5. NOTATION FOR HIGHER DERIVATIVES 285
defined so that whenever (Mn, g) is Ricci-parallel - that is, whenever the
(3, 0) -tensor V' Re vanishes identically, as happens in particular if g is Ein-
stein - the following diagram commutes:
If A is a symmetric (2, 0)-tensor, then the components of the Lichnerowicz
Laplacian 6LA in lo cal coordinates are given by the formula
(A.4)
6LA = ( 6Aij + 2gkp/q~kejApq - leRikAej -leRjkAie) dxi@dxJ.
REMARK A.2. As we saw in Chapter 3, the Lichnerowicz Laplacian is
essentially the linearization of the Ricci flow operator.
REMARK A.3. The Lichnerowicz Laplacian on symmetric 2-tensors is
formally the same as the Hodge-de Rham Laplacian on 2-forms. Indeed, an
easy application of the first Bianchi identity shows that equation (A.3) is
equivalent to
6da = ( 6aij + 2gkp lq Rik£jOpq - le Rikaej - gke Rjkaie) dxi /\ dxj,
which is formally the same as equation (A.4).
- Notation.for higher derivatives
A delicate notational issue arises when considering the operators
V'k: coo (T%Mn) -7 coo ( T%+kMn)
defined fork> 1. If A E C^00 (T%Mn) and X1, ... ,Xk are vector fields, we
adopt the convention (which is not universally followed!) that
Y'x1Y'x2···Y'xkA
is the unique (p, q)-tensor field such that
(V' x1 V' x2 · · · V' xkA) (Y1, ... , Yp; th, ... , eq)
= (Y'kA) (X1,X2, ... ,Xk, Y1, ... ,Yp; e1, ... ,eq)
for all vector fields Y1, ... , Yp and all covector fields e1, ... , eq· Note that
( V'k A) (X1, ... 'Xk) ~ V' X1 V' X2 ... V' xkA # V' X1 (V' X2 (-.. (V' xkA) ))
in general. For example, in our convention,
(A.5) Y'x (Y'vA) = Y'xY'vA + Y'vxvA = (Y'^2 A) (X, Y) + (V'A) (Y'xY).