Integral of a randomized frequency sine function in different grid resolutions

Integral of a randomized frequency sine function in different grid resolutions, tailored to test a property of the "Neural Operator: Learning Maps Between Function Spaces" paper. The dataset specifically contains a train dataset with 800 examples of a function of the type sin(c⋅2πx), where c is a randomly generated constant for each example, as well as the integral of that function from 0 to x, both in 64 grid resolution. The test dataset contains 200 examples of that same function/output pairs, but in a 1024 grid resolution. The dataset can be used to verify if a Neural Operator learning the function mapping would perform better instead of a Neural Network learning the raw values mapping when considering different train and test resolutions.