Tutorial 2: Filtering numpy arrays

In this tutorial you will learn how to apply spatial filters to a 3D field using the Scalar3D class. Make sure you have downloaded the data folder from the GitHub repository before starting.

Introduction

This example shows two use cases:

1. Filtering a synthetic 3D NumPy array

2. Filtering a DNS scalar field loaded in light_mode using pointers

The goal is to demonstrate how filtering works, and how Scalar3D simplifies operations on large DNS fields.

Import modules

First, import the modules required for this tutorial:

import numpy as np
import aPriori.DNS as DNS
from aPriori.DNS import Scalar3D

Filtering a synthetic array

Although Scalar3D is mainly used with DNS data on disk, it can also store arrays directly for convenience.

Define a small test array:

shape = [4, 2, 4]
array = np.ones(shape)
array[2:3, :, 2:3] = 2
print(array)

You should see:

[[[1. 1. 1. 1.]
  [1. 1. 1. 1.]]

 [[1. 1. 1. 1.]
  [1. 1. 1. 1.]]

 [[1. 1. 2. 1.]
  [1. 1. 2. 1.]]

 [[1. 1. 1. 1.]
  [1. 1. 1. 1.]]]

Apply a Gaussian filter:

filter_size = 2
filtered_field = DNS.filter_3D(array, filter_size,
                               favre=False, filter_type='Gauss')
print(filtered_field)

Filtering DNS temperature data

To make a more realistic example, let’s filter a temperature field loaded from the DNS dataset using Scalar3D in light mode.

Locate the data directory and temperature file:

import os

# Change this to the appropriate path on your machine
directory = os.path.join("..", "data", "Lifted_H2_subdomain")
T_path = os.path.join(directory, "data", "T_K_id000.dat")

print(f"\nChecking the path '{T_path}' is correct...")

if not os.path.exists(T_path):
    raise ValueError(
        f"The path '{T_path}' does not exist. "
        "Ensure you have downloaded the DNS dataset."
    )

The DNS folder contains an info.json file that stores the grid shape:

import json
with open(os.path.join(directory, "info.json"), "r") as file:
    info = json.load(file)

DNS_shape = info["global"]["Nxyz"]

Create the Scalar3D object:

T = Scalar3D(shape=DNS_shape, path=T_path)

Access some values:

print("Temperature values in the cells 5:8, 5:8, 5:8")
print(T._3D[5:8, 5:8, 5:8])

Filter the scalar field using a Gaussian kernel:

filter_size = 20
T_filt = DNS.filter_3D(T._3D, filter_size,
                       favre=False, filter_type='Gauss')

print("Filtered temperature values in the cells 5:8, 5:8, 5:8")
print(T_filt[5:8, 5:8, 5:8])

You can also use a box filter:

T_filt_box = DNS.filter_3D(T._3D, filter_size,
                           favre=False, filter_type='Box')

print("Filtered temperature values in the cells 5:8, 5:8, 5:8")
print(T_filt_box[5:8, 5:8, 5:8])

Conclusion

This example showed how to:

• Apply 3D Gaussian and box filters to NumPy arrays

• Load DNS temperature data using Scalar3D in lightweight mode

• Filter DNS fields without loading the entire dataset into memory

To better visualize the effects of filtering, consider completing the next two tutorials, where filtering is applied and plotted on real DNS fields.