Ly_contours=L)
if __name__ == "__main__":
# Solvers
α = .9
# α = .000000001
α = .8
dt = -np.log(α)
solver_cbs = solvers.CbsSolver(
dt=dt,
# dt=.2,
# frac_min=45/100,
# frac_max=50/100,
# dt=.02,
parallel=True,
beta=beta,
dirname=m.__name__ + f"/test-simulation",
# adaptive=False,
adaptive=True,
# dirname=m.__name__ + f"/simulation-{isimul}",
opti=True)
solver, plot_step = solver_cbs, 1
vec = L * np.ones(m.n)
# ensembles = -vec + 2*vec*np.random.rand(J, m.n)
ensembles = 3 * np.random.randn(J, m.n)
niters = 80
# # plt.savefig(f"figures/{m.__name__}.png")
matplotlib.rc('font', size=20)
matplotlib.rc('font', family='serif')
matplotlib.rc('text', usetex=True)
matplotlib.rc('figure', figsize=(14, 8))
matplotlib.rc('savefig', bbox='tight')
matplotlib.rc('figure.subplot', hspace=.0)
matplotlib.rc('figure.subplot', wspace=.03)
# Solvers
α = .8
solver_cbs = solvers.CbsSolver(
dt=-np.log(α),
# frac_min=50/100,
# frac_max=55/100,
parallel=True,
beta=.5,
adaptive=False,
dirname=m.__name__,
opti=False)
if __name__ == "__main__":
# Plots
plotter = m.Plotter(m.ip, show_weights=True, adapt_size=True,
contours=True, Lx=1, Ly=1, Lx_contours=5, Ly_contours=40)
# Number of particles
J = 1000
# ensembles_x = np.random.randn(J)
import matplotlib.pyplot as plt
# Forward model and solver
import solvers
import model_elliptic_2d_constrained as m
# Set seed to zero
np.random.seed(0)
# plotter = m.AllCoeffsPlotter(m.ip, show_weights=True)
# plotter = m.MainModesPlotter(m.ip, show_weights=True)
plotter = m.AllCoeffsPlotter(m.ip)
solver_cbs = solvers.CbsSolver(dt=np.inf,
parallel=True,
adaptive=True,
opti=True,
dirname=m.__name__)
solver_cbo = solvers.CboSolver(dt=.1,
parallel=True,
beta=100,
lamda=1,
sigma=.3,
dirname=m.__name__ + "_small_epsilon",
epsilon=.5)
solver_eks = solvers.EksSolver(dt=10,
reg=True,
noise=True,
parallel=True,
import numpy as np
import matplotlib.pyplot as plt
import scipy.linalg as la
# Forward model and solver
import model_elliptic_1d as m
import solvers
# Solvers
solver_cbs = solvers.CbsSolver(dt=1,
frac_min=50 / 100,
frac_max=55 / 100,
parallel=True,
adaptive=True,
dirname=m.__name__,
opti=False)
solver_cbo = solvers.CboSolver(dt=.1,
parallel=True,
adaptive=True,
beta=1,
lamda=1,
sigma=.1,
dirname=m.__name__)
solver_eks = solvers.EksSolver(dt=1,
reg=True,
noise=True,
parallel=True,
adaptive=True,
dirname=m.__name__)
import numpy as np
import matplotlib.pyplot as plt
# Forward model and solver
import model_ackley as m
import solvers
# Set seed to zero
np.random.seed(0)
# Solvers
solver_cbs = solvers.CbsSolver(dt=np.inf,
parallel=True,
beta=1,
adaptive=False,
dirname=m.__name__,
opti=True)
solver_cbo = solvers.CboSolver(dt=.005,
parallel=True,
adaptive=False,
beta=1000,
lamda=1,
sigma=.1,
dirname=m.__name__,
epsilon=1)
solver_eks = solvers.EksSolver(dt=.01,
dt_max=10,
reg=True,
noise=False,
Ly=.01,
Lx_contours=3,
Ly_contours=3)
nsimul = 10
if __name__ == "__main__":
for isimul in range(nsimul):
# Solvers
solver_cbs = solvers.CbsSolver(
dt=np.inf,
# dt=.02,
parallel=True,
beta=beta,
adaptive=True,
dirname=m.__name__ + f"/adaptive-simulation-{isimul}",
# adaptive=False,
# dirname=m.__name__ + f"/simulation-{isimul}",
opti=True)
spread = 4
ensembles_x = spread * (np.random.randn() + np.random.randn(J))
ensembles_y = spread * (np.random.randn() + np.random.randn(J))
ensembles = np.vstack((ensembles_x, ensembles_y)).T
solver, plot_step = solver_cbs, 1
for i in range(n_iter):
print("Iteration {:04d}".format(i))
data = solver.step(m.op,
ensembles,