Given a covariance matrix cov, return the weight vector minimising w' @ cov @ w subject to sum(w) == 1:

• that is inv(cov) @ 1 / (1' @ inv(cov) @ 1)
• compute it with a linear solve, not an explicit inverse

Input

[[0.04 0.01]
 [0.01 0.09]]

Output

[0.72727273 0.27272727]