Python Data Analysis Cookbook

Using arbitrary precision for optimization

The intended readers of this book should be aware of floating point number issues. I will remind you that we are not able to represent floating point numbers exactly. Even integer representation is limited. For certain applications, for instance financial calculations or work involving known analytic expressions, we need a higher precision than available with numerical software such as NumPy. The Python standard library provides the Decimal class, which we can use to achieve arbitrary precision. However, the specialized mpmath library is a better fit for more advanced use.

Temperature follows a seasonal pattern, so a model involving the cosine seems natural. We will apply such a model. The nice thing about ...