X  Delta_i(f) by direct conditional expectation = polynomial table at q in {1/8, 3/25, 2/15}: 48972 comparisons, all equal
D  G_n(c) = tau max_f c . Delta(f) exactly: 72 random (u, c), n = 3, 4: all equal
S  G_n(c)^2 <= T4^2 |c|^2 exactly at u in {2.543, 2.6, 2.7, 2.8, 2.910}: 2405 vectors, no violation; equality in 5 cases (all c = (1,1,1,1)); largest ratio of squares 1.000000000000
A  q = 0.105573, n = 3: max ratio 0.9682458366 at [0.5774 0.5774 0.5774]
A  q = 0.105573, n = 4: max ratio 1.0000000000 at [0.5 0.5 0.5 0.5]
A  q = 0.105573, n = 5: max ratio 1.0000000000 at [0.4472 0.4472 0.4472 0.4472 0.4472]
A  q = 0.119774, n = 3: max ratio 0.9838667697 at [0.5774 0.5774 0.5774]
A  q = 0.119774, n = 4: max ratio 1.0000000000 at [0.5 0.5 0.5 0.5]
A  q = 0.119774, n = 5: max ratio 1.0000000000 at [0.5 0.5 0.5 0.5 0. ]
A  q = 0.133975, n = 3: max ratio 1.0000000000 at [0.5774 0.5774 0.5774]
A  q = 0.133975, n = 4: max ratio 1.0000000000 at [0.5 0.5 0.5 0.5]
A  q = 0.133975, n = 5: max ratio 1.0000000000 at [ 0.5774  0.5774  0.5774 -0.     -0.    ]
E  q4 + 0.0001: G_3(1,1,1)/(T4 sqrt3) = 1.00011548339; q5 - 0.0001: G_5(1,...,1)/(T4 sqrt5) = 1.0001118034; q4 + 1.0e-7: G_3(1,1,1)/(T4 sqrt3) = 1.00000011547; q5 - 1.0e-7: G_5(1,...,1)/(T4 sqrt5) = 1.0000001118
N  n = 3: H_f > 0 on [q5, q4 + 1/1000] fails for 2 f; with 16 - 1e-6 on [q5, q4] fails for 2 f; with the k = 3 target 12 (1-q)^4 fails for 2 f; H_f/q^j > 0 on [0, q5] fails for 0 f (informational)
N  n = 4: H_f > 0 on [q5, q4 + 1/1000] fails for 8 f; with 16 - 1e-6 on [q5, q4] fails for 10 f; with the k = 3 target 12 (1-q)^4 fails for 68 f; H_f/q^j > 0 on [0, q5] fails for 0 f (informational)
RESULT: ALL TESTS AS EXPECTED (44s)

real	0m44.368s
user	0m45.302s
sys	0m0.200s
