Classes of stable two-layer schemes 421
As before we assume the existence of an inverse operator B-^1 (t), which
assures us ofsolvability of problem (1) for any input data y 0 and ip(t). This
family obviously contains the primary family introduced in Section 2.2.
The method of energy inequalities shows in such a setting that the
conditions
(64)
(65)
B(t) > 0.5 T A(t)
B(t) > sE+0.5rA(t)
for all t E wT ,
for all t E wT , 0 < c < 1 ,
turn out to be sufficient for the stability of scheme (1) with variable oper-
ators A(t) and B(t). The norms 11 · llA' 11 · llA-' the1nselves happen to be
dependent on the variable t:
Therefore, it makes sense to speak about stability in the space HA(t) (in-
stead of HA) and HB(t)·
The energy identity (13) with A= A(t) is the starting point in special
investigations. To obtain a recurrence inequality, we should modify the
express10n
(Ay, y) = (A(t) y(t), y(t)) = (A(t - r) y(t), y(t))
- ((A(t) - A(t - r)) y(t), y(t))
and estimate the second sum111and on the right-hand side with the aid of
inequality (62):
(A(t)y(t),y(t)) < (1 + rc 3 )(A(t - r)y(t),y(t)).
Upon substituting this esti1nate into (13) we obtain the energy inequality
'
(66) 2r ((B(t) - 0.5 T A(t)) y(t), y(t)) + E(t + r)
< (1+TC 3 )£(t)+2r (ip(t),y(t)) 1
where
E(t + r) = (A(t) y(t + r), y(t + r)) = llY(t + r)ll~(t).
Inequality ( 66) with the member ip = 0 i1nplies that