THE COMPOSITIONAL INVERSE OF A CLASS OF OVER A

ROBERT S. COULTER AND MARIE HENDERSON

Abstract. A new class of bilinear permutation polynomials was recently identified. In this note we determine the class of permu- tation polynomials which represents the functional inverse of the bilinear class.

1. Introduction and Main Result e Throughout Fq denotes the finite field of q = p elements for some prime p and positive integer e with Fq [X] representing the of poly- nomials in the indeterminate X over Fq . For polynomials f, g ∈ Fq [X], we write f ◦ g = f(g(X)) for the functional composition of f with g. A permutation over Fq is a polynomial which, under eval- uation, induces a permutation of the elements of Fq . Clearly, permu- tation polynomials are the only polynomials which have a (functional) inverse with respect to composition, id est for a permutation polyno- −1 mial f ∈ Fq [X] there exists (a unique) f ∈ Fq [X] of degree less than q such that f(f −1(X)) ≡ f −1(f(X)) ≡ X mod (Xq − X). We call f −1 the compositional inverse of f (or vice versa). The problem of discovering new classes of permutation polynomials is non-trivial and has generated much interest, see the surveys and open problems given in [3, 4, 6]. Discovering classes where the inverse polynomials can also be described seems to be even more difficult: there are very few known classes of permutation polynomials for which their compositional inverses are also known. To the authors knowledge, the classes with explicit formulae for inverses are:

(1) The linear polynomials: X + a where a ∈ Fq is trivially a permutation polynomial of Fq with the inverse polynomial being X − a.

1991 Mathematics Subject Classification. 11T06. ∗THIS RESEARCH WAS PARTIALLY SUPPORTED BY THE AUS- TRALIAN RESEARCH COUNCIL 1 Bull. Austral. Math. Soc. 65 (2002), 521-526 Corollary The [1 F The and Set 2 W example form [1 inte class Theorem is inte F or Q ]. ]. B e The Recen a . (3) (2) α ger ger or note F any ulas ( p Here p of rom X em olynomial ermutation p The The tation Chapter in see X only i and ) p tly olynomials α that linearised satisfying = v erm for n 1, Theorem erse w , ∈ [5, 1. is X k 2. monomials: i set e Dic ar a if =1 − the F utation there A ob 1 giv new p e ( L Chapter q F n, of α kson C olynomial Q et \ viously p g or 3]. ( in e i ermutation X α q { T D = p q v class − 0 and a T are r olynomial = 1 ) α R. erses 1 f n , = f ( In q r p 1) α = ( 1 X p description ≤ α ( n w A X ∈ olynomials COUL ( } X olynomials. 2 . classes X C ) sub-linearised 3]. e suc , w C = α i the , k 2 of X F Denote ) i the k a ere ) ≤ ha k are + − for q 1. = − = o ) h n = p 1 TER v 1 v \ k − B monomial is erm is ( er X e cases, p α In sho 2 p X kno X { − α some olynomial k for olynomials over 2 a the D 0 − k F + T 2 is suc AND , 1 1 − utation nk wn p m the − q r wn. 1 , 1 for whic of X erm i ( +2 − } ( the if follo define X h   X α 1 , for inte F q k the to tr M. + and cases, ) − ( the , + the Q + utation n p h 1 ac a c + X j − X . − wing α omp a olynomials) b ) ger HENDERSON =1 1 . in i 1) 1st p e 2 − e ( . m p where only ∈ k comp / F α olynomials . v over 1 − mapping olynomials 2 p or + erses the ( where 1 ositional k + + { kind: erm X − ob . 0 X 1 p 1 1) , α T T if L vious olynomial ositional F comp .  q r utation r et X nm n can Q ( ( as ( 1 − n, X X D nm . 2 } 1 n . , but fr ) ) n q ab b ≡ ositional 2 inverse corollary the b ( 2 w + om k e g X e ≡ ove − − as α 1 determined X 1 that an , p , ) in 1 1) a F comp mo o olynomials in 2 as ) mo v ) v Q and o 2 tro erse er = is 2 dd d no of to j define . d k in a 1, F ( − duced ositional q ( for v p f 2 explicit F q p − 2 q of ositive erse α see q erm if i − −   over d by this and (for any 1). 1), [5, . u- in in in of Bull. Austral. Math. Soc. 65 (2002), 521-526 (iv) tional T mo ing ily prop (v) prop (ii) Pr Prop we The wher (1) Prop f r(T α Our o ( (iii) (iv) dulo (ii) have in of. g This statemen Squaring (i) Using e ositions, ositions. α r( pro T osition osition ( terms c in x = THE X atten (i) r( T A T C α )) of ( v )) r r X A is i X α 2 2 = erse ( ( Squaring ( = α of B A = the X INVERSE Q mo immediate + ( ( tion α α α X ts of T + ( ) the basically ) ) A c W of (2 α 4. 3. d r( definition ≡ )) α ≡ ≡ of α X the k (2  e x S ( − ( f from k X = ≡ ≡ follo Using C F C X 1 X ). ) α 0 i α b − Theorem − ( or 2. . k i i 2 2 egin 2 is X Q mo 2 ) 1) t − C T α C − k OF nk This w Our − = 1) 1 r (2 + ) giv k i wing C A The + 1 o − ( ( this − as d k − = w in X X + 1 k 2 α α − A X 1 the 1 1) b − T p es (T e , ( 1 v T 2 of ) y pro ( 1) T 1 X olynomials − B r + ) CLASS iden k olving n 2 / ( obtain r ( p +2 X 1) r( r k j − result Pr C X collecting ( α α ( / X Q 1, same A − =1 ( α oin α X 1) T ( X X of 1 2 k k 2 ∈ α − α + ) − C / − r( + tit whic + oof + ) 2 2 ( 1 i T ) 2 t 1 2 k in k − F ( X X T 2 X ( − k α r( y − OF X + 1) nk 1 is 2 closer X Q − r v 1 is the notation ) 1 − ) ) X 1 ( + − olv is [ T (1 2 h + 1) . directed 1 2 X X giv and 1 of k mo PERMUT tedious nk ) r ) − 1 is , used + ( T )) + A ] es 1 iden some + X en + r = and d α mo to Theorem α examination α ( mo X ) establishing X (2 ( α (2 and α in + X tit sho 2 k man d 2 k ) d 2 as − ) k C − useful k α X to A but Q + − Theorem y: 1 ( 1 − = ( i − TION X 2 w − 2 i X B + ab 2 − 1) T X as y 1) establishing + i ) 0. Q =1 α 1 k that 2 Q r T ove seemingly X ) 2 . times 2 1 + + j r define k mo + k iden T ) ( ( POL 1 − F , X 1 X X r i X or then − of ( g and a . d ) X 1 )) ) α 1, ) 2 tities. YNOMIALS . set ( in k n , f d ) is X − whic α 2 1 in o k the Q ) ◦ the of − dd, necessary the 2 g + The k α sequen h − follo , comp X i w reduced S remain- primar- or ) e i ( em X wing ha osi- tial . ) v   1 e 3 Bull. Austral. Math. Soc. 65 (2002), 521-526 F then Next B Using ( (2) sum 4 W Pr Finally and (3) As X rom α f e o ( Q α S of. split X a ( using + i in 2 ( g +1 a put )T Prop X b α α X ( , 1 By b ( b ( X ( = ) (2 1 put r( X X 1 the ) X ( ( k = ) b A X where − X )) Prop ( ) expanding 1 osition ) b = A b C 1 ( α 2 + = ) terms − 2 = ) b X mo ( ( k α 2 ( 2 1) = X X X − then X C ( ( S ) α C X X 1 osition ) k )) d i 2 X ( = ) (2 + i ( a i − =2 α ) k = ) X k = ( ( 1 3 + = of (2 − c + T X X = ( w  ), R. α 1 X α (i), k k r( i ( ( − =1 this − ( Q ) e ( − B α α ( α X f α 1) + from 1 A COUL 1 α = (2 arriv α − + 3 α +  (2 + T α C 2 ( 1) k ( 1)( + nk (iv), k X ( α A r( − X sum X k 1) 2 − 1) X 1 + − (2 − 1) α X 1 − ))T ) X TER ) Prop e 1 − 1 ( B k )) X k 1) i α α T 1)( ◦ − B X ) =1 1) − = at α B 2 so 2 i + r( r( + + + 1 g α − 2 k ( )T + − α X α ( X ( AND X 1) C A α that osition X 1 α α ( ( 1 1 r( i α X X 2 2 − α ). ) X + k (2 i + nk T ). =1 2 ( − 1 (2 A k k ). ) X T r( − 1 1 Using − − M. k 1) α f Then − 1 r( (with 1 i  C X )) ( − Iden T α i − X + T − X HENDERSON i ( 1) r( 1 3 ) T + X 1) r( (2 )) )) T  r( and X (i) ) X k ) tically ( T r( / Prop + − X ◦ α ) g mo ( r( ) 2 2 g X α α ( w (2 k k ) + α as α ( X − − (2 e )) k ( X d + 1 1 + − X k 1)( osition − ) ha c − + ( 2 ) (2 i mo 1 α 1) X ) k ) 2 ) k S − = v A k α = = − − A i Q e i 2 and d ) +1 α 2 (2 2 1 S A α 2 − k ( + a c k ( − i ( i X ( − α α ( X ( ) 3 i X X X 1 S α X X ) ( ) re-writing − S X (ii) Q i (2 ) ) +1 ) 1) + ) ) i . ( + k and . . ) + T − X ( B + b 1 and X r( X − ( ) α 2 X . X B 1) ) ( b ) . X ( . ) α + )) X (iii), ( mo )) X α ) the . − = )) d 1 Bull. Austral. Math. Soc. 65 (2002), 521-526 So F b and The Prop Pr f Pr W it ( α or X e ( follo o o from g of. of then ) β α it result ( S X i ( = = = β osition =1 of ∈ k β i THE ws follo As ( X X β b c X k i F Equations The ha =2 T 1 α − ) X q ( )) directly 2 r( 1 T X X , i nk ws no v INVERSE =1 ) c k r( β from = e, or − α ) = w X β 1 T 5. + T em using β X + r( ( ) r( follo n f (2 b X j − X ) F α T X 2 Equation =1 k 1: that 1) ( ( or = r( − ) X g 2 ) / ws (2 OF 2 α Prop (2 2 β k β F and ) β k ( − β k X − or X T − i 2 f from A ) ∈ 2 r 2 2 ) β α )) k j ( k 2 x − osition k CLASS ( 2 − 3 X F k − i g k − i as ) ∈ q − i ) 1 α w 1 S ), ( S i ( calculating X then S + e i F i x required. ( it ( i X Q )) ha T X ( T X OF , is r( 4 ) r ) v = if ( f . ) + X and e simple β α PERMUT = = T x X ) ( c r( . g α + β X ) α i these T x Supp =1 k ( = X X r( ) i the β =1 k = to X β X 2 β ) 2 A ( T )) 2 ) k ose 0 sum iden X see TION 2 (2 T r( j then k = k r( 2 X − − nk T i tities, X 2 − β ) a − k r( 1 (2 POL − f 1 ( ) X x from 1 = α (2 k + ) − ( ) k S ) g 2 YNOMIALS − β T = 1 k α that + 2 2 − ( r( ( k k X i X − Prop 1 − ) b X S i i ( ) ) )) S for X i S ) for + ( i 2 . i X ( ). k ( X osition − c X x ) β α 1 . ) ) S ) ∈ . . ∈ k ( F F X Q   q 4 5 ) , . Bull. Austral. Math. Soc. 65 (2002), 521-526 [6] [5] [4] [3] 6 Using [2] As elemen X f where using [1] The m fields. α utation ). ( G.L. R. R. W. A. and T graphs field?, the language cations amongst pro g w ec α determination e ( f Lidl, Lidl ceedings hnical, Blokh the Bosma, elements y Prop t α a β ha This Mullen, )) ( y survey II g v , ∈ and (D. p MA α and = ∈ , , e When G.L. uis, the ( olynomial osition Amer. J. F x assumed Essex, result F β J. Jungnic Surv )) q GMA of Sym Q G.L. f of , Demb Permutation of R.S. Cannon, Mullen, α = = = = do and satisfying the ( the r eys Math. g b e es x x x x England, c Mullen, α olic underlines 4 owski-Ostr ent Coulter, Fifth R. ( k field? of + + + + algebra a in x el T that classes )) p COUL Comput. r( c c c c and r and and the olynomial Pure Mon α α α α esults x ,    (1 = In ) Amer. p When x x x T G. ternational C. T 1993. H. olynomials: in + thly TER β 2 2 2 = M. and r( r nk nk nk , pac om References x v ( Pla are T Finite Niederreiter, T y − − − x that, erse 1 24 urn Henderson, ) = r 1 1 1 ) 100 Appl. k Math. do AND p ( over y 6= for + + + age olynomials x = oust, not (1997), w y es )) 1 1 1 ald, . 0 class Fields (1993), 1 in . some a a + + + M. Conference can [2] Th Math., simple a then Mon The finite p general, Dickson X ( ( olynomial matrix i HENDERSON n n us =1 to k 235–265. X X j j − − giv eds.), b Appl. =1 =1 thly and Magma 1) 1) , 71–74. e x ( again generate f field / / n Finite v en written α X 2 2 j − to ol. ∈ =1 ( analo X x 95 C.M. 1) i g 2001, =0 1 Polynomials k 2 on in / p in α F describ 65, 2 2 ermute over (1995), j ( (1988), algebr Q ( ( k f v X Fields x x Finite gue this erses α . + Longman O’Keefe, pp. ( )) + + in examples g By a x of α a x x ≡ 2 e. finite the ( article 37–42. the 242–258. (2 system 243–246. 2 2 x and schur’s Fields for ) ) j Prop X )) , − 2 2 2 2 elements 1) Pitman form j j k k Scien k = kno mo Applications: field − −  1 1 I: osition and x − − relied for c d wn i i onje . The tific   y p ( ermute Ev Mono- Appli- X small = of ctur p user q and ery β er- the on  5, + x e Bull. Austral. Math. Soc. 65 (2002), 521-526 Technology, land, Computer Inf E-mail E-mail Centre orma St. THE addr addr Lucia, tion Science f INVERSE or ess ess GPO Discrete : : Security [email protected] [email protected] Queensland, and Bo OF x Electrical A 2434, Ma Resear CLASS thema Brisbane, 4072, ch OF u. du tics Engineering, au .au Centre, A PERMUT ustralia and Queensland, Computing, A Queensland TION The POL University 4001, YNOMIALS Dep University A ar ustralia tment of Queens- of of 7