[Python] Multi-dimensional scipy.optimize.LinearConstraint?

My linear constraint for scipy.optimize.minimize is

ones = np.ones_like(x)
np.outer(x, ones) - np.outer(ones, x) > something


where something is a given matrix. (Mathematically, a_ij < x_i - x_j < a_ji).

How do I express this in terms of scipy.optimize.LinearConstraint which wouldn't accept a 3d "matrix" as a 1st argument?

Continue reading...