Distribution theory extends the idea of a function so singular objects and weak derivatives can be handled rigorously. It gives precise mathematical meaning to objects such as the Dirac delta, derivatives of discontinuous functions and point sources that ordinary classical functions cannot represent directly.
The core move is subtle and powerful: instead of asking what a generalized function does at each point, ask how it acts on a carefully chosen family of smooth test functions. Singular behaviour is transferred from pointwise values into linear functional action.
Series route: Mathematics Learning Hub → How Mathematics Works → Distribution Theory. Useful foundations include Real Analysis, Functional Analysis, Harmonic Analysis and Partial Differential Equations.
1. Classical functions can be too narrow
A classical derivative requires local pointwise smoothness. Yet many equations in physics and engineering naturally involve discontinuities, impulses or concentrated sources.
Distribution theory enlarges the space of admissible objects so differentiation survives beyond classical smoothness.
2. Test functions are smooth probes
The standard test-function space D(Ω) consists of infinitely differentiable functions with compact support inside an open set Ω.
Compact support keeps each probe local and prevents boundary-at-infinity complications.
3. A distribution acts on test functions
A distribution T assigns a scalar T(φ) to every test function φ and does so linearly and continuously in the topology of the test-function space.
The generalized object is therefore defined by its responses to all allowable probes.
4. Ordinary locally integrable functions generate distributions
If f is locally integrable, define T_f(φ)=∫f(x)φ(x)dx.
This embeds a large class of ordinary functions into distribution space.
5. Different functions can represent the same distribution almost everywhere
If two locally integrable functions differ only on a set of measure zero, their integrals against every test function agree.
Distribution theory therefore inherits the measure-theoretic idea that pointwise changes on null sets can be invisible.
6. The Dirac delta is evaluation at a point
The Dirac distribution at x₀ is defined by δ_{x₀}(φ)=φ(x₀).
It behaves like a unit point mass under integration but is not an ordinary function with an infinite spike.
7. The delta is rigorous because its action is finite
The misleading picture “zero everywhere and infinity at one point” is unnecessary.
All required meaning is encoded by the rule φ↦φ(x₀).
8. Distributional differentiation moves derivatives onto the test function
The derivative T′ of a distribution T is defined by T′(φ)=−T(φ′).
The minus sign comes from integration by parts and makes the generalized definition agree with classical differentiation when the classical derivative exists.
9. Every distribution can be differentiated indefinitely
Unlike classical functions, distributions do not lose differentiability after a finite number of derivatives.
The derivative remains another continuous linear functional on test functions.
10. A jump creates a delta derivative
Let H be the Heaviside step function, zero for negative x and one for positive x under a conventional definition at zero.
Its classical derivative vanishes away from the jump and fails to exist at zero. Distributionally, H′=δ.
The lost derivative has become a concentrated source at the discontinuity.
11. Piecewise smooth functions acquire singular derivative terms
If a function has jumps, its distributional derivative contains its ordinary derivative away from the jumps plus delta terms weighted by jump sizes.
This is a precise mathematical version of “all the missing change occurred at the jump.”
12. Higher derivatives create derivatives of delta
The derivative δ′ is defined by δ′(φ)=−φ′(0).
Successive derivatives encode increasingly singular local response while remaining rigorous linear functionals.
13. Multiplication by smooth functions is well defined
If a is smooth and T is a distribution, define (aT)(φ)=T(aφ).
This makes variable coefficients compatible with generalized solutions.
14. Arbitrary multiplication of distributions is not generally defined
There is no universal product extending ordinary multiplication to all distributions while preserving every desired property.
Expressions such as δ² therefore require a more specialised framework rather than naive symbolic multiplication.
15. Convolution smooths distributions
Convolution of a distribution with a smooth compactly supported function produces a smooth function under the standard construction.
This lets singular objects be approximated by smooth representatives.
16. Mollifiers create controlled smooth approximations
A mollifier is a smooth compactly supported approximate identity scaled to a small length ε.
Convolving with a mollifier replaces sharp structure by a smooth local average while recovering the original object in an appropriate limiting sense.
17. Approximate identities explain the delta intuitively
A family of narrow unit-mass smooth functions can converge to δ distributionally.
The limit is not pointwise convergence to an ordinary function. It is convergence of integrals against every test function.
18. Support records where a distribution is active
The support of a distribution is the smallest closed set outside which it acts as zero on all test functions.
δ has support at a single point.
19. Tempered distributions permit Fourier analysis
Tempered distributions are continuous linear functionals on the Schwartz space of rapidly decreasing smooth functions.
They include many slowly growing functions and singular objects while remaining compatible with the Fourier transform.
20. Fourier transformation extends by duality
The Fourier transform of a tempered distribution is defined through its action on transformed test functions.
This preserves the elegant rules linking differentiation and multiplication by frequency.
21. The Fourier transform of delta is constant
Under standard transform conventions, a point impulse has uniform frequency content.
Conversely, a constant function transforms into a concentrated mass at zero frequency up to convention-dependent factors.
22. Fundamental solutions turn point sources into PDE kernels
A fundamental solution E for a differential operator L satisfies LE=δ in the distributional sense.
Convolving E with a forcing term can then construct solutions of Lu=f under suitable conditions.
23. Green’s functions are source-response machines
Boundary-value problems often use Green’s functions adapted to both the differential operator and boundary conditions.
A point-source response can be integrated against general forcing to reconstruct the full solution.
24. Weak solutions arise naturally from distributions
A PDE solution need not have classical derivatives everywhere if the equation can be interpreted distributionally.
This expands the solution class enough to include physically relevant shocks, interfaces and low-regularity data.
25. Weak derivatives lead directly to Sobolev spaces
A function has weak derivative g when its distributional derivative is represented by the locally integrable function g.
Sobolev spaces collect functions whose weak derivatives belong to specified Lp spaces.
Continue through Sobolev Spaces.
26. Distributional equations preserve conservation laws across jumps
Conservation-law PDEs can develop discontinuous shock solutions even from smooth initial data.
Weak formulations integrate the conservation law against test functions so the equation remains meaningful across the shock.
27. Weak solutions may not be unique
Enlarging the solution class can admit multiple weak solutions.
Additional entropy conditions or admissibility principles may be needed to select the physically relevant solution.
28. Distribution theory is linear by design
Distributions form a vector space and interact beautifully with linear differential operators.
Nonlinear operations require more care because products and compositions can fail to be defined on singular distributions.
29. Singular support records where smoothness fails
The singular support of a distribution is the set outside which it can be represented by a smooth function.
This separates where an object exists from where its nonsmoothness exists.
30. Microlocal analysis refines singularity by direction
Wavefront sets record not only where a distribution is singular but also in which frequency directions the singularity occurs.
This becomes essential in propagation-of-singularities theory and advanced PDE analysis.
31. A worked mechanism: derivative of the Heaviside step
Let H(x)=0 for x<0 and H(x)=1 for x>0. To compute its distributional derivative, test against φ.
- By definition, H′(φ)=−H(φ′).
- H(φ′)=∫_0^∞ φ′(x)dx.
- Compact support makes φ vanish far away.
- Thus ∫_0^∞φ′(x)dx=−φ(0).
- Therefore H′(φ)=φ(0)=δ(φ).
So H′=δ distributionally.
32. A worked mechanism: delta as a point source
Suppose −u″=δ on the real line in a simplified one-dimensional model.
- Away from zero, u″=0, so u is linear on each side.
- The derivative u′ must jump across zero so its derivative produces a delta.
- The jump size is fixed by integrating the equation across a small interval around zero.
- The singular forcing becomes a matching condition between ordinary solutions on either side.
33. Common distribution-theory failure modes
- Delta-as-function confusion: treating δ as an ordinary infinite-valued function.
- Pointwise-limit confusion: interpreting distributional convergence as ordinary pointwise convergence.
- Product overreach: multiplying arbitrary distributions as if ordinary function algebra still applied.
- Weak=approximate confusion: assuming a weak derivative is merely an imprecise derivative.
- Test-space blindness: changing the class of test functions without noticing the dual space changes too.
- Weak-solution uniqueness: assuming distributional existence automatically gives a unique physically relevant solution.
34. Distribution theory as a mathematical machine
Singular or Low-Regularity Object → Test-Function Action → Distribution → Weak Differentiation → Convolution/Fourier Transform → Weak PDE Formulation → Regularity or Physical Interpretation.
35. What mastery looks like
- understand distributions as continuous linear functionals on test spaces;
- embed ordinary locally integrable functions into distribution space;
- treat the Dirac delta through evaluation rather than an infinite-spike picture;
- compute distributional derivatives by integration by parts;
- recognise jump terms and delta derivatives;
- use mollification and convolution correctly;
- extend Fourier analysis to tempered distributions;
- connect weak derivatives to Sobolev spaces;
- separate existence of weak solutions from uniqueness and admissibility.
36. Further reading
- Encyclopedia of Mathematics: Generalized Function
- Encyclopedia of Mathematics: Generalized Functions, Space Of
- AMS Mathematics Subject Classification
37. Conclusion
Distribution theory works by replacing pointwise evaluation with action on smooth probes. That shift makes point sources, impulses and jump derivatives mathematically legitimate. Differentiation becomes universal. Fourier transforms extend to singular objects. PDEs gain weak solutions beyond classical smoothness.
Classical analysis asks what a function does at each point. Distribution theory asks what an object does to every smooth local test—and gains enough flexibility to keep differentiation alive when pointwise calculus breaks.
How Mathematics Works | Batch 13
- How Mathematics Works | Convex Analysis
- How Mathematics Works | Nonlinear Analysis
- Distribution Theory — this article
- How Mathematics Works | Sobolev Spaces
Return to the Mathematics Learning Hub or the How Mathematics Works root.