Internat. J. Math. & Math. Sci. Vol. 8 No. 2 (1985) 373-388 373 ON FUNDAMENTAL SETS OVER A FINITE FIELD YOUSEF ABBAS and JOSEPH J. LIANG Department of Mathematics University of South Florida Tampa, Florida 33620 U.S.A. (Received March 3, 1983) A partition over finite field is defined and each equivalence class is constructed and represented by a set called the fundamental set. If a primitive ABSTRACT. element is used to construct the addition table over these fundamental sets then The number of partitions is given all additions over the field can be computed. for some finite fields. KEY WORDS AND PHRASES. Finite field, primitive polynomial. 19B0 MATHEmaTICS SUBJECT CLASSIFICATION CODE. 12C99. i. INTRODUCTION Throughout p,q will be fixed but arbitrary primes. n > 1 and define A A E GF(p n) {a, a+l, +2, then B E Define (p-l)A such that 1,2,...,p- I, thus Let A, A A. As As. DEFINITION Note that if 8 6 DEFINITION. , Ae, Aa p= p/x {x A P+ (p-l)} Aap the values of , 1 , then ( Since pn =i AS" AC for {a p3 A n-l|| +bla Z*p b EXa, Z . p * (A a pj + -1 or A A8" b. we have Apn A +I, Therefore ’ This implies 8C P ,Ap and j E Z } n X A--B. LEMMA i.i. If 8 then PROOF. If 8 6 A then there exist a 6 Z 8 P ,P so, A( GF(p n) for every DEFINITION. X p 0,i 2 , Ap Aap a Since If A} in the definition can be limited to 0,1,2,...,n-i. DEFINITION. thus {x/x P-IU= DA *A =+ (p-l)}. A 2A, 3A P b 6 Z (Zp,+,.) P and J Z n such that is a field and a 0 both a Y. ABBAS AND J. J. LIANG 374 Since pn A such that A will be called a Fundamental Set. A DEFINITION. E GF(p n), there A for every a pm aA for some a Let u,8 E DEFINITION. 6 Z* GF(pn). From Lemma i.i. in GF(p n) as Aa AB is an equivalence relation. the relation will partition the field GF(p n) into equivalence This equivalence relation classes A n *A We define the relation 8 iff <_ p It follows that A a THEOREM i.I. exists a least positive integer m m-i and each class is represented by a fundamental set. A will be called a Fundamental Class. DEFINITION. pn_ i Let 8 be a primitive element in , GF(pn). 8p Since, -I is primitive in Z P where i < k < p- i. So, k can be determined there exists k a a a p set A are expressed as powers of 8, then fundamental of the elements easily. If the can be expressed from A as powers of 8. So to calculate the addition table of Z then for every a Aa it is sufficient to have the addition table of GF(pn), are all the fundamental classes in addition tables over A GF(pn). , Z P B integer for every as Z and b a Therefore, if AI,Aa2,...,A=s, it will be enough to tabulate only the with respect to 8 to do all the calculations over ,...,A GF(pn), If for some a for some a i A. apm m is the least positive integer such that a+ b then it is true that m will be the smallest positive P Bpm such that aB+ b’ where b’E Z P This will be shown below. aa+ b for GF(pn), Let a DEFINITION. ap m some a nd coefficient of Z* P we say b a has index 0 with coefficient a if m is the least positive integer such that Z P and a is the then m is called the index of If has an index m with coefficient a. Z P we say i. A has the If a has index m with coefficient a, then every 8 LEMMA 1.2. = same index m and the same coefficient a. 6 Z and Z There exist E p P P m pj pj pm p + 6 b) + (aa + 6) (a + 6. This implies 8 Z such that 8 n Hence A Therefore s _< m. But from Lemma i.i. we have + c where c 8. a. coefficient with m index has < s, which implies 8 A. Therefore m PROOF j a a with index s and Let B B Z PJ As Zp. DEFINITION The fundamental set A has index m with coefficient a if and only pm if m is the least positive integer such that A THEOREM 1.2. If aA. has index m with coefficient a coefficient a and each fundamental set A C 8 Aa then A has index m with has the same index m with the same coefficient a. PROOF. Follows from Lemma 12 From the above theorem we can define the index A to be the index of any element or any fundamental set included in A Now, we want to discuss some properties of the index and the coefficient of the fundamental set. 375 FUNDAMENTAL SETS OVER A FINITE FILED In GF(p n), if A has index m and coefficient a then m dfvldes THEOREM 1.3. n/m n and a I. mpm m+ Let n PROOF a(m) + l)b A pr aA which A So bA, b r Hence ml for some bE Z m r a r n/m . a -m + (a we have p -i + a -2 + i. Let km am+ 6 and mp pm m such that If + r, 0 _< b+ 6’. But Therefore This implies r > m. 6". r < m. Since all finite fields of the same order are Note: b. and a i E GFI(P n) isomorphic, if am+ b, (mp )p A, we have m < r , r P (ako + 6 )P + 6’ akm p + r (P )P 0. n and a then m divides P Since m is the index of km b+ 6’ m 0 and a implies r A, then there exist 6, 6’ Z If a p Let A be a fundamental set with index m and coefficient a. /m b. E Z* PROOF. mp + (a -I + a -2 +...+ l)b and m THEOREM 1.4. p Since m r, 0 < r < m. having index m with coefficient a, is the image of a. There2 GF2(pn) under an isomorphism o, then 2 has index m with coefficient fore; GFl(pn) and GF2(pn) have the same partitions with respect to the equivalence relation GF(52) EXAMPLE: Let F the primitive polynomial two equivalence classes, Z 8 3 817 + 4 2A 8 3A 4A and A 2. 4A 8 5 0 8 8228/ 815 , Z P GF(pm) in GF(pn). 822 i Since 8 815 + 2 86 3 82 812 and + 8 2, then 4 87 + 2 8 4 87 + 4 =821 87 + 1 88} {019 019 + 3 016 019 + i 0 I0 019 + 4 821 019 + {813,813 + 4 810,813 + 3 83, 813 + 2 814, 813 + 8 (a) + 82 817. SOLUTIONS OF EQUAl IONS OF THE FORM Ap solutions of: - 8 {8 7 To study the fundamental sets in 6 {8, A8 and So, we have 8 Therefore, 818. 5 and 8 be a primitive element in F such that 8 satisfies 2 x + 4x + 2. The field F has polynomial], P(x) indexing pm x Note: ax + 6; a 2 I 06 85 m aA. GF(p n) with index m < n, we have to study the Z* # i, 6 E Z a P P and (b) x pm x + 6; 0 in (b), then all the elements of the subfield If 6 will satisfy (b). We will now consider the solutions of x where , a 6 Zp, a LEMMA 2 i. n’m n and a # 1 and 6 p ax + (2.1) 6 Zp Equation (2 i) has a solution e Z p with index m only if m divides i. The proof is a direct application of Theorem 1.3 and Theorem m1.4. ay. we have x + LEMMA 2.2. In equation (2.1), for y a-1 PROOF yP m )pm=xpm + a-i 6 ax + 6 + a--- a(y- a_l yP (x + ay. a + a-i 376 Y. ABBAS AND J. J. LIANG Therefore, to study the solutions of (2.1), it is sufficient to study the solutions of ypm ay. LEMMA 2.3. -a x is a solution of equation If 0 is a primitive element in GF(pn), (2.1). then the following statements are true: p -i P-I (a) 0 (b) Z Z * there exists an integer ka such that a p If a i, ka # 0. For every a where 0 < k < p-i. a LEMMA 2.4. n p -I k p For every m such that m divides n and n m 0 n p-i, then + I. n m Let 0(mod k). LEMMA 2.5. If a k n PROOF. n/m n a _p -i 1 m n/m p -i n p -i p Pn-! -ka .nm i p -I Therefore, p-i k a m p Given x ax 0(a) i and where m n, a If P r Z (a) If x (b) Since x n k 1 a p-1 m (c) Or ax. x ax and a # i, is a solution of then x n p 1 m p -i It follows from Lemma 2.5 a p-i + (pro) k-2 is an integer. pn-l_.ka implies So p-i k a Z (mod P is primitive in GF(pn), pn -!). m p -I P k a p-i O. _GF(pm). (2.2) then (or) pm-I =- pnp-ii ka pn_ 1). (mod r(pm- i) r k 0 mod(pn-l) Therefore x m p n/m m then xp P I n m (2.2) 0 is the only solution of (2.2) in Z PROOF. which implies n m in p -I x i Therefore, i 0 p- Also, a GF(p n) and r O O a solution of (2 2) where a is i then Equation (2.2) has solution If I p i. THEOREM 2.1. (c) p-1 is an integer. n/m i 1,2 n -i then p m p -i Im_(n/m) ka n n p- i 0 (a) (b) m l(mod k) for j 1 and i, a p-i But, from Lemma 2.4 (pro)k- 1 pm -i a pn_ i pm 1 n m Hence j=O n/m a pJ l(mod k), we have k-i (pm)j (pro) k -I m p -i -= Since p pn -1 l(mod k) and So, p k. Pm p is an integer, PROOF. 1 p pn-i =a= 8 (mod pn p-i k a I) m p -i that the above congruence is meaningful. Thus, then FUNDAMENTAL SETS OVER A FINITE FIELD a P Let r PROOF n pn_ i -k o p-I i I a (Or) p So, a. O m 0,1,2,.. j n (8p-l)j a pn_ i) r is a solution of p -i a p-I + j pm -I -i m p -i m_l pn k (mod I p p -I (2.2). (Sr)P k n p -i m For every r THEOREM 2.2. 377 then m aO. Now consider the solutions of x where a # i, Z a we have Theorem 2.2, m + ax n m and 0(a) (2.3) 6 From Lemma 2.2, Theorem 2.1 and is a solution of (2.3), where k n p -I r Zp, mln 6 r O + -a 6 p p a "p-i (mod m p -i is the only solution in Z and m p -i P Let A be a fundamental set in GF(pn) with index m and coefficient n and If A is not included in any proper subfield of GF(p n), then O(a) m THEOREM 2.3. i a n l(P- i). m Let 0(a) PROOF. p Thus, s coefficient a. dm d p a + a s GF(pdm) for some 6 6 Z + 1)6 + dl and a has an index m with Hence P Therefore s 6GF(p s. din), which implies and d COROLLARY 2.1. such that a + as (ed-1 + d-2 GF(pn) It follows A. d and u m Let X mln, p + 6 aX m (2.4) , then equation (2.4) has solutions GF(p n) and all the solutions are included in PROOF. GF(p is a subfield of GF(pn). Equation (2.4) satisfies the conditions in i, 6 Z P a 1 and O(a) GF(pm). m) of Theorem 2.1 up m a u + (a over GF(p -1 + a-2 m) and if u is a solution, then + 1)6 + + . O GF(pm) Therefore p -1 Note: in y r’ If 8 is a primitive element in GF(pm). + P r’ m k’a 1 p-1 m p -1 a 0 So, k a 8 r p-i k’. a + i_-, then y 0 is a primitive element Therefore, the solutions of equation (2.4) are of the form: where pn-I GF(pn), p m -i k pn-I a ,we have 8 p-I (mod pm a (8 and a m 7 p 1 a Since p -1 n p -1 k _p.m_ 1 k’ 1 m p -i p m -I p-l’ k a pn -1 -I k’a Op Therefore, the solutions of (2.4) over GF(pn) are of the form where r n P -I m p -i n r p -1 m p -i k a p-l- (mod pn !) m p -i _ 378 Y. ABBAS AND J. J. LIANG m LEMMA 2.6. Equation (2.4) has p solutions in GF(pn). PROOF. Corollary 2.1 insures that equation (2.4) has solution in GF(p p -I + 0 solution of (2.4). O(a)[ and Theorem 2.2 is a solution and if a a a p-i m From the previous note, Theorem 2.1 and a m). H It is clear that H has we have 0 8p -1 lGF(pm)} {0 B + pm elements. then a is a THEOREM 2.4. For every m divides n and every a E Z 1 such that a P there exists a fundamental set in GF(p n) with index m and coefficient a. n m PROOF. By Lemma 2.6, x p ax has solutions in n p_ -1 primitive element in is a solution. an index m with coefficient a. To prove this, we assume coefficient b, therefore by Theorem 1.4 and Theorem 2.2 solution of x p bx and it is of the form e If is a ok 0 p -1 6F(p n) then GF(pn). es We claim that has an index we conclude that n where s has m with p -I p -I p-I nod is a pn_ 1) p -i So, there exists j where j > 0 such that: n k p -1 _< then 0 Since j p-1 p -1 pm -1 p-1 pro_ llp_ 1 pm_ i_ > P ka p -i < p- 1 and i _< <_ J < 0 we will have p_-- P and this is a contradiction. Hence A--A is a fundamental set with index m and coefficient a. COROLLARY 2.2. The minimum subfield which contains all the solutions of GF(pm). equation (2.4) is The proof is a direct application of Theorem 1.3 and Theorem 2.4. PROOF. We will now review some known facts about the solutions of x the field j p m x + b, + where b EGF(p n) n g.c.d. (m, LEMMA 2.7. If x b in d in GF(pn). x 0 0 is a solution of + 8p (2.5) The following theorems were given in [i]. n) and r i, 2,..., p -i, element x GF(pn). Let x and d p -i (2.5) in is a solution of GF(pn), then for every (2.5) where 8 is a primitive FUNDAMENTAL SETS OVER A FINITE FIELD PROOF. m n Xo + 8p 379 d m x -i + (O p m p d i n_ (x THEOREM 2.5. THEOREM 2.6. 0+b) + e p -I Xo 1 + ep + b (2.5) in GF(p n) is ether 0 or p Equation (2.5) has solutions in GF(p n) if and only if The number of solutions of d r-i If we assume m divides n and b Z , in the equation P x p m x + (2.6) b then we can conclude the following: (a) (b) d g- c- d(m.n), m for every b E Z P we have n E pm b=--nb=O m .ffiO if and only if p divides n m Now, we can restate the following theorems: (a) Theorem 2.7. Equation (2.6) has solution in GF(p n) if and only if p divides (b) n m Theorem 2.8. If equation (2.6)has a solution in solutions, also if x 0 is a solution then x 0 GF(pn), it ham pm + 0 p -1 0,1,...,p -i is a solution where 0 is primitive in GF(pn). 2.9. If (2.6) has a solution in GF(pn),then the minimum subfield that THEOREM j , contain all .the solutions is PROOF. p divides Since equation GF(ppm). (2.6) has a solution and m divides n then by Theorem 2.7 therefore GF(p pro) is a subfield of m then we have a aP a + b, which implies aP GF(pn). If a is a solution of (2.6), pm = + pb = so = GF(p pro) and GF(pm)[b#0] Let GF(p Z) be the minimum subfield which contains all the solutions of (2.6) therefore GF(p) and # m. This implies . [pm. GF(ppm) But since equation (2.6) has p and by Theorem 2.5 and Theorem 2.6, Therefore, pm C g c d(m,) m. If a is a solution of (2.6) in GF(p n) where i. coefficient a then s divides m and a LEMMA 2.8. m solutions in GF(p ) Hence, has an index s with Y. ABBAS AND J. J. LIANG 380 Theorem 1 4 PROOF P e Let m_ SIR. implies s + as + 6 for r Hence a b. E Z SORe and 1 therefore P r-i r-2 + a a P P + + Assumes a # 1 and r. s m rs a r ar-i + ar-2+ ...+i) + , 0, but thls 0 which implies b 1 0 of equation (2.6) contradicts the condition b there In GF(p n), for every m divides n and p divides pm m GF(pm). A and A A that such 6 A exist (p-l)p elements e where COROLLARY 2.3 Zp* m PROOF. Equation (2.6) has p solutions over GF(pn) for fixed b and there are p-i different values for b. In COROLLARY 2.4. GF(pn), index m with coefficient a PROOF. This is a direct consequence of Lemma 2.8. is a solution of (2.6) and PROOF. then (2.6) has p Since p has p --_m (b), s hence s solutions. Let 1 where lm and p {Sl, s2, solutions of (2.6) such that 1 _< n m i x then a satisies ps x < sj m Sl + p s2 + P < P . + has (2.7) i equation (2.7) By Theorem 2.7 and Theorem 2.8 st} s Also by Lemma 2.8, if solutions. siR, be the set of all the indices of the < R for every I < j + p Since s _< we have is the greatest integer less than or equal to where where then there exists an has an index m with coefficient i. an index s with coefficient a s __n In equation (2.6) if p divides THEOREM 2.10. where 6 if m is a prime, then all solutions of (2.6) have i. s < p + p2 + p p-1 But since P k+l p-I p < p2k p for k > i and p k+l p-I p R p-i P < p 2k + i we have p<p and this is a contradiction. COROLLARY 2.5. In GF(pn), for every m divides n, and p divides I. exists a fundamental set with index m and coefficient a 3. n there THE TOTAL NUMBER OF FUNDAMENTAL CLASSES IN SOME FINITE FIELDS. In this section we will investigate the total number of fundamental classes for some special finite fields From the previous study we conclude that If p there exists a fundamental set with index m and coefficient 1 in GF(p n). In GF(pn), for every m divides n, and every a exists a fundamental set with index n c g p-l) # i. R Z* P and coefficient a. a # 1 where Since a m n m a 1 there 1 then the d( If A is a fundamental set with index m then A has p(p- l)m elements. FUNDAMENTAL SETS OVER A FINITE FIELD 381 In this section we will use the following notations: Number of fundamental classes in GF(pn). (a) 0(p n) Number of fundamental classes in GF(pn) but not in any proper n) (b) A(p GF(pn). subfield of Number of elements in GF(pn) with index m and coefficient a GF(pn). and none of these elements belonging to any proper subfield of m p {x ax + a Z 1}. Z and 0 if a E(m,a) %(pn,m,a) (c) . (d) (e) . " p P SE(m,a,n) is the set of all solutions of the equations of E(m,a) in GF(pn) but not in We shall investigate in the following the number of fundamental classes in GF(p q ), where q p-l, and q @ p. LEMMA 3.1. PROOF. If qp- l,then q t+l pq divides t [q-l] 0,i,2, By Fermat Theorem the len,a We will prove this 1emma by induction. O. is true for t Assume it is true for t pq [q-l]_ (pq s [q-l] s then (pqS[q_l]) i [ 1) i for every t s (pq [q-l]) 1 s q-i q s-2 s + (pq [q-l]) + + eve J 0,I I] (pq [q-1]) [pq [n-ll, 11 ]--1 pq Since s s [q-l] i rood q, then s (pq [q-l]) q E i (mod q) for q s S (pq [q-l]) So, q s -= (mod q) 0 (mod q) which implies q s+2 divides j=l [pq s+l [q_l] 1]. LEMMA 3.2. t+l where qp-1 # p, we have and q t t A(p In GF(p q t+l) Pq [pq [q-l]_l] t+l p(p-l)q PROOF. Since g d(q,p-l) c l,then for every a 6 , Zp, where a s @ I, a q I for every s O, therefore equation (2.6) and equation (2.7) have no solutions in t+l t+l q q h but not in for every m GF(p h > 0. Hence every element in GF(p q t+l t q t+l but not which mplies that every fundamental set in GF(p GF(p q has ndex q t t+l q So: in GF(p also has index q A(pt+l From Lemma 3.1 and g COROLLARY 3.1 t pq t+l p_q t pq.t [_pq_ [q-if_ t+l p (p-l)q c d(p-l,q) In GF(p q) where O(p q) i p (p-l)q 11 t+l 1 we have A(p gp-l, +---P---= p(p- l)q t+l) is an integer. p we have: q i p(pq-l_ + p (p-l)q I) (3.1) 382 Y. ABBAS AND J. J. LIANG Since O(p q t+l O(p q t t+l + A(p q we conclude that: -1 O(p q 1 A(p q + t+l t--O pqt [pq -i i + [q-l] P(P-l)qt+l t=0 where t qp-i, q # p. s We shall now study the number of fundamental classes in GF(p p ). By Theorem 2.7, Theorem 2.8 and Theorem 2.9, the equation x pp t + x where b t+l p GF(p b Z , and t < s has P ). LEMMA 3.3. PROOF. Let k < t and a pp pP (3.2) t solutions and all solutions are included in All the solutions of (3.2) have an index p be a solution of (3.2) and has index m. We have m k + c for some c t So x(PP t+l p For a fixed b E Z PROOF. k for some P (3.2). t+l p COROLLARY 3.2 In GF(p pP p t not a solution of with index t , t I) , P but not in GF(p p COROLLARY 3.3. equation (3.2) has If e E GF(p t+l p t pp t (p-l) elements in Z p pP ), there are (p-l)p p elements t solutions and we have (p-l) GF(p t) then but has index p t+l or p t where t > I. COROLLARY 3.4. For t > i; A’PP( t+l) (p_l)p p pp t+l_pp t _(p_l)pP t + p(p-l)p pp p t+l pp In GF(pP), where b 6 Z , P the equation x p t + p(p-l)p pp t -t-i t+l ppt (P-I)-I-I (p-l) t -t-I [pP (P-I)-I_I ] (p-l) x + + i b LEMMA 3.4. GF(pP)-z P wlth index p. O(p p) i (3.3) (3.4) has p solutions and each solution has index p. p(p-l) elements in t Therefore there are This implies: pP-p(p-l)-p_ + p (p-l) + p(p-l)p (3.5) FUNDAMENTAL SETS OVER A FINITE FIELD COROLLARY 3.5. s-i O(p p + 2 + A(p p 383 t+l (3.6) tl s In what follows we will study the number of fundamental classes in GF(p p "q qp-I, where Since s p p and qp-i s, I. > eve then for m divides p s -q and eve E Zp a with a i, we "q m have a a q i. # 1 and every s implies that SE(m, a, p q So, Theorem 1.3 m lp s ) for every q By Theorem 2.7, Theorem 2.8 and Theorem 2.9, we conclude the following: LEMMA 3.5. For every t and h; 0 < t < s-i, 0 < h < SE(ptq h, I, pSq) has t (p-l)p GF(p p , h p "q and SE(ptq h, LEMMA 3.6. SE(ptqh . elements, all of them are contained in the minimum subfield t+l h q PROOF. is included in Let u q SE which implies u t GF(p p "q h For every t, h such that 0 < t < s-i, 0 < h < -i, pSq4) 1 pSq)N i, (ptqh, pp t "q h + I I, SE(ptqh+l 1 pSq) ptqh pSq) Then u p + [q u ], where [x] a + [q- ] Since , P’ is the least nonnegative residue SE(p t q h+l i, p Sq). It is clear that if u SE(pt q 1, p Sq) ,then u has an index p t some 0 < r < h. Since, SE(p i, pSq A) C SE(p i pSq) C C SE of x mod p. q Z 6 for some 0 then h, t, tq, q r for (ptqh+l, i p Sq) we conclude that X(P p t+l h+l "q (p_l)pP Therefore, in GF(.p p t+l I) t "q hi (p-l)p p "q h+l t h t pp qh [q-l] (3.7) h+l q there are h t t t h+l p q h t (p_l)pP "q [pP "q [q-l]_l ]= P(P-I)pt’q h+l ppt_ h "q [pp p t t+l h .q [q-l] "q (3.8) h+l pt.q h+l fundamental classes with index From Lemma 3.1 and the fact that t p t + i, the number given in (3.8) is an integer. t+l h+l t h+l "q Since any proper subfield of GF(p p is a subfield of GF(p p "q or GF(p p t+l index p "q h t+l pp So, in GF(p p ). "q t+l h+l .q is equal h+l _pp t .q t+l "q h+l ), the number of fundamental classes t6 h+l h t t h h _pp.t+l .q +pp .q... _(p_l)pP .q.. p(p-l)p pP t+l .q h [pp with t+l .q h t+l .q [q-l]_l]_pp p (p-l) pt+l Therefore, we have the following t [pp-q h [q-l] h+l t -q q h t [pp-q h+l h [q-l] -1]- I:’ (3.9) 384 Y. ABBAS AND J. J. LIANG s In GF(p p "q ), THEOREM 3.11. (PP t+l "q n+l (3.8) + (3.9). From it follows that O(pP t+l "q h+l t+l (pP t + O(P p "q "q h+l + O(p p h+l t+l t O(P p "q "q h h (3.10) We now study the general case, i.e. the number of fundamental classes in First, let n i 1,2 ql r qr 2 q2 i such that i giP-I # 0 and qi # p GF(pn). for every r. Let N(s,p) be the number of elements in GF(p s) but not in any proper subfield of GF (pS). THEOREM 3.12.. (i)q j The number N(s,p) i-j=s (See [2]). Since for every m divides n and every a E Z where is the Mblus functions. . , the set SE(m,a,n) It EGF (pk) where GF (pk) is a subfield of GF(p n) and doesn’t belong to any proper subfield of GF(p then = has an index So, we have the following implies that if P k), THEOREM 3.13. If GF(p k) pk. C GF(p n) then A(p N(k,p) k) p(p-l)p Now let n qi2P-I k p ql for every i LEMMA 3 7 i q2 2 qs s p tl t ql t2 t i’ Since i LEMMA 3.8. for i 1,2 then SE (mqrl GF(ppm), p ,s and for some t LEMMA 3.9. If COROLLARY 3.6. ,l,n) # pk r t t "ql t r SE(mqrl,l,n) SE(mqr # 0 and . . t 1 < qs s where 0 < t < k-I, 0 < t r -i, SE(m,l,n) divides m, and m divides n then If i (p-l)p elements, all of them are SE(m,l,n) N GF(pm) we have SE(m,l,n) SE(m,l,pm). For every m SE(mqr,l,n). " where 0 < t < k-l, pro) SE(m,l,n)C p’ t qs s m q2 _< i _< 1,2,...,s then SE(m,l,n) has contained in the minimum subfield GF(p and 0 # O, qi # such that k 1,2,...,s. For every m k # and i is a proper subset of If SE(mqr2,1;n) 1 n) 2 pt SE(m,l,n) f SE(r,l,n) divides m but not r, then The proof is a direct application of Lemma 3.7. COROLLARY 3.8. p If for some 0 < t < k-i ptlm and p t+lr then SE(m,l,n) where (re,r)= g- c SE(.r,l,n) i SE(m,l,n) for some r # r # I 2 The proof is an immediate application of Lemma 3.7. COROLLARY 3.7. < t Ir but pt+l 2m and SE((m,r),l,n) d(m,r). The proof is a direct application of Lemma 3.7, Corollary 3.6 and . FUNDAMENTAL SETS OVER A FINITE FIELD 3.8. Corollary We know from Theorem 2.10 SE(m,l,n) such that a SE(m,l,n) 0 _< _< t 385 Let m k- i, t # 0 Ir p tll t12 t .ql qtlh "q12 I l,...,h and h < s. for r factorization of n such that qr qlr @ then there is an We want to find the number of elements in a has an index m. with index m. SE(m,l,n) if m divides n and for r and m divides n where q’s If we rearrange the 1,2,. in the ,1,n) ,h, then SE( h for every i i=iN 1,2,...,h, and 1,n)qi, SE(ql q2m’’’qh SE(--" l, n). We shall use the following notations for the remaining of this section: m p t t t i "ql q2 t 2 qh h 1 < t. h h and m m I qi i=l 2 qiqj i<j h r qil qi2 .= 1 < i qi i I < i 2 r r || || qi j=l j =l j q. for i If R[SE(m,I,n)] R[SE(m,I,n)] O. Therefore m r h we conclude the following. Also, if B m or pm. is the number of elements in m m and + If r > h then we define m From previous Lemmas, h r r + I m m 2 3 p tl-i t2-1 t SE(m,l,n) qh with index m then + (-i) hmh + (-l)rm + + th-1 "q2 "ql r GF(p pm) and not im any proper subfield of Hence we have the following. GF(ppm), then (3.11) B has index THEOREM 3.]4. &(ppm) R[Sm(m,l,pm)] (pm,p)-R[Sm(m,l,pm)] p(p-l)m p (p-l)pm s To determine the number of equivalence classes in GF(p q to study SE(m,a,q SE(M,l,q s) , a Z p if a s for all m dividing q for every q S/m mlq s. i then SE(m,a a’s in Z Z* P and a qS) has p(pm-l) can one find? order q v PROOF. b k In Z p’ if qV Z P mlq s we need and every a # I, One question we will try a # I such that a qs/m I E(m,a,q ). divides (p-l) then there are q (q-l) elements of Let b be primitive element in Z Assume for some k; p q i. This implies k-q 0 mod(p-l). So, k t- is a solution x qlp-l, Another question is for fixed m and P a, how many m’ s, a’ s are there such that SE(m’,a’,q LEMMA 3.10. . elements m, is there an a where From Theorem 2.7 we have Also we know that for every to answer first is that for given and then how many such s (3 12) k P-__!_ q p I, for some 386 Y. ABBAS AND J. J. LIANG p-1 for every But also if r 1,2,...,q-I we have q r and I b # 1 which implies that in Z we have (q-l) elements of P order q. Let p i q h where the g c d(p-l,h) I, then for every r such that r 0 mod(q -v h) and 0 < r < qV we have b r is a solution of x I, which implies that in Z there are qV- I elements such that x qv i and x For v 1 I, P v 2 we will have q 2 I- (q-l) q(q-1) elements of order q 2 The same for v t I, v t we will have i elements of order 3.11. LEMMA In Z if for a fixed b where O(b) and i < p < there exists p an a of order qV where _> v > and a satisfies x qv- b (3.13) then there are q v- distinct elements in Z * of order qV satisfying (3.]3) P PROOF. Note that equation (3.13) has no repeated roots. If a is a solution, we 1,2,. t -(br) q ,q-l. ’p-I" b qv qt claim that a,a qp+l a (qt-l-l) 2q+l (qV--l)’q-i a prove our claim, first since O(a) distinct c,l,.nrs in Z g d(rqU+ l,q) c r Also (a P 1 and O(a) THEOREM 3.]5. q If b f Z v q v then we ,ch that are distinct solutions of (3.13). To riq+l a q+l qV- qt. qt-l(q-l) qP r-q a aqV- .ve qP- b i O(a rq+l) O(b) q v-U_l are 1,2 for r. v q 0 < v < b. Since v , t|n for every v, P v v-D e]rlnts there ord,r are q _> of a and satisfying (3.]3). v-i v PROOF. In Z we have q (q-l) el,ments of order q Let c e z and 0(c) p P p vq q and let c d so d # 1 and d i. Furthermore, the order of d is q By , v - Lemma 3.11, _q v-i q we have q q-l(q-l) v- elements satisfying x qv-v q distinct d’s of order PROOF. If q divides p-i, then q t v But there are d for each d. for which the above equation is solvable and that is exactly the total number of elements in Z LEMMA 3.12. q divides pqt -i P having order qV. for every t > 0. By induction. In GF(p q) where q divides p-l, the set SE(l,a,q) # if and only if O(a) q. q, the set (q-l) elements of order q and for a fixed b with O(b) SE(I b,q) has p(p-l) elements, therefore GF(p q) has p(p-l)(q-l) (q-l) fundamental p (p-l) Since there are classes with index i. So we conclude: O(p q) Since P -= (q-l) + pq-p-p(p-l)(q-l)+ (q-l) + (--pq-l-l) (q-l) + pq-2 +pq-3 + p (p-l) -q I + (p-l)q +1+1 (3.14) q I sod q. q i + >pq-r 1 + (q-l) sod q 0 sod q. r=2 So, (3.14) is an iteger. THEOREM 3.16. In GF(p q where q divides (p-l), if O(a) t+v t, where 0 < t < s-v we have SE( v+t=l t SE(qt ,a,q Sm(q a,q v+t) qt ,a, qS) SE( qt ,a,q # q and V then for every 387 FUNDAMENTAL SETS OVER A FINITE FIELD By Corollary 2.1 we have PROOF. SE ,a, Theorem 2.4 and Corollary q solution a n with index m set of this equation is H (D GF(p q H will imply that X t+v-i 2.2 t {s GF(Pq n and + O pq t aX + b; b GF(p q t and it is clear that It is possible that in some cases This theorem is not true in general. the minimum subfield that contains all the solutions of x pm subfield which contains some of thse solutions. s q From ’Fheorem 3. conclude that in CF(p ), w]ere t subset of SE(q ,a,q s) if and only if v > qV’ O(a) q v t t’ + v’ t + v and + has a proper qlp-l, SE(q t’ t-t’ q b ,b,q s) is a a, wlere v Z * with order q where p Also for a fixed a s-v. ax O, Theorem 3.15 insures the existence of exactly q elements "b" in Z such P v+l q v q and b a. Hence, for a fixed a with order q and fixed t" that 0(b) < s-v there are q elements b in Z such that: p SE(qt-i ,b,q s) Zp Therefore, if we start with a 0 a Z where O(a i)__ a n SE(qt- (i+l) and has a P i-a NOTE: O(b) Z By Theorem 2.1 the solution ). / # ,a,q qV+n ai+ I, q s) >_ SE(q t-i (ai+l)q and ,ai, q s qV ao So, we have a.. for every i then there are 0,1,2 ,n-1 where t-n (p-i). Let p-1 every t q p such that 0( v+i s t SE(q ,a,q ). g q h where g i such that O(a) qV , d(h,q) c Then for every a i. i < v < E and t + SE(qt,a,qS), In elements with index q t (pq there are exactly and coefficient a. From this lemma, we conclude that in GF(p q t (Pq -l)-(Pq P a # 0 i and v < s we have the following Lemma. LEMMA 3.13. Z >_ t -1)p-(p q t-I -1)p- g t+v there are t-i -i) (p-l) -q q t qV-l(q_l (3.15) fundamental classes and each class has an index q t and a coefficient with order qv Lemma 3.12 insures that (3.15) has an integer value. We will use the notation: F(m,t,v) (3.15), where m It is clear now that in GF(p q (pq s t + v, 1 < v < and t > i s where s > i, we have -pq s-i )-p (pq s-i-i) (q-l) p(p-l)q s (3.16) Y. ABBAS AND J. J. LIANG 388 fundamental classes and each class has index q Therefore, for s > 2, O(q s) O(q s-l) + (3.16) + 2 F(s s-v v) + -Pq 6 ( s-6-1(q (3.17) p (p-l)q v=l If with coefficient 1. > 2 we conclude that: s-6-1 where s-6-1 S min{-l, s-l} and 6 max{O,s-}. 1 then we have: s-i 0(q s) 0(q s-I + (3 16) + p(pq -i) (q-l) p(p-1)q s-i REFERENCES i. LIANG, Joseph J., On the solutions of trinomial equations over a Finite Field. Bull. Calcutta Math. Soc. 70(1978), no. 6, pp. 379-382. 2. LONG, Andrew F., Factorization of irreducible Polynomials over a Finite Field with the substitution xq-x for X. Acta Arith. XXV 1973, pp. 65-80. 3. BERT, A.A., Fundalnental Concets of Highher lgebra. Phoenix Science Series, The University of Chicago Press, 1956. 4. ALAHEN, J.D. and KNUTH, Donald E., Tables of Finite Fields. Sankhya, The Indian Journal of Statistics, Series A, Vol. 26, Dec. 1964, Part 4, pp. 305-328.