FACTORIZATION OF LARGE NUMBERS USING NUMBER FIELD SIEVE: SIEVING STEP BY SUSHMA GUDAVALLI TEJA VALURUPALLI FACTORIZATION Factoring an integer ‘n’ means finding integers x and y such that n=x.y Prime factorization of a number, written as a product of prime numbers is unique. Factoring a large number is believed to be a very hard problem. FACTORIZATION IN CRYPTOGRAPHY The security of RSA relies on the difficulty of factorization. It relies on the fact that it is computationally difficult to factor a large integer. Factorization Product Message Encryption Encrypted Message Two Prime Factors Decryption Message FACTORING METHODS General algorithms( usable for RSA) : Continued Fractions (CFRAC) Quadratic Sieve (efficient up to 110 digits) Number field Sieve (efficient beyond 110 digits) NUMBER FIELD SIEVE The NFS algorithm consists of five steps which are mutually dependent: Polynomial step Finding Factor Bases Sieving Linear Algebra (Matrix step) Square root NFS ALGORITHM Polynomial Step: Chooses a good polynomial f(x) from a large set of usable polynomials. Factor Bases: Chooses factor bases for the sieving step. RFB: ( p0, p0mod m), (p1, p1mod m),…… AFB: (p0, r0), (p1, r1),… Such that f(ri) = 0 mod pi NFS ALGORITHM (contd…) Sieving: The purpose of the sieving step is to find usable relations (ai, bi) Linear algebra: This step finds combination of elements from the relation set which has a product that is a square. Square root: The purpose of this step is to find rational square root and algebraic square root for the solutions obtained from the matrix step. SIEVING The task of Sieving step is to find usable relations (ai, bi) from many possible pairs. These relations should have the following three properties: gcd(a,b)=1 a+bm is smooth over RFB bdeg(f) f(a/b) is smooth over AFB ( f(x) is the polynomial, m f(m) = 0 mod n) SIEVING (contd..) What is meant by ‘being smooth over factor base?’ Example: RFB : { (2, 1), (3, 0), (11, 7),…} Let a+bm = 264; 264 = 23. 3. 11 Thus (a,b) are smooth over RFB Similar is the case with AFB SIEVING (contd..) An initial pair (a, b) which is ‘likely’ to be a relation is found on the sieving line. All possible relation pairs on the sieving line are found out by adding log2pi ( pi є RFB) to the location of the initial pair (a,b) For each ‘b’ fixed a Pairs of elements likely to be relations Sieving Line log2p log2p SIEVING (contd..) The sorted pairs from the Sieving line are checked with the above mentioned properties. Thus the Relations are obtained. Briefly, the Sieving step involves: Input: RFB, AFB, f(x), m, sieving interval (u) Output: List of Relations:{ (a0, b0),…(at, bt)} MAGMA Radically new system to solve computationally hard problems in algebra, number theory, geometry. It is both computer algebra system and a programming language. SPECIALITY: Provision of mathematical data types such as groups, rings, fields, sets, sequences, mappings etc. Large collection of functions for performing standard tasks in algebra. QUESTIONS ???