$(7$ points$)$

$($a$)$ While using the numba module to implement parallel computing, you can decorate the function to be parallelized either ty all or by anjit or by@cuda.jit.

What is the major difference between the three?

$($b$)$ Complete the following program which uses numba and GPU to implement the Monte Carlo method of a calculation parallelly $($NOTE: please remember to transfer the data from the host memory to the device memory before calling the kernel function, and then transfer it back to the host memory after calculation$).$