Creating From Values

A new fld_data.FldData instance can be created directly from NumPy arrays using fld_data.FldData.fromvalues() method. This is demonstrated in examples/demo_fromvalues_tofile.py, found in the repo and described below.

First, we set the size of the problem, including the dimensions, number of GLL points, and number of elements:

ndims = 3
nx1 = 3
ny1 = nx1
nz1 = nx1
nelgt = 10
nelt = nelgt

We can optionally specify the datatypes of integers and floats in the file. By default, fld_data.FldData.fromvalues() uses 32-bit ints and floats.

float_type = np.dtype(np.float32)
int_type = np.dtype(np.int32)

We must explicitly set global element IDs. In many downstream use cases, this is arbitrary.

glel = np.arange(1, nelt + 1)

To create the gridpoints, you can use a variety of NumPy trickery. For demonstration purposes, we will start with simple array of shape (nelt, ndim) and use it to construct an NumPy array of shape (nelt, ndim, nx1 * ny1 * nz1). The latter is the expected shape for FldData.coords.

We start with these points:

pt =[[0.0, 0.0, 0.0],
     [0.5, 0.0, 0.0],
     [1.0, 0.0, 0.0],
     [0.0, 0.5, 0.0],
     [0.5, 0.5, 0.0],
     [1.0, 0.5, 0.0],
     [0.0, 1.0, 0.0],
     [0.5, 1.0, 0.0],
     [1.0, 1.0, 0.0],
     [0.0, 0.0, 0.5],
     [0.5, 0.0, 0.5],
     [1.0, 0.0, 0.5],
     [0.0, 0.5, 0.5],
     [0.5, 0.5, 0.5],
     [1.0, 0.5, 0.5],
     [0.0, 1.0, 0.5],
     [0.5, 1.0, 0.5],
     [1.0, 1.0, 0.5],
     [0.0, 0.0, 1.0],
     [0.5, 0.0, 1.0],
     [1.0, 0.0, 1.0],
     [0.0, 0.5, 1.0],
     [0.5, 0.5, 1.0],
     [1.0, 0.5, 1.0],
     [0.0, 1.0, 1.0],
     [0.5, 1.0, 1.0],
     [1.0, 1.0, 1.0]]

And create a coordinate array of the proper shape:

nxyz = nx1 * ny1 * nz1
coords = np.empty((nelt, ndims, nxyz), dtype=float_type)
for j in range(nelt):
    for i in range(nxyz):
        coords[j][0][i] = pt[i][0]
        coords[j][1][i] = pt[i][1]
        coords[j][2][i] = pt[i][2] + j  #offset in the z-direction by the element number

Next, we create a velocity field with some dummy values. Like coordinates, FldData.u must have shape (nelt, ndim, nx1 * ny1 * nz1):

u = np.empty((nelt, ndims, nxyz), dtype=float_type)
for j in range(nelt):
    for i in range(nxyz):
        u[j][0][i] = j + 1.0
        u[j][1][i] = (j + 1.0) * 2.0
        u[j][2][i] = (j + 1.0) * 3.0

Next, we create pressure and velocity fields with dummy values. The expected shape of scalar fields is (nelt, nx1 * ny1 * nz1).

i = 1.0
p = np.arange(i, i + nelt * nxyz, dtype=float).reshape(nelt, nxyz)

i = i + nelt * nxyz
t = np.arange(i, i + nelt * nxyz, dtype=float).reshape(nelt, nxyz)

Finally, we instantiate an fld_data.FldData and write it to file:

data = FldData.fromvalues(nx1=nx1, ny1=ny1, nz1=nz1, nelgt=nelgt, nelt=nelt,
                         glel=glel, coords=coords, p=p, u=u, t=t)

data.tofile('fdtf0.f00001')