Merge 2a54499 into b7ae604

README.md

@@ -46,7 +46,7 @@ Y2.cache_filter_function(omega)
 X.cache_filter_function(omega)
 
 hadamard = Y2 @ X           # equivalent: ff.concatenate((Y2, X))
-hadamard.is_cached('F')
+hadamard.is_cached('filter function')
 # True  (filter function cached during concatenation)
 ```
 

doc/source/examples/examples.rst

@@ -8,7 +8,7 @@ This directory contains static examples that can also be run interactively from
    getting_started
    periodic_driving
    advanced_concatenation
-   calculating_error_transfer_matrices
+   calculating_quantum_processes
    extending_pulses
    qutip_integration
    quantum_fourier_transform

doc/source/examples/getting_started.ipynb

@@ -31,7 +31,7 @@
     "    {H}_n &= \\sum_\\alpha s_\\alpha(t) b_\\alpha(t) B_\\alpha \\\\\n",
     "\\end{align*}\n",
     "\n",
-    "where $A_i$ and $B_\\alpha$ are the control and noise operators, respectively, $a_i(t)$ the control strength of $A_i$ at time $t$, $s_\\alpha(t)$ the sensitivity at time $t$ of the system to noise source $\\alpha$, and $b_\\alpha(t)$ classically fluctuating noise variables. Since the noise is captured in a spectral density function that accounts for different realizations of the random noise, the $b_\\alpha(t)$ are not required at instantiation of a `PulseSequence` instance. Note that we always calculate in units where $\\hbar\\equiv 1$.\n",
+    "where $A_i$ and $B_\\alpha$ are the control and noise operators, respectively, $a_i(t)$ the control strength of $A_i$ at time $t$, $s_\\alpha(t)$ a deterministic time dependence of the noise operators to model, for instance, non-linear coupling to the noise sources, and $b_\\alpha(t)$ classically fluctuating noise variables. Since the noise is captured in a spectral density function that accounts for different realizations of the random noise, the $b_\\alpha(t)$ are not required at instantiation of a `PulseSequence` instance. Note that we always calculate in units where $\\hbar\\equiv 1$.\n",
     "\n",
     "The `PulseSequence` class requires three positional arguments at instantiation; `H_c`, `H_n`, and `dt`, with `dt` the time-deltas of piece-wise constant control. The former two represent the control and noise Hamiltonians and are passed in the same nested-list-of-lists structure of operators and coefficient lists similar to that required by [QuTiP](http://qutip.org/) (the difference being that QuTiP requires implicit functions for calculating the coefficients instead of explicit values). Optionally, we pass unique identifiers for each operator as a third element of the list. That is, \n",
     "\n",

doc/source/examples/periodic_driving.ipynb

@@ -183,7 +183,7 @@
     {
      "data": {
       "text/plain": [
-       "<matplotlib.legend.Legend at 0x7fb642ac4fa0>"
+       "<matplotlib.legend.Legend at 0x220172b7490>"
       ]
      },
      "execution_count": 6,
@@ -249,7 +249,7 @@
     {
      "data": {
       "application/vnd.jupyter.widget-view+json": {
-       "model_id": "30b23e89df10479c8c339c7abd7c19ff",
+       "model_id": "04190ea704d242d58228b8357a015184",
        "version_major": 2,
        "version_minor": 0
       },
@@ -296,16 +296,16 @@
      "output_type": "stream",
      "text": [
       "===========================================\n",
-      "ATOMIC initialization\t\t : 0.0013 s\n",
-      "ATOMIC filter function\t\t : 0.0395 s\n",
-      "NOT concatenation (periodic)\t : 0.0837 s\n",
-      "NOT concatenation (standard)\t : 5.5189 s\n",
-      "ECHO concatenation\t\t : 0.0384 s\n",
+      "ATOMIC initialization\t\t : 0.0018 s\n",
+      "ATOMIC filter function\t\t : 0.0354 s\n",
+      "NOT concatenation (periodic)\t : 0.1257 s\n",
+      "NOT concatenation (standard)\t : 5.8858 s\n",
+      "ECHO concatenation\t\t : 0.0454 s\n",
       "-------------------------------------------\n",
-      "Total (periodic)\t\t : 0.1628 s\n",
-      "Total (standard)\t\t : 5.5980 s\n",
+      "Total (periodic)\t\t : 0.2083 s\n",
+      "Total (standard)\t\t : 5.9684 s\n",
       "===========================================\n",
-      "Total (brute force)\t\t : 148.03 s\n",
+      "Total (brute force)\t\t : 242.65 s\n",
       "===========================================\n"
      ]
     }
@@ -353,7 +353,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.1"
+   "version": "3.8.3"
   },
   "widgets": {
    "application/vnd.jupyter.widget-state+json": {

doc/source/examples/quantum_fourier_transform.ipynb

@@ -1094,7 +1094,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.1"
+   "version": "3.8.3"
   },
   "widgets": {
    "application/vnd.jupyter.widget-state+json": {

doc/source/examples/qutip_integration.ipynb

@@ -138,7 +138,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.1"
+   "version": "3.8.3"
   }
  },
  "nbformat": 4,

examples/qft.py

@@ -34,6 +34,7 @@
 import filter_functions as ff
 import qutip as qt
 from qutip import qip
+from qutip.qip.algorithms.qft import qft as qt_qft
 
 # %% Define some functions
 
@@ -147,7 +148,7 @@ def QFT_pulse(N: int = 4, tau: float = 1):
 prop = ff.util.mdot(swaps) @ QFT.total_Q
 qt.matrix_histogram_complex(prop)
 print('Correct action: ',
-      ff.util.oper_equiv(prop, qip.algorithms.qft.qft(N), eps=1e-14))
+      ff.util.oper_equiv(prop, qt_qft(N), eps=1e-14))
 
 fig, ax, _ = ff.plot_filter_function(QFT, omega)
 # Move the legend to the side because of many entries

examples/randomized_benchmarking.py

@@ -23,6 +23,7 @@
 between gates captured in the filter functions.
 """
 import time
+from typing import Dict, Sequence
 
 import numpy as np
 import qutip as qt
@@ -37,15 +38,15 @@
 # %%
 
 
-def fitfun(m, A, B):
-    return A*m + B
+def fitfun(m, A):
+    return 1 - A*m
 
 
 def state_infidelity(pulse: PulseSequence, S: ndarray, omega: ndarray,
-                     ind: int = 2) -> float:
+                     ind: int = 3) -> float:
     """Compute state infidelity for input state eigenstate of pauli *ind*"""
     R = pulse.get_control_matrix(omega)
-    F = np.einsum('jko->jo', util.abs2(R[:, np.delete([0, 1, 2], ind)]))
+    F = np.einsum('jko->jo', util.abs2(R[:, np.delete([0, 1, 2, 3], ind)]))
     return np.trapz(F*S, omega)/(2*np.pi*pulse.d)
 
 
@@ -68,8 +69,10 @@ def find_inverse(U: ndarray) -> ndarray:
 
 
 def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
-                                omega):
-    infidelities = np.empty((N_l, N_G), dtype=float)
+                                alpha: Sequence[float],
+                                spectra: Dict[float, Sequence[float]],
+                                omega: Sequence[float]):
+    infidelities = {a: np.empty((N_l, N_G), dtype=float) for a in alpha}
     lengths = np.round(np.linspace(min_l, max_l, N_l)).astype(int)
     delta_t = []
     t_now = [time.perf_counter()]
@@ -85,9 +88,10 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
             U = ff.concatenate(cliffords[randints])
             U_inv = find_inverse(U.total_Q)
             pulse_sequence = U @ U_inv
-            infidelities[l, j] = state_infidelity(
-                pulse_sequence, S, omega
-            ).sum()
+            for k, a in enumerate(alpha):
+                infidelities[a][l, j] = state_infidelity(
+                    pulse_sequence, spectra[a], omega
+                ).sum()
 
     return infidelities, delta_t
 
@@ -97,17 +101,17 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
 H_n = {}
 H_c = {}
 dt = {}
-H_c['Id'] = [[qt.sigmax().full(), [0], 'X']]
-H_n['Id'] = [[qt.sigmax().full(), [1], 'X'],
-             [qt.sigmay().full(), [1], 'Y']]
+H_c['Id'] = [[qt.sigmax().full()/2, [0], 'X']]
+H_n['Id'] = [[qt.sigmax().full()/2, [1], 'X'],
+             [qt.sigmay().full()/2, [1], 'Y']]
 dt['Id'] = [T]
-H_c['X2'] = [[qt.sigmax().full(), [np.pi/4/T], 'X']]
-H_n['X2'] = [[qt.sigmax().full(), [1], 'X'],
-             [qt.sigmay().full(), [1], 'Y']]
+H_c['X2'] = [[qt.sigmax().full()/2, [np.pi/2/T], 'X']]
+H_n['X2'] = [[qt.sigmax().full()/2, [1], 'X'],
+             [qt.sigmay().full()/2, [1], 'Y']]
 dt['X2'] = [T]
-H_c['Y2'] = [[qt.sigmay().full(), [np.pi/4/T], 'Y']]
-H_n['Y2'] = [[qt.sigmax().full(), [1], 'X'],
-             [qt.sigmay().full(), [1], 'Y']]
+H_c['Y2'] = [[qt.sigmay().full()/2, [np.pi/2/T], 'Y']]
+H_n['Y2'] = [[qt.sigmax().full()/2, [1], 'X'],
+             [qt.sigmay().full()/2, [1], 'Y']]
 dt['Y2'] = [T]
 
 # %% Set up PulseSequences
@@ -134,7 +138,7 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
 # %% Construct Clifford group
 tic = time.perf_counter()
 cliffords = np.array([
-    Id,                                 # Id
+    Y2 @ Y2 @ Y2 @ Y2,                  # Id
     X2 @ X2,                            # X
     Y2 @ Y2,                            # Y
     Y2 @ Y2 @ X2 @ X2,                  # Z
@@ -166,8 +170,10 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
 # %% Run simulation
 
 eps0 = 2.7241e-4
-# Scaling factor for the noise so that alpha = 0 and alpha = 0.7 have the same
-# power
+
+alpha = (0.0, 0.7)
+# Scaling factor for the noise so that alpha = 0 and alpha = 0.7 give the same
+# average clifford fidelity
 noise_scaling_factor = {
     0.0: 0.4415924985735799,
     0.7: 1
@@ -176,37 +182,36 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
 state_infidelities = {}
 clifford_infidelities = {}
 
-for i, alpha in enumerate((0.0, 0.7)):
-    S0 = 1e-13*(2*np.pi*1e-3)**alpha/eps0**2*noise_scaling_factor[alpha]
-    S = S0/omega**alpha
+spectra = {}
+for i, a in enumerate(alpha):
+    S0 = 4e-11*(2*np.pi*1e-3)**a/eps0**2*noise_scaling_factor[a]
+    S = S0/omega**a
+    spectra[a], omega_twosided = util.symmetrize_spectrum(S, omega)
 
     # Need to calculate with two-sided spectra
-    clifford_infidelities[alpha] = [
-        ff.infidelity(C, *util.symmetrize_spectrum(S, omega)).sum()
+    clifford_infidelities[a] = [
+        ff.infidelity(C, spectra[a], omega_twosided).sum()
         for C in cliffords
     ]
 
-    print('=============================================')
-    print('\t\talpha = {}'.format(alpha))
-    print('=============================================')
-    state_infidelities[alpha], exec_times = run_randomized_benchmarking(
-        N_G, N_l, m_min, m_max, omega
-    )
+state_infidelities, exec_times = run_randomized_benchmarking(
+    N_G, N_l, m_min, m_max, alpha, spectra, omega_twosided
+)
 
 # %% Plot results
 fig, ax = plt.subplots(1, 2, sharex=True, sharey=True, figsize=(8, 3))
 
-fidelities = {alpha: 1 - infid for alpha, infid in state_infidelities.items()}
-for i, alpha in enumerate((0.0, 0.7)):
+fidelities = {a: 1 - infid for a, infid in state_infidelities.items()}
+for i, a in enumerate(alpha):
 
-    means = np.mean(fidelities[alpha], axis=1)
-    stds = np.std(fidelities[alpha], axis=1)
+    means = np.mean(fidelities[a], axis=1)
+    stds = np.std(fidelities[a], axis=1)
 
-    popt, pcov = optimize.curve_fit(fitfun, lengths, means, [0, 1], stds,
+    popt, pcov = optimize.curve_fit(fitfun, lengths, means, [0], stds,
                                     absolute_sigma=True)
 
     for j in range(N_G):
-        fid = ax[i].plot(lengths, fidelities[alpha][:, j], 'k.', alpha=0.1,
+        fid = ax[i].plot(lengths, fidelities[a][:, j], 'k.', alpha=0.1,
                          zorder=2)
 
     mean = ax[i].errorbar(lengths, means, stds, fmt='.', zorder=3,
@@ -217,16 +222,16 @@ def run_randomized_benchmarking(N_G: int, N_l: int, min_l: int, max_l: int,
     # sequence length and r = 1 - F_avg = d/(d + 1)*(1 - F_ent) the average
     # error per gate
     exp = ax[i].plot(lengths,
-                     1 - np.mean(clifford_infidelities[alpha])*lengths*2/3,
+                     1 - np.mean(clifford_infidelities[a])*lengths*2/3,
                      '--', zorder=4, color='tab:blue')
-    ax[i].set_title(r'$\alpha = {}$'.format(alpha))
+    ax[i].set_title(r'$\alpha = {}$'.format(a))
     ax[i].set_xlabel(r'Sequence length $m$')
 
 handles = [fid[0], mean[0], fit[0], exp[0]]
 labels = ['State Fidelity', 'Fidelity mean', 'Fit',
           'RB theory w/o pulse correlations']
 ax[0].set_xlim(0, max(lengths))
-ax[0].set_ylim(.993, 1)
+# ax[0].set_ylim(.9, 1)
 ax[0].set_ylabel(r'Surival Probability')
 ax[0].legend(frameon=False, handles=handles, labels=labels)
 

filter_functions/__init__.py

@@ -25,7 +25,7 @@
 from .numeric import (error_transfer_matrix, infidelity,
                       liouville_representation)
 from .plotting import (
-    plot_bloch_vector_evolution, plot_error_transfer_matrix,
+    plot_bloch_vector_evolution, plot_cumulant_function,
     plot_filter_function, plot_pulse_correlation_filter_function,
     plot_pulse_train)
 from .pulse_sequence import (PulseSequence, concatenate, concatenate_periodic,
@@ -34,7 +34,7 @@
 __all__ = ['Basis', 'PulseSequence', 'analytic', 'basis', 'concatenate',
            'concatenate_periodic', 'error_transfer_matrix', 'extend',
            'infidelity', 'liouville_representation', 'numeric',
-           'plot_bloch_vector_evolution', 'plot_error_transfer_matrix',
+           'plot_bloch_vector_evolution', 'plot_cumulant_function',
            'plot_filter_function', 'plot_pulse_correlation_filter_function',
            'plot_pulse_train', 'plotting', 'pulse_sequence', 'remap', 'util']
 

filter_functions/basis.py

@@ -619,7 +619,7 @@ def expand(M: Union[ndarray, Basis], basis: Union[ndarray, Basis],
                     {\mathrm{tr}\big(C_j^\dagger C_j\big)}.
 
     """
-    coefficients = np.einsum('...ij,bji->...b', np.asarray(M), basis)
+    coefficients = np.tensordot(M, basis, axes=[(-2, -1), (-1, -2)])
 
     if not normalized:
         coefficients /= np.einsum('bij,bji->b', basis, basis).real