Generalized Kronecker and permanent deltas, their spinor and tensor equivalents — Reference Formulae R L Agacy L 1 42 Brighton Street, Gulliver, Townsville, QLD 4812, AUSTRALIA Abstract T h e work is essentially divided into two parts. T h e purpose of t h e first part, applicable t o n-dimensional space, is t o (i) introduce a generalized permanent delta (gpd), a symmetrizer, on a n equal footing with t h e generalized Kronecker delta (gKd), (ii) derive combinatorial formulae for each separately, and then in combination, which then leads us t o (iii) provide a comprehensive listing of these formulae. For t h e second part, applicable t o spinors in t h e mathematical language of General Relativity, t h e purpose is t o (i) provide formulae for combined g K d / g p d spinors, (ii) obtain and tabulate spinor equivalents of gKd and gpd tensors and (iii) derive and exhibit tensor equivalents of gKd and gpd spinors. 1 Introduction T h e generalized Kronecker delta (gKd) is well known as an alternating function or antisymmetrizer, eg S^Xabc = d\X[def\. In contrast, although symmetric tensors are seen constantly, there does not appear t o be employment of any symmetrizer, for eg X(abc), in analogy with t h e antisymmetrizer. However, it is exactly t h e combination of both types of symmetrizers, treated equally, that give us the flexibility t o describe any type of tensor symmetry. We define such a 'permanent' symmetrizer as a generalized permanent delta (gpd) below. Our purpose is t o restore an imbalance between the gKd and gpd and provide a consolidated reference of combinatorial formulae for them, most of which is not in t h e literature. T h e work divides into two Parts: t h e first applicable t o tensors in n-dimensions, t h e second applicable t o 2-component spinors used in t h e mathematical language of General Relativity. In P a r t I, section 2 gives definitions, illustrations and tables of combinatorial formulae of t h e gKd and gpd. T h e large section 3 derives an array of formulae for b o t h a single gKd and products of such, which are collected together in Appendix A. A similarly large section 4 does the same for gpd's — the formulae here are not in t h e literature anywhere — they are tabulated in Appendix B. In section 5 simplifying formulae for combined gKd's and gpd's are derived, and presented in Appendix C. In P a r t IE, in section 5 we determine the spinor equivalents of all gKd's and gpd's u p t o 8 (ie 4 up, 4 down) indices, t h e most useful for the tensor/spinor languages of Relativity — these are listed in Appendix E, I and I I respectively. In section 6, in reverse, we derive t h e tensor equivalents of gKd and gpd spinors and sums of products of t h e m — t h e results are collated in Appendix F . Conventions are as follows. In Part I, general indices range from 1 , . . . , n . B o l d e d i n d i c e s are t o b e regarded as fixed. An index set {k} stands for {fcj,... , k } (p < n) and an index set {ki} stands for {fcn,... , &i } etc, all indices ranging over 1 , . . . , n. T h e index summation convention is understood. All index sets are permutations of each other. In P a r t II, two-component spinor indices, are in capital lower case roman. p p PART I 2 Definitions and illustrations of the gKd and gpd Complete symmetry of (a tensor's) indices, as opposed t o total antisymmetry, is manifested by all positive signs in any p-linear expression. Whereas t h e gKd's an­ tisymmetry comes about through a determinant (interchanges of rows/columns or indices changes t h e sign), total positive or pure or permanent symmetry, as we term it, comes about through the use of a permanent Then in complete analogy t o t h e gKd we introduce t h e generalized permanent delta or gpd. This is defined, like t h e gKd determinant of a matrix, except that we take all positive signs. W e use t h e kernel letter 7r t o denote a gpd a n d double vertical lines for t h e permanent of t h e defining matrix. T h e gKd and gpd are completely complementary t o each other and are defined, for p ( < n ) distinct indices, respectively by, ft 3p 7T, 31-3p 31 -3v 4 3p where t h e first is a determinant and t h e second a permanent. T h e gKd has value + 1 (-1) depending on whether ( j i , . . . , j ) is a n even (odd) permutation of ( t i , . . . ,i ). T h e g p d has t h e permanent value + 1 for any permutation of t h e index sets. Note t h a t 7r£ =<5". T h e gpd is a complete symmetrizer (permanent symmetrizer) and now one easily sees t h a t ir^Xabe = 3!X(«je/). In terms of bracketed and parenthesized notation, t h e gKd and gpd are p p ™n-3p v 3\-3p n P-^Uin 3Y P A simple b u t important interaction between t h e gKd and g p d is t h a t in any expression containing t h e m as a product, where there is a summation on a pair of indices between them, such expression vanishes. For ,.o...6.. 7T...o...b... = 6 ..a...b.. ...6...0.. ..0...6.. 7T...0...6... = 0. From this it also obviously follows t h a t t h e product of any gKd and any gpd with two or more common contracted indices vanishes. If A = [a*-] is a n n x n matrix its d e t e r m i n a n t and p e r m a n e n t are: detA=- aV n! T a'j3n 3 1 3n in' l—3n SI«1—*n T h e Riemann tensor Robed a n d Lanczos tensor Ly* are Young tableau tensors in their algebraic symmetries and obey t h e partial symmetries and antisymmetries as determined by t h e tableau representing t h e m (Agacy ). And so here t h e interplay of symbols for both permanent symmetries (ir) a n d antisymmetries (6) manifests itself in t h e one-line definition of each of t h e Riemann and Lanczos tensors in respect of their algebraic symmetries 1 1 12 1 — o ab cd V v n qs "fpk Limn- ^-efgh 3 3.1 Generalized Kronecker 8 Single generalized Kronecker 6 In this section many of t h e derived formulae have been presented (but without deriva­ tion) long ago in Veblen . In t h e following definitions of numerical symbols t h e range of indices is from 1 t o n . I n any set of p indices . . . , i } , it will b e understood t h a t for p < n t h e indices are all distinct T h e generalized Kronecker 6 (gKd) may b e defined by 2 p % • • • 6 i n-3p T h e interchange of any two superscripts (or subscripts) interchanges the correspond­ ing rows (or columns) of t h e determinant thus changing its sign. Additionally if any particular superscript (or subscript) is not contained in t h e set of subscripts (or su­ perscripts) then t h a t particular row (or column) consists of zeros, t h e determinant vanishing. Hence only if t h e set of superscripts a n d subscripts are permutations of each other does one obtain a non-zero result. T h e gKd can then equally well b e described a s +1 —1 0 if i\,. •. , i is a n even permutation of j \ , . . . , j if i\,... , i is an odd permutation of j % , . . . ,j if t i , . . . , i is not a permutation of j i , . . . , j P p p p p p Unless specifically stated or obvious from t h e context we will in future take it t o be t h e case t h a t t h e sets of superscripted a n d subscripted indices are permutations of each other. Note t h a t t h e indices i , j range from 1 t o n b u t t h a t there are p (< n) indices. W h e n p = n ( = dim V ) , and only then, we will also write P s ^l.-.tn — °ii...in a n a € "~ ° 1 n and refer t o e a s t h e alternating or permutation symbol (and in General Rela­ tivity (GR) t h e Levi-Civita symbol as well). I t takes t h e value + 1 when t h e set { » i , . . . ,i } is a n even permutation of { 1 , . . . , n } a n d —1 when ,i } is a n odd permutation of { 1 , . . . , n } . T h e gKd is a tensor (transforms like a tensor) n n 1 n whereas and e* •* are not We modify Einstein's summation convention by indicating fixed, repeated indices by b o l d t y p e so t h a t there is no summation over these indices, which will occur as matching superscripts and subscripts. This notation makes clearer the derivation of most of t h e following formulae. Unbolding or unfixing means summations go ahead. In considering a product of t h e two gKd's 0 h-3p °ki...k p there is no summation over t h e k ' s . For brevity we sometimes denote a set of summed, repeated indices {ki,... , k } by {k} and a set of fixed, repeated indices by { k } . Clearly for a non-zero result above t h e sets {i}, {j} and {k} must all be permutations of each other. As t h e set {k} is a fixed permutation of the set {i} and t h e set {j} a permutation of {k} then t h e resultant value of t h e product is just t h e value of a generalized Kronecker 6 as t h e product of those permutations. Our permutation multiplication convention is from right t o left p /ki...k \ (i ...i \ _ p x v /ii...i \ p V i i \ k i . . . k j \ji...jp) Thus k i . . . k , cti...tp cki..Jtp _ °h-ip °k!...kp - cti-tp 'ii... j - c c P In particular we have a result used often Formula k l . 0 il-in - °l-..n °jl...jp - 6 e 3i-3n- Formula k 2 . r ffi-h-ih 3l-jp-l3p _ u ( Z-/^ I \P+s fii. eh -i—ip ' 3p 3\-3p-\ v v 8=1 where a caret ( i ) over an index indicates its omission from t h e list. Imagine t h e gKd as a p x p matrix with t h e indices . . . ,i signifying rows and t h e indices j i , . . . , j signifying columns. Expand t h e g K d by t h e last column j . Note t h a t in t h e determinantal expansion, t h e sign associated with t h e element in row 1 a n d column p is ( — l ) , and so on. One then has p p p p+1 CfcS = ( - i r ^ i & t . 2 + (-ir ^CA+• • • + ( - i ) and so we have t h e result p r«i...ip_i»p 3\-3p-\3p i v-^ / — 1 V + * tf? X«i- «-H> Z~J^ ' 3p 3\-3p-\ 8=1 p + p ^::fcl Now consider t h e expansion of t h e gKd ^ " ^ j ^ Formula 2). W e get by its last (p 4- l ) t h row (see fcSK - ( - i r ^ c t * + ( - i r ^ ^ ^ + • • •+(-ij^&itt 3 1 where a caret ( j ) over a n index means t h a t it is excluded from t h a t set. There are p + 1 terms in t h e sum. Then if we let j =i so t h a t there is a summation, we get t h e single common index formula for a gKd p + i p + i Using this result, now for two repeated indices, gives fcttStt - (— * - UfcttS - (« - p - 1 ) ( « - p)t.%• Continuing t o a d d repeated summed indices leads t o t h e most general result for a single g K d Formula k3 *i!:::tfev:<: - ( » - « + 1 ) •••(» - p ) # : J = o < P < 9 < „ which clearly gives t h e correct result for p = q. This formula is equation (8.1) of Veblen . A particular case of Formula 3, when q = n, is Formula k 4 2 c*l...*pip+i...i,» .M...i i +i...i , = p p .. n£ . — Another particular case of Formula 3 , when p — 0, should b e noted Formula k5 2 This formula is equation (8.2) of Veblen . A useful specialization, when q = 1, is t h e simple result Formula k6 $ = n. Another specialization of Formula 3, when p = 0, q = n, is Formula k7 2 This formula is equation (8.3) of Veblen . 3.2 Products of gKd's Consider t h e following product of two gKd's gil-ipip+i-U £Jp+i-jq Jl—ipjp+l—Jq kp+l...k ' q Note t h a t t h e fixed set J = { j + i , • • • , j } must take on t h e values of t h e specific set K = {kp+i,... ,k }. Unbolding t h e indices of J so t h a t there are (q — p)\ per­ mutations of it a n d hence (q — p)l summations, gives t h e key result — (q — p) is t h e number of common indices Formula k 8 p q q C*l...tptp+l...t, °h-3pjp+l~3 cjp+1-..jq _ / \i °fcp+l-.fc, ~ ~ W q ctl...tptp+l...t, P>' ° h - 3 p h + l - V 2 This formula is equation (8.6) of Veblen . Letting t h e indices k i,... ,k = i , i , and using Formula k 3 for (q — p) common indices on t h e R.H.S. gKd, results in t h e first partial common indices product formula Formula k 9 p+ lplp "3l-3p3p+l-3q "tp+l-tq + 1 q p + i q ...t, V V« V)- ji... 3pip+l—*q (n - p)\ .tp (n-q) 2 This formula is equation (8.7) of Veblen . A second formula for t h e product of two gKd's comes from unbolding two sets of indices in t h e obvious relation (a right t o left multiplication of two permutations on reversing t h e order of t h e two L.H.S. terms) ctl...ipip+1-..lq rJl--JpJp+l---Jq Jl--Jpjp+1---Jq *l...fcplp+l...iq _ _ £>l...tpi p + l...iq _ cil...ip m fcl...fcpip+l...l kl...kp q Some care is required now. Consider unbolding t h e set { i + i , . . . , i } . Each index ranges over not n, b u t (n — p) distinct values, values complementary t o t h e values of {*!,... , i } = {hi,... ,k }. There are (,!*) ways of selecting a set of values for i p + i , . . . ,i in a n y order from t h e set of (n — p) values. B u t there a r e (q — p)\ permutations of these for t h e summation. Unbolding thus introduces t h e factor (q — P)' (g-p) (nlg)i — equivalent of selecting (q — p) ordered values for ip+i,... ,i from a set of n — p values. An alternative idea is t o consider t h e product of t h e two gKd's by focussing purely on t h e summation of i i , . . . ,i with (n — p) for n, and (q — p) for q (see k5) a s = ^J^Z^lp^ t o obtain t h e factor. This gives t h e intermediate result p p q p q = q p + fii...ipip+ ...i 1 oJi..JpJp+i..J q ji..jp+iJp+i...j fci...j^,«p+i...t, q q _ q (n ^ — p)\ n _ rfi...tp °fci...* p Now unbolding t h e set { j i , . . . , j } introduces q\ permutations and we get q Formula klO i ...i i ..,i h-jpip+i f 1 u r p+1 q n ch-jpif+i-U —in *i -kpip+i ...t, 1 \ ( ~ P) - ch-ip H• ^ _qy ki ...ftp • v n A specialization of this result, when p = g, is Formula k l l Equating t h e indices {k\,... gives Formula k l 2 ,k ) = ( i i , . . . , i ) a n d then employing Formula 5 p p n ! fci&"*'=p\ A specialization of Formula 10, when q = n , is °ii...ipip+i...in °fci...fc t i...t„ P n ! l p+ n P)-%...fcp- Two further specializations of this last can b e applied here. First, when p — 0 we get Formula k l 3 Second, when p = n w e get Equating t h e indices (fci,... , k ) = («i,.. - , i ) confirms the first specialization. n n We now examine product formulae involving the alternating symbol. For t h e first, it is easy t o see Formula k l 4 e « l . . . * p t p . . . t „ gip+l.-jn _ + 1 *p+l»-*n / > _ \j ii...i j ...jn e p p+1 * ' as there are (n — p)\ permutations of the common indices. Secondly, unbolding the (n — p) indices in < c»l...tpl +l...i p °l...p,p+l...n n cl...p,p+l...n °h...j l + ...l p p l u cM...tpip+i...i ~ °Ji...ipip+i...i n B which then involves (n—p)\ summations, reproduces an earlier result Formula 4. This formula is equation (8.5) of Veblen 2 Formula k l 5 h...i i ,+i...i e p f ,. n 1 .. = (n — vV S*. "*? 2 This formula is equation (8.4) of Veblen . A little consideration is now given t o products of more than two gKd's. For a compact notation we use {A;} t o stand for {k\,... , kp} (p < n ) and {ki} t o stand for {kn,... ,k\ } etc. p T h e earlier relation o^"'*^ 6^.'"^ can be abbreviated t o Then more generally, a relation of t h e type %...ap °bi...bp- 0 °ji...j p — h-jp- can b e compacted t o Formula k ! 6 W %} ~ %} I" °ii...J )' fi {k } P 2 Each of t h e sets {k,} = { k n , . . . , k } is a permutation of {i} = is t h e set {j} = {ji,..,j }. I p ... ,i } as also p p By now having familiarity with unbolding fixed repeated indices and counting t h e summations we should be able t o see t h e result °ki...k °h...3p p P-°3\-3P or briefly From this we have for a product of three 5's and more generally Formula k l 7 All combinatorial formulae for t h e gKd and alternating symbol are collected to­ gether and numbered as above, for reference in Appendix A. 4 Generalized Permanent 8 There are notable differences between gKd and gpd combinatorial formulae. For example 6jj = n — n whereas 7Ty = n + n. In analogy with t h e gKd it is possible t o define a symmetric (or permanental) symbol p in terms of a gpd by p* -** = ^.'".n ^ Pn...»« = Like t h e alternating symbol e, t h e symmetric symbol p is not a tensor. Unlike t h e alternating symbol which satisfies = it is not t h e case t h a t = p '"**Pji.„j - For example, with n = 2, 7r|j = 6, but p**'py = 2. In general p* "'* p« ...» = n!. As it is not possible t o match a product of contracted p's with t h e gpd it is convenient t o define a new symbol 7r, like a factorial function, t o accommodate such a product. Formally we define a factorial symmetric symbol (also tensor) by Formula p i 2 2 1 11 n 1 n 1 n &£&-<p+i)&S <-*«s-i) and t h a t it is symmetric in all its upper indices and all its lower indices. We specify t h a t t h e factorial symmetric symbol 7r be linked t o t h e symmetric symbol p by t h e relation Formula p2 0*1 —in 7r t , il-i»=P i Ph-Jn- Symmetry in upper and lower indices is consistent. T h e first equation shows t h e factorial nature of the symbol Formula p 3 oti...ip T h e kernel letter 7r in n is used t o indicate permanent symmetries like t h e gpd; however, 7r is not a permanent t o be evaluated from a matrix. T h e symmetric symbol p is useful for total symmetrizations, eg of functions, as in ffi-^ffa,... , * J or pM-Zifo) • • • / „ ( x i n ) but does not play any role in Relativity and is not presented in any Relativity for­ mulae, being otherwise included for completeness. 4.1 Single generalized permanent 6 It is now time t o establish formulae for t h e gpd corresponding t o those for t h e gKd. Proceeding in a similar manner t o t h e gKd, and fortunately because we don't have t o worry about signs with regard t o a gpd, we can expand t h e gpd by t h e column j t o get p Formula p 4 p il-ip-lh 3l-3p-l3p _ V " * ci. Jp n ti...t....ip 31-Jp-l * «=1 Let j p = ip a n d also change p t o p + 1 t o obtain T h e result can b e extended beyond one common index, so t h a t more generally we have Formula p5 *i-«p»p+i-«« _ '31 -ipip+i•••»« (w + q — l)t M - i p (n-fp—1)! A particular case, when q = n , is Formula p6 h...i i +i...i p p _ n "ii-ipip+i-•*» (2n 1)! (n + p - ii...«p 1)! Another particular case, when p = 0, is Formula p 7 = (n + q - 1)! (n - 1)! A specialization, when g = 1 is t h e simple result Formula p 8 ttJ = n = Another specialization of Formula 7, when q = n, is Formula p 9 (n-1)!" 4.2 Products of gpd's T h e reasoning for a product of two gpd's is identical with t h a t for gKd's. Thus we have t h e analogous result t o Formula k l l Formula plO 1T 7r 3i-3p ki...k p - P- ^...kp- Equating t h e indices ( k i , . . . , k ) = Formula p l l . . . , i ) a n d using Formula p 7 gives p p e3l-3v - ^ =*i—*p p! ( n + p :.? ) ! A specialization of this t o t h e case p = n gives Formula p l 2 j ^ <ti~.t» i:=n! 3i...jn ( 7 r n ! f( n"- 1^) !. Arguing in exactly t h e same manner as for t h e derivation of Formula k8, b u t for t h e g p d now, gives t h e analogous formula Formula p l 3 Specializing t o (fej,... , k ) = (ii,... Formula p l 4 , i ) gives p p ti...tp» i...t, ^ p + i . . . i p _ , i p+ "31«.3pip+l-j«"ip+l •*« ( n n + g +p *i...*V. _ 1)J 31-ip' Following t h e more involved reasoning for t h e gKd leading t o Formula klO pro­ duces t h e analogous result for gpd's Formula p l 5 h...ipi i...i ii-ipip+i-iq h-ipip+\-U"fci...fc *> i...t J)+ _ q | (n + q — 1)! ii...*p « • ^ p _ -g j "a* • n P + 4 n A specialization of this result, when (fci,... , fc ) = ( i j , . . . , i ) is p 1 t * "« 9 p (n-1)! a result already obtained as Formula p l l . T h e further specialization, when q = n, gives _*i...**_n...j» _ "h-in^i-in 2 n n \ ( ' ( n n 1 1 ~ ) _ x ) ! confirming again Formula p l 2 . Finally for a product formula of t h e symmetric symbol with a gpd, it is easy t o see Formula p l 6 pti...»pip l...t„ Jp+l"in _ + 7 r £ n _pjjp*l...tpjp+l-Jn as there a r e (n — p)\ possible permutations of t h e common indices. For a product of p's with repeated indices we have Formula p l 7 . . . n\ 0*1—H> W h e n there is summation on all indices we get t h e specialization Formula p l 8 Formulae for products of several gpd's are obtained by t h e same reasoning as for gKd's. Thus we have Formula p l 9 Unbolding t h e bolded indices so t h a t summations occur gives u s Formula p 2 0 All combinatorial formulae for t h e gpd a n d symmetric symbol are collected to­ gether a n d numbered as above, for reference in Appendix B. 5 Combined gKd and gpd Here we derive formulae for combined products of gKd's a n d gpd's where a t least one factor is a gpd or symmetric symbol and another is a gKd or alternating symbol. We saw earlier t h e useful result t h a t t h e product of any gKd a n d any gpd with two or more summed common indices vanishes. Consequently for a combined product we can have at most one matching index pair for superscript and subscript a n d another matching index pair for subscript a n d superscript. Consider a combined product of a gpd a n d a gKd with (necessarily) at most one superscript matching a subscript as i i ^ « . ^ ^ 2 » . i * * - Expanding t h e gpd by its (first) row i\ leads t o m x Formula c l *ita-..tV rfcifca.-.k, _ r r t ! ia...»V _ ~ P V • riiia.-.ip , , c*i « . • • » > l cfcika-*, *a o*ik .»*« *h..-5-~iv i.h...i ' 2 1 0 q 8=1 T h e unexpanded product can, a t t h e same time, also have a t most one gpd sub­ script matching a gKd superscript as in ^l^.'.^ii^."!*'1 For t h e combined product P* "^^^"^, i\ we have Formula c 2 expanding t h e gKd b y its first column * 1 «.»n£fcl*2...kn _ * l — * n " *li2—in " 0 1 a+1 5 E(- ) * fc^» 0 fcn .8=1 Consider t h e converse relation ^ "'* T^^".'j i\ is a summation with all positive signs Formula c 2 il n n n t c 5.1 *l...tn_klka...kn *u'a...in /l • T h e expansion of t h e gpd b y column n \ ^ ,k«*3—*n- kl— * « — * n Z-/ » •>» ' 8=1 r _ Idempotent and orthogonal relations We now enter a certain class of identities combining gKd's and gpd's whose origin stems from orthogonal idempotents in t h e group algebra of t h e symmetric group S . T h e derivation of, rather t h a n t h e background to, these identities is what concerns us here. T h e orthogonal identities are visually obvious by having blanks (dots) in gKd's and gpd's, eg 7r^, where arbitrary free indices could b e placed. In order t o prove t h e identities it is necessary t o call on some simpler identities, which, although established in generality earlier a n d listed in tables a t t h e end, is pertinent t o b e stated here for specialized values. Expansion of a gKd or gpd as a sum of terms comes about by considering it as an expansion of a determinant or permanent by a row or column. T h e basic formulae below can all b e inverted, that is, all superscripts lowered and subscripts raised. Reference t o a basic formula implies either its use or its inversion. n (1) tJ 6 ir- = 0 (repetition of two indices in a gKd a n d gpd) T h e fourth t e r m S^S^S^S^^ = 46£6&ir%ir$ using (2). Interchange dummy indices i, k a n d j , I, a n d as a gpd is totally index symmetric, t h e value of t h e term > is unchanged; its index rearrangement becomes 46^^7rf/7r?/ which is t h e new first term. Hence t h e s u m of t h e first a n d fourth terms in X is %&%&Mtfk tfi • Interchange of *, k a n d j , I in t h e third term S^SQS^Sj^n^itji which does not change its value, converts it t o t h e second term, so t h a t X includes t h e expression 2^c^M^^r«7r« A certain facility is acquired after some manipulations with gKd's and gpd's. W e see t h a t we have 1 t Interchange of indices i,k a n d j , I converts t h e third and fourth terms t o t h e second and first terms respectively, so t h a t = - «<f < - *S<f< + * t f "Si) Adding this t o t h e sum of t h e first a n d fourth terms previously leads t o X 12$&«8*«*$fc so t h a t ^ « < <T jg ft 7# TTjf = 12«««. This result is recorded as F o r m u l a 13 XP9 JCr« ~ma c*j ski ef ah _ nff^fTi^f gh Derivation of Formula i4 + CP«!*S = 6£?<f + M2?*g 16 (using (2)) (using (5) and (4)) = Derivation of Formula i2 = 6 ^ < 9 = - (flrt? - tfifiKg " 2#r£* +M},r£ (using (3)) (using (5)) This result is recorded as Formula i2 Derivation o f Formula i3 We have This result is t h e most complicated we encounter. = - Write t h e R.H.S. asX = 6%S£ nfi, where Y are t h e four terms in parenthesis. They can b e expanded and then compacted Y = nff^JL + - n # j « , - *J#ii* Now reconstruct X a s a sum of four terms, which will b e treated individually. T h e first term is T h e fourth t e r m S^S^S^S^iif = 4$*jE&it%'ii$ using (2). Interchange dummy indices i, k a n d j , I, and as a gpd is totally index symmetric, the value of the term is unchanged; its index rearrangement becomes 45jj,5^7T^7r^j which is t h e new first h term. Hence t h e sum of t h e first and fourth terms in X is BS^Sj^ir^i^. 7 1 Interchange of i, k a n d j , I in t h e third t e r m ^ ^ w ^ i ^ p i * * ^ i * which does not change its value, converts it t o t h e second term, so t h a t X includes the expression ^oc^M^fw^rg^i* "Rji- A certain facility is acquired after some manipulations with gKd's and gpd's. We see t h a t we have Interchange of indices i, k and j , I converts the third and fourth terms t o t h e second and first t e r m s respectively, so t h a t + <f *£ + <M? - "« « Adding this t o t h e sum of t h e first and fourth terms previously leads t o X 1 2 ^ S « so t h a t *™ « C <f *ff = 124*^ < « « This result is recorded as Formula i3 n °ab °cd pr Kqs °tu °uw ™ik ^jl ~ ^"ab^cd^pr^a • D e r i v a t i o n o f F o r m u l a i4 = 6«X ; + Cr<C< = ME}*J + 2£f (using (2)) (using (5) and (4)) = This result is recorded as Formula i4 cpvwtu cqefnh °abe "pd°tvw qu n _ o cpc/^jh ~ °°ol>c dp7r Derivation of Formula o l 5 27r ^^^7= « ST = 0 (using (5), then obvious). This result is recorded as Formula o l Derivation of Formula o 2 O f f * =- 2 ^ * 5 - 0 (using (4), then obvious). This result is recorded as Formula o 2 Derivation of Formula o3 «2<'«Si = «S«gS-0 (using(7)). This result is recorded as Formula o 3 Derivation of Formula o 4 = 2 « * ( ^ + ^ . ) = 0 (obvious). This result is recorded as Formula o 4 Derivation of Formula o5 = OT(2<>$ - 2nSJ^) = W ( « + = 0 (obvious). This result is recorded as Formula o5 Derivation of Formula 0 6 = 0 (obvious). This result is recorded as Formula 0 6 Derivation of Formula o7 = 0 (obvious). " (using (5)) - This result is recorded as Formula o 7 Derivation of Formula 0 8 /ctvuw — l°o6cd = 0 ctwuv °o6cd — cuvtw 1 cuuitw\_*i_fcl °abcd + °obcd J^tv^uv) — (obvious). This result is recorded as Formula 0 8 Derivation of Formula o9 = = 0. ( ™ * « (8)) This result is recorded as Formula o 9 9h ?!. %i n 9 = 0. P Derivation of Formula m l Depending on t h e manner of verification one may b e led t o other unobvious iden­ tities. We give a few. T h e first is = 2 [ M ^ - « , ' + = 2[(w - TO€f « f ) - (W - = + («Stf-<«+nS«2)] = = 2 ( « - « + * M ) 2<C / + (W - This result is recorded as Formula m l a 7V °abc dqr ps ~ zo obc "dq • Derivation of Formula m 2 For t h e second, consider ^S^it^.n^. Interchanging dummy indices p,q and r,s will not change its value. However t h e expression now becomes S ^S^ir {'n ^. This result is recorded as Formula m 2 P t q Derivation of Formula m 3 For t h e third, consider S^S^S^T^TTQ. p We have = 0 (obvious). This result is recorded as Formula m 3 All combinatorial formulae for t h e combined gKd a n d gpd are collected together and numbered as above, for reference in Appendix C. P A R T II In this part we first examine t h e 2-component gKd a n d gpd spinors which can be useful for t h e mathematical spinor algebra of General Relativity. Secondly we find and tabulate t h e spinor equivalents of gKd's u p t o four double-index (4up, 4down) indices. Thirdly, t h e other way round, we derive a n d tabulate t h e tensor equivalents of gKd and g p d spinors. In t h e first situation since any product of a gKd and a gpd with two (or more) summed indices vanishes, a non-zero result can only b e obtained if there is at most one summed index (up a n d down) a n d a different summed index (down and up). For convenience, products of gKd's are simplified; similarly for gpd's. All these results, easily obtained are listed in Appendix D. 6 Spinor equivalents of gKd and gpd tensors T h e spinor equivalent of t h e basic Kronecker 6, 6% = ir% is SgSg,. From this we can construct t h e spinor equivalent of, for example, t h e gKd A 2 = 6%% and express t h e result in terms of spinor gKd's and gpd's, eg 8&$ Sfrg 6§6g b%b% A = ^ = = 2 1 _ rcAB—A'B' , — 21rCLl^C'D' + -ABcA'B'-i *CD°C<D<\' Similarly we can easily construct t h e spinor equivalent of t h e tensor gpd II2 = 7T^. One has n 2 = 7$ 1 _ tcAB cA'B' — ~2[°CD°C'D' J 1 = sgsg ftsg, , 7r ^AB-A'B'-i CD C'£)'J7r Spinor equivalents of all gKd's a n d gpd's u p t o four double-index gKd's a n d gpd's are collected together for reference in Appendix E. 7 Tensor equivalents of gKd and gpd spinors Whilst most attention focuses on spinor equivalents of tensors, it is also of interest t o exhibit t h e tensor equivalents of certain numerical spinors. T h e various derived formulae are collected together in a table a t t h e end. Multiplying t h e fundamental spinor identity £ £ CD EF by e A E e B F £ + £CE FD "I" &CF&DE = 0 we obtain t h e most useful result for t h e numerical gKd spinor £ £cd — o . CD 3,4 T h e following spinor equivalents are well-known , in mixed-mode form &b e a b = e <*i(6£6*6*6§: ab a i - cd 6*6*6*6%) and in covariant form as 9ab <=> Sated £ab£a'b> i(£AC£BD£A'D'£B'C' ~ CD £Ad£bc£A'C'£b'D')- CD Multiplying t h e first of this latter pair by e e ' 6 Sab£ £A'B' g** gives — °AB°A'B' Q 9ab- Here we have t h e tensor equivalent of t h e product of two gKd spinors. In using spinors here, it is convenient t o employ t h e abbreviations a' = $g it = *#g. We may therefore write t h e above tensor equivalent as AA ' sf*9cd (a A A ' spinor). We go on t o obtain t h e tensor equivalents of An', HA', I H T spinors. Expansion of a A n ' is #g *$& = W&g - + = 6^6*6^16*, — 6^6*6^,6*, — kAcA' cBcB' = 0 0 ,0 0 C C D D I s&sg) + 6^6*6^,6*, JiAcA'cBcB'.cAcBcA'cB' — 0 0 , 0 0 , + O O b ,0 i D D C C c D D C — 6^6*6^16*, — cAcBcA'cB' OpOcO&Ojy. T h e first two terms comprise A2 t o give t h e tensor equivalent 6c6c<6*6*y - 6p6o'6*6*, 6$ and t h e last two terms, with a factor i, form the mixed alternating tensor. Combining, we have t h e tensor equivalent of a A n ' ab An' = 6**rt:*; &6%-ie . i cd Taking t h e conjugate gives AB TTA' _ -*1 1 A = TT CD RA'B' d , , c D O cab , • at d + IE cdc d Adding these confirms t h e result for A . Subtracting t h e m gives a 'new look' beautiful appearance for t h e spinor equivalent of t h e mixed alternating tensor, easily remem­ bered, 2 L 2°A<n' — W i ) ~ 2 D d ' e °c'D') cd- We now have tensor equivalents for A A ' , A l l ' , IIA'. To determine t h e remaining product, a I I I I ' , expand it directly Now we have t h e spinor equivalent for n 2 T h e R.H.S. is t h e s u m of t h e first a n d fourth terms in t h e I I I I ' equation preceding it. We substitute for this s u m and write (with abuse of language by mixing tensor and spinor notations) 1 -AB-A'B „ab _ cA cB cA' cB' , cA cB cA' cB' C D C ' D ' ~ cd— % " D £ ) ' C ' + D C°C' D'- 7 I 7 I n 6 d d d /i\ \ ) d l Having reached here we still do not know any tensor equivalent of t h e last two terms in t h e above. Holding this latest nn' equation in abeyance we proceed on another tack. Using t h e 2-index g K d a n d g p d we can isolate a product of two single 5's Taking t h e conjugate a n d then interchanging indices A',B', gives * l XA'B' , ^A'B'\ 2\~ C'D' + C'D') d lr _ cA'cB' °D'°C'- Multiplying these last two brings in t h e AA', An', nA', A nn' 1 * l X & JsA'B , cAB —A'B' ^AB cA'B' , —AB ^.A'B'\ -£\—O 0 , i -f- 0 TT , i — TT Oi i -f- licD ^C'D') CD c D C D c D CD c D _ cAcBcA'cB' ~ °C D D'°C' • 0 0 Except for t h e last nn' term in parenthesis, we have previously obtained tensor equivalents for all other terms, AA', An' — nA'. Thus \(-g*g« - 216°" * + nn') = 6£6Sfi£sg. (2) Take t h e conjugate a n d get al \(-9 '9 i + a ab 2ie cd + UIl') =6*6*8*6%. Add these two, a n d then substitute for t h e two R.H.S terms in (1) t o get i(-p*0«i+nir) = n f f - i £ . Solving for II IT we finally obtain its tensor equivalent nir = ^ : ^ 2 ^ - ^ c d . B If we then use this result back in (2) we find t h a t t h e spinor 8^6*6*,8 .' equivalent I has t h e tensor We tabulate in Appendix F , t h e spinor/tensor equivalents of t h e numerical tensors/spinors we have come across. Mixed mode spinor and tensor indices can be lowered by t h e £43, £a'B'> 9ab- T h e covariant form of t h e mixed mode spinors is in the line directly below each of them. 7.1 Appendix A. gKd combinatorial formulae Definition formulae Single g K d formulae ^ = n = i c f c t ( 0 * * * °il~ip*P+l-*, ^ 4 K (n-ff)l A~4>' = jftjj, c a r e t d e n o t e s o * * °il...i i i...t P k 6 . < = n, p + B k7. m i - t t e d ^ I™ = e P)'°ii...3 P " " ^ = nl Product of gKd's °ii K A U •ipip+i-i, C * - 0 3\...jpip+l...J, W *i...*pi -l...t, PJ- (n-,)i - lH y-(n-g)!%...fcp ! k 1 2 kn. C:£ <::t=*> CX> kl4. *" L U 1 t j fc ' kl6. +1 e* • V>«j>+i"*»S .» "i d { k j } t d { k a } n • d m 1, = Jl-Jp»p+l-«n - 6 - **£ t i = p ^ (n-p)\e* -Wp+i-in v V* m r)' j\...j v {- 0 ^ ), h p kl7. 6 {ki} d { f c 2 } - (J>!j 0 ^ 7.2 gpd combinatorial formulae Appendix B . Definition formulae oil-in - f oh-in , Single g p d formulae o»l...*p P - ^...ip = P 3 . _c ! U...ip^ip ™ ii-ipip+i-t, _ ~ ii...*....^ (n+g-l)l (caret denotes omitted index) U-tp - "ii-ipVn-** ~~ (n+p-i)l ^ i i - V P°' f P * ^ii...*, (n-l)! — ' ii.-.tptp+i.-tn 7l P ° * ii.«j>vn-»« P°* 7 1 — - °t) t f P il-ip J^tv^l}^ _ — J'i- ip (n+p-i)! ' ^ . . . i n :: - (n-l)l P r o d u c t o f gpd's p l U . 7rjj...j„7r^ ...fc 1 1 tt* "*" 7 r ? " nl2 P A t f A O 7r jl-.3pjp+l..j, % « . ki...kpip ...i +1 q A P 1 7r 7r - ii...ip ii...J w W 1 Pi' (n+p-l)J 1 1 = jjajlj, 1 3l Ipitf. ll 9 7 7 TT W { k i } 77 T T j p+1 { k l } { k j } n ... 7 < nT- {kr> 0 } ! (n-l)! n h - 3 p — ? • (n+p-l)l ^fci.-fcp 1 il P ...jpfcp+i p* ' " ^ ' " ^ T r j ^ J ' j = ( n - p ) ^ " ^ " * P I T . P "^ "^P ... pi ...i _ p 1 W lr * ^Si-ipip+i-iq pl6. 1 ^ k p + i - q A 2n = n> ( ~ ) i B - ^h-ipjp+\:.j P ** P 1 - P ! 7Tfci...Jfcp) I> - 7 T 0} r— * ' TT* Jl—ip'' { i } 1 1 = n! P l 8 . p^Pir...^ { } { l } { f c p } I>20 - * 7T 7 T{ i } — - fljlV P { *'! } 7^T{ f *e } • • • ^ IPO 7T / U { 7.3 Appendix C. 7.3.1 C A P r o d u c t s of gpd's, gKd's, alternating a n d symmetric symbols ' c2. Combined gKd, gpd formulae ^j\h-3p°i\h.-l ~ L>*=1 q P 7.3.2 ^ 6*£± n 0 3l-3.-3p 3sh = c = i (-i) a + V- Iq i a Identities w i t h combined gKd's a n d gpd's o 1 11. ASwjag-nS = 12. ^ *3 5 - i4. «T * 5 C l < tt^ 1 = 8 8%<K'£ = 8«2f< = - o2. = 0 % TTjlp 4 o - ^ « ? ^ « / h « . = o5. ml- m2. m3. = 2 S % < = Q i r ^= 0 0 0 fi^^TrrjTr^^Tr-^O <*' ^ ^ ^ f r = 0 o7. ^ ^ ^ C ^ ^ - = h 0 8 . 6^ 6%6^-n^ w o9. # * $ f i i * « 0 = 0 0 7.3.3 A l t e r n a t i n g a n d s y m m e t r i c s y m b o l p r o d u c t formulae c £ ^ 0 tabcd ~ °abcd efBh p Pabcd efad p Pabcd = Kobcd = = 2!# /> "pa6cd = e/C 12*2 e e abcd = 3!^ P^Pabed £ Cabal = 4! P ^ Poked = 4! = 24*1 7.4 Appendix D . 7lB = gAB °CD *B fiCE . FG f*CD ^FG ' cAB °CD £ ^ A B B Spinor Product Formulae $D ^ F G — ~AB CD n ^FG „CE _ _ B qr-AE I ~A "FG ~ "D "FG + "D AB CD CD"FG 26 B n F -BE "FG AB FG = 27T B —-AE ~A —-BE — TV TTpQ — II 71 p ~CE "FG D D G £ c d e A 7.5 B Appendix E. Spinor equivalents for the gKd and gpd Reference formulae for t h e spinor equivalents of t h e gKd and gpd are given below. It should b e mentioned t h a t there is a good deal of interplay between t h e gKd and gpd in such specifications, there being a variety of ways t o express expansions of some gKd's, gpd's (and also their spinor equivalents). Simple examples at t h e end of t h e section exhibit application of t h e gKd and gpd. 7.5.1 I. Spinor equivalents of the gKd's _ A = Sf = 2 d 6*6$ l CD^C'D' % % «J A s = 6£ = f 6* 61 6) 6* 61 6} — °d°ef °e°df + °f°de d * ^D^D'i^EF^E'F' ^F^F'i^DE^D'E' + + 7 r BF^E'F') — ^E^E'i^DF^D'F' ^DF^D'F') ^DE^D'E')]' W e can expand t h e gKd with 4 (upper/lower) indices by a Laplace expansion of its first two rows a n d complementary minors. T h e result is A4 = « 61 6\ 61 6i = 61 6) 6) 6) 6- 61 6) 61 61 81 6\ 6% T h e R . H . S . can b e w r i t t e n as t h e sum/difference of permanents 6% VH 6f 6°* °gh f T h e spinor equivalent of 8^^ gab "eg fid "eg 6% 6f 1 T is 7r ~(6EG E'G> + ^EG^E'G^i^FH^F'H' ^FH^F'H') 7r ~K^£ff E'if' + ^EH^E'H^i^FG^F'G' +(6pQWpiQ, C 7 I ^i^GH^G'H' S + + ^G8P,Q,)(8EJJ'K ,,JJ, ~(^FH F'H' of g K d ' s , if desired *FG F'G>) +T^EH^E'H') ^H^F'Hdi^EG^E'G' "^EG^E'G') 1 ~^ ^H^G'H'^EF^E'F' + "^EF^E'F )]• 7.5.2 II. Spinor equivalents of the gpd's Iii = ^<5£ A n =7r£ 2 6 6% = 6%6% 9 $ 3 9 3 i^^O'C^SF^B'F' + 7 r £F 7 r A A B B B ' F ' ) + S 6 ,(6 p8 ,p, ) ) ^CF^D'F') Since p e r m a n e n t s only involve positive signs, t h e following Laplace expansion is also clear 61 6 l 6 l 04 = 7 $ $ = •7 * T h e R.H.S. can b e written as t h e sum of permanents of permanents, if desired + + T h e spinor equivalent of T T ^ ^ is ^JSH^E'irX^G^F'G' •'"(^FG^F'G' 7 r 7 r 7 r N FG F'G')(^BH^B'H' r r 7 r , 7 r GH G'H')(^EF^B'F' + JR ' EH' E'H') LT +(^FH^F'H' +' FH' F'H')(^BG^B G' "K^Gff^G'tf' + 7 r FG F'G') 7R ' EG E'G') 7 r 7 r BF E'F')] 7.6 Appendix F. Spinor^Tensor equivalents T h e covariant f o r m o f t h e m i x e d m o d e s p i n o r s is in t h e line d i r e c t l y b e l o w e a c h o f them. Spinor Tensor 6$ sab 0afc CABGcD — CAC&BD £ac£bd + ~ | EAD&BC £ad£bc mo +*3S- e €AC BD£A'C'£B'D' MSS # = +nA') ^AD^BC^A'D^B'C' 5 1 SACSBDeA'C'SB'D + £ac£bd — <7ad<7bc i(AA'+nn') ffac&d + ^AD^BC^A'D'SB'C' 9ad9bc £cd °* = £ ° * erf *{£ac€bdGa'D'£b'c 1 — eADGBceA'cea'D ) £abcd ^ & & = AA' Raided ^AC^BD&A'C'^B'jy — ^AC^BD^A'iy^B'C' GAD€BC£A'D'£b'C' = SAB^CDi^A'C'SB'D' ^ g ^ 1 — {SACeBD (sacSbd + — —eAcSBDSA'D'SB'C 1 SAD^BC^A'C^B'D 1 SA'D'Sb'C ) = n A ' sacSbdSa'c'Sb'd "^cd^c'd 1 ~ + + = ^ad^bcSa'd'Sb'c' <7ac<7bd — 9ad9bc + *So6ed €Ad£bc£A'C'£b'D' £ADSBc)eA'B'£C'D' 2*3 - g^ga nn' + sadSbc) x (ex'c'fiB'B' 4- e x ' z y e B ' c ) 29ae9bd + 29ad9bc — 9ab9cd ^(SocPbd + ffodSbc — gab9cd — i£abcd) 1 R.L. Agacy, Generalized Kronecker, permanent delta and Young tableaux applications to tensors and spinors; Lanczos-Zund spinor classification and general spinor factorizations, PhD Thesis, London University (1997). 2 0 . Veblen, Invariants versity Press, (1927). of Quadratic Differential Forms, Cambridge Uni­ 3 R~ Penrose and W. Rindler, Spinors and space-time, Cambridge Univer­ sity Press, Vol. 1, (1984). 4 F . Trautman, F. Pirani, and H. Bondi, Lectures on General Brandeis Summer Institute in Theoretical Physics, (1964). Relativity, PHYSICS AUXILIARY PUBLICATION SERVICE Document No: JMAPAQ-40-033903-35 Journal Reference: Journal of Mathematical Physics Vol. 40, No. 4 - April 1999 [p.2055-2063] Title: Generalized Kronecker delta and permanent deltas, their spinor and tensor equivilents and applications. P A P S Title: Reference Formulae Authors: R. L. Agacy For further information: e-mail: paps@aip.org or fax: 516-576-2223 (see http://ojps.aip.org), via the web (http://www.aip.org/pubservs/epaps.html) o r from ^tg^a^gjorg^jn^h^directory^ A service of the American Institute of Physics 2 Huntington Quadrangle-Suite 1N01 Melville, New York 11747-4502