1. tan(alpha_i/2) with signed angles (atan2) vs half-angle form (r_i r_{i+1} - D_i)/A_i: 1000 cases, n=4..8, max relative difference of the weights 3.1e-12
2. 60-digit MVC of 24 real points (4 quadrilaterals, x inside and outside): max |F_P(lambda)| / sum|terms| = 2.4e-58; formula F_P(lambda) = prod_eta(...) at perturbed points: max relative error 5.9e-48
3. Lemma 4.1, n=5, V=[(9, 15), (8, 12), (-10, 18), (11, -16), (8, -11)]: N_P(w0+te)/t^2 for t=1e-15,1e-25,1e-35,1e-45: ['-4.475768311378e+20', '-4.475768311378e+20', '-4.475768311378e+20', '-4.475768311378e+20']; converges to a nonzero limit (order exactly 2): True
3. Lemma 4.1, n=6, V=[(-8, -9), (-8, -3), (20, 20), (3, 20), (-7, -14), (-15, -7)]: N_P(w0+te)/t^2 for t=1e-15,1e-25,1e-35,1e-45: ['-1.764460173574e+46', '-1.764460173574e+46', '-1.764460173574e+46', '-1.764460173574e+46']; converges to a nonzero limit (order exactly 2): True
3. Lemma 4.1, n=7, V=[(16, -7), (-13, 4), (-8, -7), (3, 18), (1, -13), (18, 7), (-16, 0)]: N_P(w0+te)/t^2 for t=1e-15,1e-25,1e-35,1e-45: ['2.561285675022e+110', '2.561285675024e+110', '2.561285675024e+110', '2.561285675024e+110']; converges to a nonzero limit (order exactly 2): True
4. Proposition 4.2(a), n=5: N_P(w) vs s^(2^(n-1)) prod l_eta,x(w) on K(x): relative difference 8.7e-79
4. Proposition 4.2(a), n=6: N_P(w) vs s^(2^(n-1)) prod l_eta,x(w) on K(x): relative difference 7.7e-78
ALL OK: True
