29.7. THE DIVERGENCE THEOREM 1027

Letting xp = y+gi (x1, · · · ,xp−1) and changing the variable, this equals

=∫

Bi

∫ 0

−∞

p

∑j=1

D j (ψ i f )(x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)) ·

w jdydx1 · · ·dxp−1

=∫

Ai

∫ 0

−∞

p

∑j=1

D j (ψ i f )(x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)) ·

w jdydx1 · · ·dxp−1

Recall Ai is all of Bi except for the set of measure zero where the derivative does not exist.Also D j refers to the partial derivative taken with respect to the entry in the jth slot. In thepth slot is found not just xp but y+ gi (x1, · · · ,xp−1) so a differentiation with respect to x jwill not be the same as D j. In fact, it will introduce another term involving gi, j. Thus fromthe chain rule,

=∫

Ai

∫ 0

−∞

p−1

∑j=1

∂x j(ψ i f (x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)))w j−

Dp (ψ i f )(x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)) ·gi, j (x1, · · · ,xp−1)w jdydx1 · · ·dxp−1

+∫

Ai

∫ 0

−∞

Dp (ψ i f )(x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1))wpdydx1 · · ·dxp−1 (29.7.32)

Consider the term∫Ai

∫ 0

−∞

p−1

∑j=1

∂x j(ψ i f (x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)))w jdydx1 · · ·dxp−1

This equals

∫Bi

∫ 0

−∞

p−1

∑j=1

∂x j(ψ i f (x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1)))w jdydx1 · · ·dxp−1,

and now interchanging the order of integration and using the fact that spt(ψ i) ⊆ Qi, itfollows this term equals zero. The reason this is valid is that

x j→ ψ i f (x1, · · · ,xp−1,y+gi (x1, · · · ,xp−1))

is the composition of Lipschitz functions and is therefore Lipschitz. Therefore, this func-tion can be recovered by integrating its derivative, Lemma 26.2.6.

Then, changing the variable back to xp it follows 29.7.32 reduces to

−∫

Ai

∫ gi(x1,··· ,xp−1)

−∞

 ∑p−1j=1 Dp (ψ i f )(x1, · · · ,xp−1,xp)

·gi, j (x1, · · · ,xp−1)w j

dxpdx1 · · ·dxp−1

29.7. THE DIVERGENCE THEOREM 1027Letting x» = y+ gi (x1,--- ,Xp—1) and changing the variable, this equals0 ?P[ [Xeni oe apy tsi lai tp-1)):i7—8 j=]w jdydx, -++dxXp—|0 P= I / VY Dj (Wi) G1. Xp. + 81 (M1, -+* xpi) +i ~o j=]w jdydx, -++dxp—|Recall A; is all of B; except for the set of measure zero where the derivative does not exist.Also D, refers to the partial derivative taken with respect to the entry in the j'” slot. In the" slot i is found not just x, but y+ g;(x1,--: ,xp—1) so a differentiation with respect to x;will not be the same as Dj. In fact, it will introduce another term involving g;_;. Thus fromthe chain rule,0 ¥ 1=|, JX if (X15°°* Xp—15Y +81 (X15 °° -Xp—1))) WiDp (Wf) Oy Xp YF Bi (AI * -Xp—1)) * Bij O19 Apa) Wjdydx ---dxp—10+/ / Dp (Wf) (x1,°°° Xp-1,Y + 8i (x1,°°° Xp—1)) Wpdydxy ++dXp—1 (29.7.32)Consider the term0 ¥ 1in» if (X15°°* Xp—1,Y +i (X1,°+* .Xp—1))) widydx1 ---dxp—1This equals0 ¥ 1i> Y ie (Wah Cs pty tb Be (1 xpi) widydny dpa,jand now interchanging the order of integration and using the fact that spt(w;) C Qj, itfollows this term equals zero. The reason this is valid is thatxj => Wf (a1,°°° Xp-1, V+ Bi (x1,°°° Xp—1))is the composition of Lipschitz functions and is therefore Lipschitz. Therefore, this func-tion can be recovered by integrating its derivative, Lemma 26.2.6.Then, changing the variable back to x, it follows 29.7.32 reduces to-1gi(X1 Xp i Dp (Wif) 15++* Xp—15Xp)-[f[ “Bij (X1,°°* Xp-1) Wj dX pdx, +--AXp—|