'x0')
of = fit.Parameter(args[1], 'of')
a = fit.Parameter(args[2], 'a')
fitfunc_str = '(1-of) + a (x-x0)**2'
def fitfunc_parabolic(x):
return (1. - of()) + a() * (x - x0())**2
fit_result = fit.fit1d(x,
y,
None,
p0=[of, a, x0],
fitfunc=fitfunc_parabolic,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
plot.plot_fit1d(fit_result,
np.linspace(x[0], x[-1], 201),
ax=ax,
plot_data=False,
print_info=False)
return fit_result
def fit_rabi_damped_exp(g_f, g_A, g_a, g_tau, *arg):
"""
fits a cosine thats damped exponentially,
y(x) = a + A * exp(-x/tau) * cos(f*x + phi)
Initial guesses, in this order:
g_f : frequency
g_A : initial amplitude of the oscillation
g_a : offset1
### g_phi : phase
g_tau : decay constant
"""
fitfunc_str = "a + A * exp(-x/tau) * cos(f*x + phi)"
f = fit.Parameter(g_f, 'f')
A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
phi = fit.Parameter(0., 'phi')
tau = fit.Parameter(g_tau, 'tau')
p0 = [f, A, a, tau, phi]
#print tau
def fitfunc(x):
return a() + A() * exp(-x / tau()) * cos(2 * pi *
(f() * x + phi() / 360.))
return p0, fitfunc, fitfunc_str
def fit_linear(folder=None, ax=None, a_guess=1., b_guess=0.):
"""
fit a x + b from Sequence in folder
"""
a, ax, x, y = plot_result(folder, ax)
a = fit.Parameter(a_guess, 'a')
b = fit.Parameter(b_guess, 'b')
fitfunc_str = 'a x + b'
def fitfunc(x):
return a() * x + b()
fit_result = fit.fit1d(x,
y,
None,
p0=[a, b],
fitfunc=fitfunc,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
plot.plot_fit1d(fit_result,
np.linspace(x[0], x[-1], 201),
ax=ax,
plot_data=False,
print_info=False)
return fit_result
def fit_calibration(self, V, freq, fixed):
guess_b = -21.3
guess_c = -0.13
guess_d = 0.48
guess_a = freq[0] +3*guess_b-9*guess_c+27*guess_d
a = fit.Parameter(guess_a, 'a')
b = fit.Parameter(guess_b, 'b')
c = fit.Parameter(guess_c, 'c')
d = fit.Parameter(guess_d, 'd')
p0 = [a, b, c, d]
fitfunc_str = ''
def fitfunc(x):
return a()+b()*x+c()*x**2+d()*x**3
fit_result = fit.fit1d(V,freq, None, p0=p0, fitfunc=fitfunc, fixed=fixed,
do_print=False, ret=True)
if (len(fixed)==2):
a_fit = fit_result['params_dict']['a']
c_fit = fit_result['params_dict']['c']
return a_fit, c_fit
else:
a_fit = fit_result['params_dict']['a']
b_fit = fit_result['params_dict']['b']
c_fit = fit_result['params_dict']['c']
d_fit = fit_result['params_dict']['d']
return a_fit, b_fit, c_fit, d_fit
def fast_pi2():
folder = toolbox.latest_data('cal_fast_pi_over_2')
a = mbi.MBIAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='adwindata')
a.get_electron_ROC()
#a.plot_results_vs_sweepparam(ax=ax, name='adwindata')
x = a.sweep_pts
y = a.p0.reshape(-1)
u_y = a.u_p0.reshape(-1)
n = a.sweep_name
x2 = x[::2]
y2 = y[1::2] - y[::2]
u_y2 = np.sqrt(u_y[1::2]**2 + u_y[::2]**2)
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4), sharex=True)
ax1.errorbar(x2, y2, yerr=u_y2, fmt='o')
ax1.set_xlabel(n)
ax1.set_title('Difference btw. Pi/2-Pi and Pi/2' + '\n' + a.timestamp)
ax1.set_ylabel('Difference')
m = fit.Parameter((y[-1] - y[0]) / (x[-1] - x[0]), 'm')
x0 = fit.Parameter(x2.mean(), 'x0')
p0 = [m, x0]
def ff(x):
return m() * (x - x0())
fitfunc_str = 'm * (x - x0)'
fit_result = fit.fit1d(x2,
y2,
None,
p0=p0,
fitfunc=ff,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
ax2.errorbar(x2, y[0::2], yerr=u_y[0::2], fmt='o', label='Pi/2')
ax2.errorbar(x2, y[1::2], yerr=u_y[1::2], fmt='o', label='Pi/2 - Pi')
ax2.legend(frameon=True, framealpha=0.5)
ax2.set_ylabel('P(0)')
ax2.set_xlabel(n)
if fit_result != False:
ax2.axvline(x0(), c='k', lw=2)
ax2.axhline(0.5, c='k', lw=2)
ax2.set_title('X marks the spot')
plot.plot_fit1d(fit_result,
np.linspace(x2[0], x2[-1], 201),
ax=ax1,
plot_data=False,
print_info=True)
return x0(), 0
def fit_5msmt_proc_fid(g_A0, g_offset0, g_t, g_p):
'''
g_offset0 - Offset, derived from the 0 msmt case. Fixed.
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 5 Zeno-measurements and fit the process fidelity with a prederived function. That depends on one parameter. The fidelity of the parity measurement.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
offset0 = fit.Parameter(g_offset0, 'offset0')
### Ramsey
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, offset0, t, p]
def fitfunc(x):
decay = -np.exp(-(x**2 / (2. * (t()**2)))) / 64.
decay2 = 6. * np.exp((5. / 18.) * (x**2 / ((t()**2))))
decay3 = 15. * np.exp((4. / 9.) * (x**2 / ((t()**2)))) * (p() - 1.)**5
bracket1 = 10 + (-5 + p()) * p()
bracket2 = (bracket1 * p() - 10) * p() + 5
bracket3 = (bracket2 * 5 * p() - 21) * 2. / 64.
Hahn = np.exp(-(x / 604.)**1.6162)
#### decaying parts + the constant part which arises from the measurement echo.
Fx = decay * ((p() - 1)**5 *
(1 + decay2) + decay3) - (bracket3 + 0.5) * Hahn + 0.5
Fz0 = 2. * offset0()
Con0 = 2. * (Fz0 - 0.5)
Conp = Con0 * (1 - p())**5
Fzp = Conp / 2. + 0.5
# am = (2*A0()+1)/2.
# return (2.*am*Fx+Fzp-1.)/2.
### commented out because apparently wrong.
Cx = 2 * Fx - 1.
Cxred = 2 * A0() * Cx
Fx = (Cxred + 1) / 2.
return (2. * Fx + Fzp - 1.) / 2.
return p0, fitfunc, fitfunc_str
def calibrate_pi2_noMBI(folder):
a = sequence.SequenceAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results('ssro')
a.get_electron_ROC()
x = a.sweep_pts
y = a.p0
u_y = a.u_p0
n = a.sweep_name
a.finish()
x2 = x[::2]
print x2
y2 = y[1::2] - y[::2]
u_y2 = np.sqrt(u_y[1::2]**2 + u_y[::2]**2)
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4), sharex=True)
ax1.errorbar(x2, y2, yerr=u_y2, fmt='o')
ax1.set_xlabel(n)
ax1.set_title('Difference btw. Pi/2-Pi and Pi/2')
ax1.set_ylabel('Difference')
m = fit.Parameter((y[-1] - y[0]) / (x[-1] - x[0]), 'm')
x0 = fit.Parameter(y2.mean(), 'x0')
p0 = [m, x0]
def ff(x):
return m() * (x - x0())
fitfunc_str = 'm * (x - x0)'
fit_result = fit.fit1d(x2,
y2,
None,
p0=p0,
fitfunc=ff,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
ax2.errorbar(x2, y[0::2], yerr=u_y[0::2], fmt='o', label='Pi/2')
ax2.errorbar(x2, y[1::2], yerr=u_y[1::2], fmt='o', label='Pi/2 - Pi')
ax2.legend(frameon=True, framealpha=0.5)
ax2.set_ylabel('P(0)')
ax2.set_xlabel(n)
ax2.axvline(x0(), c='k', lw=2)
ax2.axhline(0.5, c='k', lw=2)
ax2.set_title('X marks the spot')
plot.plot_fit1d(fit_result,
np.linspace(x2[0], x2[-1], 201),
ax=ax1,
plot_data=False,
print_info=True)
fig.savefig(os.path.join(folder, 'pi2_calibration.png'))
def fit_ramsey_gaussian_decay(g_tau, g_a, *arg):
"""
fitfunction for a ramsey modulation, with gaussian decay,
y(x) = a + A*exp(-(x/tau)**2) * mod,
where:
mod = sum_i(cos(2pi*f_i*x +phi) - 1)
Initial guesses (in this order):
g_tau : decay const
g_A : Amplitude
g_a : offset
For the modulation:
an arbitrary no of tuples, in the form
(g_f, g_A)[i] = (frequency, Amplitude, phase)[i]
"""
fitfunc_str = 'y(x) = a + exp(-(x/tau)**2)*('
no_frqs = len(arg)
if no_frqs == 0:
print 'no modulation frqs supplied'
return False
tau = fit.Parameter(g_tau, 'tau')
# A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
p0 = [tau, a]
print 'fitting with %d modulation frequencies' % no_frqs
frqs = []
amplitudes = []
phases = []
print 'arg2 =' + str(arg)
for i, m in enumerate(arg):
fitfunc_str += 'A%d*cos(2pi*f%d*x+phi%d)' % (i, i, i)
frqs.append(fit.Parameter(m[0], 'f%d' % i))
phases.append(fit.Parameter(m[2], 'phi%d' % i))
amplitudes.append(fit.Parameter(m[1], 'A%d' % i))
p0.append(frqs[i])
p0.append(amplitudes[i])
p0.append(phases[i])
fitfunc_str += ')'
def fitfunc(x):
prd = exp(-(x / tau())**2)
mod = 0
for i in range(no_frqs):
mod += amplitudes[i]() * (cos(2 * pi * frqs[i]() * x +
phases[i]()))
return a() + prd * mod
return p0, fitfunc, fitfunc_str
def fit_parabolic(folder=None,
ax=None,
x0_guess=0.,
of_guess=0.5,
a_guess=1.,
**kw):
"""
fit (1-of) + a (x-x0)**2 from Sequence in folder
"""
do_print = kw.pop('do_print', False)
fixed = kw.pop('fixed', [])
a0, ax, x, y = plot_result(folder, ax, **kw)
x0 = fit.Parameter(x0_guess, 'x0')
of = fit.Parameter(of_guess, 'of')
a = fit.Parameter(a_guess, 'a')
x_min = kw.pop("x_min", None)
x_max = kw.pop("x_max", None)
fit_x = x
fit_y = y
if x_min is not None:
mask = fit_x > x_min
fit_x = fit_x[mask]
fit_y = fit_y[mask]
if x_max is not None:
mask = fit_x < x_max
fit_x = fit_x[mask]
fit_y = fit_y[mask]
fitfunc_str = '(1-of) + a (x-x0)**2'
def fitfunc_parabolic(x):
return (1. - of()) + a() * (x - x0())**2
fit_result = fit.fit1d(fit_x,
fit_y,
None,
p0=[of, a, x0],
fitfunc=fitfunc_parabolic,
fitfunc_str=fitfunc_str,
do_print=do_print,
ret=True)
fit_result['u_y'] = a0.u_p0
plot.plot_fit1d(fit_result,
np.linspace(np.min(fit_x), np.max(fit_x), 201),
ax=ax,
plot_data=False,
**kw)
return fit_result
def fit_8msmt_state_fid(g_A0, g_t, g_p, contrast=False):
'''
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 8 Zeno-measurements and fit the state fidelity with a prederived function.
That depends on one parameter. The Probability of the parity measurement to mix the state.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
### Ramsey (decay time divided by sqrt(2).)
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, t, p]
def fitfunc(x):
decay = np.exp(-(x**2 / (2. * (t()**2)))) / 512.
decay2 = 9. * np.exp((16. / 81.) * (x**2 / ((t()**2)))) * (p() - 1.)**8
decay3 = 36. * np.exp(
(28. / 81.) * (x**2 / ((t()**2)))) * (p() - 1.)**8
decay4 = 84. * np.exp(
(36. / 81.) * (x**2 / ((t()**2)))) * (p() - 1.)**8
decay5 = 126. * np.exp(
(40. / 81.) * (x**2 / ((t()**2)))) * (p() - 1.)**8
#### decay to 0.5 only for 6 measurements no other constant parts of the function due to echo effects.
Fx = 0.5 + decay * ((p() - 1)**8 + decay2 + decay3 + decay4 + decay5)
Cx = A0() * 2. * (Fx - 0.5)
Fx = Cx / 2. + 0.5
if contrast: return Cx
else: return Fx
return p0, fitfunc, fitfunc_str
def fit_3msmt_proc_fid(g_A0, g_offset0, g_t, g_p):
'''
g_offset0 - Offset, derived from the 0 msmt case. Fixed.
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 3 Zeno-measurements and fit the process fidelity with a prederived function. That depends on one parameter. The fidelity of the parity measurement.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
offset0 = fit.Parameter(g_offset0, 'offset0')
### Ramsey
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, offset0, t, p]
def fitfunc(x):
decay = -0.0625 * np.exp(-(x**2 / (2. * (t()**2))))
decay2 = np.exp((3. / 8.) * (x**2 / ((t()**2))))
Hahn = np.exp(-(x / 604.)**1.6162)
Fx = -Hahn * ((-11. + 3 * p() *
(3 + (3 - p()) * p())) / 16. + 0.5) + decay * (
(p() - 1.)**3 + 4. * decay2 * (p() - 1.)**3) + 0.5
Fz0 = 2. * offset0()
Con0 = 2. * (Fz0 - 0.5)
Conp = Con0 * (1 - p())**3
Fzp = Conp / 2. + 0.5
# am = (2*A0()+1)/2.
# return (2.*am*Fx+Fzp-1.)/2.
### commented out because apparently wrong.
Cx = 2 * Fx - 1.
Cxred = 2 * A0() * Cx
Fx = (Cxred + 1) / 2.
return (2. * Fx + Fzp - 1.) / 2.
return p0, fitfunc, fitfunc_str
def fit_5msmt_state_fid(g_A0, g_t, g_p, contrast=False):
'''
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 5 Zeno-measurements and fit the process fidelity with a prederived function. That depends on one parameter. The fidelity of the parity measurement.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
### Ramsey
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, t, p]
def fitfunc(x):
Hahn = np.exp(-(x / 604.)**1.6162)
decay = -np.exp(-(x**2 / (2. * (t()**2)))) / 64.
decay2 = 6. * np.exp((5. / 18.) * (x**2 / ((t()**2))))
decay3 = 15. * np.exp((4. / 9.) * (x**2 / ((t()**2)))) * (p() - 1.)**5
bracket1 = 10 + (-5 + p()) * p()
bracket2 = (bracket1 * p() - 10) * p() + 5
bracket3 = (bracket2 * 5 * p() - 21) * 2. / 64.
#### decaying parts + the constant part which arises from the measurement echo.
Fx = decay * ((p() - 1)**5 *
(1 + decay2) + decay3) - (bracket3 + 0.5) * Hahn + 0.5
Cx = A0() * 2. * (Fx - 0.5)
Fx = Cx / 2. + 0.5
if contrast: return Cx
else: return Fx
return p0, fitfunc, fitfunc_str
def fit_4msmt_proc_fid(g_A0, g_offset0, g_t, g_p):
'''
g_offset0 - Offset, derived from the 0 msmt case. Fixed.
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 4 Zeno-measurements and fit the process fidelity with a prederived function. That depends on one parameter. The fidelity of the parity measurement.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
offset0 = fit.Parameter(g_offset0, 'offset0')
### Ramsey
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, offset0, t, p]
def fitfunc(x):
decay = np.exp(-(x**2 / (2. * (t()**2)))) / 32.
decay2 = 5. * np.exp((8. / 25.) * (x**2 / ((t()**2)))) * (p() - 1.)**4
decay3 = 10. * np.exp(
(12. / 25.) * (x**2 / ((t()**2)))) * (p() - 1.)**4
Fx = 0.5 + decay * ((p() - 1)**4 + decay2 + decay3)
Fz0 = 2. * offset0()
Con0 = 2. * (Fz0 - 0.5)
Conp = Con0 * (1 - p())**4
Fzp = Conp / 2. + 0.5
# am = (2*A0()+1)/2.
# return (2.*am*Fx+Fzp-1.)/2.
### commented out because apparently wrong.
Cx = 2 * Fx - 1.
Cxred = 2 * A0() * Cx
Fx = (Cxred + 1) / 2.
return (2. * Fx + Fzp - 1.) / 2.
return p0, fitfunc, fitfunc_str
def calibrate_epulse_rabi(folder, ax, *args, **kws):
fit_x0 = kws.pop('fit_x0', True)
fit_k = kws.pop('fit_k', True)
double_ro = kws.pop('double_ro', 'False')
guess_x0 = kws.pop('guess_x0', 0)
a = mbi.MBIAnalysis(folder)
a.get_sweep_pts()
a.get_readout_results(name='adwindata')
a.get_electron_ROC()
a.plot_results_vs_sweepparam(ax=ax, name='adwindata')
x = a.sweep_pts.reshape(-1)[:]
if double_ro == 'electron':
y = a.p0[:, 0]
elif double_ro == 'nitrogen':
y = a.p0[:, 1]
else:
y = a.p0.reshape(-1)[:]
f = fit.Parameter(args[0], 'f')
A = fit.Parameter(args[1], 'A')
x0 = fit.Parameter(guess_x0, 'x0')
k = fit.Parameter(0, 'k')
p0 = [f, A]
if fit_x0:
p0.append(x0)
if fit_k:
p0.append(k)
fitfunc_str = '(1 - A) + A * exp(-kx) * cos(2pi f (x - x0))'
def fitfunc(x):
return (1.-A()) + A() * np.exp(-k()*x) * \
np.cos(2*np.pi*f()*(x - x0()))
fit_result = fit.fit1d(x,
y,
None,
p0=p0,
fitfunc=fitfunc,
fitfunc_str=fitfunc_str,
do_print=True,
ret=True)
plot.plot_fit1d(fit_result,
np.linspace(x[0], x[-1], 201),
ax=ax,
plot_data=False,
print_info=False)
return fit_result
def fit_bi_exp(g_A1, g_A2, g_t1, g_t2,*arg):
fitfunc_str = 'A1 *exp(-x/t1) + A2 *exp(-x/t2)'
A1 = fit.Parameter(g_A1, 'A1')
A2 = fit.Parameter(g_A2, 'A2')
t1 = fit.Parameter(g_t1, 't1')
t2 = fit.Parameter(g_t2, 't2')
p0 = [A1,A2,t1,t2]
def fitfunc(x):
return A1()*np.exp(-x/t1()) + A2()*np.exp(-x/t2())
return p0, fitfunc, fitfunc_str
def fit_rabi_fixed_upper(g_f, g_A, g_phi, g_k, *arg):
fitfunc_str = '(1-A) + A * exp(-kx) *cos(2pi*(f*x + phi/360))'
f = fit.Parameter(g_f, 'f')
A = fit.Parameter(g_A, 'A')
phi = fit.Parameter(g_phi, 'phi')
k = fit.Parameter(g_k, 'k')
p0 = [f, A, phi, k]
def fitfunc(x):
return (1. - A()) + A() * exp(-k() * x) * cos(2 * pi *
(f() * x + phi() / 360.))
return p0, fitfunc, fitfunc_str
def fit_population_vs_detuning(g_a, g_A, g_F, g_x0, *arg):
fitfunc_str = 'a + A * F**2/(F**2+(x-x0)**2) * sin(pi/(2F) * sqrt(F**2+(x-x0)**2))**2'
A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
F = fit.Parameter(g_F, 'F')
x0 = fit.Parameter(g_x0, 'x0')
p0 = [a, A, F, x0]
def fitfunc(x):
return a() + A() * F()**2 / (F()**2 + (x - x0())**2) * sin(
pi / F() / 2. * sqrt(F()**2 + (x - x0())**2))**2
return p0, fitfunc, fitfunc_str
def fit_ramsey_hyperfinelines_fixed(g_tau, g_A, g_a, g_det, g_hf, g_phi1,
g_phi2, g_phi3, *arg):
"""
fitfunction for a gaussian decay,
y(x) = a + A*exp(-(x/tau)**2)
Initial guesses (in this order):
g_tau : decay constant
g_A : amplitude
g_a : offset
"""
tau = fit.Parameter(g_tau, 'tau')
A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
det = fit.Parameter(g_det, 'det')
hf = fit.Parameter(g_hf, 'hf')
phi1 = fit.Parameter(g_phi1, 'phi1')
phi2 = fit.Parameter(g_phi2, 'phi2')
phi3 = fit.Parameter(g_phi3, 'phi3')
p0 = [tau, A, a, det, hf, phi1, phi2, phi3]
fitfunc_str = 'sumf of three cos and decay'
def fitfunc(x):
return a() + A() * exp(-(x / tau())**2) * (
cos(2 * pi * det() * x + phi1()) + cos(2 * pi *
(det() - hf()) * x + phi2())
+ cos(2 * pi * (det() + hf()) * x + phi3())) / 3.
return p0, fitfunc, fitfunc_str
def fit_general_exponential(g_a, g_A, g_x0, g_T, g_n):
fitfunc_str = 'a + A * exp(-((x-x0)/T )**n)'
a = fit.Parameter(g_a, 'a')
A = fit.Parameter(g_A, 'A')
x0 = fit.Parameter(g_x0, 'x0')
T = fit.Parameter(g_T, 'T')
n = fit.Parameter(g_n, 'n')
p0 = [a, A, x0, T, n]
def fitfunc(x):
return a() + A() * np.exp(-(x-x0())**n()/(T()**n()))
return p0, fitfunc, fitfunc_str
def fit_rabi_multiple_detunings(g_A, g_a, g_F, g_tau, *arg):
"""
fitfunction for an oscillation that drives several transitions
(several nuclear lines, for instance)
y(x) = a + A * sum_I[ F**2/(F**2 + delta_i**2) *
(cos(sqrt(F**2 + delta_i**2 + phi_i)) - 1) * exp(-x/tau)
Initial guesses:
g_A : full Rabi amplitude
g_a : offset
g_F : Rabi frequency
g_tau : exp decay constant
For the driven levels:
all successive args are treated as detunings for additional
levels, (delta_i, g_phi_i) -- given as tuples!
detuning is not a free param, but is given exactly.
"""
fitfunc_str = "a + A * sum_I[ F**2/(F**2 + delta_i**2) * (cos(sqrt(F**2 + delta_i**2 + phi_i)) - 1) * exp(-x/tau)"
no_detunings = len(arg)
A = fit.Parameter(g_A, 'A')
a = fit.Parameter(g_a, 'a')
F = fit.Parameter(g_F, 'F')
tau = fit.Parameter(g_tau, 'tau')
p0 = [A, a, F, tau]
detunings = []
phases = []
for i, d in enumerate(arg):
fitfunc_str += '\ndetuning d%d := %f' % (i, d[0])
detunings.append(d[0])
#phases.append(fit.Parameter(d[1], 'phi%d'%i))
#p0.append(phases[i])
def fitfunc(x):
val = a()
for i, d in enumerate(detunings):
f2 = F()**2 + d**2
val += A() * (F()**2 / f2) * (cos(2 * pi * sqrt(f2) * x + 0.) - 1)
return val * exp(-x / tau())
return p0, fitfunc, fitfunc_str
def fit_exp(x=[], y=[], plot_fits=False):
guess_b = 10.
guess_c = 0.
b = fit.Parameter(guess_b, 'b')
def fitfunc(x):
return np.exp(-np.abs((x) / b()))
xx = np.linspace(0, x[-1], 1000)
fit_result = fit.fit1d(x,
y,
None,
p0=[b],
fitfunc=fitfunc,
do_print=False,
ret=True)
b_fit = fit_result['params_dict']['b']
err_b = fit_result['error_dict']['b']
if plot_fits:
plt.figure()
plt.plot(x * 1000, y, '.b')
plt.plot(xx * 1000, np.exp(-np.abs((xx) / b_fit)), 'r')
plt.xlabel('time [msec]')
plt.show()
return b_fit, err_b
def g2_hist(self, range=(-520,520), bins=2*52):
# NOTE bins should be an even number!
self.peaks = {}
self.amplitudes = []
self.normpeaks = {}
for i,d in enumerate(self.deltas):
hy,hx = np.histogram(self.coincidences[i]+self.offset,
bins=bins, range=range)
self.peaks[d] = (hy,hx)
self.fitdt = hx[1]-hx[0]
if d != 0:
A = fit.Parameter(max(hy))
def fitfunc(x):
return A()*np.exp(-abs(x)/self.tau)
fit.fit1d(hx[:-1]+(hx[1]-hx[0])/2., hy, None, fitfunc=fitfunc, p0=[A], do_print=True,
fixed=[])
self.amplitudes.append(A())
# normalize peaks
self.meanamp = np.mean(self.amplitudes)
for d in self.peaks:
self.normpeaks[d] = (self.peaks[d][0]/self.meanamp, self.peaks[d][1])
return True
def fit_16msmt_state_fid(g_A0, g_t, g_p, contrast=False):
'''
g_A0 - Amplitude derived from the 0 msmt case. Fixed.
g_p - Probability for faulty parity measurement
g_t - Decay of Ramsey. Fixed.
The function should take the data set for 16 Zeno-measurements and fit the state fidelity with a prederived function.
That depends on one parameter. The Probability of the parity measurement to mix the state.
'''
fitfunc_str = '''analyitcal solution'''
### Parameters
A0 = fit.Parameter(g_A0, 'A0')
### Ramsey (decay time divided by sqrt(2).)
t = fit.Parameter(g_t, 't')
### Zeno
p = fit.Parameter(g_p, 'p')
p0 = [A0, t, p]
def fitfunc(x):
N = 16
Coefficients = []
for k in range(N + 1):
Coefficients.append(
sp.special.binom(N + 1, k) *
np.exp(-(((N + 1 - 2 * k) * x /
(N + 1) / t())**2) / 2.) / 2**(N + 1))
Cx = A0() * (1 - p())**N * (
np.sum(Coefficients, axis=0)
) # WATCH OUT, need to sepcify axis (the axis of coefficients.)
Fx = Cx / 2. + 0.5
if contrast: return Cx
else: return Fx
return p0, fitfunc, fitfunc_str
