Fast dot product over finite field

نویسندگان

  • Jérémy JEAN
  • Jean-Louis Roch
چکیده

Finite fields have great applications in various areas as cryptography, that is why it is important to have fast ways of computation to manipulate them. A first approach developed in this report lies in representing integers of the field using floating-point numbers, which lead to efficient computations. Operations in our case are done by restricting the characteristic p of the field to a floating-point mantissa: p−1 < 2M−1. Taking advantage of error-free transformations on modern architectures, one can manage quite large finite fields exactly with floating-point arithmetic. After returning back to the basic of floating-point numbers, we introduce slightly different approaches to compute the dot product in an efficient way. In a second part, we have the same calculations done in a Residue Number System (RNS) over both integer and floating-point numbers. We show how this system can be efficient for well-chosen basis and present experimental results. Finally, we discuss how we parallelized our algorithms on a GPU card. Résumé Les corps finis ont des applications particulièrement intéressantes dans beaucoup de domaines comme la cryptographie, et il est important d’avoir des modes de calculs rapides pour les manipuler. La première approche choisie consiste à représenter les grands entiers du corps dans des nombres flottants sur lesquels les calculs sont efficaces. Les calculs sont conduits en limitant la caractéristique p du corps à une mantisse flottante: p− 1 < 2M−1. En utilisant des algorithmes de transformations exactes sur des architectures récentes, on voit qu’il est possible de gérer des corps finis relativement grands de manière exacte. Après un rappel sur l’arithmétique flottante, nous présenterons dans ce rapport deux méthodes légèrement différentes pour le calcul du produit scalaire de deux vecteurs chacune visant à réduire le temps de calcul. Dans un deuxième temps, nous montrerons comment les même calculs peuvent être fait en arithétique flottante et entière en utilsant un système modulaire de représentation des nombres (RNS). Nous montrons que ce système peut s’avérer très efficace pour peu que la base soit choisie correctement. Finalement, nous exposons comment nous avons parallélisé nos algorithmes sur GPU. Sammanfattning Ändliga kroppar har intressanta tillämpningar i flera omr̊aden, s̊asom kryptografi, och det är viktigt att snabbt kunna utföra beräkningar för att manipulera dem. Tillvägag̊angssättet som utvecklas i denna rapport g̊ar ut p̊a att representera element i kroppen genom att använda flyttal, vilket leder till effektiva beräkningar. Beräkningarna är utförda genom att begränsa karaktäristiken p av kroppen till en flyttalsmantissa: p− 1 < 2M−1. Genom att använda exakta transformationer p̊a modern arkitektur, kan man hantera stora ändliga kroppar exakt med flyttalsaritmetik. Efter en genomg̊ang av flyttal, kommer vi att introducera ett n̊agot annorlunda tillvägag̊angssätt för att beräkna skalärprodukten p̊a ett effektivt sätt. I en andra del, har vi samma beräkningar som gjorts i en Residue Number System (RNS) över b̊ade heltal och flyttal. Vi visar hur detta system kan vara effektivt för väl vald bas och presenterar experimentella resultat. Slutligen diskuterar vi hur vi parallelliserade v̊ara algoritmer p̊a en grafikprocessor (GPU). Acknowledgments Thanks to Stef Graillat for his numerous re-readings and Johan H̊astad and Torbjörn Granlund for their advises.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Fast Dot Product over Finite Field

Finite fields have great applications in various areas as cryptography, that is why it is important to have fast ways of computation to manipulate them. A first approach developed in this report lies in representing integers of the field using floating-point numbers, which lead to efficient computations. Operations in our case are done by restricting the characteristic p of the field to a float...

متن کامل

Dot Product Representations of Graphs

We introduce the concept of dot product representations of graphs, giving some motivations as well as surveying the previously known results. We extend these representations to more general fields, looking at the complex numbers, rational numbers, and finite fields. Finally, we study the behavior of dot product representations in field extensions.

متن کامل

Efficient dot product over word-size finite fields

We want to achieve efficiency for the exact computation of the dot product of two vectors over word size finite fields. We therefore compare the practical behaviors of a wide range of implementation techniques using different representations. The techniques used include floating point representations, discrete logarithms, tabulations, Montgomery reduction, delayed modulus.

متن کامل

Pinned distance sets, k-simplices, Wolff’s exponent in finite fields and sum-product estimates

An analog of the Falconer distance problem in vector spaces over finite fields asks for the threshold α > 0 such that |∆(E)| & q whenever |E| & q, where E ⊂ Fq , the d-dimensional vector space over a finite field with q elements (not necessarily prime). Here ∆(E) = {(x1 − y1) 2 + · · · + (xd − yd) 2 : x, y ∈ E}. The fourth listed author and Misha Rudnev ([20]) established the threshold d+1 2 , ...

متن کامل

Classical Wavelet Transforms over Finite Fields

This article introduces a systematic study for computational aspects of classical wavelet transforms over finite fields using tools from computational harmonic analysis and also theoretical linear algebra. We present a concrete formulation for the Frobenius norm of the classical wavelet transforms over finite fields. It is shown that each vector defined over a finite field can be represented as...

متن کامل

Classical wavelet systems over finite fields

This article presents an analytic approach to study admissibility conditions related to classical full wavelet systems over finite fields using tools from computational harmonic analysis and theoretical linear algebra. It is shown that for a large class of non-zero window signals (wavelets), the generated classical full wavelet systems constitute a frame whose canonical dual are classical full ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2010