Documentation of decision vector variable types ordering

[juan11iguel]

Reading the documentation, it does state with mathematical notation ($\mathbf{lb}, \mathbf{ub} \in \mathbb R^{n_{cx}} \times \mathbb Z^{n_{ix}}$) that the decision vector is composed by real variables followed by the integer part, however I think it would be good to explicitly state this with words in the docs.

To verify the order matters (not that I did it by mistake at first and this is how I realised):

class Problem:
	def get_bounds(self, ) -> tuple[np.ndarray, np.ndarray]:

		# Some unoptimized code
		box_bounds_lower = [0, 0, ..., 0, 0.0, ..., 0.0]
		box_bounds_upper = [1, 1, ..., 1, 1.0, ..., 1.0]

		return box_bounds_lower, box_bounds_upper

# Integer dec. variables before real raises value error
prob = pg.problem(Problem())

# Output:
ValueError: 
function: check_problem_bounds
where: /root/install/pagmo2-2.19.0/src/problem.cpp, 104
what: An upper bound of the integer part of the decision vector is: 35.182177 and is not an integer.

By changing the bounds order, the error goes away:

	...
	def get_bounds(self, ) -> tuple[np.ndarray, np.ndarray]:
		return box_bounds_lower[::-1], box_bounds_upper[::-1]

prob = pg.problem(Problem())

# Output:
Problem name: SolarMED MINLP problem
	C++ class name: pybind11::object

	Global dimension:			78
	Integer dimension:			24
	Fitness dimension:			11
	Number of objectives:			1
	Equality constraints dimension:		0
	Inequality constraints dimension:	10
	Tolerances on constraints: [0, 0, 0, 0, 0, ... ]
	Lower bounds: [0, 0, 0, 0, 0, ... ]
	Upper bounds: [35.3206, 35.286, 35.1822, 0, 0, ... ]
	Has batch fitness evaluation: false

Also, I wonder if since the data types of the box-bounds already need to match the data type of the decision vector, if it would not be possible to support an arbitrary ordering of the decision vector.

Best regards,
Juanmi