From 88873b0c140edef7e9d08ef36d31029d084f5c71 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?K=C3=BChle=2C=20Laura=20Christine=20=28lakue103=29?=
<laura.kuehle@uni-duesseldorf.de>
Date: Wed, 24 Nov 2021 16:53:28 +0100
Subject: [PATCH] Added documentation to 'ANN_Data_Generator'.
---
ANN_Data_Generator.py | 119 ++++++++++++++++++++++++++++++++++++++++++
1 file changed, 119 insertions(+)
diff --git a/ANN_Data_Generator.py b/ANN_Data_Generator.py
index c8a4e91..5706196 100644
--- a/ANN_Data_Generator.py
+++ b/ANN_Data_Generator.py
@@ -9,6 +9,7 @@ TODO: Improve verbose output -> Done
TODO: Change order of methods -> Done
TODO: Fix bug in initialization of input matrix -> Done
TODO: Improve function selection (more even distribution) -> Done
+TODO: Add documentation -> Done
"""
@@ -21,8 +22,45 @@ import timeit
class TrainingDataGenerator(object):
+ """Class for training data generator.
+
+ Generate random training data for given initial conditions.
+
+ Attributes
+ ----------
+ smooth_functions : list
+ List of smooth initial/continuous conditions.
+ troubled_functions : list
+ List of discontinuous initial conditions.
+ data_dir : str
+ Path to directory in which training data is saved.
+
+ Methods
+ -------
+ build_training_data()
+ Builds random training data.
+
+ """
def __init__(self, initial_conditions, left_bound=-1, right_bound=1, balance=0.5,
stencil_length=3, directory=None):
+ """Initializes TrainingDataGenerator.
+
+ Parameters
+ ----------
+ initial_conditions : list
+ List of names of initial conditions for training.
+ left_bound : float, optional
+ Left boundary of interval. Default: -1.
+ right_bound : float, optional
+ Right boundary of interval. Default: 1.
+ balance: float, optional
+ Ratio between smooth and discontinuous training data. Default: 0.5.
+ stencil_length : int, optional
+ Size of training data array. Default: 3.
+ directory : str, optional
+ Path to directory in which training data is saved. Default: 'test_data'.
+
+ """
self._balance = balance
self._left_bound = left_bound
self._right_bound = right_bound
@@ -49,6 +87,21 @@ class TrainingDataGenerator(object):
os.makedirs(self._data_dir)
def build_training_data(self, num_samples):
+ """Builds random training data.
+
+ Creates training data consisting of random ANN input and saves it.
+
+ Parameters
+ ----------
+ num_samples : int
+ Number of training data samples to generate.
+
+ Returns
+ -------
+ data_dict : dict
+ Dictionary containing input (normalized and non-normalized) and output data.
+
+ """
tic = timeit.default_timer()
print('Calculating training data...\n')
data_dict = self._calculate_data_set(num_samples)
@@ -60,6 +113,22 @@ class TrainingDataGenerator(object):
return data_dict
def _calculate_data_set(self, num_samples):
+ """Calculates random training data of given stencil length.
+
+ Creates training data with a given ratio between smooth and discontinuous samples and
+ fixed stencil length.
+
+ Parameters
+ ----------
+ num_samples : int
+ Number of training data samples to generate.
+
+ Returns
+ -------
+ dict
+ Dictionary containing input (normalized and non-normalized) and output data.
+
+ """
num_smooth_samples = round(num_samples * self._balance)
smooth_input, smooth_output = self._generate_cell_data(num_smooth_samples,
self._smooth_functions, True)
@@ -84,6 +153,28 @@ class TrainingDataGenerator(object):
'normalized_input': norm_input_matrix}
def _generate_cell_data(self, num_samples, initial_conditions, is_smooth):
+ """Generates random training input and output.
+
+ Generates random training input and output for either smooth or discontinuous
+ initial_conditions.
+
+ Parameters
+ ----------
+ num_samples : int
+ Number of training data samples to generate.
+ initial_conditions : list
+ List of names of initial conditions for training.
+ is_smooth : boolean
+ Flag whether initial conditions are smooth.
+
+ Returns
+ -------
+ input_data : np.array
+ Array containing input data.
+ output_data : np.array
+ Array containing output data.
+
+ """
troubled_indicator = 'without' if is_smooth else 'with'
print('Calculating data ' + troubled_indicator + ' troubled cells...')
print('Samples to complete:', num_samples)
@@ -131,6 +222,20 @@ class TrainingDataGenerator(object):
return input_data, output_data
def _build_stencil(self):
+ """Builds random stencil.
+
+ Calculates fixed number of cell centers around a random point in a given 1D domain.
+
+ Returns
+ -------
+ interval : np.array
+ List containing left and right bound of interval.
+ stencil : np.array
+ List of cell centers in stencil.
+ grid_spacing: float
+ Length of cell in grid.
+
+ """
# Calculating Cell centers for a given 1D domain with n elements, and
# Calculating Corresponding Legendre Basis Coefficients for given polynomial_degree
# Create stencil and basis_coefficients for smooth_function mapped onto stencil
@@ -156,6 +261,19 @@ class TrainingDataGenerator(object):
@staticmethod
def _normalize_data(input_data):
+ """Normalize data.
+
+ Parameters
+ ----------
+ input_data : np.array
+ Array containing input data.
+
+ Returns
+ -------
+ np.array
+ Array containing normalized input data.
+
+ """
normalized_input_data = input_data
for i in range(len(input_data)):
max_function_value = max(max(np.absolute(input_data[i])), 1)
@@ -163,6 +281,7 @@ class TrainingDataGenerator(object):
return normalized_input_data
def _save_data(self, data):
+ """Saves data."""
print('Saving training data.')
for key in data.keys():
name = self._data_dir + '/' + key + '_data.npy'
--
GitLab