Once the input file is ready, there are two main steps in the construction of the Taylor integrator.
First, we should ask taylor to produce the code to compute the jet of derivatives and the automatic step size (and degree) control. This code is arithmetic-independent, in the sense that the real numbers are declared as MY_FLOAT (type to be defined later) and the arithmetic operations have been replaced by C macros. Hence, the selection of basic arithmetic only depends on the definition of these macros.
The second step is to ask taylor to produce a header file with a concrete definition of the type MY_FLOAT and the macros that define the basic arithmetic. This file is included by the previous C file with the jet of derivatives and the step size control, so that the C preprocessor can substitute these macros by the code for the desired arithmetic.
We provide header files to use some extended precision arithmetics (later we give a concrete list), but none of these libraries is included in the taylor package; the user is supposed to retrieve and install them separately.
To use an arithmetic different from the ones mentioned here, the user only needs to introduce the corresponding function calls in the header file.
The two previous steps can be combined in a single one, by asking taylor to put everything (jet, step size and headers) in a single file.
The next section contains a simple example, using the standard double precision of the computer. Next, in Section 5.2, we will show all the options of the taylor program.