MacRobert E function

In mathematics, the E-function was introduced by Thomas Murray MacRobert (1937–1938) to extend the generalized hypergeometric series _pF_q(·) to the case p > q + 1. The underlying objective was to define a very general function that includes as particular cases the majority of the special functions known until then. However, this function had no great impact on the literature as it can always be expressed in terms of the Meijer G-function, while the opposite is not true, so that the G-function is of a still more general nature. It is defined as:

$E(p;\alpha _{r};\rho _{s};z)\equiv {\frac {\Gamma (\alpha _{q+1})}{\prod _{k=1}^{q}\Gamma (\rho _{k}-\alpha _{k})}}\prod _{\mu =1}^{q}\int _{0}^{\infty }\lambda _{\mu }^{\rho _{\mu }-\alpha _{\mu }-a}(\lambda _{\mu }+1)^{-\rho _{\mu }}d\lambda _{\mu }\prod _{\nu =2}^{p-q-1}\int _{0}^{\infty }\lambda _{q+\nu }^{\alpha _{q+\nu }-1}\exp(-\lambda _{q+\nu })d\lambda _{q+\nu }\int _{0}^{\infty }\lambda _{p}^{\alpha _{p}-1}\exp(-\lambda _{p})\left[{\frac {\prod _{k=q+2}^{p}\lambda _{k}}{z\prod _{k=1}^{q}\lambda _{k}+1}}+1\right]d\lambda _{p}$

Definition

There are several ways to define the MacRobert E-function; the following definition is in terms of the generalized hypergeometric function:

when p ≤ q and x ≠ 0, or p = q + 1 and |x| > 1:

E\!\left(\left.{\begin{matrix}\mathbf {a_{p}} \\\mathbf {b_{q}} \end{matrix}}\;\right|\,x\right)={\frac {\prod _{j=1}^{p}\Gamma (a_{j})}{\prod _{j=1}^{q}\Gamma (b_{j})}}\;_{p}F_{q}\!\left(\left.{\begin{matrix}\mathbf {a_{p}} \\\mathbf {b_{q}} \end{matrix}}\;\right|\,-x^{-1}\right)

when p ≥ q + 2, or p = q + 1 and |x| < 1:

E\!\left(\left.{\begin{matrix}\mathbf {a_{p}} \\\mathbf {b_{q}} \end{matrix}}\;\right|\,x\right)=\sum _{h=1}^{p}{\frac {\prod _{j=1}^{p}\Gamma (a_{j}-a_{h})^{*}}{\prod _{j=1}^{q}\Gamma (b_{j}-a_{h})}}\Gamma (a_{h})\;x^{a_{h}}\;_{q+1}F_{p-1}\!\left(\left.{\begin{matrix}a_{h},1+a_{h}-b_{1},\dots ,1+a_{h}-b_{q}\\1+a_{h}-a_{1},\dots ,*,\dots ,1+a_{h}-a_{p}\end{matrix}}\;\right|\,(-1)^{p-q}\;x\right).

The asterisks here remind us to ignore the contribution with index j = h as follows: In the product this amounts to replacing Γ(0) with 1, and in the argument of the hypergeometric function this amounts to shortening the vector length from p to p − 1. Evidently, this definition covers all values of p and q.

Relationship with the Meijer G-function

The MacRobert E-function can always be expressed in terms of the Meijer G-function:

E\!\left(\left.{\begin{matrix}\mathbf {a_{p}} \\\mathbf {b_{q}} \end{matrix}}\;\right|\,x\right)=G_{q+1,\,p}^{\,p,\,1}\!\left(\left.{\begin{matrix}1,\mathbf {b_{q}} \\\mathbf {a_{p}} \end{matrix}}\;\right|\,x\right)

where the parameter values are unrestricted, i.e. this relation holds without exception.

gollark: https://www.sbert.net/examples/applications/semantic-search/README.html is kind of like what you want.

gollark: Instead of recomputing the embeddings every time a new sentence comes in.

gollark: The embeddings for your example sentences are the same each time you run the model, so you can just store them somewhere and run the cosine similarity thing on all of them in bulk.

gollark: Well, it doesn't look like you ever actually move the `roberta-large-mnli` model to your GPU, but I think the Sentence Transformers one is slow because you're using it wrong.

gollark: For the sentence_transformers one, are you precomputing the embeddings for the example sentences *then* just cosine-similaritying them against the new sentence? Because if not that's probably a very large bottleneck.

References

Andrews, L. C. (1985). Special Functions for Engineers and Applied Mathematicians. New York: MacMillan. ISBN 0-02-948650-5.CS1 maint: ref=harv (link)
Erdélyi, A.; Magnus, W.; Oberhettinger, F. & Tricomi, F. G. (1953). Higher Transcendental Functions (PDF). Vol. 1. New York: McGraw–Hill.CS1 maint: ref=harv (link) (see § 5.2, "Definition of the E-Function", p. 203)
Gradshteyn, Izrail Solomonovich; Ryzhik, Iosif Moiseevich; Geronimus, Yuri Veniaminovich; Tseytlin, Michail Yulyevich; Jeffrey, Alan (2015) [October 2014]. "9.4.". In Zwillinger, Daniel; Moll, Victor Hugo (eds.). Table of Integrals, Series, and Products. Translated by Scripta Technica, Inc. (8 ed.). Academic Press, Inc. ISBN 0-12-384933-0. LCCN 2014010276. ISBN 978-0-12-384933-5.CS1 maint: ref=harv (link)
MacRobert, T. M. (1937–38). "Induction proofs of the relations between certain asymptotic expansions and corresponding generalised hypergeometric series". Proc. Roy. Soc. Edinburgh. 58: 1–13. JFM 64.0337.01.CS1 maint: ref=harv (link)
MacRobert, T. M. (1962). "Barnes integrals as a sum of E-functions". Mathematische Annalen. 147 (3): 240–243. doi:10.1007/bf01470741. Zbl 0100.28601.CS1 maint: ref=harv (link)

External links

Weisstein, Eric W. "MacRobert's E-Function". MathWorld.

This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.