def apply_field1(mesh):
sim = Sim(mesh, name='dyn')
sim.set_tols(rtol=1e-10, atol=1e-10)
sim.alpha = 0.02
sim.gamma = 2.211e5
sim.Ms = 8.0e5
sim.set_m(np.load('m0.npy'))
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
demag = Demag()
sim.add(demag)
mT = 0.001 / mu0
print("Applied field = {}".format(mT))
zeeman = Zeeman([-24.6 * mT, 4.3 * mT, 0], name='H')
sim.add(zeeman, save_field=True)
ts = np.linspace(0, 1e-9, 201)
for t in ts:
sim.run_until(t)
print('sim t=%g' % t)
def apply_field1(mesh):
sim = Sim(mesh, name='dyn')
sim.driver.set_tols(rtol=1e-10, atol=1e-10)
sim.driver.alpha = 0.02
sim.driver.gamma = 2.211e5
sim.Ms = 8.0e5
sim.set_m(np.load('m0.npy'))
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
demag = Demag()
sim.add(demag)
mT = 0.001 / mu0
print("Applied field = {}".format(mT))
zeeman = Zeeman([-24.6 * mT, 4.3 * mT, 0], name='H')
sim.add(zeeman, save_field=True)
ts = np.linspace(0, 1e-9, 201)
for t in ts:
sim.run_until(t)
print('sim t=%g' % t)
def test_sim_single_spin(do_plot=False):
mesh = CuboidMesh(nx=1, ny=1, nz=1)
sim = Sim(mesh, name='spin')
alpha = 0.1
gamma = 2.21e5
sim.alpha = alpha
sim.gamma = gamma
sim.mu_s = 1.0
sim.set_m((1, 0, 0))
H0 = 1e5
sim.add(Zeeman((0, 0, H0)))
ts = np.linspace(0, 1e-9, 101)
mx = []
my = []
mz = []
real_ts = []
for t in ts:
sim.run_until(t)
real_ts.append(sim.t)
print(sim.t, abs(sim.spin_length()[0] - 1))
mx.append(sim.spin[0])
my.append(sim.spin[1])
mz.append(sim.spin[2])
mz = np.array(mz)
# print mz
a_mx, a_my, a_mz = single_spin(alpha, gamma, H0, ts)
print(sim.stat())
if do_plot:
ts_ns = np.array(real_ts) * 1e9
plt.plot(ts_ns, mx, ".", label="mx", color='DarkGreen')
plt.plot(ts_ns, my, ".", label="my", color='darkslateblue')
plt.plot(ts_ns, mz, ".", label="mz", color='m')
plt.plot(ts_ns, a_mx, "--", label="analytical", color='b')
plt.plot(ts_ns, a_my, "--", color='b')
plt.plot(ts_ns, a_mz, "--", color='b')
plt.xlabel("time (ns)")
plt.ylabel("m")
plt.title("integrating a macrospin")
plt.legend()
plt.savefig("single_spin.pdf")
print(("Max Deviation = {0}".format(
np.max(np.abs(mz - a_mz)))))
assert np.max(np.abs(mz - a_mz)) < 5e-7
def test_sim_single_spin(do_plot=False):
mesh = CuboidMesh(nx=1, ny=1, nz=1)
sim = Sim(mesh, name='spin')
alpha = 0.1
gamma = 2.21e5
sim.alpha = alpha
sim.gamma = gamma
sim.mu_s = 1.0
sim.set_m((1, 0, 0))
H0 = 1e5
sim.add(Zeeman((0, 0, H0)))
ts = np.linspace(0, 1e-9, 101)
mx = []
my = []
mz = []
real_ts = []
for t in ts:
sim.run_until(t)
real_ts.append(sim.t)
print sim.t, abs(sim.spin_length()[0] - 1)
mx.append(sim.spin[0])
my.append(sim.spin[1])
mz.append(sim.spin[2])
mz = np.array(mz)
# print mz
a_mx, a_my, a_mz = single_spin(alpha, gamma, H0, ts)
print sim.stat()
if do_plot:
ts_ns = np.array(real_ts) * 1e9
plt.plot(ts_ns, mx, ".", label="mx", color='DarkGreen')
plt.plot(ts_ns, my, ".", label="my", color='darkslateblue')
plt.plot(ts_ns, mz, ".", label="mz", color='m')
plt.plot(ts_ns, a_mx, "--", label="analytical", color='b')
plt.plot(ts_ns, a_my, "--", color='b')
plt.plot(ts_ns, a_mz, "--", color='b')
plt.xlabel("time (ns)")
plt.ylabel("m")
plt.title("integrating a macrospin")
plt.legend()
plt.savefig("single_spin.pdf")
print("Max Deviation = {0}".format(
np.max(np.abs(mz - a_mz))))
assert np.max(np.abs(mz - a_mz)) < 5e-7
def relax_system():
mesh = CuboidMesh(nx=1, ny=1, nz=1)
sim = Sim(mesh, name='relax')
sim.driver.set_tols(rtol=1e-10, atol=1e-10)
sim.driver.alpha = 0.5
sim.set_m((1.0, 0, 0))
sim.add(Zeeman((0, 0, 1e5)))
ts = np.linspace(0, 1e-9, 1001)
for t in ts:
sim.run_until(t)
def run(integrator, jacobian):
name = "sim_" + integrator
if integrator == "sundials":
name += "_J1" if jacobian else "_J0"
sim = Sim(mesh, name, integrator, use_jac=jacobian)
sim.Ms = 0.86e6
sim.driver.alpha = 0.5
sim.set_m((1, 0, 1))
sim.add(UniformExchange(A=13e-12))
sim.add(Demag())
ts = np.linspace(0, 3e-10, 61)
for t in ts:
sim.run_until(t)
def relax_system():
mesh = CuboidMesh(nx=1, ny=1, nz=1)
sim = Sim(mesh, name='relax')
sim.driver.set_tols(rtol=1e-10, atol=1e-10)
sim.driver.alpha = 0.5
sim.set_m((1.0, 0, 0))
sim.add(Zeeman((0, 0, 1e5)))
ts = np.linspace(0, 1e-9, 1001)
for t in ts:
sim.run_until(t)
def run(integrator, jacobian):
name = "sim_" + integrator
if integrator == "sundials":
name += "_J1" if jacobian else "_J0"
sim = Sim(mesh, name, integrator, use_jac=jacobian)
sim.Ms = 0.86e6
sim.alpha = 0.5
sim.set_m((1, 0, 1))
sim.add(UniformExchange(A=13e-12))
sim.add(Demag())
ts = np.linspace(0, 3e-10, 61)
for t in ts:
sim.run_until(t)
def test_sim_pin():
mesh = CuboidMesh(nx=3, ny=2, nz=1)
sim = Sim(mesh)
sim.set_m((0, 0.8, 0.6))
sim.alpha = 0.1
sim.gamma = 1.0
sim.pins = pin_fun
anis = UniaxialAnisotropy(Ku=1, axis=[0, 0, 1], name='Dx')
sim.add(anis)
sim.run_until(1.0)
print(sim.spin)
assert sim.spin[0] == 0
assert sim.spin[2] != 0
def test_sim_pin():
mesh = CuboidMesh(nx=3, ny=2, nz=1)
sim = Sim(mesh)
sim.set_m((0, 0.8, 0.6))
sim.alpha = 0.1
sim.gamma = 1.0
sim.pins = pin_fun
anis = UniaxialAnisotropy(Ku=1, axis=[0, 0, 1], name='Dx')
sim.add(anis)
sim.run_until(1.0)
print sim.spin
assert sim.spin[0] == 0
assert sim.spin[2] != 0
def excite_system(mesh, beta=0.0):
# Specify the stt dynamics in the simulation
sim = Sim(mesh, name='dyn_%g' % beta, driver='llg_stt_cpp')
sim.driver.set_tols(rtol=1e-12, atol=1e-12)
sim.driver.alpha = 0.1
sim.driver.gamma = 2.211e5
sim.Ms = 8.6e5
# sim.set_m(init_m)
sim.set_m(np.load('m0.npy'))
# Energies
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
anis = UniaxialAnisotropy(5e4)
sim.add(anis)
# beta is the parameter in the STT torque
sim.a_J = global_const * 1e11
sim.p = (1, 0, 0)
sim.beta = beta
# The simulation will run for 5 ns and save
# 500 snapshots of the system in the process
ts = np.linspace(0, 0.5e-9, 21)
xs = []
thetas = []
for t in ts:
print('time', t)
sim.run_until(t)
spin = sim.spin.copy()
x, theta = extract_dw(spin)
xs.append(x)
thetas.append(theta)
sim.save_vtk()
np.savetxt('dw_%g.txt' % beta, np.transpose(np.array([ts, xs, thetas])))
def excite_system(mesh, beta=0.0):
# Specify the stt dynamics in the simulation
sim = Sim(mesh, name='dyn_%g'%beta, driver='llg_stt_cpp')
sim.driver.set_tols(rtol=1e-12, atol=1e-12)
sim.driver.alpha = 0.1
sim.driver.gamma = 2.211e5
sim.Ms = 8.6e5
# sim.set_m(init_m)
sim.set_m(np.load('m0.npy'))
# Energies
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
anis = UniaxialAnisotropy(5e4)
sim.add(anis)
# beta is the parameter in the STT torque
sim.a_J = global_const*1e11
sim.p = (1,0,0)
sim.beta = beta
# The simulation will run for 5 ns and save
# 500 snapshots of the system in the process
ts = np.linspace(0, 0.5e-9, 21)
xs=[]
thetas=[]
for t in ts:
print('time', t)
sim.run_until(t)
spin = sim.spin.copy()
x, theta = extract_dw(spin)
xs.append(x)
thetas.append(theta)
sim.save_vtk()
np.savetxt('dw_%g.txt'%beta,np.transpose(np.array([ts, xs,thetas])))
def excite_system(mesh, time=5, snaps=501):
# Specify the stt dynamics in the simulation
sim = Sim(mesh, name='dyn', driver='llg_stt')
# Set the simulation parameters
sim.set_tols(rtol=1e-12, atol=1e-14)
sim.alpha = 0.05
sim.gamma = 2.211e5
sim.Ms = 8.6e5
# Load the initial state from the npy file saved
# in the realxation
sim.set_m(np.load('m0.npy'))
# Add the energies
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
anis = UniaxialAnisotropy(5e4)
sim.add(anis)
# dmi = DMI(D=8e-4)
# sim.add(dmi)
# Set the current in the x direction, in A / m
# beta is the parameter in the STT torque
sim.jx = -1e12
sim.beta = 1
# The simulation will run for x ns and save
# 'snaps' snapshots of the system in the process
ts = np.linspace(0, time * 1e-9, snaps)
for t in ts:
print('time', t)
sim.run_until(t)
sim.save_vtk()
sim.save_m()
def excite_system(mesh, time=5, snaps=501):
# Specify the stt dynamics in the simulation
sim = Sim(mesh, name='dyn', driver='llg_stt')
# Set the simulation parameters
sim.set_tols(rtol=1e-12, atol=1e-14)
sim.alpha = 0.05
sim.gamma = 2.211e5
sim.Ms = 8.6e5
# Load the initial state from the npy file saved
# in the realxation
sim.set_m(np.load('m0.npy'))
# Add the energies
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
anis = UniaxialAnisotropy(5e4)
sim.add(anis)
# dmi = DMI(D=8e-4)
# sim.add(dmi)
# Set the current in the x direction, in A / m
# beta is the parameter in the STT torque
sim.jx = -1e12
sim.beta = 1
# The simulation will run for x ns and save
# 'snaps' snapshots of the system in the process
ts = np.linspace(0, time * 1e-9, snaps)
for t in ts:
print 'time', t
sim.run_until(t)
sim.save_vtk()
sim.save_m()
def excite_system(mesh):
sim = Sim(mesh, name='dyn')
sim.driver.set_tols(rtol=1e-10, atol=1e-14)
sim.driver.alpha = 0.01
sim.driver.gamma = 2.211e5
sim.Ms = spatial_Ms
# sim.set_m(init_m)
sim.set_m(np.load('m0.npy'))
A = 1.3e-11
exch = UniformExchange(A=A)
sim.add(exch)
demag = Demag(pbc_2d=True)
sim.add(demag)
mT = 795.7747154594767
sigma = 0.08e-9
def gaussian_fun(t):
return np.exp(-0.5 * (t / sigma)**2)
zeeman = TimeZeeman((80 * mT, 0, 0), time_fun=gaussian_fun, name='hx')
#zeeman = Zeeman((100*mT,0,0), name='hx')
sim.add(zeeman, save_field=True)
ts = np.linspace(0, 1e-9, 501)
for t in ts:
print('time', t)
print('length:', sim.spin_length()[0:200])
sim.run_until(t)
sim.save_vtk()
def excite_system(mesh):
sim = Sim(mesh, name='dyn', driver='llg_stt')
sim.set_tols(rtol=1e-8, atol=1e-10)
sim.alpha = 0.5
sim.gamma = 2.211e5
sim.Ms = 8.6e5
sim.set_m(np.load('m0.npy'))
exch = UniformExchange(A=1.3e-11)
sim.add(exch)
dmi = DMI(D=-4e-3)
sim.add(dmi)
zeeman = Zeeman((0, 0, 4e5))
sim.add(zeeman, save_field=True)
sim.jx = -5e12
sim.beta = 0
ts = np.linspace(0, 0.5e-9, 101)
for t in ts:
print 'time', t
sim.run_until(t)
sim.save_vtk()
def excite_system(mesh):
sim = Sim(mesh, name='dyn', driver='llg_stt')
sim.set_tols(rtol=1e-8, atol=1e-10)
sim.alpha = 0.5
sim.gamma = 2.211e5
sim.Ms = 8.6e5
sim.set_m(np.load('m0.npy'))
exch = UniformExchange(A=1.3e-11)
sim.add(exch)
dmi = DMI(D=-4e-3)
sim.add(dmi)
zeeman = Zeeman((0, 0, 4e5))
sim.add(zeeman, save_field=True)
sim.jx = -5e12
sim.beta = 0
ts = np.linspace(0, 0.5e-9, 101)
for t in ts:
print 'time', t
sim.run_until(t)
sim.save_vtk()
# Interactive mode (this needs so set up a proper backend
# when importing matplotlib for the first time)
plt.ion()
# Set False to avoid the execution of the following code
plt.show(False)
# ---------------------------------------------------------------------
# Now run the simulation printing the energy
for time in times:
if not run_from_ipython():
print 'Time: ', time, ' s'
print 'Total energy: ', sim.compute_energy(), ' J'
print '\n'
sim.run_until(time)
# Update the vector data for the plot (the spins do not move
# so we don't need to update the coordinates) and redraw
m = np.copy(sim.spin)
# reshape rows, transpose and filter according to top layer
m = m.reshape(3, -1).T[top_z]
quiv.set_UVC(m[:, 0], m[:, 1], m[:, 2])
# Update title
ttime.set_text('Time: {:.4f} ns'.format(time * 1e9))
tenergy.set_text('Energy: {:.6e} ns'.format(sim.compute_energy()))
# fig.show()
fig.canvas.draw()
# Interactive mode (this needs so set up a proper backend
# when importing matplotlib for the first time)
plt.ion()
# Set False to avoid the execution of the following code
plt.show(False)
# ---------------------------------------------------------------------
# Now run the simulation printing the energy
for time in times:
if not run_from_ipython():
print 'Time: ', time, ' s'
print 'Total energy: ', sim.compute_energy(), ' J'
print '\n'
sim.run_until(time)
# Update the vector data for the plot (the spins do not move
# so we don't need to update the coordinates) and redraw
m = np.copy(sim.spin)
# reshape rows, transpose and filter according to top layer
m = m.reshape(3, -1).T[top_z]
quiv.set_UVC(m[:, 0], m[:, 1], m[:, 2])
# Update title
ttime.set_text('Time: {:.4f} ns'.format(time * 1e9))
tenergy.set_text('Energy: {:.6e} ns'.format(sim.compute_energy()))
# fig.show()
fig.canvas.draw()
# Finite difference mesh.
mesh = FDMesh(nx=1, ny=1, dx=10, dy=10, unit_length=1e-9)
sim = Sim(mesh)
sim.Ms = Ms
sim.alpha = alpha
sim.gamma = gamma
sim.add(Zeeman((0, 0, H)))
sim.set_m((1, 0, 0)) # initial magnetisation
# Sampling time steps.
t_array = np.arange(0, 5e-9, 0.01e-9)
mx_simulation = []
for t in t_array:
sim.run_until(t)
m = sim.spin.reshape((len(sim.spin) / 3, 3))
mx_simulation.append(m[:, 0][0])
###################
# Analytic solution
###################
mx_analytic = macrospin_analytic_solution(alpha, gamma, H, t_array)
###################
# Plot comparison.
###################
plt.figure(figsize=(8, 5))
plt.plot(t_array / 1e-9, mx_analytic, "o", label="analytic")
plt.plot(t_array / 1e-9, mx_simulation, linewidth=2, label="simulation")