The engine performs a single Heston Closed Form solution for a European Call Option.
It is template to generate either a floating point (Float-32) kernel or a double (Float-64) kernel. The double kernel will be more accurate but takes up more FPGA resource. The code contains some HLS directives which ensure the design is pipeline and parallel in order to improve its performance. It uses complex number arithmetic which is implemented in wrapper functions by using the HLS math library.
The integration is performed using the trapezoidal rule with a configurable dw (the integration time step) and Wmax (the upper limit of the integral). These parameters are configurable at run time and passed to the kernel along with the Heston parameters.