def run_cmaes_p_concat(v, k, r, seed, generations, sig=None):
suzuki = approx.suzuki_vals(k)
if sig == None:
sig = 1e-5 / len(suzuki)
chain = hchain.HeisenbergChain(len(v), v)
random.seed(seed)
np.random.seed(seed)
# Error from target
def target_error(ind):
if NORMALISE:
norm_ind = norm_f(ind)
else:
norm_ind = ind
final_ind = approx.r_copies(approx.expand_vals(ind), r)
return approx.error(chain, final_ind, t=2 * chain.n),
toolbox = base.Toolbox()
toolbox.register("evaluate", target_error)
strategy = cma.Strategy(centroid=suzuki, sigma=sig)
toolbox.register("generate", strategy.generate, creator.Individual)
toolbox.register("update", strategy.update)
hof = tools.HallOfFame(1)
stats = tools.Statistics(lambda ind: ind.fitness.values)
stats.register("avg", np.mean)
stats.register("std", np.std)
stats.register("min", np.min)
stats.register("max", np.max)
pop, log = algorithms.eaGenerateUpdate(toolbox,
ngen=generations,
stats=stats,
halloffame=hof,
verbose=True)
return pop, log, hof
def evaluate_rows(input_df, start_row, end_row, output_folder_path):
df = input_df.copy()
if not os.path.isdir(output_folder_path):
os.makedirs(output_folder_path)
logs_path = os.path.join(output_folder_path, 'logs')
if not os.path.isdir(logs_path):
os.makedirs(logs_path)
output_data_path = os.path.join(output_folder_path, 'output_{}_{}.csv'.format(start_row, end_row))
for row_id in range(start_row-1, end_row):
row = df.loc[row_id]
n = int(row['n'])
v = row['v_1':'v_%d' % n]
k = int(row['k'])
r = int(row['r'])
seed = int(row['seed'])
gens = int(row['gens'])
chain = hchain.HeisenbergChain(n, v)
t = time.time()
pop, log, hof = run_cmaes_p(v=v, k=k, r=r, seed=seed, generations=gens)
t = time.time() - t
best = hof[0]
suz_err = approx.error(chain,approx.suzuki_solution(k, r),t=2*n)
op_err = approx.error(chain,approx.r_copies(best, r),t=2*n)
perc = 100*op_err/suz_err
for j in range(1,5**(k-1)+1):
df.loc[row_id,'p_%d'%j] = best[j-1]
df.loc[row_id,'optimised error'] = op_err
df.loc[row_id, 'suzuki error'] = suz_err
df.loc[row_id, '% error'] = perc
df.loc[row_id, 'time'] = t
pd.DataFrame(log).to_csv(os.path.join(logs_path, 'row_{}.csv'.format(row_id + 1)), index=False)
df.dropna(inplace=True)
df.to_csv(output_data_path, index=False)
def run(input_file, start, end):
if not os.path.isdir(OUTPUT_DIR):
os.makedirs(OUTPUT_DIR)
input_file_prefix = os.path.splitext(os.path.basename(input_file))[0]
output_file = os.path.join(OUTPUT_DIR, input_file_prefix + '_output_{}_{}.csv'.format(start, end))
header, all_rows = read_all(input_file)
with open(output_file, 'w') as outcsv:
output_header = header + ['suzuki error', 'optimised error', '%']
writer = csv.DictWriter(outcsv, output_header)
writer.writeheader()
for row in all_rows[start-1:end]:
n = int(row['n'])
k = int(row['k'])
r = int(row['r'])
t = int(row['t'])
p_vect = {k: v for k, v in row.items() if k.startswith('p_')}
v_vect = {k: v for k, v in row.items() if k.startswith('v_') and v != ''}
v_values = [float(v) for v in v_vect.values()]
p_values = [float(p) for p in p_vect.values()]
chain = hchain.HeisenbergChain(n, v_values)
suz_err = approx.error(chain, approx.suzuki_solution(k, r), t)
op_err = approx.error(chain, approx.r_copies(p_values, r), t)
percentage = 100 * op_err / suz_err
row['suzuki error'] = suz_err
row['optimised error'] = op_err
row['%'] = percentage
writer.writerow(row)
def plot(folder_path):
df = pd.read_csv(os.path.join(folder_path, 'exp16_results.csv'))
v = list(df.loc[0, 'v_1':'v_5'])
chain = hchain.HeisenbergChain(5, v)
suz_errs = [
approx.error(chain, approx.suzuki_solution(2, r), 10)
for r in range(25, 401, 25)
]
r_counts = range(25, 401, 25)
for i in range(0, len(df), 16):
df_r = df.loc[i:i + 15]
orig_r = int(df_r.loc[i, 'original_r'])
plt.plot(r_counts,
df_r['error'],
color=(1.0, 1 - orig_r / 400, 1 - orig_r / 400, 0.8),
marker='.',
linewidth=1.5,
markersize=4,
label='_nolegend_')
plt.plot(r_counts,
suz_errs,
color='blue',
marker='.',
linewidth=2,
markersize=7,
label='Suzuki')
plt.yscale(value='log')
plt.legend(fontsize=16)
plt.xlabel('r', fontsize=18)
plt.ylabel('Absolute Error', fontsize=18)
plt.xticks(fontsize=16)
plt.yticks(fontsize=16)
plt.tight_layout()
plt.show()
def test_unitary(self):
small_chain = hchain.HeisenbergChain(2, [1] * 2)
actual = small_chain.exponential(1j)
expected = np.array([[
-0.6536 - 0.7568j, 0.0000 + 0.0000j, 0.0000 + 0.0000j,
0.0000 + 0.0000j
],
[
0.0000 + 0.0000j, 0.2720 + 0.5944j,
-0.6882 + 0.3149j, 0.0000 + 0.0000j
],
[
0.0000 + 0.0000j, -0.6882 + 0.3149j,
0.2720 + 0.5944j, 0.0000 + 0.0000j
],
[
0.0000 + 0.0000j, 0.0000 + 0.0000j,
0.0000 + 0.0000j, 1.0000 + 0.0000j
]])
self.assertTrue(np.allclose(actual, expected, atol=0.001))
def greedy():
""" Run greedy hillclimber. Enumerate neighbourhood and choose best. """
random.seed(1234)
chain = hchain.HeisenbergChain(num_qubits, v)
suzuki = np.array(approx.suzuki(k))
suzuki_error = approx.error(chain, approx.r_copies(suzuki, r), t)
print('Suzuki error: {}'.format(suzuki_error))
# Start at Suzuki
current = suzuki
current_error = suzuki_error
for gen in range(1, gens + 1):
current_error_percent = 100 * (current_error / suzuki_error)
print('Gen: {} Best: {} Percent: {} '.format(gen, current_error,
current_error_percent))
first = random.randint(1, num_p_values)
second = random.randint(1, num_p_values)
neighbour = current.copy()
neighbour[first - 1] = neighbour[first - 1] + step_size
neighbour[second - 1] = neighbour[second - 1] - step_size
neighbour_err = approx.error(chain, approx.r_copies(neighbour, r), t)
neighbour_err_percent = 100 * (neighbour_err / suzuki_error)
print('Step {} error: {}, error %: {}'.format(gen, neighbour_err,
neighbour_err_percent))
if neighbour_err < current_error:
current = neighbour
current_error = neighbour_err
current_error_percent = 100 * (current_error / suzuki_error)
print('New best: error: {} percent: {} solution: {})'.format(
current_error, current_error_percent, current))
def compare_lambda_over_n(input_filename, input_df, start_row, end_row,
output_folder_path):
df = input_df.copy()
df.insert(1, 'optimised error', None)
df.insert(1, 'suzuki error', None)
df.insert(1, '%', None)
if not os.path.isdir(output_folder_path):
os.makedirs(output_folder_path)
input_filename_prefix = os.path.splitext(input_filename)[0]
output_data_path = os.path.join(
output_folder_path,
input_filename_prefix + '_output_{}_{}.csv'.format(start_row, end_row))
for row_id in range(start_row - 1, end_row):
row = df.loc[row_id]
n = int(row['n'])
v = list(map(float, list(row['v_1':'v_%d' % n])))
k = int(row['k'])
r = int(row['r'])
lam = list(row.loc['lambda_1':'lambda_%d' % (r * 5**(k - 1))])
chain = hchain.HeisenbergChain(n, v)
# Hardcode time
suz_err = approx.error(chain, approx.suzuki_solution(k, r), t=SIM_TIME)
op_err = approx.error(chain, lam, t=SIM_TIME)
perc = 100 * op_err / suz_err
df.loc[row_id, 'suzuki error'] = suz_err
df.loc[row_id, 'optimised error'] = op_err
df.loc[row_id, '%'] = perc
df.dropna(inplace=True)
df.to_csv(output_data_path, index=False)
def setUp(self):
self.chain = hchain.HeisenbergChain(self.n, self.v)
def calc_values():
results_path = os.path.join('..', 'results')
if not os.path.isdir(results_path):
results_path= os.path.join('..', '..', 'results')
if not os.path.isdir(results_path):
results_path = os.path.join('..','..', '..', 'results')
if not os.path.isdir(results_path):
results_path = os.path.join('..','..','..', '..', 'results')
path = os.path.join(results_path, 'perm_analysis')
if not os.path.isdir(path):
os.makedirs(path)
np.random.seed(1294635)
n = 5
chain = hchain.HeisenbergChain(n, np.random.uniform(-1,1,n))
group_perm = [4 * i for i in range(chain.n)] + [4 * i + 1 for i in range(chain.n)] \
+ [4 * i + 2 for i in range(chain.n)] + [4 * i + 3 for i in range(chain.n)]
can_perm = [i for i in range(4*chain.n)]
rand_perms = [np.random.permutation(4*chain.n) for _ in range(20)]
cols = ['r', 'gate count', 'error']
group_k2 = pd.DataFrame(columns=cols)
group_k3 = pd.DataFrame(columns=cols)
can_k2 = pd.DataFrame(columns=cols)
can_k3 = pd.DataFrame(columns=cols)
rand_k2 = pd.DataFrame(columns=cols)
rand_k3 = pd.DataFrame(columns=cols)
k=2
for r in range(25,401,25):
print(r)
suz = approx.suzuki_solution(k,r)
group_k2.loc[len(group_k2)] = [r, approx.gate_count(chain,len(suz),permutation=group_perm),
approx.error(chain,suz,t=2*chain.n, permutation=group_perm)]
can_k2.loc[len(can_k2)] = [r, approx.gate_count(chain,len(suz),permutation=can_perm),
approx.error(chain,suz,t=2*chain.n, permutation=can_perm)]
mean_gate_count, mean_error = 0, 0
for perm in rand_perms:
mean_gate_count += approx.gate_count(chain, len(suz), permutation=perm)
mean_error += approx.error(chain, suz, t=2*chain.n, permutation=perm)
mean_gate_count = mean_gate_count/len(rand_perms)
mean_error = mean_error / len(rand_perms)
rand_k2.loc[len(rand_k2)] = [r, mean_gate_count, mean_error]
group_k2.to_csv(os.path.join(path,'Grouped k=2.csv'))
can_k2.to_csv(os.path.join(path, 'Canonical k=2.csv'))
rand_k2.to_csv(os.path.join(path, 'Random k=2.csv'))
k=3
for r in range(5,61,5):
print(r)
suz = approx.suzuki_solution(k, r)
group_k3.loc[len(group_k3)] = [r, approx.gate_count(chain,len(suz), permutation=group_perm),
approx.error(chain, suz, t=2*chain.n, permutation=group_perm)]
can_k3.loc[len(can_k3)] = [r, approx.gate_count(chain, len(suz), permutation=can_perm),
approx.error(chain, suz, t=2*chain.n, permutation=can_perm)]
mean_gate_count, mean_error = 0, 0
for perm in rand_perms:
mean_gate_count += approx.gate_count(chain, len(suz), permutation=perm)
mean_error += approx.error(chain, suz, t=2 * chain.n, permutation=perm)
mean_gate_count = mean_gate_count / len(rand_perms)
mean_error = mean_error / len(rand_perms)
rand_k3.loc[len(rand_k3)] = [r, mean_gate_count, mean_error]
group_k3.to_csv(os.path.join(path, 'Grouped k=3.csv'))
can_k3.to_csv(os.path.join(path, 'Canonical k=3.csv'))
rand_k3.to_csv(os.path.join(path, 'Random k=3.csv'))
df = pd.DataFrame(columns=['n'] + ['v_%d' % i for i in range(1, chain.n+1)])
df.loc[len(df)] = [chain.n] + list(chain.v)
df.to_csv(os.path.join(path, 'chain.csv'))
def greedy():
""" Run greedy hillclimber. Enumerate neighbourhood and choose best. """
chain = hchain.HeisenbergChain(num_qubits, v)
steps = [-step_size, 0, step_size]
neighbourhood_size = len(steps)**num_p_values
suzuki = np.array(approx.suzuki(k))
suzuki_error = approx.error(chain, approx.r_copies(suzuki, r), t)
print('Suzuki error: {}'.format(suzuki_error))
# Start at Suzuki
# current = suzuki
# current_error = suzuki_error
# Start at 1/p_length
# current = np.array([0.2] * 5)
# current_error = approx.error(chain, approx.r_copies(current, r), t)
import random
random.seed(1234)
suzuki_coeff = approx.suzuki(k)
random.shuffle(suzuki_coeff)
current = np.array(suzuki_coeff)
current_error = approx.error(chain, approx.r_copies(current, r), t)
for gen in range(1, gens + 1):
current_error_percent = 100 * (current_error / suzuki_error)
print('Gen: {} Best: {} Percent: {} '.format(gen, current_error,
current_error_percent))
best_neighbour = None
best_neighbour_error = 1e6
for neighbour_index in range(1, neighbourhood_size):
move = [
step_size * (coeff - 1)
for coeff in ternary(neighbour_index, num_p_values)
]
neighbour = current + move
neighbour_err = approx.error(chain, approx.r_copies(neighbour, r),
t)
neighbour_err_percent = 100 * (neighbour_err / suzuki_error)
print('Step {} Neighbour {}/{} error: {}, error %: {}'.format(
gen, neighbour_index, neighbourhood_size, neighbour_err,
neighbour_err_percent))
if neighbour_err <= best_neighbour_error:
best_neighbour = neighbour
best_neighbour_error = neighbour_err
if best_neighbour_error < current_error:
current = best_neighbour
current_error = best_neighbour_error
current_error_percent = 100 * (current_error / suzuki_error)
print('New best: error: {} percent: {} solution: {})'.format(
current_error, current_error_percent, current))
elif best_neighbour_error >= current_error:
best_neighbour_percent = 100 * (best_neighbour_error /
suzuki_error)
print('Best result: error: {} percent: {} solution: {}'.format(
best_neighbour_error, best_neighbour_percent, best_neighbour))
return