“…That is, if both functions f 1 (P i , x i ) = log 2 1 + |hi| 2 (Pi−xi) W σ 2 and f 2 (t, x A , x B ) = tlog 2 1 + Kī t a j x A |h A | 2 + a k x B |h B | 2 + K 1 , (i,ī = A or B), are concave, then P 5 is convex. Taking the secondorder derivative of f 1 (P i , x i ), the Hessian matrix is given by As for f 2 (t, x A , x B ), the Hessian matrix is given by (6) at the top of this page, where cī A = Kīa j |h A | 2 , cī B = Kīa k |h B | 2 and cī 1 = KīK 1 ,ī ∈ {A, B}. Since the second and third order leading principle minors are 0, ∂ 2 f2(t,xA,xB) ∂(t,xA,xB) 2 is negative semidefinite and f 2 (t, x A , x B ) is concave.…”