TimeSeries functionality and parallel computing

1

Hi,

I usually use TimeSeries functionality to store data for later computations:

Ufe_timeseries = TimeSeries('Data_Ufe')
Ufe_timeseries.store(u_n.vector(), t)

And to work with them in an another script I use:

timeseries_T = TimeSeries('Data_Ufe')
timeseries_T.retrieve(u_n.vector(), time)
u_n1, u_n2, u_n3  = split(u_n)

However, when I work with the split functions, I realized that the data is disorganized.

This is what I get:

2020-04-25_12h33_32
2020-04-25_12h33_321800×454 25.3 KB
Figure 1
This is what I was expecting:
2020-04-25_12h34_09
2020-04-25_12h34_091802×449 18 KB
Figure 2
I guess this is due to parallel computing, although I would expect eight partitions and not seven. In any case, there is a problem storing this data for later reuse.

Second, to produce Figure 1 I had to do the following:

class MyExpression(UserExpression):
    def __init__(self, u1, **kwargs):
        self.u1 = u1
        super().__init__(**kwargs)
    def eval(self, values, x):
        values[0] = self.u1(x)/1.0
    def value_shape(self):
        return ()

f0 = MyExpression(u_n3, degree=1)
V = FunctionSpace(mesh, 'P', 1)
u_n3_exp = interpolate(f0, V)

vtkfile = File("u_n3.pvd")
vtkfile << u_n3_exp

I don’t understand why I cannot write a new file just doing this:

vtkfile = File("U_n3.pvd")
vtkfile << u_n3

Because I get the following:
AttributeError: 'Indexed' object has no attribute '_cpp_object'

I really appreciate your help.

Regards,
Santiago