We derive the Cox-de Boor recurrence from the Leibniz rule for divided differences. Fix , let , and set
Away from the spline knots, . Introduce
The defining divided-difference recurrence says . Since and the first factor is linear, the Leibniz rule for divided differences gives
Multiplication by consequently yields
Replace and by the corresponding lower-order B-splines divided by their support widths. This proves
The denominators are positive for the distinct spline knots of this question. The recursion starts with , the interval indicator on . For , values at the spline knots follow by the continuous extension of the left side; the usual half-open convention for the order-one factors gives the same result. This proves the recurrence rather than assuming it as a definition.
Write the strictly increasing local spline knots as , . The explicit formula for a divided difference gives
Each summand is a truncated power function of : it is a polynomial on either side of its knot , and for is globally . Hence is a piecewise polynomial function of degree at most , with these spline knots and at least this global smoothness.
For , all summands vanish. For , all knot values agree with those of the ordinary polynomial in the divided-difference variable . Its order- divided difference is zero, because its degree is less than . Thus also vanishes to the left of .
The closed support is exactly the indicated interval, rather than merely contained in it. For , subtracting the omitted term from the zero divided difference of gives
For , only the last truncated-power term survives, giving
There are therefore nonzero values arbitrarily close to either endpoint, proving
At each simple knot, the st derivative has a nonzero jump from exactly one truncated-power summand, so this is also the exact global smoothness. For order , the function is instead the normalized interval indicator; there is no assertion of classical continuity, and the notation is only the customary formal spline smoothness notation. This distinction is part of simple-knot B-spline regularity.
Take . In the normalization used here, the Dirichlet kernel has the finite expansion
Averaging a finite number of the integral formulas for the Fourier partial sums is legitimate by linearity of the integral. Hence the Fejér sum is convolution with
To obtain the nonnegative form of the Fejér kernel, expand a squared geometric sum:
The coefficient counts pairs of indices whose difference is . Dividing by and summing the geometric progression gives
At the ratio has its continuous limiting value . Thus this is a continuous, nonnegative trigonometric polynomial, not a kernel with genuine singularities.
Integration over a full period kills every nonconstant cosine term, so
Translation invariance of integration over the circle consequently gives, for every ,
The first step is the integral triangle inequality, the second uses nonnegativity, and the last uses the mass just computed. Taking the supremum norm proves
In particular, Fejér summation is a uniform-norm contraction. The factor , rather than , is essential for the half-normalized Dirichlet kernel and Fejér kernel in this problem.
A local Maxwellian can be written as , with and coefficients depending on . Its logarithm is a linear combination of collision invariants, so and at each spatial point.
Comparing powers of in gives
Here is the symmetric part of a matrix formed by the coefficients of the quadratic polynomial. For constants , choose , and . All four coefficient equations hold. The resulting expanding Gaussian solution of the Boltzmann equation is
Direct verification is particularly simple: and are constant along free characteristic curves, so the transport derivative is zero. At fixed its logarithm is the displayed Maxwellian quadratic, so the collision term is also zero. Its spatial dependence is nonconstant, it is positive and rapidly decaying, and its Gaussian velocity integral is finite because . It is not a compactly supported example, nor does this final part require one.
Strict positivity throughout phase space is incompatible with compact support. The following calculation uses positive smooth rapidly decaying solutions with all displayed integrals justified; nonnegative cases use the corresponding entropy limits where justified.
Write and use integration over . The two symmetrized weak collision identities are
and
They follow from particle interchange and the angular exchange for elastic collisions. They are the weak identities needed here; no invalid fixed-angle Jacobian is used.
The collision invariants , and obey . Momentum follows from , and energy from . Hence
For the Boltzmann H functional , the spatial transport contribution is a boundary divergence and the term is zero. Set , . The second weak identity with gives the Boltzmann H theorem
Indeed this integrand with its negative sign is for , which is nonpositive. The kinetic H functional decreases; the physical entropy with opposite sign increases.
Interpret as the velocity operator appearing in the equation and in the stated quadratic-form hypothesis. Let . Smoothness turns the distributional equation into the usual equation. Multiply it by and integrate; for complex functions use the real part of the inner product. Spatial integration by parts gives , while the coercivity hypothesis at each gives
Thus is nonincreasing. The coercive energy estimate for kinetic transport is
No division by is required, so the zero solution causes no exception. Compact spatial support, or sufficient decay to eliminate the boundary flux, is enough; compact support in time is unnecessary.
The shear is a smooth diffeomorphism with determinant one and sends compact sets to compact sets. Thus are locally integrable functions. On a compact time interval and compact label set , the integral is finite. The Fubini theorem, followed by a countable exhaustion of time intervals and label sets, shows that for almost every .
To transform the weak equation, choose , where and . Including a velocity cutoff in ensures the required compact support. The transport derivative of this test function is . Change variables to the characteristic labels to obtain
The fundamental test-function identity makes the bracket zero almost everywhere. Choose a countable dense family of time test functions on each bounded interval and extend by continuity; there is then a single negligible label set outside which as one-dimensional distributions.
For each remaining label, subtract the primitive . The distributional derivative of is zero, so it is a constant almost everywhere. Choose the resulting absolutely continuous representative along free characteristics, . The fundamental theorem of calculus gives
The representative qualification is essential: an arbitrary locally integrable representative can be modified on one time slice without changing the distributional equation, and would not satisfy the identity for every pair of times.
The reference to part (f) within this part is a printed self-reference; the needed product estimate is part (e). Since both masses are one, subtracting the Bobylev identities gives, for ,
The normalized angular average and part (e) give
For completeness, the Duhamel principle for this scalar equation yields . Apply the Gronwall inequality to to obtain the Fourier nonexpansion for Maxwell molecules, . This estimate is nonexpansion; by itself it does not prove strict decay or convergence to a specified equilibrium.
Expansion of the two squares gives
Let be the Fourier distance of order two, with the supremum taken over . Both Fourier transforms have modulus at most one. Add and subtract , then use the triangle inequality:
Dividing by proves the bound by , and splitting the last expression into its two weighted terms gives the requested intermediate inequality. If one of vanishes, its unweighted difference is zero by equal mass; its weighted term is interpreted as zero, avoiding a quotient.
Equal mass and first moment also explain finiteness of this distance: subtract the constant and linear Taylor terms in the Fourier integral and use . This bounds the difference by . No direction-independent extension of the quotient at zero is required.
Apply the spherical identity of part (b) with , and . The angular exponential in part (c) can be replaced, after angular integration, by . The full phase is then
The Fubini theorem factors the two velocity integrals into Fourier transforms. With , the gain integral is exactly .
The loss Fourier transform is . Dividing the gain by the sphere area gives the Bobylev identity for the Maxwell molecule collision operator:
The initial Fourier datum is . The surface measure on a sphere here is two-dimensional surface area, not the restriction of ambient three-dimensional Lebesgue measure, which would give the sphere measure zero.
The printed hint with unchanged is not a valid change of variables: at fixed , the outgoing velocities forget the direction of . A correct proof uses the angular exchange for elastic collisions.
Set and , where and . Then , while and . The gain integral becomes
Exchange the two independently integrated angular variables and . The measure is unchanged, and reverting to , gives
The measure-zero set causes no difficulty. This proves the requested identity without invoking the false fixed- Jacobian assertion.
If either vector vanishes, both arguments are zero. Otherwise choose a rotation matrix sending to . The surface measure on a sphere is rotation invariant. Changing variables gives
Thus the two spherical integrals are equal. The property extends to complex continuous by treating real and imaginary parts separately, as needed for the Fourier exponential.
Nonnegativity and unit mass give, by the triangle inequality for the Fourier transform, . The finite second moment implies a finite first absolute moment, so differentiating under the Fourier integral is justified and
The exponent convention has no , so no such factor occurs in this derivative.
The orthogonal projection satisfies , hence and . Set . Then and . Crucially, applying the energy identity of part (e) to gives
Solving this scalar equation and taking square roots proves
This argument uses (e) explicitly, rather than bypassing the requested energy method with an explicit solution formula.
Using and the orthogonal projection identity , we have . The orthogonal decomposition then gives
The sign is negative, as in the PDF; the converted TeX loses it at the last equality. This establishes the exact spectral gap of normalized velocity relaxation: on the zero-mass subspace , with gap one rather than the weaker half-gap estimate.
Since , . Also exactly when , which is exactly the range of . Thus
A bounded self-adjoint idempotent is an orthogonal projection: if and , then . Here the orthogonal complement is precisely the zero-mass subspace .
The weighted inner product satisfies . Therefore
The identity operator and are linear, bounded and self-adjoint, hence so is . Directly, ; equality holds at . Thus .
For the normalized velocity-reset collision operator,
This bounded everywhere-defined operator is self-adjoint. Put . Expanding the square with the probability measure gives
It follows that
Replacing the difference by gives the other printed integral expression. For real functions the modulus squares are ordinary squares; the modulus version also proves the complex-space statement.
Use the weighted Hilbert structure of velocity-reset relaxation, with inner product and norm . The Cauchy-Schwarz inequality gives
Thus is an absolutely convergent integral and a bounded linear functional. Since , the normalized velocity-reset collision operator obeys
This establishes boundedness. The subsequent orthogonal decomposition improves the operator norm to exactly one.
Let . The damped free-transport evolution preserves nonnegativity, and a nonnegative kernel makes preserve it as well. Every term in is therefore nonnegative. By linearity, the difference solves the fixed point equation with initial value . The assumed uniqueness identifies it with this nonnegative sum on every finite interval. Hence the order preservation for the linear Boltzmann equation gives
Continuity upgrades pointwise limits in the series to the stated continuous solution; no pointwise differentiability or minimum principle is needed.

Pinned article: Introduction to the OurBigBook Project

Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
We have two killer features:
  1. topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculus
    Articles of different users are sorted by upvote within each article page. This feature is a bit like:
    • a Wikipedia where each user can have their own version of each article
    • a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
    This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
    Figure 1.
    Screenshot of the "Derivative" topic page
    . View it live at: ourbigbook.com/go/topic/derivative
  2. local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:
    This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
    Figure 5. . You can also edit articles on the Web editor without installing anything locally.
    Video 3.
    Edit locally and publish demo
    . Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.
  3. https://raw.githubusercontent.com/ourbigbook/ourbigbook-media/master/feature/x/hilbert-space-arrow.png
  4. Infinitely deep tables of contents:
    Figure 6.
    Dynamic article tree with infinitely deep table of contents
    .
    Descendant pages can also show up as toplevel e.g.: ourbigbook.com/cirosantilli/chordate-subclade
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact