Definition
Order of a distribution
The order of a distribution is the least derivative degree needed in uniform seminorm estimates on compact supports.
Definition
Let be a distribution on an open set . It has order at most if for every compact there is such that
for every with support in , where has nonnegative integer entries and . Its order is the least such nonnegative integer ; if no single works for all compact , its order is infinite. Thus order records how many derivatives of a test function are needed to control the distribution.
Examples and differentiation
A distribution represented by a locally finite measure has order . The Dirac distribution has order , while has order . More generally, distributional differentiation raises the order by at most the number of derivatives:
These estimates follow directly from the definition of distributional derivative Hörmander, §2.3.
Local and global scope
Continuity of on each fixed-support test-function space implies a finite-order estimate on every compact , but the required integer may grow with . Consequently every distribution has finite order locally, while not every distribution has one finite global order on a noncompact open set. A compactly supported distribution does have finite order Folland, Chapter 9.
Conventions
Some texts say “order ” when they only mean “order at most .” Here “order” means the least admissible . The zero distribution is bounded by every order estimate; assigning it order or is a convention that should be stated when formulas depend on the value.
References
- Lars Hörmander, The Analysis of Linear Partial Differential Operators I: Distribution Theory and Fourier Analysis, 2nd ed., Springer, 2003. Springer DOI record. Relevant: §2.3, distributions of finite order and compact support.
- Gerald B. Folland, Real Analysis: Modern Techniques and Their Applications, 2nd ed., Wiley, 1999. Wiley publisher record. Relevant: Chapter 9, distributions and their order.