Skip to content

Commit

Permalink
removed bosonic fitting tests
Browse files Browse the repository at this point in the history
  • Loading branch information
@gsuarezr
gsuarezr committed Nov 9, 2023
1 parent 9699687 commit 77f23a4
Showing 1 changed file with 4 additions and 240 deletions.
244 changes: 4 additions & 240 deletions qutip/tests/solver/heom/test_bofin_baths.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,9 +17,6 @@
FermionicBath,
LorentzianBath,
LorentzianPadeBath,
FitCorr,
FitSpectral,
OhmicBath,
)


Expand Down Expand Up @@ -313,7 +310,8 @@ def test_create(self):
)
def test_terminator(self, combine):
Q = sigmaz()
op = -2 * spre(Q) * spost(Q.dag()) + spre(Q.dag() * Q) + spost(Q.dag() * Q)
op = -2 * spre(Q) * spost(Q.dag()) + \
spre(Q.dag() * Q) + spost(Q.dag() * Q)

bath = DrudeLorentzBath(
Q=Q,
Expand Down Expand Up @@ -388,7 +386,8 @@ def test_create(self):
)
def test_terminator(self, combine):
Q = sigmaz()
op = -2 * spre(Q) * spost(Q.dag()) + spre(Q.dag() * Q) + spost(Q.dag() * Q)
op = -2 * spre(Q) * spost(Q.dag()) + \
spre(Q.dag() * Q) + spost(Q.dag() * Q)

bath = DrudeLorentzPadeBath(
Q=Q,
Expand Down Expand Up @@ -662,238 +661,3 @@ def test_create(self):
sigma_bar_k_offset=-1,
tag="bath1",
)


class TestFitUtils:
bath = FitUtils()

def spectral_density(self, w, a, b, c):
tot = 0
for i in range(len(a)):
tot += (
2
* a[i]
* b[i]
* w
/ (((w + c[i]) ** 2 + b[i] ** 2) * ((w - c[i]) ** 2 + b[i] ** 2))
)
return tot

def test_pack(self):
n = np.random.randint(100)
before = np.random.rand(3, n)
a, b, c = before
assert len(self.bath.pack(a, b, c)) == n * 3
assert (self.bath.pack(a, b, c) == before.flatten()).all()

def test_unpack(self):
n = np.random.randint(100)
before = np.random.rand(3, n)
a, b, c = before
assert (self.bath.unpack(self.bath.pack(a, b, c)) == before).all()

def test_rmse(self):
lam = 0.05
gamma = 4
w0 = 2
func = lambda x, lam, gamma, w0: np.exp(-lam * x) + gamma / w0
x = np.linspace(1, 100, 10)
y = func(x, lam, gamma * 4, w0 * 4)
N = 0
assert np.isclose(self.bath._rmse(func, x, y, lam, gamma, w0, N), 0)

def test_leastsq(self):
t = np.linspace(0.1, 10 * 5, 1000)
a, b, c = [list(range(2))] * 3
N = 2
C = self.spectral_density(t, a, b, c)
sigma = 1e-4
C_max = abs(max(C, key=abs))
wc = t[np.argmax(C)]
guesses = self.bath.pack([C_max] * N, [wc] * N, [wc] * N)
lower = self.bath.pack([-100 * C_max] * N, [0.1 * wc] * N, [0.1 * wc] * N)
upper = self.bath.pack([100 * C_max] * N, [100 * wc] * N, [100 * wc] * N)
a2, b2, c2 = self.bath._leastsq(
self.spectral_density,
C,
t,
guesses=guesses,
lower=lower,
upper=upper,
sigma=sigma,
)
C2 = self.spectral_density(t, a2, b2, c2)
assert np.isclose(C, C2).all()

def test_fit(self):
a, b, c = [list(range(2))] * 3
w = np.linspace(0.1, 10 * 5, 1000)
J = self.spectral_density(w, a, b, c)
rmse, [a2, b2, c2] = self.bath._fit(self.spectral_density, J, w, N=2)
J2 = self.spectral_density(w, a2, b2, c2)
assert np.isclose(J, J2).all()
assert rmse < 1e-15


class TestFitSpectral:
alpha = 0.005
wc = 1
T = 1
bath = FitSpectral(T, sigmax(), nk=2)

def test_spectral_density_approx(self):
J = 0.4
a, b, c = [list(range(2))] * 3
w = 1
assert self.bath.spectral_density_approx(w, a, b, c) == J
a, b, c = [list(range(3))] * 3
w = 2
assert self.bath.spectral_density_approx(w, a, b, c) == J

def test_spec_spectrum_approx(self):
a, b, c = [list(range(2))] * 3
w = np.linspace(0.1, 10 * self.wc, 1000)
J = self.bath.spectral_density_approx(w, a, b, c)
pow = J * ((1 / (np.e ** (w / self.T) - 1)) + 1) * 2
rmse, self.bath.params_spec = self.bath._fit(
self.bath.spectral_density_approx, J, w, N=2, label="spectral"
)
pow2 = self.bath.spec_spectrum_approx(w)
assert np.isclose(pow, pow2).all()

def test_get_fit(self):
a, b, c = [list(range(5))] * 3
w = np.linspace(0.1, 10 * self.wc, 1000)
J = self.bath.spectral_density_approx(w, a, b, c)
self.bath.get_fit(J, w)
a2, b2, c2 = self.bath.params_spec
J2 = self.bath.spectral_density_approx(w, a2, b2, c2)
assert np.sum(J - J2) < (1e-12)
assert self.bath.spec_n <= 5

def test_matsubara_coefficients_from_spectral_fit(self):
ckAR, vkAr, ckAI, vkAI = (
[
(0.29298628613108785 - 0.2097848947562078j),
(0.29298628613108785 + 0.2097848947562078j),
(-0.01608448645347298 + 0j),
(-0.002015397620259254 + 0j),
],
[(1 - 1j), (1 + 1j), (6.283185307179586 + 0j), (12.566370614359172 + 0j)],
[(-0 + 0.25j), -0.25j],
[(1 - 1j), (1 + 1j)],
)
self.bath.params_spec = np.array([1]), np.array([1]), np.array([1])
self.bath._matsubara_spectral_fit()
ckAR2, vkAr2, ckAI2, vkAI2 = self.bath.cvars
assert np.isclose(ckAR, ckAR2).all()
assert np.isclose(vkAr, vkAr2).all()
assert np.isclose(ckAI, ckAI2).all()
assert np.isclose(vkAI, vkAI2).all()


class TestFitCorr:
alpha = 0.005
wc = 1
T = 1
bath = FitCorr(T, sigmax())

def test_corr_approx(self):
t = np.array([1 / 10, 1 / 5, 1])
C = np.array(
[
-0.900317 + 0.09033301j,
-0.80241065 + 0.16265669j,
-0.19876611 + 0.30955988j,
]
)
a, b, c = -np.array([list(range(2))] * 3)
C2 = self.bath.corr_approx(t, a, b, c)
assert np.isclose(C, C2).all()

def test_fit_correlation(self):
a, b, c = -np.array([list(range(2))] * 3)
t = np.linspace(0.1, 10 / self.wc, 1000)
C = self.bath.corr_approx(t, a, b, c)
self.bath.fit_correlation(t, C, Nr=3, Ni=3)
a2, b2, c2 = self.bath.params_real
ai2, bi2, ci2 = self.bath.params_imag
C2re = self.bath.corr_approx(t, a2, b2, c2)
C2imag = self.bath.corr_approx(t, ai2, bi2, ci2)
assert np.isclose(np.real(C), np.real(C2re)).all()
assert np.isclose(np.imag(C), np.imag(C2imag)).all()

def test_matsubara_coefficients(self):
ans = np.array(
[
[(0.5 + 0j), (0.5 + 0j), (0.5 - 0j), (0.5 - 0j)],
[(-1 - 1j), (-1 - 1j), (-1 + 1j), (-1 + 1j)],
[-0.5j, -0.5j, 0.5j, 0.5j],
[(-1 - 1j), (-1 - 1j), (-1 + 1j), (-1 + 1j)],
]
)
self.bath.params_real = [[1] * 2] * 3
self.bath.params_imag = [[1] * 2] * 3
self.bath._matsubara_coefficients()
assert np.isclose(np.array(self.bath.matsubara_coeff), ans).all()

def test_corr_spectrum_approx(self):
self.bath.params_real = [[1] * 2] * 3
self.bath.params_imag = [[1] * 2] * 3
self.bath._matsubara_coefficients()
S = self.bath.corr_spectrum_approx(np.array([1]))
assert np.isclose(S, np.array([-0.8 + 0j]))


class TestOhmicBath:
alpha = 0.5
s = 1
wc = 1
T = 1
Nk = 4
bath = OhmicBath(T, sigmax(), alpha, wc, s, Nk, method="spectral", rmse=1e-4)

def test_ohmic_spectral_density(self):
w = np.linspace(0, 50 * self.wc, 10000)
J = self.bath.ohmic_spectral_density(w)
J2 = self.alpha * w * np.exp(-abs(w) / self.wc)
assert np.isclose(J, J2).all()

def test_ohmic_correlation(self):
t = np.linspace(0, 10, 10)
C = self.bath.ohmic_correlation(t, s=3)
Ctest = np.array(
[
1.11215545e00 + 0.00000000e00j,
-2.07325102e-01 + 3.99285208e-02j,
-3.56854914e-02 + 2.68834062e-02j,
-1.02997412e-02 + 5.98374459e-03j,
-3.85107084e-03 + 1.71629063e-03j,
-1.71424113e-03 + 6.14748921e-04j,
-8.66216773e-04 + 2.59388769e-04j,
-4.81154330e-04 + 1.23604055e-04j,
-2.87395509e-04 + 6.46270269e-05j,
-1.81762994e-04 + 3.63396778e-05j,
]
)
assert np.isclose(C, Ctest).all()

def test_ohmic_power_spectrum(self):
w = np.linspace(0.1, 50 * self.wc, 10)
self.bath.s = 3
pow = self.bath.ohmic_power_spectrum(w)
testpow = np.array(
[
9.50833194e-03,
6.38339038e-01,
1.93684196e-02,
2.53252115e-04,
2.33614419e-06,
1.77884399e-08,
1.19944201e-10,
7.43601157e-13,
4.33486581e-15,
2.41093731e-17,
]
)
assert np.isclose(pow, testpow).all()

0 comments on commit 77f23a4

Please sign in to comment.