Use of Sparse and/or Complex Exponents in Batch Verification of Exponentiations
|
|
- Harriet Benson
- 6 years ago
- Views:
Transcription
1 Use of Sparse and/or Complex Exponents n Batch Verfcaton of Exponentatons Jung Hee Cheon 1 and Dong Hoon Lee 2 1 Department of Mathematcs, Seoul Natonal Unversty jhcheon@math.snu.ac.kr, 2 Natonal Securty Research Insttute (NSRI) dlee@etr.re.kr Abstract. Modular exponentaton n an abelan group s one of the most frequently used mathematcal prmtves n modern cryptography. Batch verfcaton s to verfy many exponentatons smultaneously. We propose two fast batch verfcaton algorthms. The frst one makes use of exponents wth small weght, called sparse exponents, whch s asymptotcally 10 tmes faster than the ndvdual verfcaton and twce faster than the prevous works wthout securty loss. The second one s appled only to ellptc curves defned over small fnte felds. Usng sparse Frobenus expanson wth small nteger coeffcents, we propose a complex exponent test whch s four tmes faster than the prevous works. For example, each exponentaton n one batch requres asymptotcally 9 ellptc curve addtons n some ellptc curves for 2 80 securty. Keywords: Batch verfcaton, modular exponentaton, sparse exponent, Frobenus map. 1 Introducton Batch verfcaton s an algorthm to verfy many exponentatons smultaneously: Let G be an abelan group wth a generator g. Gven a batch nstance of n exponentaton pars {(x 1, y 1 ), (x 2, y 2 ),..., (x n, y n )} wth x Z and y G, the algorthm checks at once f g x = y for all. The small exponent test s to check f g n =1 x s = n =1 ys for randomly chosen l bt ntegers s when the screenng parameter s l [2]. We extend the small exponent test to more general exponent sets and mprove the performance usng sparse exponents. Our dea s to take random exponents whch have smaller Hammng weghts and longer bt szes. The probablty that a wrong batch nstance passes the proposed test wth an exponent set S s at most 1/ S, as we wll show, when all elements n S are dstnct modulo the order of a gven group. Therefore t s mportant how to take a exponent set S and how to
2 maxmze the cardnalty of elements that are dstnct modulo the order of the group. If we take S to be a set of non-negatve ntegers n Z p wth Hammng weght k 19 from Z p for an 160 bt prme p, S s greater than In ths case, our sparse exponent test requres at most n multplcatons to verfy n exponents wth securty 2 80, whle the small exponent test requres n multplcatons at average. Another varant of the algorthm s to use sparse sgned bnary exponents. It s proper for verfyng scalar multplcatons on ellptc curves where a subtracton s as effcent as an addton. More savngs are obtaned by usng even larger exponents than the order of the underlyng group. It s applcable to sgnature schemes the order of whose underlyng group s secret such as RSA and Gullou-Qusquater (GQ) sgnatures [9]. We can generalze the exponents set to complex endomorphsms n ellptc curves. When an ellptc curve s defned over a small fnte feld F q, we have a very effcent complex endomorphsm, called the Frobenus map. We take as an exponent set S a polynomal of the Frobenus map wth small nteger coeffcents. In most prevous works on Frobenus expansons, the set of coeffcents have at most q cardnalty due to the unqueness of representaton (For more references, refer to [14]). If we consder ths coeffcent set, we have the same effcency wth the sparse exponent set. In our method, however, we enlarge the coeffcent set to all ntegers whose absolute values are less than q 2 /2 and coprme to q and prove that each element of S s dstnct as an endomorphsm on the cyclc subgroup of the ellptc curve f each element of S has no adjacent non-zero coeffcents. As a result, we obtan the batch verfcaton four tmes faster than the prevous works. For example, each exponentaton requres asymptotcally 9 ellptc curve addtons n Kobltz curves for 2 80 securty. Batch Cryptography Batch cryptography was frst ntroduced by Fat [8]. He ntroduced an algorthm to obtan two exponentatons by one full exponentaton and several small exponentatons. More technques on batch computatons can be found n [3, 12]. Batch verfcaton of exponentatons was frst proposed by Naccache et al. [15] to verfy effcently modfed DSA sgnatures. They used the subset consstng of e-bts prme numbers for small e as an exponent set. Yen and Lah [18] mproved the test by adoptng small exponents set {0, 1} l for the screenng parameter l. Bellare et al. [2] gave systematc approaches wth two more tests: the atomc test uses S = {0, 1} and the
3 bucket test s the combnaton of the atomc test and the small exponent test. Later, Boyd and Pavlovsk [4] ndcated the weakness of the tests when they are used n non-prme order subgroups. Current batch verfcatons are effcent for verfyng many sgnatures by one sgner at once. The applcatons nclude: 1)an authentcated database n whch each data s sgned by a data owner [13], 2) electronc commerce where shops or banks need to verfy all the cons, usually ssued by one bank, very effcently, and 3) votng protocols where the tally needs to authentcate huge number of votes and an effectve vote should be accompaned by the sgnature sgned by the votng center. Organzaton Ths paper s organzed as follows: In Secton 2, we ntroduce general approach for batch verfcatons and dscuss ts securty. In Secton 3, we propose a batch verfcaton algorthm usng sparse bnary exponents and ts varants usng sparse sgned bnary exponents and sparse long exponents. In Secton 4, we consder complex exponents on ellptc curves. We conclude n Secton 5. 2 Batch Verfcaton For the defnton of batch verfcaton, we wll follow the notaton n [2]. Let G be a cyclc group of order p wth a generator g and R( ) a Boolean relaton on the set of nstances {(x, y) x Z p, y G}. We say that R(x, y) = 1 or the nstance (x, y) s correct f and only f g x = y. A batch nstance for relaton R s a sequence nst 1,, nst n of nstances for R. We say that the batch nstance s correct (R(nst 1,..., nst n ) = 1) f R(nst ) = 1 for all = 1,..., n, and ncorrect (R(nst 1,..., nst n ) = 0) f there s some {1,..., n} for whch R(nst ) = 0. Defnton 1. A batch verfer for relaton R s a probablstc algorthm V that takes a batch nstance X = (nst 1,..., nst n ) for R and a screenng parameter l, and satsfes: 1. If X s correct then V outputs If X s not correct then the probablty that V outputs 1 s at most 2 l. In ths case, we say the error of V s 2 l. Now we consder a subset S of Z such that the dfference of any two elements of S s coprme to p. Then we defne a batch verfer V S as Fgure 1.
4 Input: A batch nstance {(x 1, y 1),..., (x n, y n)} wth x Z p and y G. Check: That g x = y for all 1 n. 1. Choose n elements s 1,..., s n randomly from S. 2. Compute g n =1 s x and n =1 ys. 3. Accept f they concde, else reject. Fg. 1. The Batch Verfer V S wth a Exponent Set S Note that any correct batch nstance always passes the test, but even a wrong nstance may pass the test wth some probablty. If we let α Z p be such that g α = y /g x, the test s passed f and only f g n =1 s α = 1. Thus the error of the batch verfer V S s Err(V S ) = max α {(s 1,..., s n) g s α = 1 for s S}, (1) {(s 1,..., s n) s 1,..., s n S} where α = (α 1,..., α n ) runs through all but (0, 0,, 0) n Z n p, f all elements of S are equally lkely to be chosen n the test. Now we ntroduce the formula for Err(V S ). Theorem 1. Let S be a subset of nonnegatve ntegers such that the dfference of any two elements n S s coprme to p. Then we have Err(V S ) 1/ S. Proof. Consder an nstance of n exponentaton pars (x 1, y 1 ),..., (x n, y n ) wth x Z p and y G where y = g x +α for some α Z p. Gven a test parameter (s 1,..., s n ) S n, the nstance passes the test f and only f g s 1α 1 + +s nα n = 1. If the nstance contans an ncorrect par, there must be at least one such that α 0 mod p. In that case, we have at most one s satsfyng s α n j s jα j mod p, because any two elements s, s satsfyng the equaton should have gcd(s s, p) 1 whch contradcts the assumpton of S. Hence, for any gven α, we have {(s 1,..., s n) s α 0 mod p for s S} S n 1. From Equaton (1), we have Err(V S ) 1/ S. In [2], Bellare et al. suggested three tests. In the atomc test, S s taken to be {0, 1}. In the small exponent test, S s taken to be {s Z 0 s < 2 l } for a gven screenng parameter l. The bucket test s the combnaton of the atomc test and the small exponent test. In ths paper, we consder dfferent types of a set conssts of sparse exponents.
5 2.1 How to Select Screenng Parameters The securty level of today s sgnature schemes s usually set on 2 80, whch s equvalent to that of 1024-bt RSA or 160-bt ellptc curve cryptosystem. But, the screenng level of batch verfer can be dfferent because even the owner of sgnng key can produce a set of wrong sgnatures that passes the verfers, wth at most 2 l probablty for the screenng parameter l. In [2] and [4], l = 60 was proposed for most practcal use. Usually, selecton of the screenng parameter depends on the stuaton where each applcaton deploys. If an applcaton performs only small number of verfcatons, t would be very unlkely to accept the wrong nstance n ts lfetme even wth the small screenng parameter. When we take the small screenng parameter, the batch verfcaton algorthm s more effcent. Hence t s desrable to set the screenng parameter as small as possble n tolerable ranges for each applcaton. We may consder an teraton of several weak batch verfcatons to acheve stronger batch verfcatons. If we perform t tmes the batch verfcaton wth screenng level l ndependently, then the screenng level becomes tl. Ths enables us a parallelzaton of batch verfcaton. Moreover we can desgn a cascade flter of sgnatures consstng of successve weak batch verfers. If the test fals n the frst flter, the set of sgnatures should be moved to the step of dentfyng wrong sgnatures. If the test succeeds, t goes to the second flter and so on untl the ntended screenng level s acheved. Ths would be useful for fast screenng DoS attacks. Note that when there s only one nvald sgnature n the set of n nstances and the batch verfer fals the verfcaton, we can dentfy the nvald one wth at most log 2 n + 1 batch verfcatons by bnary tree search technque. If the number of wrong sgnatures are k, the maxmum number of batch verfcatons becomes k( log 2 n log 2 k )+2 log 2 k Sparse Exponent Test 3.1 Basc Sparse Exponent Test Let l be the screenng parameter and let the order of G be a prme p. In sparse exponent test, we consder the set S of exponents S = {s Z 0 s < 2 log p, wt(s) k}, where wt(s) s the Hammng weght of s n the bnary representaton. Then every element s dstnct modulo p and so we have Err(V S ) 1/ S
6 by Theorem 1. To bound Err(V S ) by 2 l, k must satsfy S = k ( ) log p 2 l. Table 1 gves mnmum k values for varous screenng parameters l that satsfy the above equaton when log p = 160. Observe that k s much less than l/4. In partcular, k l/5 for l 50. Table 1. k values for varous screenng parameters when log p = 160 l k Our sparse exponent test s descrbed n Fgure 2. In order to perform the test effcently, we use the technque of smultaneous multplcatons as usual. Input: (x, y ) for 1 n. Check: That g x = y for all 1 n. 1. Choose n random elements s 1, s 2,..., s n from S defned n Equaton (3.1) 2. y 1 and y 0 g. 3. s 0 n =1 sx mod p. 4. For j = log p to 1, do (a) For = 0 to n, y yy f the j-th bt of s s one. (b) y y 2 unless j = Accept f y s one, else reject. Fg. 2. Sparse exponent test usng smultaneous multplcatons Table 2 gves the average numbers of operatons whch are requred for verfyng n nstances for the naïve test, the small exponent test (SE), and our sparse exponent test (SPET). Naïve test verfes n nstances ndependently. In the table, Exp, Mul, and Sqr represent the exponentaton, the multplcaton, and the squarng on G respectvely. P denotes a prme feld (Sqr=0.8 Mul) and B-PB and B-NB denote a bnary feld wth a polynomal bass (Sqr=0.1 Mul) and a bnary feld wth a normal bass (Sqr s for free), respectvely. Multplcaton modulo p operatons are gnored n total estmatons.
7 Table 2. Average number of operatons for verfyng n-nstances when log p = 160. Naïve SE SPET(ours) Exp n 1 0 Mul 0 ln/2 kn + log p /2 Sqr 0 l log p Total P 208n 10n n (n Mul) B-PB 96n 10n n + 96 l = 20 B-NB 80n 10n n + 80 Total P 208n 40n n (n Mul) B-PB 96n 40n n + 96 l = 80 B-NB 80n 40n n + 80 Compared wth the small exponent test, our method removes one exponentaton and reduces the number of multplcatons at the cost of ncreasng only a small number of squarngs (whch s ndependent of n). Moreover squarng s generally more effcent than multplcaton [5, 10]. 3 From Table 2, we can see that our test s expected to be about 2 3 tmes faster than the small exponent test as n grows. The sparse exponent test can be appled to sgnature schemes based on modular exponentatons. Especally t s useful for modfed DSA as n [15]. But we need to take prmes p and q such that (p 1)/(2q) has no dvsor less than q, and regard two group elements to be equal f they are the same up to the sgn due to the attack by Boyd and Pavlovsk [4]. 3.2 Bucket Test based on Sparse Exponents The bucket test s a combnaton of the atomc test and the small exponent test [2]. We can get another bucket test based on sparse exponents by replacng the small exponent test by the sparse exponent test appearng n the bucket test: Gven an nstance of n exponentaton pars (x, y ), n pars are randomly parttoned nto 2 m groups and the x s and y s n the same partton are added and multpled to form a new exponentaton par. Then the 2 m pars are verfed usng the sparse exponent test wth the screenng parameter m. Ths procedure s repeated ndependently 3 Accordng to [5, 10], the speed rato of the squarng over the multplcaton s for prme felds and for bnary felds.
8 l/(m 1) tmes for the screenng parameter l. The m s chosen to optmze the performance of the test. We compare the performance of the bucket test based on sparse expoents wth the prevous tests n Table 3. In the table, we set p to be 160 bt prmes and l = 60 as n [2]. Hence k = 13 s enough for our test. K denotes the unt of Table 3. Comparson of Average Numbers of Multplcatons for Dfferent Tests n Naïve Small Exp Bucket Sparse Exp Buc.+Spar. 5 1 K 0.41 K 4.35 K K 2.99 K K 3.26 K 5.78 K 1.48 K 4.08 K K 6.26 K 7.17 K 2.78 K 5.08 K K 15.3 K 10.8 K 6.68 K 8.08 K 1K 200 K 30.2 K 16.6 K 13.2 K 13.1 K 10K 2000 K 300 K 100 K K 85.6 K Observe that our sparse exponent test mproves the small exponent test by a factor of about two and our bucket test based on sparse exponent test gves better performance than the orgnal bucket test. 3.3 A Group wth Secret Order: a Composte Modulus Let N be a product of two strong prmes p N and q N such that p N = 2p N + 1 and q N = 2q N + 1 for two prmes p N and q N. We assume that N s hard to factorze. Then we may take even larger exponents than N for batch verfcaton. We consder S = {s [0, 2 L ) wt(s) k}, where L s a sze of the exponent set S. We may assume that the dfference of any two elements n S s coprme to φ(n)/4 = p N q N. Otherwse, we can get a nontrval dvsor of p N q N or a multple of φ(n) whch gves a factorzaton of N [11]. Hence we may assume that the batch verfer wth exponent set S has the error at most 1/ S by Theorem 1. The cardnalty of S s gven by S = k ( ) L.
9 For the screenng parameter l, k s the mnmum nteger satsfyng S > 2 l. Table 4 gves several k values for varous exponent szes L and screenng parameters l. Table 4. k values for varous exponent szes L and screenng parameters l L k when l = k when l = k when l = The sparse long exponent test s smlar to the sparse exponent test (Fgure 2). The dfference s that the order p of G s secret and the bt length of exponents s longer. Snce p s unknown, Step 3 of Fgure 2 uses normal nteger operatons. Ths s useful for verfyng the modfed GQ sgnatures. We need to modfy the scheme a lttle bt as n the DSA scheme [15, 4]. Let N be the product of two prmes p N and q N and e be a 160 bt nteger. Take random J and compute a J e 1 mod N. Then the publc key s (N, J, e) and the prvate key s (p N, q N, a). The GQ sgnature for a message m Z N s (r = (k e mod N), σ = (ka h mod N)) for randomly chosen k Z N and h a hash output of m and r. The verfcaton s to check f σ e J h r mod N. Assume we are gven n sgnatures (m, r, σ ). We take n random s from S to apply our batch verfcaton and computes ( n =1 σ s ) e J n =1 s h and n =1 r s, (L+160) log n 2 +2kn+80 and verfy f they are equal modulo N. It requres multplcatons and (L + 160) log n + 2L squarngs at average. It can be also appled to the RSA, but t s effcent only when relatvely large publc exponent s used. Some applcatons may requre relatvely large publc exponent e [17] to reduce the prvate exponent. Also the smlar technque can be appled to sgnatures based on strong RSA problems. One example s one of two verfcaton equaltes n the verfcaton n the Cramer-Shoup sgnature [6]. 3.4 Sparse Sgned Bnary Exponent Tests In ellptc curves, subtractons are as effcent as addtons. In ths case, we may consder sgned bnary representaton of exponents. Let S be a
10 subset of nonnegatve ntegers S = {s = log p 1 s 2 0 s = 0, ±1, s s +1 = 0}, where the number of nonzero a s at most k. That s, an element of S has a non-adjacent representaton of the weght k and the most sgnfcant bt s one. Note that an nteger has the unque non-adjacent form (NAF) [16]. By modfyng the proof, we can show that all elements n S are dstnct modulo p. Lemma 1. Any two elements n S are dstnct modulo p. Proof. Suppose that two elements x = log p 1 x 2 and y = log p 1 y 2 n S are congruent modulo p and have dfferent representaton n NAF. Then two elements are between zero and p 1 and so t should be dentcal n Z. Let j be the frst bt from the least sgnfcant bt such that x j y j. Then (x y)/2 j 0 mod 4. Ths s a contradcton. The cardnalty of S s gven n the next theorem. Theorem 2. S = k ( ) log p =1 Proof. We partton S nto two dsjont subsets S 0 and S 1, where log p 1 S 0 = { s 2 S s log p 1 = 0} and S 1 = S \ S 0. Frst, we count the cardnalty of S 0. We choose postons from log p postons, each poston except the frst poston from the most sgnfcant bt s flled wth 01 or 0 1, and the frst poston s flled wth 01 because an element of S should be postve. Snce they are all dstnct and covers all elements of weght n S 0, we have k ( ) log p S 0 = =1 When we count the cardnalty S 1, we frst fx the frst bt of each element by one. Then by smlar argument but wthout the constrant on the frst poston, we have ( k 1 S 1 = log p 1 ) 2 = ( k =1 log p 1 ) 2 1.
11 Usng the well-known formula ( a b 1) + ( a b ) = ( a+1 b ), we have the lemma. In Table 5, we compare the k values for sparse bnary exponents and sparse sgned bnary exponents. Table 5 shows that f we use sparse sgned bnary exponents, then the batch verfcaton could be more effcent when the screenng parameter l 40. ECDSA that s the ellptc curve analogue of DSA s a good example that admts a sparse sgned bnary exponent tests [7]. We have to slghtly modfy the ECDSA for batch verfcaton n the same way [15]. Remark that f we take an ellptc curve whose order s prme, the Boyd and Pavlovsk attack [4] can not be appled. Table 5. k values for varous screenng parameters when log p = 160 l k (Sparse Bnary) k (Sparse Sgned Bnary) Complex Exponent Test Consder an ordnary ellptc curve E defned over F q wth #E(F q ) = q + 1 t and gcd(q, t) = 1. The Frobenus map φ s defned as follows: φ : E(F q ) E(F q ); (x, y) (x q, y q ), where F q s the algebrac closure of F q. The Frobenus map φ satsfes φ 2 tφ + q = 0 on E(F q ). The endomorphsm rng of E contans Z[φ] = Z[x]/(x 2 tx + q). Then the norm functon N( ) on Z[φ] s defned by N(a + bφ) := (a + bφ)(a + b φ) = a 2 + tab + qb 2, a, b Z, where φ s the dual of φ correspondng to the complex conjugate of φ n C. (We may regard φ as a complex number (t + t 2 4q)/2 and Z[φ] as a subset of C.) We denote E(F q m) by the subgroup of E(F q ) consstng of F q m-ratonal ponts. Let G be the subgroup of E(F q m) generated by P wth a prme order p satsfyng p 2 #E(F q m) and p E(F q ). Then we have φ m = 1 and φ m 1 + φ m φ + 1 = 0 on the group G snce G E(F q ) = {O}.
12 In ths secton, the exponent sets can be extended to ths endomorphsm rng, that s why we call complex exponents. Now we consder two canddates for the exponent set. S 1 = { S 2 = { e a φ a Z, a q 1, a a +1 = 0}, e a φ a Z, q a, a < q 2 /2, a a +1 = 0}, where the number of non-zero a s at most k. We note that each S s a set of generalzed sgned φ-adc NAF expansons. To count the number of elements n S whch are dstnct as endomorphsms of G, we need the followng lemma. Lemma 2. Let e 5 and f(x) = e a x be a polynomal n Z[x] wth a M. If 30q e+1 M 2 p, f(φ)p = 0 mples that x 2 tx + q dvdes f(x) n Z[x]. Proof. By dvson algorthm, we have g, r Z such that f(x) = (x 2 tx + q)(g e 2 x e g 1 x + g 0 ) + r 1 x + r 0. Let M 0 = 0, M 1 = M and M = M + t M 1 + qm 2 for 2. Then, by equatng the coeffcents of x, we have g e 2 = a e M 1, g e 3 = a e 1 + tg e 2 M + t M 1 = M 2, g e 1 = a e +1 + tg e qg e +1 for 2 e 1. Also we have M + t M 1 + qm 2 = M r 1 = a 1 + tg 0 qg 1 M e, r 0 = a 0 qg 0 M + qm e 1. If we take M = M + M ( 0) for M = M/( t + q 1), we have M = t M 1 + qm 2 for 2. Ths recurrence relaton has the unque soluton M = c 1β1 +c 2β2 where β 1 and β 2 are the solutons of x 2 t x+q n C and c 1 = M β 2 +M M β 1 β 2 and c 2 = M β 1 +M M β 2 β 1. Thus we have for 4 M max{c 1, c 2 } β 1 + β 2 + M 21 8 q/2 M,
13 because and c ( ) M q + 1 4q t 2 t + q M β 1 + β 2 = q + 1 #E 1 (F q ) 2 q by Hasse-Wel Theorem, where E 1 s E f t > 0 and the twst of E (wth q t F q -ratonal ponts) f t < 0. Hence we have and r qe/2 M r 0 M + qm e 1 ( q(e+1)/2 )M 22 8 q(e+1)/2 M. Usng ths bound, we have N(f(φ)) = N(r 1 φ+r 0 ) = r1 2q+tr 1r 0 +r0 2 < 30q e+1 M 2. Suppose f(φ)p = O. Then N(f(φ))P = f( φ)f(φ)p = O and N(f(φ)) Z, and so N(f(φ)) s dvsble by p. Therefore, f we take 30q e+1 M 2 p, then N(f(φ)) = 0 and so r 1 = r 0 = 0, whch completes proof. Now we can show that each element of S s dstnct as an endomorphsm of G under some condton. log p 2 log M 5 Theorem 3. If 5 e log q, then each element of S s a dstnct endomorphsm of G where M = 2(q 1) for S 1 and M = 2 q2 1 2 for S 2. Proof. Suppose that f 1 (φ)p = f 2 (φ)p for two dfferent polynomals f 1 (x) = e a x and f 2 (x) = e b x n Z[x] wth a M/2 and b M/2. Then f 1 (x) f 2 (x) s dvsble by x 2 tx + q by Lemma 2. Let j be the smallest ndex such that a j b j. Then g(x) = (f 1 (x) f 2 (x))/x j = e j c x Z[x] s dvsble by x 2 tx + q. So c 0 must be a multple of q. Let c 0 = c 0/q Z. Then (g(x) c 0 (x2 tx + q))/x = e j =3 c x 1 + (c 2 c 0 )x + (c 1 + tc 0 ) s also dvsble by x2 tx + q, hence c 1 + tc 0 must be dvded by q agan. That s, q2 dvdes qc 1 + tc 0. If ether a j = 0 or b j = 0, then q does not dvde c 0 by the defnton of S 1 and S 2. If both of a j and b j are non-zero, then a j+1 = b j+1 = 0 and q 2 qc 1 + tc 0 = tc 0. In both cases, g(x) can not be dvded by x 2 tx + q, whch s a contradcton. The cardnalty of S s gven n the next theorem.
14 Theorem 4. The cardnalty of S s gven by 1. S 1 = k 2. S 2 = k ( e+2 ( e+2 ) (2q 2), ) (q 2 q). Proof. At frst, we wll count the cardnalty of S 1. The proof s smlar to the proof of Theorem 2. The dfference s that a negatve coeffcent s allowed n the most sgnfcant nonzero poston. The number of elements whose most sgnfcant bt s zero s k ( ) e + 1 (2q 2). The number of elements whose most sgnfcant bt s nonzero s k 1 ( ) e (2q 2) (2q 2) 1. By summng up the above values, we have S 1 = k ( ) e + 2 (2q 2). For the cardnalty of S 2, t s essentally same as the above case. Only dfference s the number of coeffcents, whch s q 2 q. Our complex exponent test usng Frobenus maps s descrbed n Fgure 3. In order to perform the test effcently, we utlze BGMW multplcatons [1]. In Table 6, gven a power of prme q we present the maxmum e such that all elements of S are dstnct as endomorphsms of G. Also we compute the maxmum weght k satsfyng that the cardnalty S s greater than 2 l for varous screenng parameter l. For example, when q = 8, the maxmum e becomes 47 and we need only 9 ellptc curve addtons for each exponentaton when the number of batch nstances n s large. More precsely, to verfy n exponentatons, we need 9n + 80 ellptc curve addtons when a normal bass s used for representaton of feld elements.
15 Input: (x, Q ) for 1 n. Check: x P = Q for all 1 n where P s a generator of E. 1. Choose n random elements σ 1, σ 2,, σ n from S 1(e, k, q) or S 1(e, k, q) Denote that σ = e j=0 cjφj and ɛ j = c j/ c j for nonzero c j for each. 2. σ n =1 σx mod (φm 1 + φ m φ + 1). 3. R[] O for 0 M 4. For j = 0 to e, do (a) For = 0 to n, do f c j of σ s not zero, then R[ c j ] R[ c j ] + ɛ jφ j (Q ). 5. Q R[M], T R[M] 6. For = M 1 to 1 do (a) T T + R[] (b) Q Q + T 7. Accept f Q = σp, else reject. Fg. 3. Complex exponent test usng Frobenus maps Table 6. The maxmum weght k n each S 1 and S 2 l S 1 (q = 2, e = 153) S 2 (q = 2, e = 153) S 1 (q = 4, e = 74) S 2 (q = 4, e = 73) S 1 (q = 8, e = 49) S 2 (q = 8, e = 47) l S 1 (q = 3, e = 95) S 2 (q = 3, e = 94) S 1 (q = 5, e = 64) S 2 (q = 5, e = 62) S 1 (q = 7, e = 52) S 2 (q = 7, e = 51)
16 5 Concluson In ths paper, we developed two batch verfcaton algorthms usng sparse exponents wth small weghts. The frst one makes use of exponents of (sgned) bnary representaton. We can take exponents whose weght s less than l/4 wth bnary expanson and l/5.5 wth sgned bnary expanson whle the average weght of exponents s l/2 n the small exponent test. Hence we expect that our test s about twce faster. If we use a group whose order s secret, e.g. RSA group, we can extend our algorthm to allow longer exponents to reduce the weght of exponents. The second one can be appled only to specal famly of ellptc curves defned over small fnte felds. It utlzes the exponents wth sparse φ-adc expanson. By enlargng the coeffcents set of the expanson, we obtaned a complex exponent test much faster than sparse exponent test. References 1. E. Brckell, D. Gordon, K. McCurley, and D. Wlson, Fast Exponentaton wth Precomputaton, Proc. of Eurocrypt 92, LNCS Vol. 658, pp , Sprnger- Verlag, M. Bellare, J. Garay, and T. Rabn, Fast Batch Verfcaton for Modular Exponentaton and Dgtal Sgnatures, Proc. of Eurocrypt 98, LNCS Vol. 1403, pp , Sprnger-Verlag, Full verson s avalable va 3. M. Beller and Y. Yacob, Batch Dffe-Hellman Key Agreement Systems and ther Applcaton to Portable Communcatons, Proc. of Eurocrypt 92, LNCS Vol. 658, pp , Sprnger-Verlag, C. Boyd and C. Pavlovsk, Attackng and Reparng Batch Verfcaton Schemes, Proc. of Asacrypt 2000, LNCS Vol. 1976, pp , Sprnger-Verlag, M. Brown, D. Hankerson, J. López, and A. Menezes, Software Implementaton of the NIST Ellptc Curves over Prmes Felds, Proc. of CT-RSA 2001, LNCS, Vol. 2020, pp , Sprnger-Verlag, R. Cramer and V. Shoup, Sgnature Schemes based on the Strong RSA Assumptons, ACM TISSEC, Vol. 3, No. 3, pp , A prelmnary verson appeared n Proc. of Crypto 89, LNCS Vol. 435, pp , Sprnger-Verlag, Publc Key Cryptography for the Fnancal Servces Industry: The Ellptc Curve Dgtal Sgnature Algorthm (ECDSA), ANSI X9.62, approved January 7, A. Fat, Batch RSA, J. Cryptology, Vol. 10, No. 2, pp , Sprnger-Verlag, A prelmnary verson appeared n Proc. of Crypto 89, LNCS Vol. 435, pp , Sprnger-Verlag, L. Gullou and J. Qusquater, A Practcal Zero-knowledge Protocol ftted to Securty Mcroprocessor Mnmzng both Transmsson and Memory, Proc. of Eurocrypt 88, LNCS Vol. 330, pp , Sprnger-Verlag, D. Hankerson, J. Hernandez, and A. Menezes, Software Implementaton of Ellptc Curve Cryptography Over Bnary Felds, Proc. of CHES 2000, LNCS, Vol. 1965, pp. 1 24, Sprnger-Verlag, 2000.
17 11. A. May, Computng the RSA Secret Key s Determnstc Polynomal Tme equvalent to Factorng, Proc. of Crypto 2004, LNCS Vol. 3152, pp , Sprnger- Verlag, D. M Rath and D. Naccache, Batch Exponentaton - A Fast DLP based Sgnature Generaton Strategy, ACM Conference on Computer and Communcatons Securty, pp , ACM, E. Mykletun, M. Narasmha and G. Tsudk, Authentcaton and Integrty n Outsourced Databases, Proc. of ISOC Symposum on Network and Dstrbuted Systems Securty (NDSS04), V. Muller, Fast Multplcaton on Ellptc Curves over Small Felds of Characterstc Two, J. of Cryptology, Vol. 11, pp , Sprnger-Verlag, D. Naccache, D. M Rath, S. Vaudenay, and D. Raphael, Can D.S.A be Improved? Complexty trade-offs wth the Dgtal Sgnature Standard, Proc. of Eurocrypt 94, LNCS Vol. 950, pp , Sprnger-Verlag, J. Solnas, An Improved Algorthm for Arthmetc on a Famly of Ellptc Curves, Proc. of Crypto 97, pp , Sprnger-Verlag, Full verson s avaable at H. Sun, W. Yang, and C. Lah, On the Desgn of RSA wth Short Secret Exponent, Proc. of Asacrypt 99, pp , Sprnger-Verlag, S. Yen and C. Lah, Improved Dgtal Sgnature sutable for Batch Verffcaton, IEEE Trans. on Computers, Vol. 44, No. 7, pp , 1995.
Speeding up Computation of Scalar Multiplication in Elliptic Curve Cryptosystem
H.K. Pathak et. al. / (IJCSE) Internatonal Journal on Computer Scence and Engneerng Speedng up Computaton of Scalar Multplcaton n Ellptc Curve Cryptosystem H. K. Pathak Manju Sangh S.o.S n Computer scence
More informationTHE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens
THE CHINESE REMAINDER THEOREM KEITH CONRAD We should thank the Chnese for ther wonderful remander theorem. Glenn Stevens 1. Introducton The Chnese remander theorem says we can unquely solve any par of
More informationFinding Primitive Roots Pseudo-Deterministically
Electronc Colloquum on Computatonal Complexty, Report No 207 (205) Fndng Prmtve Roots Pseudo-Determnstcally Ofer Grossman December 22, 205 Abstract Pseudo-determnstc algorthms are randomzed search algorthms
More informationRSA /2002/13(08) , ); , ) RSA RSA : RSA RSA [2] , [1,4]
1000-9825/2002/13(081729-06 2002 Journal of Software Vol13, No8 RSA 1,2 1, 1 (, 200433; 2 (, 200070 E-mal: yfhu@fudaneducn http://wwwfudaneducn : RSA RSA :, ; RSA,,, RSA,, : ; RSA ; ;RSA; : TP309 : A RSA
More informationExample: (13320, 22140) =? Solution #1: The divisors of are 1, 2, 3, 4, 5, 6, 9, 10, 12, 15, 18, 20, 27, 30, 36, 41,
The greatest common dvsor of two ntegers a and b (not both zero) s the largest nteger whch s a common factor of both a and b. We denote ths number by gcd(a, b), or smply (a, b) when there s no confuson
More informationFoundations of Arithmetic
Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an
More informationFACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP
C O L L O Q U I U M M A T H E M A T I C U M VOL. 80 1999 NO. 1 FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP BY FLORIAN K A I N R A T H (GRAZ) Abstract. Let H be a Krull monod wth nfnte class
More informationProvable Security Signatures
Provable Securty Sgnatures UCL - Louvan-la-Neuve Wednesday, July 10th, 2002 LIENS-CNRS Ecole normale supéreure Summary Introducton Sgnature FD PSS Forkng Lemma Generc Model Concluson Provable Securty -
More informationLecture 5 Decoding Binary BCH Codes
Lecture 5 Decodng Bnary BCH Codes In ths class, we wll ntroduce dfferent methods for decodng BCH codes 51 Decodng the [15, 7, 5] 2 -BCH Code Consder the [15, 7, 5] 2 -code C we ntroduced n the last lecture
More informationLETTER Skew-Frobenius Maps on Hyperelliptic Curves
189 Skew-Frobenus Maps on Hyperellptc Curves Shunj KOZAKI a, Nonmember, Kazuto MATSUO, Member, and Yasutomo SHIMBARA, Nonmember SUMMARY Scalar multplcaton methods usng the Frobenus maps are known for effcent
More informationChapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems
Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons
More informationa b a In case b 0, a being divisible by b is the same as to say that
Secton 6.2 Dvsblty among the ntegers An nteger a ε s dvsble by b ε f there s an nteger c ε such that a = bc. Note that s dvsble by any nteger b, snce = b. On the other hand, a s dvsble by only f a = :
More informationThe Minimum Universal Cost Flow in an Infeasible Flow Network
Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran
More informationProblem Set 9 Solutions
Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem
More informationAPPENDIX A Some Linear Algebra
APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,
More informationCryptanalysis of pairing-free certificateless authenticated key agreement protocol
Cryptanalyss of parng-free certfcateless authentcated key agreement protocol Zhan Zhu Chna Shp Development Desgn Center CSDDC Wuhan Chna Emal: zhuzhan0@gmal.com bstract: Recently He et al. [D. He J. Chen
More informationMin Cut, Fast Cut, Polynomial Identities
Randomzed Algorthms, Summer 016 Mn Cut, Fast Cut, Polynomal Identtes Instructor: Thomas Kesselhem and Kurt Mehlhorn 1 Mn Cuts n Graphs Lecture (5 pages) Throughout ths secton, G = (V, E) s a mult-graph.
More informationThe Ramanujan-Nagell Theorem: Understanding the Proof By Spencer De Chenne
The Ramanujan-Nagell Theorem: Understandng the Proof By Spencer De Chenne 1 Introducton The Ramanujan-Nagell Theorem, frst proposed as a conjecture by Srnvasa Ramanujan n 1943 and later proven by Trygve
More informationDISCRIMINANTS AND RAMIFIED PRIMES. 1. Introduction A prime number p is said to be ramified in a number field K if the prime ideal factorization
DISCRIMINANTS AND RAMIFIED PRIMES KEITH CONRAD 1. Introducton A prme number p s sad to be ramfed n a number feld K f the prme deal factorzaton (1.1) (p) = po K = p e 1 1 peg g has some e greater than 1.
More informationCollege of Computer & Information Science Fall 2009 Northeastern University 20 October 2009
College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:
More informationCOMPLEX NUMBERS AND QUADRATIC EQUATIONS
COMPLEX NUMBERS AND QUADRATIC EQUATIONS INTRODUCTION We know that x 0 for all x R e the square of a real number (whether postve, negatve or ero) s non-negatve Hence the equatons x, x, x + 7 0 etc are not
More informationHMMT February 2016 February 20, 2016
HMMT February 016 February 0, 016 Combnatorcs 1. For postve ntegers n, let S n be the set of ntegers x such that n dstnct lnes, no three concurrent, can dvde a plane nto x regons (for example, S = {3,
More information5 The Rational Canonical Form
5 The Ratonal Canoncal Form Here p s a monc rreducble factor of the mnmum polynomal m T and s not necessarly of degree one Let F p denote the feld constructed earler n the course, consstng of all matrces
More informationThe Order Relation and Trace Inequalities for. Hermitian Operators
Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence
More informationREDUCTION MODULO p. We will prove the reduction modulo p theorem in the general form as given by exercise 4.12, p. 143, of [1].
REDUCTION MODULO p. IAN KIMING We wll prove the reducton modulo p theorem n the general form as gven by exercse 4.12, p. 143, of [1]. We consder an ellptc curve E defned over Q and gven by a Weerstraß
More informationThe internal structure of natural numbers and one method for the definition of large prime numbers
The nternal structure of natural numbers and one method for the defnton of large prme numbers Emmanul Manousos APM Insttute for the Advancement of Physcs and Mathematcs 3 Poulou str. 53 Athens Greece Abstract
More informationPerfect Competition and the Nash Bargaining Solution
Perfect Competton and the Nash Barganng Soluton Renhard John Department of Economcs Unversty of Bonn Adenauerallee 24-42 53113 Bonn, Germany emal: rohn@un-bonn.de May 2005 Abstract For a lnear exchange
More information2.3 Nilpotent endomorphisms
s a block dagonal matrx, wth A Mat dm U (C) In fact, we can assume that B = B 1 B k, wth B an ordered bass of U, and that A = [f U ] B, where f U : U U s the restrcton of f to U 40 23 Nlpotent endomorphsms
More informationComputing Correlated Equilibria in Multi-Player Games
Computng Correlated Equlbra n Mult-Player Games Chrstos H. Papadmtrou Presented by Zhanxang Huang December 7th, 2005 1 The Author Dr. Chrstos H. Papadmtrou CS professor at UC Berkley (taught at Harvard,
More informationAn efficient algorithm for multivariate Maclaurin Newton transformation
Annales UMCS Informatca AI VIII, 2 2008) 5 14 DOI: 10.2478/v10065-008-0020-6 An effcent algorthm for multvarate Maclaurn Newton transformaton Joanna Kapusta Insttute of Mathematcs and Computer Scence,
More informationMath 261 Exercise sheet 2
Math 261 Exercse sheet 2 http://staff.aub.edu.lb/~nm116/teachng/2017/math261/ndex.html Verson: September 25, 2017 Answers are due for Monday 25 September, 11AM. The use of calculators s allowed. Exercse
More informationThe Synchronous 8th-Order Differential Attack on 12 Rounds of the Block Cipher HyRAL
The Synchronous 8th-Order Dfferental Attack on 12 Rounds of the Block Cpher HyRAL Yasutaka Igarash, Sej Fukushma, and Tomohro Hachno Kagoshma Unversty, Kagoshma, Japan Emal: {garash, fukushma, hachno}@eee.kagoshma-u.ac.jp
More informationA new construction of 3-separable matrices via an improved decoding of Macula s construction
Dscrete Optmzaton 5 008 700 704 Contents lsts avalable at ScenceDrect Dscrete Optmzaton journal homepage: wwwelsevercom/locate/dsopt A new constructon of 3-separable matrces va an mproved decodng of Macula
More informationDifference Equations
Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1
More informationMaximizing the number of nonnegative subsets
Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum
More informationNP-Completeness : Proofs
NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem
More informationIntroduction to Algorithms
Introducton to Algorthms 6.046J/8.40J Lecture 7 Prof. Potr Indyk Data Structures Role of data structures: Encapsulate data Support certan operatons (e.g., INSERT, DELETE, SEARCH) Our focus: effcency of
More informationNotes on Frequency Estimation in Data Streams
Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to
More informationOn quasiperfect numbers
Notes on Number Theory and Dscrete Mathematcs Prnt ISSN 1310 5132, Onlne ISSN 2367 8275 Vol. 23, 2017, No. 3, 73 78 On quasperfect numbers V. Sva Rama Prasad 1 and C. Suntha 2 1 Nalla Malla Reddy Engneerng
More informationLectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix
Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could
More informationThe average time complexity of probabilistic algorithms for finding generators in finite cyclic groups
BULLETIN OF THE POLISH ACADEMY OF SCIENCES TECHNICAL SCIENCES, Vol 63, No 4, 2015 DOI: 101515/bpasts-2015-0112 The average tme complexty of probablstc algorthms for fndng generators n fnte cyclc groups
More informationCryptanalysis of a Public-key Cryptosystem Using Lattice Basis Reduction Algorithm
www.ijcsi.org 110 Cryptanalyss of a Publc-key Cryptosystem Usng Lattce Bass Reducton Algorthm Roohallah Rastagh 1, Hamd R. Dall Oskoue 2 1,2 Department of Electrcal Engneerng, Aeronautcal Unversty of Snce
More informationAttacks on RSA The Rabin Cryptosystem Semantic Security of RSA Cryptology, Tuesday, February 27th, 2007 Nils Andersen. Complexity Theoretic Reduction
Attacks on RSA The Rabn Cryptosystem Semantc Securty of RSA Cryptology, Tuesday, February 27th, 2007 Nls Andersen Square Roots modulo n Complexty Theoretc Reducton Factorng Algorthms Pollard s p 1 Pollard
More information3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X
Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number
More informationSection 3.6 Complex Zeros
04 Chapter Secton 6 Comple Zeros When fndng the zeros of polynomals, at some pont you're faced wth the problem Whle there are clearly no real numbers that are solutons to ths equaton, leavng thngs there
More informationSmarandache-Zero Divisors in Group Rings
Smarandache-Zero Dvsors n Group Rngs W.B. Vasantha and Moon K. Chetry Department of Mathematcs I.I.T Madras, Chenna The study of zero-dvsors n group rngs had become nterestng problem snce 1940 wth the
More informationSL n (F ) Equals its Own Derived Group
Internatonal Journal of Algebra, Vol. 2, 2008, no. 12, 585-594 SL n (F ) Equals ts Own Derved Group Jorge Macel BMCC-The Cty Unversty of New York, CUNY 199 Chambers street, New York, NY 10007, USA macel@cms.nyu.edu
More informationDiscussion 11 Summary 11/20/2018
Dscusson 11 Summary 11/20/2018 1 Quz 8 1. Prove for any sets A, B that A = A B ff B A. Soluton: There are two drectons we need to prove: (a) A = A B B A, (b) B A A = A B. (a) Frst, we prove A = A B B A.
More informationAnti-van der Waerden numbers of 3-term arithmetic progressions.
Ant-van der Waerden numbers of 3-term arthmetc progressons. Zhanar Berkkyzy, Alex Schulte, and Mchael Young Aprl 24, 2016 Abstract The ant-van der Waerden number, denoted by aw([n], k), s the smallest
More informationarxiv: v6 [math.nt] 23 Aug 2016
A NOTE ON ODD PERFECT NUMBERS JOSE ARNALDO B. DRIS AND FLORIAN LUCA arxv:03.437v6 [math.nt] 23 Aug 206 Abstract. In ths note, we show that f N s an odd perfect number and q α s some prme power exactly
More informationn α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0
MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector
More informationSubset Topological Spaces and Kakutani s Theorem
MOD Natural Neutrosophc Subset Topologcal Spaces and Kakutan s Theorem W. B. Vasantha Kandasamy lanthenral K Florentn Smarandache 1 Copyrght 1 by EuropaNova ASBL and the Authors Ths book can be ordered
More informationALGEBRA HW 7 CLAY SHONKWILER
ALGEBRA HW 7 CLAY SHONKWILER 1 Whch of the followng rngs R are dscrete valuaton rngs? For those that are, fnd the fracton feld K = frac R, the resdue feld k = R/m (where m) s the maxmal deal), and a unformzer
More informationFirst day August 1, Problems and Solutions
FOURTH INTERNATIONAL COMPETITION FOR UNIVERSITY STUDENTS IN MATHEMATICS July 30 August 4, 997, Plovdv, BULGARIA Frst day August, 997 Problems and Solutons Problem. Let {ε n } n= be a sequence of postve
More informationModule 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur
Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:
More informationISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 1, July 2013
ISSN: 2277-375 Constructon of Trend Free Run Orders for Orthogonal rrays Usng Codes bstract: Sometmes when the expermental runs are carred out n a tme order sequence, the response can depend on the run
More informationLecture 4: Universal Hash Functions/Streaming Cont d
CSE 5: Desgn and Analyss of Algorthms I Sprng 06 Lecture 4: Unversal Hash Functons/Streamng Cont d Lecturer: Shayan Oves Gharan Aprl 6th Scrbe: Jacob Schreber Dsclamer: These notes have not been subjected
More informationPolynomials. 1 More properties of polynomials
Polynomals 1 More propertes of polynomals Recall that, for R a commutatve rng wth unty (as wth all rngs n ths course unless otherwse noted), we defne R[x] to be the set of expressons n =0 a x, where a
More informationTHE SUMMATION NOTATION Ʃ
Sngle Subscrpt otaton THE SUMMATIO OTATIO Ʃ Most of the calculatons we perform n statstcs are repettve operatons on lsts of numbers. For example, we compute the sum of a set of numbers, or the sum of the
More informationSelf-complementing permutations of k-uniform hypergraphs
Dscrete Mathematcs Theoretcal Computer Scence DMTCS vol. 11:1, 2009, 117 124 Self-complementng permutatons of k-unform hypergraphs Artur Szymańsk A. Paweł Wojda Faculty of Appled Mathematcs, AGH Unversty
More informationAffine transformations and convexity
Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/
More informationBernoulli Numbers and Polynomials
Bernoull Numbers and Polynomals T. Muthukumar tmk@tk.ac.n 17 Jun 2014 The sum of frst n natural numbers 1, 2, 3,..., n s n n(n + 1 S 1 (n := m = = n2 2 2 + n 2. Ths formula can be derved by notng that
More informationLinear Approximation with Regularization and Moving Least Squares
Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...
More informationBoostrapaggregating (Bagging)
Boostrapaggregatng (Baggng) An ensemble meta-algorthm desgned to mprove the stablty and accuracy of machne learnng algorthms Can be used n both regresson and classfcaton Reduces varance and helps to avod
More informationThe Geometry of Logit and Probit
The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.
More informationKernel Methods and SVMs Extension
Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general
More informationFormulas for the Determinant
page 224 224 CHAPTER 3 Determnants e t te t e 2t 38 A = e t 2te t e 2t e t te t 2e 2t 39 If 123 A = 345, 456 compute the matrx product A adj(a) What can you conclude about det(a)? For Problems 40 43, use
More informationStanford University CS254: Computational Complexity Notes 7 Luca Trevisan January 29, Notes for Lecture 7
Stanford Unversty CS54: Computatonal Complexty Notes 7 Luca Trevsan January 9, 014 Notes for Lecture 7 1 Approxmate Countng wt an N oracle We complete te proof of te followng result: Teorem 1 For every
More informationFinding Malleability in NTRUSign
Fndng Malleablty n TRUSgn SungJun Mn, Go Yamamoto, and Kwangjo Km Auto-ID Labs Whte Paper WP-HARDWARE-33 Sungjun Mn Senor Researcher, atonal Computerzaton Agency Go Yamamoto Senor Researcher, Informaton
More informationOn the Interval Zoro Symmetric Single-step Procedure for Simultaneous Finding of Polynomial Zeros
Appled Mathematcal Scences, Vol. 5, 2011, no. 75, 3693-3706 On the Interval Zoro Symmetrc Sngle-step Procedure for Smultaneous Fndng of Polynomal Zeros S. F. M. Rusl, M. Mons, M. A. Hassan and W. J. Leong
More informationStructure and Drive Paul A. Jensen Copyright July 20, 2003
Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.
More informationMEM 255 Introduction to Control Systems Review: Basics of Linear Algebra
MEM 255 Introducton to Control Systems Revew: Bascs of Lnear Algebra Harry G. Kwatny Department of Mechancal Engneerng & Mechancs Drexel Unversty Outlne Vectors Matrces MATLAB Advanced Topcs Vectors A
More informationCHALMERS GÖTEBORGS UNIVERSITET. TDA352 (Chalmers) - DIT250 (GU) 12 Jan. 2017, 14:00-18:00
CHALMERS GÖTEBORGS UNIVERSITET CRYPTOGRAPHY TDA352 (Chalmers) - DIT250 (GU) 12 Jan. 2017, 14:00-18:00 No extra materal s allowed durng the exam except for pens and a smple calculator (not smartphones).
More informationRestricted divisor sums
ACTA ARITHMETICA 02 2002) Restrcted dvsor sums by Kevn A Broughan Hamlton) Introducton There s a body of work n the lterature on varous restrcted sums of the number of dvsors of an nteger functon ncludng
More information= z 20 z n. (k 20) + 4 z k = 4
Problem Set #7 solutons 7.2.. (a Fnd the coeffcent of z k n (z + z 5 + z 6 + z 7 + 5, k 20. We use the known seres expanson ( n+l ( z l l z n below: (z + z 5 + z 6 + z 7 + 5 (z 5 ( + z + z 2 + z + 5 5
More informationCHAPTER-5 INFORMATION MEASURE OF FUZZY MATRIX AND FUZZY BINARY RELATION
CAPTER- INFORMATION MEASURE OF FUZZY MATRI AN FUZZY BINARY RELATION Introducton The basc concept of the fuzz matr theor s ver smple and can be appled to socal and natural stuatons A branch of fuzz matr
More informationMath Review. CptS 223 Advanced Data Structures. Larry Holder School of Electrical Engineering and Computer Science Washington State University
Math Revew CptS 223 dvanced Data Structures Larry Holder School of Electrcal Engneerng and Computer Scence Washngton State Unversty 1 Why do we need math n a data structures course? nalyzng data structures
More information8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS
SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 493 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces you have studed thus far n the text are real vector spaces because the scalars
More informationCSE4210 Architecture and Hardware for DSP
4210 Archtecture and Hardware for DSP Lecture 1 Introducton & Number systems Admnstratve Stuff 4210 Archtecture and Hardware for DSP Text: VLSI Dgtal Sgnal Processng Systems: Desgn and Implementaton. K.
More informationOn the Multicriteria Integer Network Flow Problem
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of
More informationMore metrics on cartesian products
More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of
More informationOn cyclic of Steiner system (v); V=2,3,5,7,11,13
On cyclc of Stener system (v); V=,3,5,7,,3 Prof. Dr. Adl M. Ahmed Rana A. Ibraham Abstract: A stener system can be defned by the trple S(t,k,v), where every block B, (=,,,b) contans exactly K-elementes
More informationGeneralized Linear Methods
Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set
More informationLINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity
LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have
More informationPerformance Analysis of the Postcomputation- Based Generic-Point Parallel Scalar Multiplication Method
P a g e 3 Vol. 1 Issue 11 (Ver. 1.) October 1 Global Journal of Computer Scence and Technology Performance Analyss of the Postcomputaton- Based Generc-Pont Parallel Scalar Multplcaton Method Tur F. Al-Soman
More informationThe Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction
ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also
More informationLOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin
Proceedngs of the 007 Wnter Smulaton Conference S G Henderson, B Bller, M-H Hseh, J Shortle, J D Tew, and R R Barton, eds LOW BIAS INTEGRATED PATH ESTIMATORS James M Calvn Department of Computer Scence
More informationDirichlet s Theorem In Arithmetic Progressions
Drchlet s Theorem In Arthmetc Progressons Parsa Kavkan Hang Wang The Unversty of Adelade February 26, 205 Abstract The am of ths paper s to ntroduce and prove Drchlet s theorem n arthmetc progressons,
More informationCommunication Complexity 16:198: February Lecture 4. x ij y ij
Communcaton Complexty 16:198:671 09 February 2010 Lecture 4 Lecturer: Troy Lee Scrbe: Rajat Mttal 1 Homework problem : Trbes We wll solve the thrd queston n the homework. The goal s to show that the nondetermnstc
More informationProblem Do any of the following determine homomorphisms from GL n (C) to GL n (C)?
Homework 8 solutons. Problem 16.1. Whch of the followng defne homomomorphsms from C\{0} to C\{0}? Answer. a) f 1 : z z Yes, f 1 s a homomorphsm. We have that z s the complex conjugate of z. If z 1,z 2
More informationwhere a is any ideal of R. Lemma 5.4. Let R be a ring. Then X = Spec R is a topological space Moreover the open sets
5. Schemes To defne schemes, just as wth algebrac varetes, the dea s to frst defne what an affne scheme s, and then realse an arbtrary scheme, as somethng whch s locally an affne scheme. The defnton of
More informationSome Consequences. Example of Extended Euclidean Algorithm. The Fundamental Theorem of Arithmetic, II. Characterizing the GCD and LCM
Example of Extended Eucldean Algorthm Recall that gcd(84, 33) = gcd(33, 18) = gcd(18, 15) = gcd(15, 3) = gcd(3, 0) = 3 We work backwards to wrte 3 as a lnear combnaton of 84 and 33: 3 = 18 15 [Now 3 s
More informationCalculation of time complexity (3%)
Problem 1. (30%) Calculaton of tme complexty (3%) Gven n ctes, usng exhaust search to see every result takes O(n!). Calculaton of tme needed to solve the problem (2%) 40 ctes:40! dfferent tours 40 add
More informationLecture 4: November 17, Part 1 Single Buffer Management
Lecturer: Ad Rosén Algorthms for the anagement of Networs Fall 2003-2004 Lecture 4: November 7, 2003 Scrbe: Guy Grebla Part Sngle Buffer anagement In the prevous lecture we taled about the Combned Input
More informationREGULAR POSITIVE TERNARY QUADRATIC FORMS. 1. Introduction
REGULAR POSITIVE TERNARY QUADRATIC FORMS BYEONG-KWEON OH Abstract. A postve defnte quadratc form f s sad to be regular f t globally represents all ntegers that are represented by the genus of f. In 997
More informationErrors for Linear Systems
Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch
More information18.781: Solution to Practice Questions for Final Exam
18.781: Soluton to Practce Questons for Fnal Exam 1. Fnd three solutons n postve ntegers of x 6y = 1 by frst calculatng the contnued fracton expanson of 6. Soluton: We have 1 6=[, ] 6 6+ =[, ] 1 =[,, ]=[,,
More informationAlgorithms for factoring
CSA E0 235: Crytograhy Arl 9,2015 Instructor: Arta Patra Algorthms for factorng Submtted by: Jay Oza, Nranjan Sngh Introducton Factorsaton of large ntegers has been a wdely studed toc manly because of
More information18.1 Introduction and Recap
CS787: Advanced Algorthms Scrbe: Pryananda Shenoy and Shjn Kong Lecturer: Shuch Chawla Topc: Streamng Algorthmscontnued) Date: 0/26/2007 We contnue talng about streamng algorthms n ths lecture, ncludng
More information2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification
E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton
More information