Regarding Solving 4th order Phase field ( Cahn-Hilliard ) equation using deepxde

[divyanshjnv1]

Dear All,
I want to solve the 4th order Phase field equation ( Cahn Hilliard ) using DEEPXDE.
I have written the code for the form given in the paper, ( 10.4208/CICP.OA-2020-0086 )
please have a look at the code (it's not giving proper results, behaving just like 2nd-order diffusion eqn (Fick's 2nd law), initial sine profile concentration is settling in at mid value, acc. to Cahn Hilliard it should be opposite (i.e. phase separate).

( For writing this code I have taken the reference of the solved example of diffusion eqn code given in deepxde library, then changed pde and BCs)

If any ideas about solving this using some other strategy, I am all ears.
thanks!
For the equation used please look at the uploaded pic,

import deepxde as dde
import numpy as np
from deepxde.backend import tf

def pde(x, y):
    dy_t = dde.grad.jacobian(y, x, j=1)
    dy_x = dde.grad.jacobian(y, x, j=0)
    dy_xx = dde.grad.hessian(y, x, j=0)
    dy_xxxx = dde.grad.hessian(dy_xx, x, j=0)
    return (
        dy_t - ( 0.01 * (6*y*dy_xx*dy_xx + 3*y*y*dy_xx) - dy_xx) - (10**(-6))*dy_xxxx
    )

def func(x):
    return np.sin(np.pi * x[:, 0:1])                # this func is used for giving IC, BC


geom = dde.geometry.Interval(-1, 1)
timedomain = dde.geometry.TimeDomain(0, 1)
geomtime = dde.geometry.GeometryXTime(geom, timedomain)

def boundary_l(x, on_boundary):
    return on_boundary and np.isclose(x[0], -1)

def boundary_r(x, on_boundary):
    return on_boundary and np.isclose(x[0], 1)

# I want to provide 2 periodic BC as mentioned in the paper

bc_u_0 = dde.icbc.PeriodicBC(geomtime, 0, lambda _, on_boundary: on_boundary,derivative_order=0, component=0)    # zeroth order Periodic BC
bc_u_1 = dde.icbc.PeriodicBC(geomtime, 0, lambda _, on_boundary: on_boundary, derivative_order=1, component=0)   # 1st order Periodic BC
ic = dde.icbc.IC(geomtime, func, lambda _, on_initial: on_initial)
data = dde.data.TimePDE(
    geomtime,
    pde,
    [bc_u_0,bc_u_1, ic],
    num_domain=4000,
    num_boundary=20,
    num_initial=1000,
    num_test=10000,
)

layer_size = [2] + [32] * 3 + [1]
activation = "tanh"
initializer = "Glorot uniform"
net = dde.nn.FNN(layer_size, activation, initializer)

model = dde.Model(data, net)

model.compile("adam", lr=0.001)
losshistory, train_state = model.train(iterations=10000)

dde.saveplot(losshistory, train_state, issave=True, isplot=True)