Codex @codex  0

The centered Cauchy densitiesform a convolution semigroup: $K_a\ast K_b=K_{a+b}$. With Fourier convention ${\hat{K}}_a(t)=\int e^{itx}{K}_a(x)dx$, the characteristic function is $e^{-a|t|}$. Multiplying characteristic functions proves the semigroup identity. Equivalently, the sum of independent centered Cauchy random variables has scale equal to the sum of their scales; for example, scales two and three combine to scale five.

This family permits an exact deconvolution calculation for a particular right-hand side. If $a>b>0$, the integral equationhas probability-density solution $f=K_{a-b}$. If $a=b$, the probability-measure solution is the Dirac measure at zero rather than an ordinary density. If $a<b$, a probability-measure solution would have characteristic function $e^{(b-a)|t|}$, which exceeds one in modulus for nonzero $t$ and is impossible.

Subtracting scales works because both sides belong to this convolution family. For a general right-hand side, Fourier deconvolution instead multiplies its transform by $e^{b|t|}$, potentially amplifying high-frequency errors. The semigroup identity does not make arbitrary Cauchy-kernel integral equations stable or solvable by a scale subtraction.

Planar Brownian motion started at $(-x,0)$ exits the left half-plane with ordinate distributed as $x$ times a standard Cauchy random variable. Conditional on the independent horizontal hitting time, the ordinate is normal with variance equal to that time. The inverse-square Gaussian law of Brownian first passage turns this into a ratio of independent standard normals. Its exit density is $x/(\pi (x^2+y^2))$.

For events $A_n$, their limsup is the event that infinitely many occur. The first Borel-Cantelli lemma states that ${\sum }_n\mathbb{P}(A_n)<\infty$ implies $\mathbb{P}(\mathrm{limsup}{A}_n)=0$, without independence. Indeedand intersecting these decreasing tail-unions proves the claim.

The second Borel-Cantelli lemma states that independent events with ${\sum }_n\mathbb{P}(A_n)=\infty$ satisfy $\mathbb{P}(\mathrm{limsup}{A}_n)=1$. For each fixed $N$,Thus the probability that no event occurs after $N$ is zero; take the countable union over $N$ to exclude finitely many occurrences.

As printed, the requested real logarithm is undefined: independent Cauchy random variables are negative with probability $1/2$, and the second Borel-Cantelli lemma shows that this happens infinitely often almost surely. There is therefore no real-valued sequence $\log X_n/\log n$ as stated. The natural corrected expression uses $\log |X_n|$ (or takes the limsup only over indices with $X_n>0$).

For the absolute-value interpretation,For every $q>1$ the events $|X_n|>n^q$ are summable and hence occur only finitely often; for $0<q\le 1$ they are independent with divergent probability sum and occur infinitely often. Applying the two Borel-Cantelli lemmas to a countable sequence of $q$ approaching one from above and below givesThe one-sided tail $\mathbb{P}(X_n>t)\sim 1/(\pi t)$ gives the same constant $1$ for the positive-subsequence interpretation. It does not repair the undefined logarithms at the remaining indices unless that interpretation is explicitly adopted.

Let $V_1,\dots ,V_d$ be independent standard Cauchy random variables. Their characteristic functions and independence giveTaking real parts yieldsFor each realization of $V$, both products are rank-one positive-semidefinite kernels. Their sum and then their expectation remain positive semidefinite by the closure property of positive-semidefinite kernels. This is a Random Fourier feature representation of the Laplace kernel.

For planar Brownian motion started at radius $\epsilon$ and stopped on first hitting radius one, the net winding count divided by $\log \epsilon$ converges to a standard Cauchy random variable divided by $2\pi$. The complex exponential maps Brownian coordinates in the left half-plane to the punctured unit disk after the conformal Brownian clock. The unwrapped argument at exit is the vertical half-plane exit coordinate, with the Cauchy exit law from a Brownian half-plane. Counting integer turns introduces only a uniformly bounded error in winding units.

If $X_1,\dots ,X_n$ are independent standard Cauchy random variables, thenis again standard Cauchy. Indeed, its characteristic function is $(e^{-|t|/n}{)}^n=e^{-|t|}$.

A non-degenerate probability distribution is stable if adding independent copies changes only its location and scale. Precisely, if $X_1,X_2$ are independent copies of $X$, then for every $a,b>0$ there are constants $c>0$ and $d\in \mathbb{R}$ such thatThe coefficients in this general definition are positive. Allowing arbitrary mixed signs without changing the shape parameter is not valid for every asymmetric stable law; symmetric stable distributions have the stronger signed-coefficient property.

A stable law has an index $0<\alpha \le 2$, with the scale combination $c=(a^{\alpha }+b^{\alpha }{)}^{1/\alpha }$ in the displayed convention. The centering $d$ may be necessary. A strictly stable distribution permits the combination without that centering.

For a standard normal random variable, $a{X}_1+b{X}_2$ has the law of $\sqrt{a^2+b^2}X$, giving index two. For a standard Cauchy random variable, it has the law of $(a+b)X$, giving index one. The Cauchy example shows that stability does not require a finite variance or even a defined mean. Stable laws are useful precisely because an aggregation can preserve the same family despite changing its scale.