J Austral Math Soc Ser A 56 pp41--52, 1994.
(Received 7 February 1991; revised 12 November 1991)
Let u(x, t) be a smooth function in the domain Q = W×(0, L], W in Rn, let Du be the spatial gradient of u(x, t) and let Ñu = (Du, ut). If u(x, t) satisfies the parabolic equation F(u, Du, D2u) = ut, we define w(x, t) by g(w) = |Ñu|-1G(Ñu) (g is positive and decreasing, G is concave and homogenous of degree one) and we prove that w(x, t) attains its maximum value on the parabolic boundary of Q. If u(x, t) satisfies the equation Du + 2h(q2)uiujuij = ut (q2 = |Du|2, 1 +2q2h(q2) > 0) we prove that qf(u) takes its maximum value on the parabolic boundary of Q provided f satisfies a suitable condition. If u(x, t) satisfies the parabolic equation aij(Du)uij - b(x, t, u, Du) = ut (b is concave with respect to (x, t, u)) we define C(x, y, t, t) = u(z, q) - au(x, t) - bu(y, t) (0 < a, 0 < b, a+ b = 1, z = ax + by, q = at + bt) and we prove that if C(x, y, t, t) £ 0 when x, y, z Î W and one of t, t = 0, and when t, t Î (0, L], and one of x, y, z Î ¶W, then it is C(x, y, t, t) £ 0 everywhere.
1991 AMS Subject Classification: 35B50
Last Modified: Fri Jan 10 8:53:40 2003