arXiv Science⌕ Search

arXiv · 0706.3037

Implicit function density computation

Abstract

If two random variables X and A are functionally related via f(X)=A for some strictly monotone continuously differentiable function f:R->R, the distribution of X may easily be computed from the distribution of A.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Kerry Michael Soileau. 2022-08-14. Implicit function density computation. https://arxiv.org/abs/0706.3037

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

From Vieta to Chebyshev: A Unified Trigonometric Census of Polynomial Roots

This article gives a trigonometric method for counting how many roots of a polynomial are real and how many are complex, without solving the equation. For a depressed polynomial $P(t) = t^d + c_{d-2}t^{d-2} + \cdots + c_1 t + c_0$ of degree~$d$ with $c_{d-2} < 0$ (the case $c_{d-2}\geq 0$ is discussed briefly in Section~\ref{sec:cgeq0} but not resolved), the substitution $t = u\cosθ$, $u = 2\sqrt{-c_{d-2}/d}$, matches $P$ against the Chebyshev identity $T_d(\cosθ) = \cos dθ$ and reduces the classification problem to a single trigonometric equation $f(θ) = \sum_{k=1}^{d-3}α_k\cos^k\!θ+ \cos dθ+ δ= 0$ on $[0,π]$. A bijection between the zeros of~$f$ and the roots of~$P$ in $[-u,u]$ gives an exact interior root count. For roots outside $[-u,u]$, a universal lemma based on the positivity of~$P^{(d-2)}$ gives an exact parity and an upper bound of $d-2$ roots per half-line; combined with the interior count, this pins down the total number of real roots exactly whenever at most one root lies on each exterior half-line, a condition we prove for the cubic, verify computationally throughout our examples, but show by an explicit quartic counterexample to fail in general for $d\geq 4$ (Remark~\ref{rem:counterexample}). We also compare the method's cost to Sturm's theorem (Section~\ref{sec:complexity}): it is not asymptotically cheaper, but avoids the coefficient blow-up of the Euclidean algorithm and applies via one uniform template to every degree $d\geq 4$, with the classical Vieta cubic ($d=3$) as the degenerate base case. The derivation is conceptually simple, unifying and extending the quartic and quintic analyses developed individually in earlier work.

math.GM↗

The upper semi-Fredholm radius, extensions of restrictions, and subprojectivity

For a bounded operator $T$ between Banach spaces, let $d_+(T)$ be its distance to the operators that are not upper semi-Fredholm, and let $\inn(T)$ be Schechter's quantity $\inf\|T|_M\|$ over infinite-dimensional subspaces $M$. One always has $\inn(T)\le d_+(T)$, and González and Martinón showed that the two quantities are not equivalent in general. We prove that $d_+(T)$ is the infimum of the norms of operators on the whole space that extend a restriction of $T$ to some infinite-dimensional subspace. The gap between $\inn$ and $d_+$ thus becomes an extension problem, and projection constants enter naturally. If the domain is uniformly subprojective with constant $λ$, then $d_+\leλ\,\inn$. In the other direction we prove a quantitative obstruction: a complemented Hilbertian subspace all of whose infinite-dimensional subspaces have projection constant at least $Λ$ produces operators with $d_+/\inn\geΛ/2$. Consequently, on the subprojective but not uniformly subprojective space $(\bigoplus_nL_{p_n})_{\ell_2}$ with $p_n\to\infty$, there is no uniform comparison between $\inn$ and $d_+$. We conjecture that uniform comparison is equivalent to uniform subprojectivity. Finally, we compute $\inn$ for inverses and for shears in terms of the quantity $τ(T)=\sup_M j(T|_M)$, and we use the arbitrary distortability of Hilbert space to show that a rectangular diagonal operator $\operatorname{diag}(A,B)$ built from two isomorphisms with $\inn(A)=\inn(B)=1$ can have arbitrarily small $\inn$. In contrast, every operator $T$ on a complex Banach space satisfies $\inn(T)\leτ(T)$, so every automorphism satisfies $\inn(T)\,\inn(T^{-1})\le1$ and the same mechanism cannot work for endomorphisms.

math.GM↗