You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
How accurate is naive floating-point summation? Exact errors against the worst-case bound, the role of the condition number, and Kahan, pairwise, np.sum and math.fsum compared.
Selects its own solve strategy — direct, LU or least-squares — from matrix shape and condition number, with a RidgeCV fallback for extreme ill-conditioning. Returns diagnostics alongside every solution. 19 tests.
Linear system solver using Rouché–Capelli theorem and SVD rank computation — handles unique, inconsistent, and underdetermined systems with condition number checking and NHS application.