How can I store data in a file and extract in a matrix form later?

[engsbk]

Thank you for the magnificent work in py-pde!

I'm simulating the 1D wave equation (with a time function source) in py-pde, and I want to compare the results to another results I got from another python code output.
I want to know how can I use FileStorage to save the fields:
x, t , u

• How can I store the data in npz/ h5py file?
• How can I extract the data in my other code?

In my other code, I'm saving my data in an npz file:

data_file = 'mydata'
np.savez(data_file, t_input=t, x_input=x, y_input=y, u_output=exact )

and then loading them to check/plot the values:

data = np.load(data_file+'.npz')
t_i, x_i, y_i, exact_i = data["t_input"], data["x_input"], data["y_input"], data["u_output"]

How can I do something similar in py-pde?

Thanks in advance!

This my code:

from math import sqrt,pi
import pde
from pde import WavePDE, CartesianGrid, ScalarField, MemoryStorage, FieldCollection, PDE, plot_kymograph, PlotTracker

#Set the Geometry
#Generate grid
grid = CartesianGrid( [[-1, 1]] , 256 )


#Initial condition
u = ScalarField.from_expression(grid, "sin(pi*x)", label="Field $u$" )
v = ScalarField(grid, 0, label="Field $v$")
initial_state = FieldCollection([u, v])


#Boundary condition
#specified within eq definition
#eq = pde.WavePDE(speed = 0.01, bc=[{"value": 0}, {"derivative": 0}])
speed = 0.01
frequency = 10
source = f"sin(2*pi*{frequency}*t)"
eq = PDE(
    {
        "u": f"v",
        "v": f"{speed}**2 * laplace(u) + {source}"
    }, bc=[{"value": 0}, {"derivative":0}]
)

# store intermediate information of the simulation
storage = MemoryStorage()

# solve the pde
result = eq.solve(initial_state, t_range = 1, tracker=[storage.tracker(0.01)])
    
#create a simulation movie
pde.movie(storage, filename="1D_wave_movie.mov")

# plot the resulting field
result.plot()

# visualize the results as a space-time plot
plot_kymograph(storage, scalar="norm_squared")

[david-zwicker]

If you want to store the results on the hard drive, it might be best to use a FileStorage instead of the MemoryStorage in your code. py-pde will then store the results in an hdf5 file which you can read either by initializing a pde.FileStorage with it or by just reading it using h5py. Note that the information about the grid (and thus the discretization) is stored in a hdf attribute and might be difficult to reconstruct manually. I thus suggest initializing the storage using py-pde again and then access the data using the data attribute. The underlying grid can be accessed using the grid attribute. In particular, you get information about the discretization using the axes_coords or cell_coords attributes.

[engsbk]

Thanks, it worked!

I have noticed that the info recorded are only data and times as I checked for keys in the file. Is there a way to also record the values of x used?

[david-zwicker]

The discretization information is not written explicitly, but contained in an HDF attribute. When you open the data using FileStorage, the grid attribute gives you access to grid instance, from which you can read off the discretization (and other information).

[anilkunwar]

Thanks a lot for this information. The FileStorage class seems to be very useful for data storage in an external file.

The above code worked for me when I added name of the file inside the brackets of pde.FileStorage()
from pde import FileStorage
...

# store intermediate information of the simulation
# storage = MemoryStorage()
storage = FileStorage(filename="a.hdf5")