Makes all faces in a box divergence formula vanish by placing the
support strictly inside the box.
Statement
Fix
.
Let
be
and have compact topological support. Let
denote the real Fréchet derivative and
the
th
standard coordinate vector. Write
For Lebesgue measure on the coordinate space,
Assumptions
The support convention throughout this card is
This closed set is assumed compact. The differentiability hypothesis is
exactly
.
The domain has its coordinate sup norm. No externally specified
integration domain or boundary-regularity assumption is needed.
Conclusion
The divergence is integrable and its whole-space integral vanishes.
Compact support is imposed on
,
which also forces the divergence to vanish off that support.
Notes
For
the two faces are endpoints and the same formula is the fundamental
theorem of calculus. The source represents face coordinates by inserting
one fixed coordinate into an
-tuple.
Proof route
Choose a box with the support strictly inside it. Write the box
divergence formula as a sum of upper-face minus lower-face integrals.
Each face value is zero, and the divergence is zero outside the box, so
the whole-space integral is zero.
Proof steps
Enclose the support and justify integrability.
Choose
such that
If
,
some neighborhood of
is disjoint from the nonzero set of
.
On that neighborhood
,
and hence
.
Therefore
The divergence is continuous because
is
,
so it is integrable on compact
.
It is zero on
,
and is therefore integrable on
as well.
Spell out the face formula. Fix
.
Let
and write
.
Insert the missing coordinate by
For fixed
,
the scalar function
is
and
The face expression in the box divergence theorem is precisely the
following repeated-integral calculation (Fubini on the compact box,
followed by the one-dimensional fundamental theorem of calculus):
Sum over
to obtain
.
In the source this sum is obtained directly from the cited box
divergence theorem:
supplies continuity and a derivative everywhere, the exceptional set is
empty, and Step 1 supplies integrability. The calculation above explains
its face terms.
Set the face values to zero and pass to the whole
space. The
th
coordinate of
is
.
Hence
Each integrand in the last line of Step 2 is thus
.
Since Step 1 also gives
on
,
For
,
is the zero-dimensional one-point product; the formula is simply the
difference of the two endpoint values.