def generate_npz(fname):
print(fname)
base = fname[:-8]
if os.path.exists(fname + 'diag.npz'):
pass
method = base[:base.find('-')]
energy_boundaries, mean_e, my_lnw, my_system, p_exc = compute.read_file(
base)
# Create a function for the entropy
l_function, eee, sss = compute.linear_entropy(energy_boundaries, mean_e,
my_lnw)
S = normalize_S(l_function(E))
T, C = heat_capacity.data(l_function, fname=fname)
plt.figure('fraction-well')
mean_which = np.loadtxt(f'{base}-which.dat')
if os.path.exists(f'{base}-histogram.dat'):
hist = np.loadtxt(f'{base}-histogram.dat')
else:
hist = []
errors_S = []
errors_C = []
moves = []
for frame_fname in sorted(glob.glob(f'{base}/*-lnw.dat')):
frame_base = frame_fname[:-8]
frame_moves = int(frame_fname[len(base) + 1:-8])
if frame_moves < 1e4:
continue
energy_boundaries, mean_e, my_lnw, my_system, p_exc = compute.read_file(
frame_base)
l_function, eee, sss = compute.linear_entropy(energy_boundaries,
mean_e, my_lnw)
moves.append(frame_moves)
err = np.max(
np.abs(
normalize_S(l_function(E[indices_for_err])) -
correct_S_for_err))
errors_S.append(err)
T_conv, C_conv = heat_capacity.data(l_function, fname=fname)
C_mask = [t >= lowest_interestng_T for t in T_conv]
errors_C.append(np.max(np.abs(correct_C[C_mask] - C_conv[C_mask])))
np.savez(os.path.join(base) + 'diag.npz',
E=E,
T=T,
mean_e=mean_e,
mean_which=mean_which,
hist=hist,
S=S,
C=C,
moves=moves,
errors_S=errors_S,
errors_C=errors_C)
print(f'working on {base} with moves {mymove} which is {f}')
energy_b = np.loadtxt(f+'-energy-boundaries.dat')
mean_e = np.loadtxt(f+'-mean-energy.dat')
my_lnw = np.loadtxt(f+'-lnw.dat')
if energy_b.ndim == 0: #in case of a single value
energy_b = np.array([energy_b.item()])
if energy_b[0] < energy_b[-1]:
energy_b = np.flip(energy_b)
mean_e = np.flip(mean_e)
my_lnw = np.flip(my_lnw)
# Create a function for the entropy based on this number of moves:
l_function, _, _ = compute.linear_entropy(energy_b, mean_e, my_lnw)
# l_function, _, _ = compute.step_entropy(energy_b, mean_e, my_lnw)
entropy_here = l_function(E)
plt.clf()
if 'erfinv' in base:
plt.plot(E, entropy_here, label=f)
plt.plot(E, exact_entropy, '--', label='exact')
plt.ylabel('$S(E)$')
else:
plt.plot(E, np.exp(entropy_here), label=f)
plt.plot(E, np.exp(exact_entropy), '--', label='exact')
plt.ylabel('density of states')
plt.xlabel('E')
# plt.ylim(bottom=0)
plt.legend(loc='best')
frames = set()
bases = []
print('base', args.base)
#each file has different path (including extension) so concatenating is easy
for base in args.base:
if '.cbor' in base or '.yaml' in base:
base = base[:-5]
print(base)
bases.append(base)
for f in glob.glob(base+'/*.cbor'):
f = os.path.splitext(os.path.basename(f))[0]
if float(f) > 1e7:
frames.add(f)
best_energy_boundaries, best_mean_energy, best_lnw, best_system = compute.read_file(args.base[0])
unscaled_best_function, best_energy, best_entropy = compute.linear_entropy(best_energy_boundaries, best_mean_energy, best_lnw)
if args.intensive:
scale = 1/best_system['N']
best_energy = scale*best_energy
best_entropy = scale*best_entropy
best_function = lambda e: scale*unscaled_best_function(e/scale)
else:
best_function = unscaled_best_function
energies_to_compare = np.array(best_mean_energy)
energies_reference = best_function(energies_to_compare)
# Only compare energies below the energy with max entropy
# nhi = energies_reference.argmax()
# energies_to_compare = energies_to_compare[nhi:-1]
# energies_reference = energies_reference[nhi:-1]
sigma = 1
for base in bases:
print('reading', base)
energy_boundaries = np.loadtxt(base+'-energy-boundaries.dat')
de = abs(np.diff(energy_boundaries))
lnw = np.loadtxt(base+'-lnw.dat')
mean_e = np.loadtxt(base+'-mean-energy.dat')
s_estimate = lnw[1:-1] - np.log(de)
peak_e = mean_e[np.argmax(s_estimate)]
if energy_boundaries[0] < energy_boundaries[-1]:
energy_boundaries = np.flip(energy_boundaries)
mean_e = np.flip(mean_e)
lnw = np.flip(lnw)
reference_function, reference_energy, reference_entropy = compute.linear_entropy(energy_boundaries, mean_e, lnw)
unscaled_reference_function = reference_function
if args.intensive:
energy_boundaries, mean_e, _, system, p_exc_reference = compute.read_file(base)
reference_N = system['N']
N = reference_N
reference_energy /= reference_N
reference_entropy /= reference_N
# energy_boundaries /= reference_N
# lnw /= reference_N
# peak_e /= reference_N
# s_estimate /= reference_N
# de /= reference_N
# mean_e /= reference_N
reference_function = lambda e: unscaled_reference_function(reference_N*e)/reference_N
pressure_reference, T_reference = compute.pressure_temperature(system['density'], energy_boundaries, mean_e, p_exc_reference)
base = base[:-5]
bases.append(base)
for base in bases:
print('reading', base)
energy_boundaries = np.loadtxt(base+'-energy-boundaries.dat')
de = abs(np.diff(energy_boundaries))
lnw = np.loadtxt(base+'-lnw.dat')
mean_e = np.loadtxt(base+'-mean-energy.dat')
if energy_boundaries[0] < energy_boundaries[-1]:
energy_boundaries = np.flip(energy_boundaries)
mean_e = np.flip(mean_e)
lnw = np.flip(lnw)
reference_function, _, _ = compute.linear_entropy(
energy_boundaries, mean_e, lnw)
energy_boundaries, mean_e, _, system, p_exc_reference = compute.read_file(
base)
p_reference, T_reference = compute.pressure_temperature(
system['density'], energy_boundaries, mean_e, p_exc_reference)
small_dT = 1/16
big_dT = 5
T = np.array(list(np.arange(0.125, 3 + small_dT/2, small_dT)) + list(np.arange(5, 30+big_dT/2, big_dT)))
p = canonical(T, mean_e, p_reference, reference_function)
np.savetxt(base+'-p-vs-T.dat',
np.transpose(np.vstack([T, p])),
fmt='%.4g')