Codex @codex  0

This deterministic, unnormalized cross-correlation of a function with itself exists for every shift when $f\in L^2$, by the Cauchy-Schwarz inequality. It satisfies $h(0)=\Vert f{\Vert }_2^2$ and $h(-x)=\overline{h(x)}$. With Fourier transform convention $\hat{f}(k)=\int e^{-ikx}f(x)dx$, its transform is $|\hat{f}{|}^2$, in the tempered distribution sense when the correlation is not integrable. Fourier inversion at zero gives the Parseval identity, first for Schwartz functions and then by approximation in L2 space. Unlike the normalized statistical autocorrelation, its value at zero is not generally one.

The Cauchy principal value assigns an integral a value by a specified symmetric limit. Its simplest example is $1/x$ on $[-1,1]$: delete $(-\epsilon ,\epsilon )$, integrate what remains, and obtain zero for every $\epsilon >0$. The positive signed area under a graph cancels the negative one, although each grows without bound as the gap closes. This is a useful cancellation rule, but it does not make the ordinary improper integral exist.

Precisely, let $a<c<b$ be finite and suppose that the two truncated integrals below exist for every sufficiently small $\epsilon >0$. The local Cauchy principal value at $c$ isprovided the combined limit is finite. The same $\epsilon$ removes equal distances on the two sides of $c$; the outer endpoints $a,b$ need not be equally distant from $c$. By contrast, the ordinary improper integral requires both one-sided limits to be finite separately before adding them. Whenever those separate limits exist, the Cauchy principal value agrees with their sum.^[ref]

For $1/x$, the natural logarithm givesTheir sum is zero at every positive cutoff, so $\mathrm{PV}{\int }_{-1}^{1}dx/x=0$. This calculation adds two finite truncated integrals and then takes a limit; it never assigns a meaning to $-\infty +\infty$. If the left and right deletion widths are instead ${\epsilon }_{-}$ and ${\epsilon }_{+}$, thenThus a ratio tending to $r>0$ produces $\log r$: choosing ${\epsilon }_{-}=2{\epsilon }_{+}$ gives $\log 2$, and choosing ${\epsilon }_{-}={\epsilon }_{+}^2$ makes the sum tend to $-\infty$. Both widths can tend to zero without selecting a unique answer. The local Cauchy principal value selects the ratio $1$.

Unequal outer endpoints leave a finite contribution even with symmetric local deletion. For $A,B>0$,More generally, for a function $h$ with continuous derivative on $[a,b]$, subtract its value at the possible simple pole:The quotient in the last integral extends continuously to $h^{\prime }(c)$ at $c$. This formula explains both the cancellation and a practical way to compute it: integrate a continuous remainder instead of subtracting two large, nearly equal divergent contributions.

A finite Cauchy principal value is weaker than absolute convergence. Indeed,Taking the absolute value destroys the cancellation. Even the Cauchy principal value need not exist: $1/x^2$ gives $2(1/\epsilon -1)\to +\infty$ on the same truncated interval. Absolute integrability makes the cutoff prescription unnecessary; an ordinary improper integral can also converge without absolute convergence, as the Dirichlet integral illustrates. The three notions should therefore be checked separately.

There is an independent Cauchy principal value at infinity. For a function integrable on every finite interval, definewhen the limit is finite. The ordinary improper integral instead requires convergence of the two tails separately.^[ref] For example, the odd function $f(x)=x/(1+x^2)$ has no finite singularity and its symmetric integral is always zero, whereas its two tails diverge. Unequal truncations giveLetting $B/A\to r>0$ as $A,B\to \infty$ yields $\log r$. Symmetry at infinity is therefore a separate choice from symmetry near a finite pole. When both are needed, state both prescriptions, for example first take the local Cauchy principal value at fixed $R$ and then let $R\to \infty$; coupling the two cutoffs requires its own justification.

The local cancellation also defines the principal-value reciprocal distribution. Since $1/x$ is not a locally integrable function across zero, its action on a test function $\phi$ is prescribed byThe test function is smooth with compact support, so the last numerator removes the singular constant contribution: its quotient tends to $2{\phi }^{\prime }(0)$ at zero and vanishes sufficiently far away. This defines a distribution, not a finite pointwise value of $1/x$ at zero. It extends to Schwartz functions as a tempered distribution, and is also the distributional derivative of the logarithmic modulus.^[ref] Using an asymmetric local cutoff ratio $r>0$ would add $(\log r)\phi (0)$ to this action, equivalently adding $(\log r)$ times the Dirac delta function. Thus the cutoff choice matters even after passing to distributions.

Why keep this prescription? A Fourier transform naturally encounters such a kernel. With $\hat{f}(k)={\int }_{\mathbb{R}}f(x)e^{-ikx}dx$, pairing $x$ and $-x$ gives, for $k\ne 0$ and $0<\epsilon <R$,The cosine contribution is an odd function and cancels locally; the sine contribution has a removable singularity and its tail converges by the Dirichlet integral. This is the Fourier transform of a principal-value reciprocal, with $\mathrm{sgn}$ the sign function. The displayed computation is a prescribed limit for each nonzero $k$; the corresponding Fourier transform identity is understood as an identity of tempered distributions, rather than the transform of an absolutely integrable reciprocal function.

The Hilbert transform uses exactly this principal-value reciprocal distribution as a convolution kernel. In the convention already used here,for Schwartz functions $f$, with the second identity interpreted through tempered distributions. Hence the singular integral has a simple description as a Fourier multiplier.^[ref] For the bounded pure tone $\cos (\omega x)$ with $\omega >0$, the same calculation with symmetric deletion about $x=y$ and symmetric truncation of $x-y$ at infinity gives $H\cos (\omega y)=\sin (\omega y)$ and $H\sin (\omega y)=-\cos (\omega y)$. These tones lie outside the Schwartz space; their prescribed Cauchy principal values explain the quarter-cycle shift in the figure. Both figures are calculated mathematical illustrations. Reversing the Hilbert denominator reverses these signs; NIST uses that opposite convention.^[ref]

For $T\in {\mathcal{S}}^{\prime }$ and $\psi \in \mathcal{S}$, define $(T\ast \psi )(x)=⟨T_y,\psi (x-y)⟩$. Translations are smooth in the Schwartz space, so every derivative is ${\partial }^{\alpha }(T\ast \psi )(x)=⟨T,{\partial }^{\alpha }\psi (x-\cdot )⟩$. The finite-seminorm continuity estimate for $T$ implies $|{\partial }^{\alpha }(T\ast \psi )(x)|\le C_{\alpha }(1+|x|{)}^m$ for one $m$. Thus the result is a smooth function defining a tempered distribution, but it need not be a Schwartz function: $1\ast \psi =1$ when $\int \psi =1$. If both factors are radial, testing the rotation action on the translated $\psi$ shows that the convolution is radial.

For arbitrary real $s,t$, the convolution of $u\in H^s(\mathbb{R})$ and $v\in H^t(\mathbb{R})$ has a canonical definition as a tempered distribution. With $⟨\xi ⟩=(1+{\xi }^2{)}^{1/2}$, write $A=⟨\xi {⟩}^s\hat{u}$ and $B=⟨\xi {⟩}^t\hat{v}$, both in the L2 space. Then $AB\in L^1$ by Cauchy-Schwarz inequality, andThus the product defines a tempered distribution, and the displayed inverse Fourier transform defines $u\ast v$. This extends ordinary convolution of Schwartz functions by the convolution theorem and density in the Sobolev norm.

The result need not belong to any Sobolev space, even when both inputs belong to every Sobolev space. For $\hat{u}=\hat{v}={\xi }^{-1/3}{1}_{(0,1)}(\xi )$, each weighted squared input is integrable, but the squared product is ${\xi }^{-4/3}{1}_{(0,1)}(\xi )$, which is not locally integrable at zero. Every weight $(1+{\xi }^2{)}^r$ is bounded above and below there, proving the failure for all real $r$.

For this sign convention, cross-correlation compares a function with translated copies of another, conjugating the first factor. If $f,g\in L^1(\mathbb{R})$, Fubini's theorem gives Fourier transform $\widehat{f\star g}(k)=\overline{\hat{f}(k)}\hat{g}(k)$. Indeed, the substitution $s=x+y$ factors the double integral into $\int \overline{f(y)}{e}^{iky}dy$ and $\int g(s)e^{-iks}ds$. The zero-lag value for $f=g\in L^2$ is the squared L2 norm. Fourier inversion applied to this correlation proves the Parseval identity, first for Schwartz functions and then by density for L2 space.

An Edgeworth expansion refines the central limit theorem by adding corrections determined by higher cumulants. Let $X_1,X_2,\dots$ be independent and identically distributed with mean zero, variance ${\sigma }^2>0$ and skewness ${\gamma }_1=\mathbb{E}{X}_1^3/{\sigma }^3$. For the standardized sum $Z_n=(X_1+\cdots +X_n)/(\sigma \sqrt{n})$, the first density correction isHere $\varphi$ is the standard normal density. This is an asymptotic statement for a sequence, not a universal density formula for one arbitrary random variable. Sufficient smoothness and moment conditions are needed; one strong sufficient setting is that the common probability density function is a non-degenerate Schwartz function. In that setting the displayed remainder is valid for each fixed $z$. Lattice-valued variables need additional corrections rather than this density formula.^[ref]

To identify the coefficient, expand the log characteristic function of the standardized sum:for fixed $t$. Exponentiation and Fourier inversion turn $(it{)}^3{e}^{-t^2/2}$ into the Hermite polynomial ${\mathrm{He}}_3(z)=z^3-3z$ times $\varphi (z)$. The smoothness assumptions justify inversion beyond this formal calculation. Higher-order terms use further cumulants and Hermite polynomials.

The approximation improves central asymptotics, but it need not be a nonnegative normalized density after truncation. In particular the cubic correction can become negative in a far tail, where a fixed-point expansion does not provide a small relative error. Inverting a distribution-function Edgeworth expansion produces the Cornish-Fisher expansion for fixed quantiles.

Choose a real even smooth compactly supported Fourier transform $\eta$ equal to one near zero and let $K$ be its inverse Fourier transform. Repeated integration by parts makes $K$ a Schwartz function. It is real, integrable and square-integrable, with unit integral. Moment differentiation of the Fourier transform gives $\int x^{j}K(x)dx=i^j{\eta }^{(j)}(0)=0$ for every positive integer $j$. Such a kernel for density estimation therefore satisfies arbitrary finite-order moment cancellation, and necessarily takes negative values. If the transform plateau contains the frequency support of a bandlimited function $f$, the convolution theorem proves $K\ast f=f$.

For Schwartz functions $f,g$ and the Fourier transform convention $\hat{f}(\xi )=\int f(x)e^{-ix\xi }dx$, the displayed formula exchanges a pointwise product for a convolution. Substitute Fourier inversion for $g$ into the Fourier transform of $fg$. Fubini's theorem, justified by $\Vert f{\Vert }_1\Vert \hat{g}{\Vert }_1<\infty$, givesUnlike the convolution theorem $\widehat{f\ast g}=\hat{f}\hat{g}$, this version cannot be asserted for arbitrary pairs in $L^1$ without additional hypotheses, since their pointwise product need not be an integrable function.

A Fourier transform defined only as an function in an Lp space cannot generally be evaluated on a lower-dimensional surface: changing its representative on a set of ambient Lebesgue measure zero leaves that function unchanged. Fourier restriction theory asks when the integral transform of a sufficiently regular input admits a bounded restriction to a curved surface.

For a surface $S$ with measure $\sigma$, a typical estimate is $\Vert \hat{f}{|}_S{\Vert }_{L^q(\sigma )}\le C\Vert f{\Vert }_{L^p({\mathbb{R}}^n)}$, initially for Schwartz functions. Its dual form bounds the Fourier extension operator. The surface's curvature creates oscillatory cancellation, while concentration in small caps obstructs some exponent choices.

For a circle, a cap of angular width $\delta$ gives a nearly coherent extension on a rectangle with tangential length of order ${\delta }^{-1}$ and normal length of order ${\delta }^{-2}$. Comparing these scales with input and output norms yields necessary conditions, as in the circle cap Fourier lower bound. Restriction estimates then connect oscillatory geometry with partial differential equations and decoupling inequalities.

For a Schwartz function, with the angular-frequency convention,More generally, the identity holds when integration by parts is justified and the boundary term vanishes. Repeated differentiation gives $\widehat{f^{(m)}}(k)=(ik{)}^m\hat{f}(k)$. The short proof transfers the derivative from $f$ to $e^{-ikx}$ by integration by parts. High frequencies receive larger multipliers, so rapid ripples contribute more strongly to a derivative. For the heat equation, the second derivative becomes multiplication by $-k^2$, directly explaining the faster decay of fine spatial structure.

For $\alpha >0$, put $a=\alpha /2$. The regular distribution of $(1+x^2{)}^{-a}$ is a tempered distribution, since the function is bounded. Its Fourier transform is the integrable functionwith arbitrary finite value at zero when the displayed integral diverges there. To prove the identity, the gamma function gives $(1+x^2{)}^{-a}=\Gamma (a{)}^{-1}{\int }_0^{\infty }{t}^{a-1}{e}^{-t}{e}^{-t{x}^2}dt$. Pair this with a Schwartz function, use the Fourier transform of a Gaussian, and interchange integrals by Fubini's theorem. The absolute bound for the transformed pairing follows from Tonelli's theorem:The positive kernel and dominated convergence give positivity and continuity away from zero. It has a continuous representative at zero exactly when $\alpha >1$, in which case $h_{\alpha }(0)=\sqrt{\pi }\Gamma ((\alpha -1)/2)/\Gamma (\alpha /2)$. For $\alpha \le 1$, Fatou's lemma gives $h_{\alpha }(\xi )\to \infty$ as $\xi \to 0$, so no change on a set of zero Lebesgue measure can give continuity.

For a $2\pi$ periodic distribution $u\in {\mathcal{S}}^{\prime }(\mathbb{R})$, choose a smooth periodic partition of unity $\Psi$, and define $c_n=(2\pi {)}^{-1}⟨u,\Psi e^{-inx}⟩$. A finite Schwartz seminorm bound on the fixed compact support gives $|c_n|\le C(1+|n|{)}^N$. For a Schwartz function $\varphi$, the Poisson summation formula and Fourier inversion imply $P\hat{\varphi }(x)={\sum }_n\varphi (-n)e^{inx}$. Localizing the pairing for $u$ givesproving the displayed Fourier transform formula. The series converges in tempered distributions because its coefficients have polynomial growth while $\varphi (n)$ decreases rapidly. Applying the inverse Fourier transform gives $u={\sum }_n{c}_n{e}^{inx}$ in tempered distributions. These are the smooth-window Fourier coefficients of a periodic distribution, and the transform formula shows independence of $\Psi$.

For the angular-frequency convention $\hat{f}(\xi )=\int e^{-ix\cdot \xi }f(x)dx$, the inverse-distance tempered distribution on ${\mathbb{R}}^3$ has the exact transformBoth sides are locally integrable in three dimensions and have at most polynomial growth, so they define distributions even though the original function is not integrable at infinity.

One direct proof introduces $f_{ϵ}(x)=e^{-ϵ|x|}/|x|$, with $ϵ>0$. Angular integration of the radial Fourier transform gives, for $s=|\xi |>0$,The damped sine integral follows by taking the imaginary part of $(ϵ-is{)}^{-1}$. Letting $ϵ\downarrow 0$ against a Schwartz function is justified by the locally integrable majorants $|x{|}^{-1}$ and $|\xi {|}^{-2}$ together with rapid decay. This proves the distributional identity with its constant.

The result agrees with $-\Delta (1/|x|)=4\pi {\delta }_0$: the Fourier transform of a derivative multiplies the transformed potential by $|\xi {|}^2$. Changing to a cycles-per-unit convention changes the numerical coefficient, so the transform convention is part of the statement.

Take the Gaussian function $f_a(x)=e^{-\pi a{x}^2}$ with $a>0$. Every derivative is a polynomial times the same Gaussian function, so $f_a$ is a Schwartz function. The Fourier transform of a Gaussian gives, in the cycles convention,Thus the Poisson summation formula gives the exact identityThe Gaussian Fourier transform is given in scaled form in Exercise 41 of Terence Tao's 245C, Notes 2: The Fourier transform.^[ref] The identity is also the imaginary-axis transformation of the Jacobi theta function. At $a=1$ the Gaussian function is its own Fourier transform; at reciprocal values of $a$ its spatial and spectral widths exchange.

For a short derivation of the Gaussian Fourier transform, write $H(\xi )={\hat{f}}_a(\xi )$. Differentiation under the integral and integration by parts, using $f_a^{\prime }(x)=-2\pi ax{f}_a(x)$, giveThe Gaussian integral gives $H(0)=a^{-1/2}$, so this ordinary differential equation yields the claimed formula. The decay of the Gaussian function justifies these steps and removes the boundary term.^[ref]

For $a=0.04$, the physical Gaussian function has width parameter $1/\sqrt{a}=5$, while its Fourier transform has width parameter $\sqrt{a}=0.2$. Many physical samples matter, but the dual zero sample already givesFor $a=25$, use the physical side instead:These are bounds on the full positive tails, obtained below, rather than estimates based only on the first omitted term. Ordinary floating-point arithmetic rounds these sums to $5$ and $1$; the exact real numbers are slightly larger. A Gaussian sum truncation bound lets us choose the cheaper side and certify its truncation error before evaluating it.

The Heisenberg uncertainty relation constrains the standard deviations of position and momentum in one preparation:On the real line, the position operator $x$ acts by multiplication, and the momentum operator is $p=-i\hslash d/dx$, with $\hslash$ the reduced Planck constant. Their canonical commutation relation $[x,p]=i\hslash I$ and the Robertson uncertainty principle give the bound on normalized quantum states satisfying the moment and product-domain assumptions above; Schwartz functions are a convenient sufficient class. Thus making a preparation more sharply localized in position forces a wider momentum distribution, even for ideal measurement in quantum mechanics.

An explicit saturating example is the normalized Gaussian wave packet, for $\sigma >0$,The wavefunction is a probability amplitude; its squared magnitude is the probability density function, by the Born rule. That density is a normal distribution of mean zero and standard deviation $\Delta x=\sigma$, rather than a density whose width is read directly from the amplitude's exponent. With the unitary, negative-exponent convention for the Fourier transform, its momentum amplitude isThe Gaussian integral evaluates this Fourier transform. Consequently $|{\tilde{\psi }}_{\sigma }(p){|}^2$ is another normal distribution, with mean zero and standard deviationFor example, in dimensionless units with $\hslash =1$, $\sigma =1/2,1,2$ give $(\Delta x,\Delta p)=(1/2,1),(1,1/2),(2,1/4)$. The Gaussian wave packets all saturate the same bound: narrowing one density broadens the other. More precisely, choose a reference length $L$ and set $X=x/L$, $P=pL/\hslash$; then $[X,P]=iI$, and the plotted widths are $\sigma /L$ and $L\Delta p/\hslash$.

The Heisenberg uncertainty relation connects localization to the cost in kinetic energy. In a quantum harmonic oscillator with mass $m>0$ and angular frequency $\omega >0$, the variance contribution to the expectation value of the energy is bounded using the arithmetic-geometric mean inequality:The mean-dependent contributions are nonnegative. The Gaussian eigenstate for a quadratic potential with ${\sigma }^2=\hslash /(2m\omega )$ and zero means reaches this ground state energy. This is why the tradeoff is more than a limitation on preparation: it predicts an unavoidable positive minimum energy.

Heisenberg's original article appeared in March 1927 as "Über den anschaulichen Inhalt der quantentheoretischen Kinematik und Mechanik", Zeitschrift für Physik 43, pp. 172–198. Its abstract discusses simultaneous imprecision of canonically conjugate quantities and explanatory thought experiments. The precise standard deviation formulation here distinguishes preparation spreads from a slogan about measurement disturbance. ^[ref].

With the sign conventionthe Hilbert transform is a singular integral operator. For a Schwartz function, the Cauchy principal value removes equal small intervals on both sides of $y=x$. Subtracting $f(x)$ locally makes the remaining quotient bounded there, while rapid decay controls infinity. Reversing the denominator reverses the operator sign.

For the angular-frequency Fourier transform, its Fourier multiplier is $-i\mathrm{sgn}\xi$, as derived under Hilbert-transform Fourier multiplier. Plancherel theorem therefore extends it to an isometry on $L^2(\mathbb{R})$, and $H^2=-I$ on that space because the squared multiplier is minus one almost everywhere. This frequency-dependent phase shift connects real and imaginary boundary parts of suitable holomorphic functions; Hilbert transform of the sinc function supplies an explicit example.

The transform need not preserve boundedness. For the indicator function of $[-1,1]$, direct integration givesIt diverges logarithmically near the endpoints and has tail $2/(\pi x)+O(x^{-3})$. This example also explains why a compactly supported input can produce a long-range response. The finite Hilbert transform instead restricts the integration interval and has its own endpoint and null-space issues.

With the Fourier transform convention $\hat{f}(k)={\int }_{\mathbb{R}}f(x)e^{-ikx}dx$, the inverse Fourier transform is the displayed linear operator. The integral is defined directly for an integrable function $g$. The Fourier inversion theorem ensures ${\mathcal{F}}^{-1}\hat{f}=f$ under its hypotheses, in particular for Schwartz functions, and for integrable $f$ and $\hat{f}$ at continuity points of $f$. The operation also extends to tempered distributions; the normalization and sign depend on the chosen Fourier transform convention.

If a finite measure on ${\mathbb{R}}^n$ has square-integrable Fourier transform, the Plancherel theorem gives an $L^2$ inverse transform $g$. Pairing with every Schwartz function and using Fourier inversion identifies the measure with $gdx$. This establishes absolute continuity, rather than presuming it. For a positive measure, $g\ge 0$ and $g\in L^1$ as well. Multiplication by $f\in L^{\infty }(\mu )$ yields $\Vert \widehat{fd\mu }{\Vert }_2\le \Vert \hat{\mu }{\Vert }_2\Vert f{\Vert }_{L^{\infty }(\mu )}$.

For a Schwartz function $f$, the Hilbert transform satisfies $\mathcal{H}f(x)=(\int f)/(\pi x)+o(|x{|}^{-1})$ at both infinities. Splitting the inverse Fourier transform at the sign multiplier's jump and integrating each half-line by parts extracts the zero-frequency value $\hat{f}(0)$; the remainder vanishes by the Riemann-Lebesgue lemma. A nonzero zeroth moment therefore obstructs Schwartz decay.

Fourier localization controls the variation of a demodulated envelope at reciprocal scales. Suppose the Fourier support lies in a rectangular box centered at ${\xi }_0$, with widths of order $N$ in spatial-frequency coordinates and $N^2$ in a time-frequency coordinate. Set $g(x,t)=e^{-i{\xi }_0\cdot (x,t)}f(x,t)$. Choose a Schwartz function whose transform is one on the centered box and rescale it to obtain a kernel $K_N$. Thenfor $d$ spatial coordinates. The convolution averages on spatial scale $N^{-1}$ and temporal scale $N^{-2}$. For bounded $g$, differentiating the kernel gives $\Vert {\partial }_{x_j}g{\Vert }_{\infty }\le CN\Vert g{\Vert }_{\infty }$ and $\Vert {\partial }_{t}g{\Vert }_{\infty }\le C{N}^2\Vert g{\Vert }_{\infty }$.

These are quantitative variation bounds, not literal constancy on boxes. Removing the center frequency is essential: multiplying a slowly varying envelope by $e^{iMx}$ can make its phase rotate rapidly while leaving its Fourier-support width unchanged. The principle controls a localized envelope or a smoothed magnitude, rather than forbidding carrier oscillations.