The Permutation Systems and the Determinant

Skew-symmetric systems, also known as alternating systems, are objects of immense utility and are some of the most beautiful objects in Tensor Calculus. They find crucial applications in expressing determinants, the cross product, as well as the differential operator known as the curl. Furthermore, they provide an alternative approach to constructing invariant differential operators which serves as the foundation of the subject of Differential Forms. Last but not least, skew-symmetric systems represent a celebration of the indicial notation. What '{E}lie Cartan referred to as a d\'{ebauches} of indices (translations of d\'{ebauches} range from overabundance to orgies), we will see as a dynamic choreographed dance. That said, skew symmetric systems do require utmost fluency in the tensor notation. If there are any deficiencies in your technique, this topic will provide you with a valuable opportunity to remedy them.
By definition, a system is skew-symmetric if any of its elements related by a swap of two indices have opposite values. We will consider skew-symmetric systems of various orders. For second-order systems, such as AijA_{ij}, the skew-symmetric property reads
Aij=Aji,(16.1)A_{ij}=-A_{ji},\tag{16.1}
which agrees with the familiar like-named property of matrices. For example, in four dimensions, AijA_{ij} would be represented by a skew-symmetric matrix, such as
[0235204634075670].(16.2)\left[ \begin{array} {rrrr} 0 & 2 & 3 & -5\\ -2 & 0 & -4 & -6\\ -3 & 4 & 0 & 7\\ 5 & 6 & -7 & 0 \end{array} \right] .\tag{16.2}
The conspicuous zeros on the diagonal, which correspond to i=ji=j, are an important characteristic of skew-symmetric systems.
To illustrate higher-order skew-symmetric systems, consider a fourth-order system AijklA_{ijkl}. For the time being, we are not concerned with the tensor property of AijklA_{ijkl}. Thus, the placement of the indices as subscripts is arbitrary and we could just as easily consider a fourth-order system AijklA^{ijkl} with four superscripts. We could even consider a mix of subscripts and superscripts, but that would unnecessarily complicate the notation.
For reasons that will become apparent shortly, we must require that the dimension nn of space is greater than or equal to the order of the system, i.e. n4n\geq4 in the case of AijklA_{ijkl}. For the sake of the present discussion, assume that n=6n=6.
The system AijklA_{ijkl} is skew-symmetric if, once again, any of its elements related by a swap of two indices have opposite values. This requirement can be reduced down to three identities involving swaps of consecutive indices, i.e.
Aijkl=Ajikl (switch 1st and 2nd)          (16.3)Aijkl=Aikjl (switch 2nd and 3rd)          (16.4)Aijkl=Aijlk (switch 3rd and 4th).          (16.5)\begin{aligned}A_{ijkl} & =-A_{jikl}\text{ (switch }1^{\text{st}}\text{ and }2^{\text{nd} }\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.3\right)\\A_{ijkl} & =-A_{ikjl}\text{ (switch }2^{\text{nd}}\text{ and }3^{\text{rd} }\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.4\right)\\A_{ijkl} & =-A_{ijlk}\text{ (switch }3^{\text{rd}}\text{ and }4^{\text{th} }\text{).}\ \ \ \ \ \ \ \ \ \ \left(16.5\right)\end{aligned}
Indeed, there is no need to document the swaps for all possible pairs of indices since any swap can be accomplished by swaps of consecutive indices. For example, the swap
ijklkjil (switch 1st and 3rd)(16.6)ijkl\rightarrow kjil\text{ (switch }1^{\text{st}}\text{ and }3^{\text{rd} }\text{)}\tag{16.6}
can be accomplished by three swaps of consecutive indices, i.e.
ijkljikl (switch 1st and 2nd)          (16.7)jkil (switch 2st and 3rd)          (16.8)kjil (switch 1st and 2nd).          (16.9)\begin{aligned}ijkl & \rightarrow jikl\text{ (switch }1^{\text{st}}\text{ and } 2^{\text{nd}}\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.7\right)\\& \rightarrow jkil\text{ (switch }2^{\text{st}}\text{ and }3^{\text{rd} }\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.8\right)\\& \rightarrow kjil\text{ (switch }1^{\text{st}}\text{ and }2^{\text{nd} }\text{).}\ \ \ \ \ \ \ \ \ \ \left(16.9\right)\end{aligned}
Critical to the internal consistency of the definition is the fact that any given permutation can only be represented either by an odd number of swaps or by an even number of swaps. After all, if there was a way to achieve the permutation ijklkjilijkl\rightarrow kjil by an even number swaps, then the value of AkjilA_{kjil} would be simultaneously Aijkl-A_{ijkl} and +Aijkl+A_{ijkl} depending on whether we arrive at kjilkjil by an odd or even number of swaps. Permutations that can be obtained by an odd number of swaps from the identity permutation are called odd, while those that require an even number of swaps are called even. This property of permutations is called parity.
The skew-symmetric condition imposes a severe constraint on the values of AijklA_{ijkl}. First, all elements whose indices are not all distinct must equal 00. For example, consider the element A1614A_{1614} for which the first and the third indices are equal. By the skew-symmetric condition, we must have
A1614=A1614(16.10)A_{1614}=-A_{1614}\tag{16.10}
since swapping the 1st1^{\text{st}} and the 3rd3^{\text{rd}} values in 1,6,1,41,6,1,4 yields the same combination 1,6,1,41,6,1,4. Therefore,
A1614=0.(16.11)A_{1614}=0.\tag{16.11}
Thus, in order for an element of AijklA_{ijkl} to be nonzero, its indices must have distinct values.
Second, any two elements whose indices are related by a permutation are either equal or opposite of each other. Indeed, consider the elements A1256A_{1256} and A2561A_{2561}. Since the combinations 12561256 and 25612561 are related by an odd number of swaps, e.g. 12562156251625611256\rightarrow2156\rightarrow2516\rightarrow2561, we must have
A1256=A2561.(16.12)A_{1256}=-A_{2561}.\tag{16.12}
Thus, a fourth-order skew-symmetric system in an nn-dimensional space can have at most
(n4)=n!4!(n4)!(16.13)\binom{n}{4}=\frac{n!}{4!\left( n-4\right) !}\tag{16.13}
degrees of freedom, i.e. independently assigned values. More generally, a skew-symmetric system of order mm in an nn-dimensional space can have at most
(nm)=n!m!(nm)!(16.14)\binom{n}{m}=\frac{n!}{m!\left( n-m\right) !}\tag{16.14}
degrees of freedom. In particular, and crucially for the upcoming discussion, when the order of a skew-symmetric system matches the dimension of the space, the former can only have a single degree of freedom. For example, if n=4n=4 then a system AijklA_{ijkl} has one degree of freedom represented by the element
A1234.(16.15)A_{1234}.\tag{16.15}
All of the remaining elements of AijklA_{ijkl} are either 00 or ±A1234\pm A_{1234}.
Finally, when the order of a skew-symmetric system exceeds the dimension of the space, all of its elements must vanish since the indices cannot all have distinct values.
The result of a double contraction between a skew-symmetric and a symmetric system is zero, i.e. a system in which all elements are 00. For example, if AijklA_{ijkl} is a skew-symmetric system and UijU^{ij} is a symmetric system, i.e.
Uij=Uji,(16.16)U^{ij}=U^{ji},\tag{16.16}
then
AijklUij=0.(16.17)A_{ijkl}U^{ij}=0.\tag{16.17}
We will demonstrate the above identity in a way that will make it clear that the statement holds for more complicated indicial signatures as well. Let
Xkl=AijklUij(16.18)X_{kl}=A_{ijkl}U^{ij}\tag{16.18}
and consider the combination in which the indices on UijU^{ij} are swapped, i.e.
AijklUji.(16.19)A_{ijkl}U^{ji}.\tag{16.19}
We will show that this new combination simultaneously equals XklX_{kl} and Xkl-X_{kl} which, of course, implies that Xkl=0X_{kl}=0.
Thanks to the symmetry of UijU^{ij}, we have
AijklUji=AijklUij=Xkl.(16.20)A_{ijkl}U^{ji}=A_{ijkl}U^{ij}=X_{kl}.\tag{16.20}
On the other hand, due to the skew-symmetry of AijklA_{ijkl} -- in particular, the fact that Aijkl=AjiklA_{ijkl}=-A_{jikl} -- we have
AijklUji=AjiklUji=AijklUij=Xkl.(16.21)A_{ijkl}U^{ji}=-A_{jikl}U^{ji}=-A_{ijkl}U^{ij}=-X_{kl}.\tag{16.21}
Note that in the second step in this string of identities, we simply exchanged the letters ii and jj. Thus, indeed,
Xkl=AijklUji=Xkl(16.22)X_{kl}=A_{ijkl}U^{ji}=-X_{kl}\tag{16.22}
and, therefore,
Xkl=0,(16.23)X_{kl}=0,\tag{16.23}
as we set out to show.
When both systems are second-order, this statement can be interpreted in the language of matrices. Namely, the product of a skew-symmetric and symmetric matrices has zero trace. For example, consider the product
[123245356][023204340]=[514117242283029],(16.24)\left[ \begin{array} {ccc} 1 & 2 & 3\\ 2 & 4 & 5\\ 3 & 5 & 6 \end{array} \right] \left[ \begin{array} {rrr} 0 & -2 & 3\\ 2 & 0 & 4\\ -3 & -4 & 0 \end{array} \right] =\left[ \begin{array} {ccc} -5 & -14 & 11\\ -7 & -24 & 22\\ -8 & -30 & 29 \end{array} \right] ,\tag{16.24}
and observe that the trace of the resulting matrix is 524+29=0-5-24+29=0.
Let us now turn our attention to skew-symmetric systems whose order matches the dimension of the space. We will illustrate such systems with a third-order system AijkA_{ijk} in a three-dimensional space. Since the nonzero elements of AijkA_{ijk} correspond to the permutations of the numbers 11, 22, and 33, out of a total of 2727 elements, there are only 66 potentially nonzero ones:
A123, A132, A213, A231, A312, A321.(16.25)A_{123},~A_{132},~A_{213},~A_{231},~A_{312},~A_{321}.\tag{16.25}
Denoting the sole degree of freedom shared by these elements by α\alpha, we have
A123=A231=A312=α(16.26)A_{123}=A_{231}=A_{312}=\alpha\tag{16.26}
and
A213=A132=A321=α.(16.27)A_{213}=A_{132}=A_{321}=-\alpha.\tag{16.27}
When α=1\alpha=1, the corresponding system is called a permutation system and is denoted by eijke_{ijk}. We will also consider the superscripted permutation system eijke^{ijk} defined in the exact same way. Let us summarize the values of eijke_{ijk} and eijke^{ijk} in the language that generalizes to higher and lower dimensions:
eijk,eijk={+1,if ijk is an even permutation of the numbers 1,2,31,if ijk is an odd permutation of the numbers 1,2,3+0,if ijk is not a permutation of the numbers 1,2,3.(16.28)e_{ijk},e^{ijk}=\left\{ \begin{array} {ll} \phantom{+} 1\text{,} & \text{if }ijk\text{ is an even permutation of the numbers }1,2,3\\ -1\text{,} & \text{if }ijk\text{ is an odd permutation of the numbers }1,2,3\\ \phantom{+} 0\text{,} & \text{if }ijk\text{ is not a permutation of the numbers }1,2,3. \end{array} \right.\tag{16.28}
Any third-order skew-symmetric system AijkA_{ijk} in three dimensions is a scalar multiple of eijke_{ijk}, i.e.
Aijk=αeijk.(16.29)A_{ijk}=\alpha e_{ijk}.\tag{16.29}
We must now say a few words about the placement of the indices in the permutation systems eijke_{ijk} and eijke^{ijk} since, in Tensor Calculus, indices must "earn" their placements by accurately indicating how the system transforms under a change of coordinates. While the permutation systems are independent of coordinates, we can simply assume, in the context of a coordinate analysis, that eijke_{ijk} and eijke^{ijk} have the values summarized above at all points of a Euclidean space. Then we may legitimately ask whether eijke_{ijk} and eijke^{ijk} are tensors. In other words, suppose that, in the primed coordinates ZiZ^{i^{\prime}}, eijke_{i^{\prime}j^{\prime}k^{\prime}} and eijke^{i^{\prime}j^{\prime}k^{\prime}} are defined in the exact same way as eijke_{ijk} and eijke^{ijk}. Then, are the unprimed and primed systems related by the identities
eijk=eijkJiiJjjJkk(-)e_{i^{\prime}j^{\prime}k^{\prime}}=e_{ijk}J_{i^{\prime}}^{i}J_{j^{\prime}} ^{j}J_{k^{\prime}}^{k} \tag{-}
and
eijk=eijkJiiJjjJkk?(-)e^{i^{\prime}j^{\prime}k^{\prime}}=e^{ijk}J_{i}^{i^{\prime}}J_{j}^{j^{\prime} }J_{k}^{k^{\prime}}? \tag{-}
The answer is no, but you may be surprised just how close to actually holding these identities are. In the next Chapter, we will introduce simple modifications of the permutation systems known as the Levi-Civita symbols which are, in fact, (nearly) tensors.
While the better part of this Chapter is devoted to the determinant, we would be remiss at this stage not to point out the inescapable connection between the permutation systems and the determinant. For example, for a 3×33\times3 matrix MM with entries MijM_{ij}, the determinant is given by the formula
detM=all permutations πsign(π)M1,π(1)M2,π(2)M3,π(3),(16.30)\det M=\sum\nolimits_{\text{all permutations }\pi}\operatorname{sign}\left( \pi\right) M_{1,\pi\left( 1\right) }M_{2,\pi\left( 2\right) } M_{3,\pi\left( 3\right) },\tag{16.30}
where the summation takes place over all possible permutations π\pi of the numbers 1,2,31,2,3. For example, the permutation 3,2,13,2,1 corresponds to π\pi such that
π(1)=3,    π(2)=2,    π(3)=1.(16.31)\pi\left( 1\right) =3,\ \ \ \ \pi\left( 2\right) =2,\ \ \ \ \pi\left( 3\right) =1.\tag{16.31}
The sign of the permutation, denoted by sign(π)\operatorname{sign}\left( \pi\right) is defined to be 11 if π\pi is even and 1-1 if π\pi is odd. When fully unpacked, the above formula reads
detM=M11M22M33+M12M23M31+M13M21M32                      M11M23M32M13M22M31M12M21M33.          (16.32)\begin{aligned}\det M & =M_{11}M_{22}M_{33}+M_{12}M_{23}M_{31}+M_{13}M_{21}M_{32}\ \ \ \ \ \ \ \ \ \ \\& \ \ \ \ \ \ \ \ \ \ \ \ -M_{11}M_{23}M_{32}-M_{13}M_{22}M_{31}-M_{12} M_{21}M_{33}.\ \ \ \ \ \ \ \ \ \ \left(16.32\right)\end{aligned}
From this expansion, it becomes apparent how the summation-based formula for the determinant essentially handpicks the 66 contributing terms.
However, thanks to our present experience with the tensor notation, we immediately observe that the same sum can be captured far more compactly by the indicial expression
detM=eijkM1iM2jM3k.(16.33)\det M=e^{ijk}M_{1i}M_{2j}M_{3k}.\tag{16.33}
Of course, this compactness comes at the expense of computational efficiency since the above contraction contains not 66 but 2727 terms, all of which are zero except for the 66 that matter. Nevertheless, the trade-off is well worth it. In fact, it seems almost as if the permutation systems were specifically formulated for the purpose of expressing the determinant.
Also note that for a set of three first-order systems UiU^{i}, ViV^{i}, and WiW^{i}, the combination
eijkUiVjWk(16.34)e_{ijk}U^{i}V^{j}W^{k}\tag{16.34}
corresponds to the determinant of the matrix whose columns (or rows) consist of UiU^{i}, ViV^{i}, and WiW^{i}, i.e.
eijkUiVjWk=U1V1W1U2V2W2U3V3W3.(16.35)e_{ijk}U^{i}V^{j}W^{k}=\left\vert \begin{array} {ccc} U^{1} & V^{1} & W^{1}\\ U^{2} & V^{2} & W^{2}\\ U^{3} & V^{3} & W^{3} \end{array} \right\vert .\tag{16.35}
As we observed in Chapter 3, if we think of UiU^{i}, ViV^{i}, and WiW^{i} as the components of the vectors U\mathbf{U}, V\mathbf{V}, and W\mathbf{W}, then the sign of
eijkUiVjWk(16.36)e_{ijk}U^{i}V^{j}W^{k}\tag{16.36}
tells us whether the orientation of U\mathbf{U}, V\mathbf{V}, and W\mathbf{W} is the same as that of the covariant basis Zi\mathbf{Z}_{i}.
We will leave the topic of determinants for now and will return to it later in this Chapter where we will discuss it in far greater detail.
The delta systems are objects of utmost elegance and usefulness. The delta systems are highly structured, expressive, and are governed by a tight set of algebraic rules. Thus, they go a long way towards the algebraization of our subject. Furthermore, the delta systems are tensors that have an equal number of superscripts and subscripts. As such, they introduce a great deal of balance into tensorial calculations.

16.5.1The complete delta system δrstijk\delta_{rst}^{ijk}

The complete delta system δrstijk\delta_{rst}^{ijk} is defined as the product of the permutation systems eijke^{ijk} and erste_{rst}, i.e.
δrstijk=eijkerst.(16.37)\delta_{rst}^{ijk}=e^{ijk}e_{rst}.\tag{16.37}
The term complete refers to the fact that it has as many superscripts and as many subscripts as the dimension of the space. The choice of the letter δ\delta is not arbitrary. The Kronecker delta δji\delta_{j}^{i} and the complete delta system δrstijk\delta_{rst}^{ijk} are part of the same family of related objects.
The delta system δrstijk\delta_{rst}^{ijk} has the greatest number of indices of any system we have encountered so far. Is this a d\'{ebauches} of indices or a choreographed dance? You be the judge. What is undeniable is that the delta system δrstijk\delta_{rst}^{ijk} possesses a great deal of structure that will enable us to work with it on an intuitive algebraic level.
Let us now explore its 36=7293^{6}=729 elements. Since the permutation systems consist of 1-1s, 00s, and 11s, so does the delta system δrstijk\delta_{rst} ^{ijk}. Its values can be summarized in the following table.
δrstijk={+1,if ijk and rst are permutations of 1,2,3 of like parities1,if ijk and rst are permutations of 1,2,3 of opposite parities+0,for all other combinations of indices.(16.38)\delta_{rst}^{ijk}=\left\{ \begin{array} {ll} \phantom{+} 1, & \text{if }ijk\text{ and }rst\text{ are permutations of }1,2,3\text{ of like parities}\\ -1, & \text{if }ijk\text{ and }rst\text{ are permutations of }1,2,3\text{ of opposite parities}\\ \phantom{+} 0, & \text{for all other combinations of indices.} \end{array} \right.\tag{16.38}
Thus, out of the 729729 elements, only (3!)2=36\left( 3!\right) ^{2}=36 are not zero.
Naturally, the delta system δrstijk\delta_{rst}^{ijk} is skew-symmetric in both its superscripts and its subscripts. It follows that simultaneously subjecting its superscripts and subscripts to the same permutation yields in an equivalent symbol. For example,
δrstijk=δsrtjik,(16.39)\delta_{rst}^{ijk}=\delta_{srt}^{jik},\tag{16.39}
where the first two superscripts and the first two subscripts were switched. Similarly,
δrstijk=δtrskij,(16.40)\delta_{rst}^{ijk}=\delta_{trs}^{kij},\tag{16.40}
where we moved the last superscript and the last subscript to the first position. More generally, subjecting its superscripts and subscripts to permutations of the same parity results in equivalent symbols. Finally, it is left as an exercise to demonstrate the frequently used identity
δijkijk=3!(16.41)\delta_{ijk}^{ijk}=3!\tag{16.41}
which will also be used in our upcoming discussion of the determinant.

16.5.2The relationship with the Kronecker delta

Of great utility is the following identity that expresses the delta system δrstijk\delta_{rst}^{ijk} in terms of the Kronecker deltas:
δrstijk=δriδsjδtk+δsiδtjδrk+δtiδrjδskδriδtjδskδtiδsjδrkδsiδrjδtk.(16.42)\delta_{rst}^{ijk}=\delta_{r}^{i}\delta_{s}^{j}\delta_{t}^{k}+\delta_{s} ^{i}\delta_{t}^{j}\delta_{r}^{k}+\delta_{t}^{i}\delta_{r}^{j}\delta_{s} ^{k}-\delta_{r}^{i}\delta_{t}^{j}\delta_{s}^{k}-\delta_{t}^{i}\delta_{s} ^{j}\delta_{r}^{k}-\delta_{s}^{i}\delta_{r}^{j}\delta_{t}^{k} .\tag{16.42}
It is left up to the reader to describe this identity in words. Undoubtedly, one can clearly see that the right side is related to the 3×33\times3 determinant. Indeed, the above relationship can be captured by the elegant equation
δrstijk=δriδsiδtiδrjδsjδtjδrkδskδtk.(16.43)\delta_{rst}^{ijk}=\left\vert \begin{array} {ccc} \delta_{r}^{i} & \delta_{s}^{i} & \delta_{t}^{i}\\ \delta_{r}^{j} & \delta_{s}^{j} & \delta_{t}^{j}\\ \delta_{r}^{k} & \delta_{s}^{k} & \delta_{t}^{k} \end{array} \right\vert .\tag{16.43}
Elegant though it may be, the preceding expanded form is more transparent and useful. Note that this relationship can also be adopted as the definition of the delta system δrstijk\delta_{rst}^{ijk}.
Explaining in complete detail why the identity
δrstijk=δriδsjδtk+δsiδtjδrk+δtiδrjδskδriδtjδskδtiδsjδrkδsiδrjδtk(16.42)\delta_{rst}^{ijk}=\delta_{r}^{i}\delta_{s}^{j}\delta_{t}^{k}+\delta_{s} ^{i}\delta_{t}^{j}\delta_{r}^{k}+\delta_{t}^{i}\delta_{r}^{j}\delta_{s} ^{k}-\delta_{r}^{i}\delta_{t}^{j}\delta_{s}^{k}-\delta_{t}^{i}\delta_{s} ^{j}\delta_{r}^{k}-\delta_{s}^{i}\delta_{r}^{j}\delta_{t}^{k} \tag{16.42}
is valid is left as an exercise. It is a good idea to consider a number of specific examples and to articulate the precise reason for each conclusion. To help you in this endeavor, let us consider three such examples that will shed some light on the "inner mechanics" of the expression on the right.
First, consider a typical nonzero element δ231123\delta_{231}^{123} which should equal 11 since 2,3,12,3,1 is two switches away from 1,2,31,2,3. We have
δ231123=δ21δ32δ13+δ31δ12δ23+δ11δ22δ33δ21δ12δ33δ11δ32δ23δ31δ22δ13.(16.44)\delta_{231}^{123}=\delta_{2}^{1}\delta_{3}^{2}\delta_{1}^{3}+\delta_{3} ^{1}\delta_{1}^{2}\delta_{2}^{3}+\delta_{1}^{1}\delta_{2}^{2}\delta_{3} ^{3}-\delta_{2}^{1}\delta_{1}^{2}\delta_{3}^{3}-\delta_{1}^{1}\delta_{3} ^{2}\delta_{2}^{3}-\delta_{3}^{1}\delta_{2}^{2}\delta_{1}^{3}.\tag{16.44}
Observe that the only nonzero term is δ11δ22δ33\delta_{1}^{1}\delta_{2}^{2}\delta _{3}^{3}, which indeed equals 11. For a second example, consider δ122123\delta_{122}^{123} which should equal 00 because 1,2,21,2,2 is not a permutation. We have
δ122123=δ11δ22δ23+δ21δ22δ13+δ21δ12δ23δ11δ22δ23δ21δ22δ13δ21δ12δ23,(16.45)\delta_{122}^{123}=\delta_{1}^{1}\delta_{2}^{2}\delta_{2}^{3}+\delta_{2} ^{1}\delta_{2}^{2}\delta_{1}^{3}+\delta_{2}^{1}\delta_{1}^{2}\delta_{2} ^{3}-\delta_{1}^{1}\delta_{2}^{2}\delta_{2}^{3}-\delta_{2}^{1}\delta_{2} ^{2}\delta_{1}^{3}-\delta_{2}^{1}\delta_{1}^{2}\delta_{2}^{3},\tag{16.45}
where all terms equal 00 because it is impossible to have a nonzero term when the superscripts and the subscripts are not identical sets of numbers. Finally, consider the element δ113113\delta_{113}^{113} which should also equal 00, because while the superscripts and the subscripts represent identical sets of numbers, neither is a permutation. We have
δ113113=δ11δ11δ33+δ11δ31δ13+δ31δ11δ13δ11δ31δ13δ31δ11δ13δ11δ11δ33.(16.46)\delta_{113}^{113}=\delta_{1}^{1}\delta_{1}^{1}\delta_{3}^{3}+\delta_{1} ^{1}\delta_{3}^{1}\delta_{1}^{3}+\delta_{3}^{1}\delta_{1}^{1}\delta_{1} ^{3}-\delta_{1}^{1}\delta_{3}^{1}\delta_{1}^{3}-\delta_{3}^{1}\delta_{1} ^{1}\delta_{1}^{3}-\delta_{1}^{1}\delta_{1}^{1}\delta_{3}^{3}.\tag{16.46}
This time, the two nonzero terms -- first and last -- cancel each other.
Despite its bulkiness, the identity
δrstijk=δriδsjδtk+δsiδtjδrk+δtiδrjδskδriδtjδskδtiδsjδrkδsiδrjδtk(16.42)\delta_{rst}^{ijk}=\delta_{r}^{i}\delta_{s}^{j}\delta_{t}^{k}+\delta_{s} ^{i}\delta_{t}^{j}\delta_{r}^{k}+\delta_{t}^{i}\delta_{r}^{j}\delta_{s} ^{k}-\delta_{r}^{i}\delta_{t}^{j}\delta_{s}^{k}-\delta_{t}^{i}\delta_{s} ^{j}\delta_{r}^{k}-\delta_{s}^{i}\delta_{r}^{j}\delta_{t}^{k} \tag{16.42}
finds numerous practical applications. However, its immediate theoretical impact is the crucial implication that the delta system δrstijk\delta_{rst}^{ijk} is a tensor, as it is expressed by sums and products of the tensor Kronecker delta. Additionally, the same identity shows that δrstijk\delta_{rst}^{ijk} vanishes under the covariant derivative, i.e.
pδrstijk=0.(16.47)\nabla_{p}\delta_{rst}^{ijk}=0.\tag{16.47}
In other words, the metrinilic property of the covariant derivative extends to the delta system δrstijk\delta_{rst}^{ijk}.
Interestingly, not only can the delta system be expressed in terms of the Kronecker delta, but the latter can also be expressed in terms of the former. This relationship is captured by the identity
δri=12δrjkijk(16.48)\delta_{r}^{i}=\frac{1}{2}\delta_{rjk}^{ijk}\tag{16.48}
which is one of many that will be discussed next.

16.5.3The partial delta systems

Between the Kronecker delta δji\delta_{j}^{i} and the complete delta system δrstijk\delta_{rst}^{ijk}, there exists an object with an intermediate number of indices known as the partial delta system δrsij\delta_{rs}^{ij}. In fact, when we generalize this discussion to nn dimensions, we will discover that between the Kronecker delta δji\delta_{j}^{i} and the complete delta system δj1jni1in\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}, there exists a whole family of partial delta systems δj1jki1ik\delta_{j_{1}\cdots j_{k}} ^{i_{1}\cdots i_{k}}, one for every kk between 11 and nn. The partial deltas are defined in a way that bridges the Kronecker delta and the complete delta system.
For the remainder of this Section, we will operate in three dimensions which will limit our discussion to the delta system δrsij\delta_{rs}^{ij}. However, the reader is invited to think about how all of the presented concepts generalize to nn dimensions. In describing the elements of δrsij\delta_{rs}^{ij}, we will choose our words so that the resulting definition also encompasses the Kronecker delta δji\delta_{j}^{i} and the complete delta system δrstijk\delta _{rst}^{ijk} and generalizes to nn dimensions. We have:
δrsij={+1+if the superscripts and the subscripts are identical setsof distinct numbers related by an even permutation+1,+if the superscripts and the subscripts are identical setsof distinct numbers related by an odd permutation++0,for all other combinations of indices.      (16.49)\delta_{rs}^{ij}=\left\{ \begin{array} {ll} \begin{array} {c} \\ \phantom{+} 1\text{, }\\ \\ \phantom{+} \end{array} & \begin{array} {l} \\ \text{if the superscripts and the subscripts are identical sets}\\ \text{of distinct numbers related by an \textit{even} permutation}\\ \phantom{+} \end{array} \\ \begin{array} {c} -1\text{,}\\ \\ \phantom{+} \end{array} & \begin{array} {l} \text{if the superscripts and the subscripts are identical sets}\\ \text{of distinct numbers related by an \textit{odd} permutation}\\ \phantom{+} \end{array} \\ \begin{array} {c} \phantom{+} 0\text{,}\\ \end{array} & \begin{array} {c} \text{for all other combinations of indices.}\\ \end{array} \end{array} \ \ \ \ \ \ \right.\tag{16.49}
A few examples of the values of δrsij\delta_{rs}^{ij} and the reasons for those values are given in the following table.
δ1313=+1Identical sets of distinct numbers related by the zero permutationδ1211=+0Superscripts are not distinctδ1111=+0Neither superscripts nor subscripts are distinctδ2312=+0Superscripts and subscripts are not identical setsδ3223=1Identical sets of distinct numbers related by a singles witch(16.50)\begin{array} {ll} \delta_{13}^{13}= \phantom{+} 1 & \text{Identical sets of distinct numbers related by the zero permutation}\\ \delta_{12}^{11}= \phantom{+} 0 & \text{Superscripts are not distinct}\\ \delta_{11}^{11}= \phantom{+} 0 & \text{Neither superscripts nor subscripts are distinct}\\ \delta_{23}^{12}= \phantom{+} 0 & \text{Superscripts and subscripts are not identical sets}\\ \delta_{32}^{23}=-1 & \text{Identical sets of distinct numbers related by a singles witch} \end{array}\tag{16.50}
Thus, out of a total of 8181 elements, only 66 are not zero. Specifically,
δ1212=δ2323=δ3131=1,(16.51)\delta_{12}^{12}=\delta_{23}^{23}=\delta_{31}^{31}=1,\tag{16.51}
and
δ2112=δ3223=δ1331=1.(16.52)\delta_{21}^{12}=\delta_{32}^{23}=\delta_{13}^{31}=-1.\tag{16.52}
All delta systems, from the Kronecker delta to the complete delta system are connected by various identities. Generally speaking, a contraction of a higher-order delta system yields a lower-order delta system. Meanwhile, higher-order delta systems can be expressed in terms of the Kronecker delta by determinant-like combinations. Demonstrations of all identities are left as exercises.
Let us begin by considering the result of contracting higher-order delta systems. Starting with the complete system δrstijk\delta_{rst}^{ijk}, we have
δrskijk=δrsij.(16.53)\delta_{rsk}^{ijk}=\delta_{rs}^{ij}.\tag{16.53}
This is one of the most frequently used identities in practical applications. In fact, δrsij\delta_{rs}^{ij} almost always arises as the result of contracting δrstijk\delta_{rst}^{ijk}. Furthermore, by the contraction property of tensors, the above identity immediately tells us that the partial delta system δrsij\delta_{rs}^{ij} is a tensor, leading to the conclusion that all delta systems are tensors. Additionally, thanks to the above identity, we know that δrsij\delta_{rs}^{ij} vanishes under the covariant derivative, i.e.
pδrsij=0.(16.54)\nabla_{p}\delta_{rs}^{ij}=0.\tag{16.54}
When δrsij\delta_{rs}^{ij} is itself contracted, the result is a multiple of the Kronecker delta, i.e.
δrjij=2δri.(16.55)\delta_{rj}^{ij}=2\delta_{r}^{i}.\tag{16.55}
To understand where the factor of 22 comes from, note that
δrjij=δr1i1+δr2i2+δr3i3.(16.56)\delta_{rj}^{ij}=\delta_{r1}^{i1}+\delta_{r2}^{i2}+\delta_{r3}^{i3}.\tag{16.56}
When i=ri=r and both values equal, say, 11, we have
δ1j1j=δ1111+δ1212+δ1313=0+1+1=2(16.57)\delta_{1j}^{1j}=\delta_{11}^{11}+\delta_{12}^{12}+\delta_{13}^{13}=0+1+1=2\tag{16.57}
and it is clear that the same result would be obtained for any common value of ii and rr.
In summary, the contraction of a higher-order delta system results in a lower-order delta system or a multiple thereof.
Now, let us talk about expressing higher-order delta systems in terms of the Kronecker delta. Recall that the complete delta system δrstijk\delta_{rst}^{ijk} is given in terms of the Kronecker delta by the formula
δrstijk=δriδsjδtk+δsiδtjδrk+δtiδrjδskδriδtjδskδtiδsjδrkδsiδrjδtk.(16.42)\delta_{rst}^{ijk}=\delta_{r}^{i}\delta_{s}^{j}\delta_{t}^{k}+\delta_{s} ^{i}\delta_{t}^{j}\delta_{r}^{k}+\delta_{t}^{i}\delta_{r}^{j}\delta_{s} ^{k}-\delta_{r}^{i}\delta_{t}^{j}\delta_{s}^{k}-\delta_{t}^{i}\delta_{s} ^{j}\delta_{r}^{k}-\delta_{s}^{i}\delta_{r}^{j}\delta_{t}^{k} .\tag{16.42}
Contracting on kk and tt, we discover that
δrsij=δriδsjδrjδsi.(16.58)\delta_{rs}^{ij}=\delta_{r}^{i}\delta_{s}^{j}-\delta_{r}^{j}\delta_{s} ^{i}.\tag{16.58}
This is one of the most commonly used identities, along with
δrskijk=δrsij(16.53)\delta_{rsk}^{ijk}=\delta_{rs}^{ij}\tag{16.53}
In fact, these identities are usually used in tandem: a contraction reduces δrstijk\delta_{rst}^{ijk} to δrsij\delta_{rs}^{ij} which is followed by the representation of δrsij\delta_{rs}^{ij} in terms of the Kronecker delta. Thus, for convenience, let us combine the two identities into one, i.e.
δrskijk=δriδsjδrjδsi.(16.59)\delta_{rsk}^{ijk}=\delta_{r}^{i}\delta_{s}^{j}-\delta_{r}^{j}\delta_{s} ^{i}.\tag{16.59}
Note that the expression on the right once again exhibits the familiar 2×22\times2 determinant pattern and can therefore be captured elegantly, albeit impractically, in the following way:
δrsij=δriδsiδrjδsj.(16.60)\delta_{rs}^{ij}=\left\vert \begin{array} {cc} \delta_{r}^{i} & \delta_{s}^{i}\\ \delta_{r}^{j} & \delta_{s}^{j} \end{array} \right\vert .\tag{16.60}
This completes our discussion of permutation and delta systems in three dimensions and we will now turn our attention to lower and higher-dimensional cases.
The permutation systems, the delta systems, and all of the related identities naturally generalize to any number of dimensions. The two-dimensional case is particularly important since it plays a crucial role in the study of two-dimensional surfaces.

16.6.1In two dimensions

In two dimensions, all indices assume values 11 and 22. The permutation systems eije_{ij} and eije^{ij} are given by
eij,eij={+1,if ij is an even permutation of 1,21,if ij is an odd permutation of 1,2+0,otherwise.(16.61)e_{ij},e^{ij}=\left\{ \begin{array} {ll} \phantom{+} 1\text{,} & \text{if }ij\text{ is an even permutation of }1,2\\ -1\text{,} & \text{if }ij\text{ is an odd permutation of }1,2\\ \phantom{+} 0\text{,} & \text{otherwise.} \end{array} \right.\tag{16.61}
However, it may be even easier simply to list all of the elements. For eije_{ij}, we have
e12=1,    e21=1,    and    e11=e22=0(16.62)e_{12}=1,\ \ \ \ e_{21}=-1,\ \ \ \ \text{and}\ \ \ \ e_{11}=e_{22}=0\tag{16.62}
and for eije^{ij} we have
e12=1,    e21=1,    and    e11=e22=0.(16.63)e^{12}=1,\ \ \ \ e^{21}=-1,\ \ \ \ \text{and}\ \ \ \ e^{11}=e^{22}=0.\tag{16.63}
In matrix terms, both eije_{ij} and eije^{ij} correspond to the matrix
[0110].(16.64)\left[ \begin{array} {rr} 0 & 1\\ -1 & 0 \end{array} \right] .\tag{16.64}
The complete delta system δrsij\delta_{rs}^{ij} is the product of the permutation systems, i.e.
δrsij=eijers.(16.65)\delta_{rs}^{ij}=e^{ij}e_{rs}.\tag{16.65}
Note that, generally speaking, the symbol δrsij\delta_{rs}^{ij} is ambiguous since it represents different systems in different dimensions. Thus, when working with these symbols, it is essential to indicate the relevant dimension.
This point is highlighted by the contraction property
δri=δrjij(16.66)\delta_{r}^{i}=\delta_{rj}^{ij}\tag{16.66}
satisfied by δrsij\delta_{rs}^{ij} in two dimensions. Note that the analogous identity in three dimensions read δri=2δrjij\delta_{r}^{i}=2\delta_{rj}^{ij}. On the other hand, the expression for δrsij\delta_{rs}^{ij} in terms of the Kronecker delta reads
δrsij=δriδsjδrjδsi,(16.67)\delta_{rs}^{ij}=\delta_{r}^{i}\delta_{s}^{j}-\delta_{r}^{j}\delta_{s}^{i},\tag{16.67}
which is identical to the three-dimensional case. It turns out that the form of this identity is universal for all dimensions.

16.6.2In higher dimensions

We must begin by noting that the case n>3n\gt 3 does not correspond to a geometric Euclidean space as we have defined it. Nevertheless, the analytical framework holds up perfectly for any nn. Furthermore, in the future -- in Chapter 20, to be precise -- we will generalize the concept of Euclidean spaces to higher dimensions.
In an nn-dimensional space, indices run the values between 11 and nn. The permutation systems ei1ine_{i_{1}\cdots i_{n}} and ei1ine^{i_{1}\cdots i_{n}} are defined as follows.
ei1in,ei1in={+1,if i1in is an even permutation of 1,,n1,if i1in is an odd permutation of 1,,n+0,if i1in is not a permutation of 1,,n.(16.68)e_{i_{1}\cdots i_{n}},e^{i_{1}\cdots i_{n}}=\left\{ \begin{array} {ll} \phantom{+} 1\text{,} & \text{if }i_{1}\cdots i_{n}\text{ is an even permutation of }1,\ldots,n\\ -1\text{,} & \text{if }i_{1}\cdots i_{n}\text{ is an odd permutation of }1,\ldots,n\\ \phantom{+} 0\text{,} & \text{if }i_{1}\cdots i_{n}\text{ is not a permutation of }1,\ldots,n. \end{array} \right.\tag{16.68}
Thus, each permutation system has nnn^{n} elements of which only n!n! are not zero.
The complete delta system δj1jni1in\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}} is the product of the permutation systems, i.e.
δj1jni1in=ei1inej1jn,(16.69)\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}=e^{i_{1}\cdots i_{n}} e_{j_{1}\cdots j_{n}},\tag{16.69}
and has the values
δj1jni1in={+1,if i1in and j1jn are permutationsof 1,,n of the same parity+1,if i1in and j1jn are permutationsof 1,,n of opposite parities++0,for all other combinations of indices.(16.70)\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}=\left\{ \begin{array} {cc} \begin{array} {c} \phantom{+} 1\text{,}\\ \\ \end{array} & \begin{array} {l} \text{if }i_{1}\cdots i_{n}\text{ and }j_{1}\cdots j_{n}\text{ are permutations}\\ \text{of }1,\cdots,n\text{ of the same parity}\\ \phantom{+} \end{array} \\ \begin{array} {c} -1\text{,}\\ \\ \end{array} & \begin{array} {l} \text{if }i_{1}\cdots i_{n}\text{ and }j_{1}\cdots j_{n}\text{ are permutations}\\ \text{of }1,\cdots,n\text{ of opposite parities}\\ \phantom{+} \end{array} \\ \begin{array} {c} \phantom{+} 0\text{,}\\ \end{array} & \begin{array} {c} \text{for all other combinations of indices.}\\ \end{array} \end{array} \right.\tag{16.70}
The complete delta system can be expressed in terms of the Kronecker deltas by the identity
δj1jni1in=δj1i1δjni1δj1inδjnin,(16.71)\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}=\left\vert \begin{array} {ccc} \delta_{j_{1}}^{i_{1}} & \cdots & \delta_{j_{n}}^{i_{1}}\\ \vdots & \ddots & \vdots\\ \delta_{j_{1}}^{i_{n}} & \cdots & \delta_{j_{n}}^{i_{n}} \end{array} \right\vert ,\tag{16.71}
which proves that it is a tensor, and that it vanishes under the covariant derivative, i.e.
pδj1jni1in=0.(16.72)\nabla_{p}\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}=0.\tag{16.72}
The partial delta systems δj1jki1ik\delta_{j_{1}\cdots j_{k}}^{i_{1}\cdots i_{k}} for every kk between 11 and nn can be defined by the very same language as in the two- and three-dimensional cases, i.e.
δj1jki1ik={+1+if the superscripts and the subscripts are identical setsof distinct numbers related by an even permutation+1,+if the superscripts and the subscripts are identical setsof distinct numbers related by an odd permutation++0,for all other combinations of indices.(16.73)\delta_{j_{1}\cdots j_{k}}^{i_{1}\cdots i_{k}}=\left\{ \begin{array} {ll} \begin{array} {c} \phantom{+} 1\text{, }\\ \\ \phantom{+} \end{array} & \begin{array} {l} \text{if the superscripts and the subscripts are identical sets}\\ \text{of distinct numbers related by an \textit{even} permutation}\\ \phantom{+} \end{array} \\ \begin{array} {c} -1\text{,}\\ \\ \phantom{+} \end{array} & \begin{array} {l} \text{if the superscripts and the subscripts are identical sets}\\ \text{of distinct numbers related by an \textit{odd} permutation}\\ \phantom{+} \end{array} \\ \begin{array} {c} \phantom{+} 0\text{,}\\ \end{array} & \begin{array} {c} \text{for all other combinations of indices.}\\ \end{array} \end{array} \right.\tag{16.73}
It is, once again, important to point out that this language includes the Kronecker delta δji\delta_{j}^{i} as well as the complete delta system δj1jni1in\delta_{j_{1}\cdots j_{n}}^{i_{1}\cdots i_{n}}. For every kk, the partial delta system δj1jki1ik\delta_{j_{1}\cdots j_{k}}^{i_{1}\cdots i_{k}} can be expressed in terms of the Kronecker delta by
δj1jki1ik=δj1i1δjki1δj1ikδjkik,(16.74)\delta_{j_{1}\cdots j_{k}}^{i_{1}\cdots i_{k}}=\left\vert \begin{array} {ccc} \delta_{j_{1}}^{i_{1}} & \cdots & \delta_{j_{k}}^{i_{1}}\\ \vdots & \ddots & \vdots\\ \delta_{j_{1}}^{i_{k}} & \cdots & \delta_{j_{k}}^{i_{k}} \end{array} \right\vert ,\tag{16.74}
which shows the great universality of this identity.
A contraction of any delta system results in a scalar multiple of the lower-order delta system. It is left as an exercise to show that
δj1jn1pi1in1p=δj1jn1i1in1          (16.75)δj1jn2pi1in2p=2δj1jn2i1in2          (16.76)δj1jn3pi1in3p=3δj1jn3i1in3          (16.77)\begin{aligned}\delta_{j_{1}\cdots j_{n-1}p}^{i_{1}\cdots i_{n-1}p} & =\delta_{j_{1}\cdots j_{n-1}}^{i_{1}\cdots i_{n-1}}\ \ \ \ \ \ \ \ \ \ \left(16.75\right)\\\delta_{j_{1}\cdots j_{n-2}p}^{i_{1}\cdots i_{n-2}p} & =2\delta_{j_{1}\cdots j_{n-2}}^{i_{1}\cdots i_{n-2}}\ \ \ \ \ \ \ \ \ \ \left(16.76\right)\\\delta_{j_{1}\cdots j_{n-3}p}^{i_{1}\cdots i_{n-3}p} & =3\delta_{j_{1}\cdots j_{n-3}}^{i_{1}\cdots i_{n-3}}\ \ \ \ \ \ \ \ \ \ \left(16.77\right)\end{aligned}
and so on. In general, the multiple corresponding to the contraction of the kthk^{\text{th}}-order system δj1jk1jki1ik1ik\delta_{j_{1}\cdots j_{k-1}j_{k}}^{i_{1}\cdots i_{k-1}i_{k}} is n(k1)n-\left( k-1\right) , i.e.
δj1jk1pi1ik1p=(n(k1))δj1jk1i1ik1.(16.78)\delta_{j_{1}\cdots j_{k-1}p}^{i_{1}\cdots i_{k-1}p}=\left( n-\left( k-1\right) \right) \delta_{j_{1}\cdots j_{k-1}}^{i_{1}\cdots i_{k-1}}.\tag{16.78}
For a specific example, let us consider the case n=5n=5 and denote the complete delta system by the symbol δrstuvijklm\delta_{rstuv}^{ijklm}. Then we have the following relationships
δrstupijklp=δrstuijkl,          (16.79)δrstpijkp=2δrstijk,          (16.80)δrspijp=3δrsij,          (16.81)δrpip=4δri,          (16.82)δpp=5.          (16.83)\begin{aligned}\delta_{rstup}^{ijklp} & =\delta_{rstu}^{ijkl},\ \ \ \ \ \ \ \ \ \ \left(16.79\right)\\\delta_{rstp}^{ijkp} & =2\delta_{rst}^{ijk},\ \ \ \ \ \ \ \ \ \ \left(16.80\right)\\\delta_{rsp}^{ijp} & =3\delta_{rs}^{ij},\ \ \ \ \ \ \ \ \ \ \left(16.81\right)\\\delta_{rp}^{ip} & =4\delta_{r}^{i},\ \ \ \ \ \ \ \ \ \ \left(16.82\right)\\\delta_{p}^{p} & =5.\ \ \ \ \ \ \ \ \ \ \left(16.83\right)\end{aligned}
Thus, in particular,
δijklmijklm=5!(16.84)\delta_{ijklm}^{ijklm}=5!\tag{16.84}
and, in general,
δi1ini1in=n!.(16.85)\delta_{i_{1}\cdots i_{n}}^{i_{1}\cdots i_{n}}=n!.\tag{16.85}
In Matrix Algebra, it is common to represent a square matrix AA as a sum of a symmetric matrix SS and a skew-symmetric matrix KK, i.e.
A=S+K.(16.86)A=S+K.\tag{16.86}
For example, for
A=[123456789](16.87)A=\left[ \begin{array} {ccc} 1 & 2 & 3\\ 4 & 5 & 6\\ 7 & 8 & 9 \end{array} \right]\tag{16.87}
we have
[123456789]=[135357579]+[012101210].(16.88)\left[ \begin{array} {ccc} 1 & 2 & 3\\ 4 & 5 & 6\\ 7 & 8 & 9 \end{array} \right] =\left[ \begin{array} {ccc} 1 & 3 & 5\\ 3 & 5 & 7\\ 5 & 7 & 9 \end{array} \right] +\left[ \begin{array} {rrr} 0 & -1 & -2\\ 1 & 0 & -1\\ 2 & 1 & 0 \end{array} \right] .\tag{16.88}
Such a decomposition is unique and, as it is easy to show, we must have
S=A+AT2(16.89)S=\frac{A+A^{T}}{2}\tag{16.89}
and
K=AAT2.(16.90)K=\frac{A-A^{T}}{2}.\tag{16.90}
The matrix SS is known as the symmetric part of AA and KK is known as the skew-symmetric part of AA.
In indicial notation, the same identities read
Aij=Sij+Kij,(16.91)A_{ij}=S_{ij}+K_{ij},\tag{16.91}
where
Sij=12(Aij+Aji)(16.92)S_{ij}=\frac{1}{2}\left( A_{ij}+A_{ji}\right)\tag{16.92}
and
Kij=12(AijAji).(16.93)K_{ij}=\frac{1}{2}\left( A_{ij}-A_{ji}\right) .\tag{16.93}
In the expression for KijK_{ij}, we are beginning to see the now-familiar skew-symmetric indicial pattern.
The concepts of symmetric and skew-symmetric parts generalize to higher-order systems (although it is no longer true that a system is a sum of its symmetric and skew-symmetric parts). For example, for a third-order system AijkA_{ijk}, its skew-symmetric part SijkS_{ijk} is defined as
Sijk=Aijk+Ajki+AkijAjikAikjAkji,(16.94)S_{ijk}=A_{ijk}+A_{jki}+A_{kij}-A_{jik}-A_{ikj}-A_{kji},\tag{16.94}
and it is clear how to generalize this definition to higher-order systems. The operation that converts AijkA_{ijk} into its skew-symmetric part is known as alternatization and it is valid in any dimension that is equal or greater than the order of the system. Alternatization is relevant to our narrative because it can be used as an alternative approach to constructing invariant differential operators.
As we established, the partial derivative
TiZk(16.95)\frac{\partial T_{i}}{\partial Z^{k}}\tag{16.95}
for a tensor TiT_{i} is not a tensor in its own right. We have remedied this problem by replacing the partial derivative with the combination
kTi=TiZkΓikmTm(16.96)\nabla_{k}T_{i}=\frac{\partial T_{i}}{\partial Z^{k}}-\Gamma_{ik}^{m}T_{m}\tag{16.96}
which is a tensor. Observe, however, that the combination
ΓikmTm(16.97)\Gamma_{ik}^{m}T_{m}\tag{16.97}
which represents the non-tensor portion of the partial derivative is symmetric in ii and kk. Therefore, in the alternatization of kTi\nabla_{k}T_{i}, i.e.
kTiiTk,(16.98)\nabla_{k}T_{i}-\nabla_{i}T_{k},\tag{16.98}
the terms containing the Christoffel symbol cancel each other. As a result,
kTiiTk=TiZkTkTi.(16.99)\nabla_{k}T_{i}-\nabla_{i}T_{k}=\frac{\partial T_{i}}{\partial Z^{k}} -\frac{\partial T_{k}}{\partial T_{i}}.\tag{16.99}
Thus, the alternatization
TiZkTkTi(16.100)\frac{\partial T_{i}}{\partial Z^{k}}-\frac{\partial T_{k}}{\partial T_{i}}\tag{16.100}
is also a tensor. In other words, the tensor property can be achieved by alternatization -- once again, by virtue of cancelling the non-tensor contributions. This remains true for tensors of any order as well as for partial derivatives of any order. For example, the alternatization of the variant
2TijZkZl(16.101)\frac{\partial^{2}T_{ij}}{\partial Z^{k}\partial Z^{l}}\tag{16.101}
is a tensor.
Thus, it would be of great value to express alternatization algebraically. This can be accomplished by contraction with the appropriate partial delta system. For example, consider a third-order system AijkA_{ijk} in an nn-dimensional case where n3n\geq3. Recall that SijkS_{ijk}, the skew-symmetric part, i.e. alternatized version, of AijkA_{ijk}, is given by
Sijk=Aijk+Ajki+AkijAjikAikjAkji.(16.94)S_{ijk}=A_{ijk}+A_{jki}+A_{kij}-A_{jik}-A_{ikj}-A_{kji}. \tag{16.94}
With the help of the partial delta system δijkrst\delta_{ijk}^{rst}, SijkS_{ijk} can be expressed by a simple contraction, i.e.
Sijk=δijkrstArst.(16.102)S_{ijk}=\delta_{ijk}^{rst}A_{rst}.\tag{16.102}
This follows from the formula
δijkrst=δirδjsδkt+δjrδksδit+δkrδisδjtδirδksδjtδkrδjsδitδjrδisδkt,(16.42)\delta_{ijk}^{rst}=\delta_{i}^{r}\delta_{j}^{s}\delta_{k}^{t}+\delta_{j} ^{r}\delta_{k}^{s}\delta_{i}^{t}+\delta_{k}^{r}\delta_{i}^{s}\delta_{j} ^{t}-\delta_{i}^{r}\delta_{k}^{s}\delta_{j}^{t}-\delta_{k}^{r}\delta_{j} ^{s}\delta_{i}^{t}-\delta_{j}^{r}\delta_{i}^{s}\delta_{k}^{t}, \tag{16.42}
but working out the details is left as an exercise.
Thanks to the algebraic precision of the equation
Sijk=δijkrstArst.(16.102)S_{ijk}=\delta_{ijk}^{rst}A_{rst}. \tag{16.102}
we will adopt it as the definition of alternatization, i.e. contraction on all superscripts (or subscripts) with the partial delta system of the appropriate order. For example, the alternatization of 2Tij/ZkZl\partial^{2}T_{ij}/\partial Z^{k}\partial Z^{l} is
δijklrstuTrsZtZu.(16.103)\delta_{ijkl}^{rstu}\frac{\partial T_{rs}}{\partial Z^{t}\partial Z^{u}}.\tag{16.103}
As we alluded to at the top of this Chapter, alternatization as a mechanism for achieving invariance is developed in the subject of Differential Forms, where it serves as the basis for the operation of exterior derivative.
We have already demonstrated that the determinant of a 3×33\times3 system can be expressed in an elegant and concise fashion with the help of the permutation system eijke^{ijk}, i.e.
detM=eijkM1iM2jM3k.(16.33)\det M=e^{ijk}M_{1i}M_{2j}M_{3k}. \tag{16.33}
However, in view of the literal superscripts, this expression does not live up to the tensorial standard that we have established. Therefore, further development is required in order to achieve full tensorization of the determinant. This will be the initial goal of this Section.

16.8.1Tensor expressions for the determinant

The concept of the determinant applies to second-order systems with scalar elements. For the purposes of the present discussion let us consider a mixed second-order system AjiA_{j}^{i}, even though, for the time being, we are not particularly concerned with the tensor properties of AjiA_{j}^{i}. Adjusting the above identity to the mixed indicial signature, note that detA\det A can be expressed as follows:
detA=eijkAi1Aj2Ak3.(16.104)\det A=e^{ijk}A_{i}^{1}A_{j}^{2}A_{k}^{3}.\tag{16.104}
Let us switch the indices 11 and 22 in the expression above and consider the combination
eijkAi2Aj1Ak3,(16.105)e^{ijk}A_{i}^{2}A_{j}^{1}A_{k}^{3},\tag{16.105}
Unsurprisingly, as a result of the switch, the expression changes sign thanks to the skew-symmetric nature of eijke^{ijk}. Showing this requires three formal steps. First, switch the order of the terms Ai1A_{i}^{1} and Aj2A_{j}^{2} in order to return the literal superscripts to their original order. This, of course, does not change the value of the expression, so
eijkAi2Aj1Ak3=eijkAj1Ai2Ak3.(16.106)e^{ijk}A_{i}^{2}A_{j}^{1}A_{k}^{3}=e^{ijk}A_{j}^{1}A_{i}^{2}A_{k}^{3}.\tag{16.106}
Second, switch the names of the dummy indices ii and jj, which again leaves the value of the expression unchanged, i.e.
eijkAi2Aj1Ak3=eijkAj1Ai2Ak3=ejikAi1Aj2Ak3.(16.107)e^{ijk}A_{i}^{2}A_{j}^{1}A_{k}^{3}=e^{ijk}A_{j}^{1}A_{i}^{2}A_{k}^{3} =e^{jik}A_{i}^{1}A_{j}^{2}A_{k}^{3}.\tag{16.107}
Finally, note that ejik=eijke^{jik}=-e^{ijk} by the skew-symmetric property and therefore
eijkAi2Aj1Ak3=eijkAj1Ai2Ak3=ejikAi1Aj2Ak3=eijkAi1Aj2Ak3(16.108)e^{ijk}A_{i}^{2}A_{j}^{1}A_{k}^{3}=e^{ijk}A_{j}^{1}A_{i}^{2}A_{k}^{3} =e^{jik}A_{i}^{1}A_{j}^{2}A_{k}^{3}=-e^{ijk}A_{i}^{1}A_{j}^{2}A_{k}^{3}\tag{16.108}
Thus, in summary,
eijkAi2Aj1Ak3=eijkAi1Aj2Ak3,(16.109)e^{ijk}A_{i}^{2}A_{j}^{1}A_{k}^{3}=-e^{ijk}A_{i}^{1}A_{j}^{2}A_{k}^{3},\tag{16.109}
as we set out to show.
Of course, we would reach a similar conclusion if we switched any two of the literal superscripts. In other words, we have just demonstrated that the third-order system
eijkAirAjsAkt(16.110)e^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}\tag{16.110}
is fully skew-symmetric in the superscripts rr, ss, tt. It must, therefore, be a scalar multiple of erste^{rst}, i.e.
eijkAirAjsAkt=αerst,(16.111)e^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}=\alpha e^{rst},\tag{16.111}
as we established earlier. Unsurprisingly, the scalar α\alpha is precisely detA\det A, as can be confirmed by setting rr, ss, and tt to 11, 22, and 33. Thus, we have obtained our first fully tensorial identity involving the determinant:
eijkAirAjsAkt=detA erst.(16.112)e^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}=\det A~e^{rst}.\tag{16.112}
For future reference, note that we could similarly show the identity
erstAirAjsAkt=detA eijk(16.113)e_{rst}A_{i}^{r}A_{j}^{s}A_{k}^{t}=\det A~e_{ijk}\tag{16.113}
which uses subscripted permutation systems.
In order to obtain an explicit expression for AA, contract both sides of
eijkAirAjsAkt=detA erst.(16.112)e^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}=\det A~e^{rst}. \tag{16.112}
with erste_{rst}, i.e.
ersteijkAirAjsAkt=detA ersterst.(16.114)e_{rst}e^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}=\det A~e_{rst}e^{rst}.\tag{16.114}
Recall that
ersteijk=δrstijk,(16.115)e_{rst}e^{ijk}=\delta_{rst}^{ijk},\tag{16.115}
and
ersterst=δrstrst=3!.(16.116)e_{rst}e^{rst}=\delta_{rst}^{rst}=3!.\tag{16.116}
Thus, we have arrived at the landmark formula
detA=13!δrstijkAirAjsAkt.(16.117)\det A=\frac{1}{3!}\delta_{rst}^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}.\tag{16.117}
for the determinant of AjiA_{j}^{i} that fully conforms to all rules of the tensor notation. In particular, it instantly tells us that the determinant of a mixed second-order tensor AjiA_{j}^{i} is an invariant.
Having accomplished the goal of this Section, let us give the analogous formulas for systems enumerated by two subscripts, i.e.
eijkAirAjsAkt=detA erst,(16.118)e^{ijk}A_{ir}A_{js}A_{kt}=\det A~e_{rst},\tag{16.118}
and
detA=13!eijkerstAirAjsAkt.(16.119)\det A=\frac{1}{3!}e^{ijk}e^{rst}A_{ir}A_{js}A_{kt}.\tag{16.119}
and those enumerated by two superscripts, i.e.
eijkAirAjsAkt=detA erst(16.120)e_{ijk}A^{ir}A^{js}A^{kt}=\det A~e^{rst}\tag{16.120}
and
detA=13!eijkerstAirAjsAkt.(16.121)\det A=\frac{1}{3!}e_{ijk}e_{rst}A^{ir}A^{js}A^{kt}.\tag{16.121}
Note that since -- unlike the delta systems -- the permutation systems are not tensors, we are unable to conclude that the determinant of a doubly-covariant or a doubly-contravariant tensor is an invariant.
Finally, let us generalize the formulas we have just obtained to lower and higher dimensions. In two dimensions, for second-order systems AijA_{ij}, AjiA_{j}^{i}, and AijA^{ij} we have
eijAirAjs=detA ers          (16.122)eijAirAjs=detA ers          (16.123)eijAirAjs=detA ers          (16.124)\begin{aligned}e^{ij}A_{ir}A_{js} & =\det A~e_{rs}\ \ \ \ \ \ \ \ \ \ \left(16.122\right)\\e^{ij}A_{i}^{r}A_{j}^{s} & =\det A~e^{rs}\ \ \ \ \ \ \ \ \ \ \left(16.123\right)\\e_{ij}A^{ir}A^{js} & =\det A~e_{rs}\ \ \ \ \ \ \ \ \ \ \left(16.124\right)\end{aligned}
as well as the explicit formulas
detA=12!eijersAirAjs          (16.125)detA=12!δrsijAirAjs          (16.126)detA=12!eijersAirAjs.          (16.127)\begin{aligned}\det A & =\frac{1}{2!}e^{ij}e^{rs}A_{ir}A_{js}\ \ \ \ \ \ \ \ \ \ \left(16.125\right)\\\det A & =\frac{1}{2!}\delta_{rs}^{ij}A_{i}^{r}A_{j}^{s}\ \ \ \ \ \ \ \ \ \ \left(16.126\right)\\\det A & =\frac{1}{2!}e_{ij}e_{rs}A^{ir}A^{js}.\ \ \ \ \ \ \ \ \ \ \left(16.127\right)\end{aligned}
Meanwhile, in nn dimensions, we have
ei1inAi1r1Ainrn=detA er1rn          (16.128)ei1inAi1r1Ainrn=detA er1rn          (16.129)ei1inAi1r1Ainrn=detA er1rn          (16.130)\begin{aligned}e^{i_{1}\cdots i_{n}}A_{i_{1}r_{1}}\cdots A_{i_{n}r_{n}} & =\det A~e_{r_{1}\cdots r_{n}}\ \ \ \ \ \ \ \ \ \ \left(16.128\right)\\e^{i_{1}\cdots i_{n}}A_{i_{1}}^{r_{1}}\cdots A_{i_{n}}^{r_{n}} & =\det A~e^{r_{1}\cdots r_{n}}\ \ \ \ \ \ \ \ \ \ \left(16.129\right)\\e_{i_{1}\cdots i_{n}}A^{i_{1}r_{1}}\cdots A^{i_{n}r_{n}} & =\det A~e^{r_{1}\cdots r_{n}}\ \ \ \ \ \ \ \ \ \ \left(16.130\right)\end{aligned}
as well as the explicit formulas
detA=1n!ei1iner1rnAi1r1Ainrn          (16.131)detA=1n!δr1rni1inAi1r1Ainrn          (16.132)detA=1n!ei1iner1rnAi1r1Ainrn.          (16.133)\begin{aligned}\det A & =\frac{1}{n!}e^{i_{1}\cdots i_{n}}e^{_{r_{1}\cdots r_{n}}} A_{i_{1}r_{1}}\cdots A_{i_{n}r_{n}}\ \ \ \ \ \ \ \ \ \ \left(16.131\right)\\\det A & =\frac{1}{n!}\delta_{r_{1}\cdots r_{n}}^{i_{1}\cdots i_{n}} A_{i_{1}}^{r_{1}}\cdots A_{i_{n}}^{r_{n}}\ \ \ \ \ \ \ \ \ \ \left(16.132\right)\\\det A & =\frac{1}{n!}e_{i_{1}\cdots i_{n}}e_{r_{1}\cdots r_{n}} A^{i_{1}r_{1}}\cdots A^{i_{n}r_{n}}.\ \ \ \ \ \ \ \ \ \ \left(16.133\right)\end{aligned}
The celebrated multiplicative property of determinants states that the determinant of the product ABAB of two matrices AA and BB equals the product of the determinants, i.e.
detAB=detAdetB.(16.134)\det AB=\det A\det B.\tag{16.134}
We will now use the formulas derived in the previous Section to demonstrate this property with the help of the tensor notation. Our proof will be specific to three dimensions but it will be clear that it works for any dimension.
Suppose that AjiA_{j}^{i}, BjiB_{j}^{i}, and CjiC_{j}^{i} are second-order systems, where CjiC_{j}^{i} is the "matrix product" of AjiA_{j}^{i} and BjiB_{j}^{i}, i.e.
Cki=AjiBkj.(16.135)C_{k}^{i}=A_{j}^{i}B_{k}^{j}.\tag{16.135}
Our goal is to show that
detC=detAdetB.(16.136)\det C=\det A\det B.\tag{16.136}
We have
detC=13!δrstijkCirCjsCkt.(16.137)\det C=\frac{1}{3!}\delta_{rst}^{ijk}C_{i}^{r}C_{j}^{s}C_{k}^{t}.\tag{16.137}
Substituting properly-reindexed versions of the identity Cki=AjiBkjC_{k}^{i}=A_{j} ^{i}B_{k}^{j} into the equation above, we find
detC=13!δrstijkAlrBilAmsBjmAntBkn.(16.138)\det C=\frac{1}{3!}\delta_{rst}^{ijk}A_{l}^{r}B_{i}^{l}A_{m}^{s}B_{j}^{m} A_{n}^{t}B_{k}^{n}.\tag{16.138}
With the help of the identity
erstAlrAmsAnt=detA elmn(16.113)e_{rst}A_{l}^{r}A_{m}^{s}A_{n}^{t}=\det A~e_{lmn} \tag{16.113}
note that
δrstijkAlrAmsAnt=ersteijkAlrAmsAnt=detA eijkelmn=detA δlmnijk.(16.139)\delta_{rst}^{ijk}A_{l}^{r}A_{m}^{s}A_{n}^{t}=e_{rst}e^{ijk}A_{l}^{r}A_{m} ^{s}A_{n}^{t}=\det A~e^{ijk}e_{lmn}=\det A~\delta_{lmn}^{ijk}.\tag{16.139}
or, in short,
δrstijkAlrAmsAnt=detA δlmnijk.(16.140)\delta_{rst}^{ijk}A_{l}^{r}A_{m}^{s}A_{n}^{t}=\det A~\delta_{lmn}^{ijk}.\tag{16.140}
Therefore,
detC=13!detA δlmnijkBilBjmBkn,(16.141)\det C=\frac{1}{3!}\det A~\delta_{lmn}^{ijk}B_{i}^{l}B_{j}^{m}B_{k}^{n},\tag{16.141}
and, since
13!δlmnijkBilBjmBkn=detB,(16.142)\frac{1}{3!}\delta_{lmn}^{ijk}B_{i}^{l}B_{j}^{m}B_{k}^{n}=\det B,\tag{16.142}
we arrive at the desired identity
detC=detAdetB.(16.136)\det C=\det A\det B. \tag{16.136}
We ought to take a moment to appreciate the breathtaking straightforwardness of this calculation. It is truly a testament to the elegance and the effectiveness of the tensor notation.
The related combinations
AirAjsAjrAis          (16.143)AriAsjArjAsi          (16.144)AirAjsAjrAis          (16.145)\begin{aligned}& A_{ir}A_{js}-A_{jr}A_{is}\ \ \ \ \ \ \ \ \ \ \left(16.143\right)\\& A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}\ \ \ \ \ \ \ \ \ \ \left(16.144\right)\\& A^{ir}A^{js}-A^{jr}A^{is}\ \ \ \ \ \ \ \ \ \ \left(16.145\right)\end{aligned}
for second-order systems AijA_{ij}, AjiA_{j}^{i}, and AijA^{ij} in two dimensions play an important role in a number of applications. In particular, a combination analogous to AirAjsAjrAisA_{ir}A_{js}-A_{jr}A_{is} appears in the celebrated Gauss equations
BαγBβδBβγBαδ=Rαβγδ(16.146)B_{\alpha\gamma}B_{\beta\delta}-B_{\beta\gamma}B_{\alpha\delta}=R_{\alpha \beta\gamma\delta}\tag{16.146}
in the context of surfaces. This amazing equation will be discussed in a future book while our present discussion will help lay the necessary algebraic groundwork.
Recall the equation
δrslmAliAmj=detA δrsij(16.147)\delta_{rs}^{lm}A_{l}^{i}A_{m}^{j}=\det A~\delta_{rs}^{ij}\tag{16.147}
that we have just derived for a mixed system AjiA_{j}^{i} in two dimensions. Since
δrslm=δrlδsmδrmδsl(16.58)\delta_{rs}^{lm}=\delta_{r}^{l}\delta_{s}^{m}-\delta_{r}^{m}\delta_{s}^{l} \tag{16.58}
we have
(δrlδsmδrmδsl)AliAmj=detA δrsij.(16.148)\left( \delta_{r}^{l}\delta_{s}^{m}-\delta_{r}^{m}\delta_{s}^{l}\right) A_{l}^{i}A_{m}^{j}=\det A~\delta_{rs}^{ij}.\tag{16.148}
Multiplying out the left side, we obtain.
AriAsjArjAsi=detA δrsij.(16.149)A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}=\det A~\delta_{rs}^{ij}.\tag{16.149}
This relationship, which is as stunning as it is simple, is the goal of this Section. Note that taking AjiA_{j}^{i} to be the Kronecker delta δji\delta _{j}^{i}, which corresponds to the identity matrix whose determinant is 11, yields
δriδsjδrjδsi=δrsij.(16.150)\delta_{r}^{i}\delta_{s}^{j}-\delta_{r}^{j}\delta_{s}^{i}=\delta_{rs}^{ij}.\tag{16.150}
Thus, the equation
AriAsjArjAsi=detA δrsij(16.149)A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}=\det A~\delta_{rs}^{ij} \tag{16.149}
can be seen as a generalization of the identity
δrsij=δriδsjδrjδsi,(16.58)\delta_{rs}^{ij}=\delta_{r}^{i}\delta_{s}^{j}-\delta_{r}^{j}\delta_{s}^{i}, \tag{16.58}
when interpreted in a two-dimensional space.
The equation
AriAsjArjAsi=detA δrsij(16.149)A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}=\det A~\delta_{rs}^{ij} \tag{16.149}
can also be used to express detA\det A for a 2×22\times2 matrix in terms of the traces of the matrix itself and its square. Indeed, since δijij=2\delta_{ij}^{ij} =2, contracting ii with rr and jj with ss yields
2detA=AiiAjjAjiAij(16.151)2\det A=A_{i}^{i}A_{j}^{j}-A_{j}^{i}A_{i}^{j}\tag{16.151}
and we note that the object AiiA_{i}^{i} corresponds to the trace of the matrix while AjiAijA_{j}^{i}A_{i}^{j} corresponds to the trace of its square. This result can be easily generalized to any number of dimensions. Namely, the determinant of an n×nn\times n matrix can be expressed in terms of the traces of its first nn powers.
Entirely analogous to the equation
AriAsjArjAsi=detA δrsij(16.149)A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}=\det A~\delta_{rs}^{ij} \tag{16.149}
is the subscripted version
AirAjsAisAjr=detA eijers(16.152)A_{ir}A_{js}-A_{is}A_{jr}=\det A~e_{ij}e_{rs}\tag{16.152}
for a second-order system AijA_{ij} and the superscripted version
AirAjsAisAjr=detA eijers(16.153)A^{ir}A^{js}-A^{is}A^{jr}=\det A~e^{ij}e^{rs}\tag{16.153}
for a second-order system AijA^{ij}.
In particular, for the covariant metric tensor ZijZ_{ij}, we have
ZirZjsZisZjr=Zeijers,(16.154)Z_{ir}Z_{js}-Z_{is}Z_{jr}=Ze_{ij}e_{rs},\tag{16.154}
while for the contravariant metric tensor ZijZ^{ij}, we have
ZirZjsZisZjr=1Zeijers,(16.155)Z^{ir}Z^{js}-Z^{is}Z^{jr}=\frac{1}{Z}e^{ij}e^{rs},\tag{16.155}
where, as usual, ZZ is the determinant of the covariant metric tensor ZijZ_{ij}.
We must reiterate that all identities derived in this Section are valid only in two dimensions. For a second-order system AjiA_{j}^{i} in three dimensions, the identity analogous to
AriAsjArjAsi=detA δrsij(16.149)A_{r}^{i}A_{s}^{j}-A_{r}^{j}A_{s}^{i}=\det A~\delta_{rs}^{ij} \tag{16.149}
reads
AriAsjAtk+AsiAtjArk+AtiArjAskAriAtjAskAtiAsjArkAsiArjAtk=detA δrstijk(16.156)A_{r}^{i}A_{s}^{j}A_{t}^{k}+A_{s}^{i}A_{t}^{j}A_{r}^{k}+A_{t}^{i}A_{r} ^{j}A_{s}^{k}-A_{r}^{i}A_{t}^{j}A_{s}^{k}-A_{t}^{i}A_{s}^{j}A_{r}^{k} -A_{s}^{i}A_{r}^{j}A_{t}^{k}=\det A~\delta_{rst}^{ijk}\tag{16.156}
and it is entirely apparent how to generalize this identity to any number of dimensions.
Let us return to a three-dimensional space, where the combination δrstijkAirAjsAkt\delta_{rst}^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t} produces 3!3! times the determinant of AjiA_{j}^{i}. If we drop the term AirA_{i}^{r} from that combination, we are left with
δrstijkAjsAkt,(16.157)\delta_{rst}^{ijk}A_{j}^{s}A_{k}^{t},\tag{16.157}
which is a second-order system enumerated by the free indices ii and rr. With an additional factor of 1/2!1/2!, this new combination is known as the cofactor of AirA_{i}^{r} and is denoted byAˉri\bar{A}_{r}^{i}, i.e.
Aˉri=12!δrstijkAjsAkt.(16.158)\bar{A}_{r}^{i}=\frac{1}{2!}\delta_{rst}^{ijk}A_{j}^{s}A_{k}^{t}.\tag{16.158}
The cofactor AriA_{r}^{i} has two remarkable properties. First, it represents the partial derivative of the determinant when the latter is interpreted as a function of the elements AirA_{i}^{r}, i.e.
detAAir=Aˉri.(16.159)\frac{\partial\det A}{\partial A_{i}^{r}}=\bar{A}_{r}^{i}.\tag{16.159}
Second, it equals the product of detA\det A and the matrix inverse of AriA_{r} ^{i}. It can therefore be used as an explicit algebraic expression for the matrix inverse of a second-order system. This Section is devoted to demonstrating these properties.
Let us begin by demonstrating the identity
detAAir=Aˉri.(16.159)\frac{\partial\det A}{\partial A_{i}^{r}}=\bar{A}_{r}^{i}. \tag{16.159}
Notice the tensorial consistency of the above expression. The indicial signature ir_{i}^{r} in the "denominator" is consistent with the signature of ri_{r}^{i} on the right. Since the letters ii and rr already appear as dummy indices in δrstijkAirAjsAkt\delta_{rst}^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t}, we will use two new letters ll and uu and differentiate detA\det A with respect to AluA_{l}^{u} , i.e.
detAAlu.(16.160)\frac{\partial\det A}{\partial A_{l}^{u}}.\tag{16.160}
Recall from Chapter 7 the identity
xixj=δji,(7.9)\frac{\partial x^{i}}{\partial x^{j}}=\delta_{j}^{i}, \tag{7.9}
where it was later used for quadratic form minimization in Section 8.7. In our present discussion, the independent variables, i.e. the elements AirA_{i}^{r}, are enumerated by two indices instead of one. Nevertheless, we can similarly consider the derivative of a typical one of them, say, AirA_{i}^{r} with respect to AluA_{l}^{u}:
AirAlu.(16.161)\frac{\partial A_{i}^{r}}{\partial A_{l}^{u}}.\tag{16.161}
This derivative equals 11 when AirA_{i}^{r} and AluA_{l}^{u} are one and the same variable (i.e. when i=li=l and r=ur=u) and 00 otherwise. This observation is effectively captured by the tensor identity
AirAlu=δurδil,(16.162)\frac{\partial A_{i}^{r}}{\partial A_{l}^{u}}=\delta_{u}^{r}\delta_{i}^{l},\tag{16.162}
with similar identities for AjsA_{j}^{s} and AktA_{k}^{t}, i.e.
AjsAlu=δusδjl    and    AktAlu=δutδkl.(16.163)\frac{\partial A_{j}^{s}}{\partial A_{l}^{u}}=\delta_{u}^{s}\delta_{j} ^{l}\text{ \ \ \ and \ \ \ }\frac{\partial A_{k}^{t}}{\partial A_{l}^{u} }=\delta_{u}^{t}\delta_{k}^{l}.\tag{16.163}
We are now ready to proceed with the main calculation. Differentiate both sides of the identity
detA=13!δrstijkAirAjsAkt(16.117)\det A=\frac{1}{3!}\delta_{rst}^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t} \tag{16.117}
with respect to AluA_{l}^{u}, i.e.
detAAlu=13!δrstijk(AirAjsAkt)Alu.(16.164)\frac{\partial\det A}{\partial A_{l}^{u}}=\frac{1}{3!}\delta_{rst}^{ijk} \frac{\partial\left( A_{i}^{r}A_{j}^{s}A_{k}^{t}\right) }{\partial A_{l} ^{u}}.\tag{16.164}
By the product rule,
detAAlu=13!δrstijk(AirAluAjsAkt+AirAjsAluAkt+AirAjsAktAlu).(16.165)\frac{\partial\det A}{\partial A_{l}^{u}}=\frac{1}{3!}\delta_{rst} ^{ijk}\left( \frac{\partial A_{i}^{r}}{\partial A_{l}^{u}}A_{j}^{s}A_{k} ^{t}+A_{i}^{r}\frac{\partial A_{j}^{s}}{\partial A_{l}^{u}}A_{k}^{t}+A_{i} ^{r}A_{j}^{s}\frac{\partial A_{k}^{t}}{\partial A_{l}^{u}}\right) .\tag{16.165}
Since
AirAlu=δurδil,    AjsAlu=δusδjl,    and    AktAlu=δutδkl,(16.166)\frac{\partial A_{i}^{r}}{\partial A_{l}^{u}}=\delta_{u}^{r}\delta_{i} ^{l}\text{,\ \ \ \ }\frac{\partial A_{j}^{s}}{\partial A_{l}^{u}}=\delta _{u}^{s}\delta_{j}^{l}\text{,\ \ \ \ and\ \ \ \ }\frac{\partial A_{k}^{t} }{\partial A_{l}^{u}}=\delta_{u}^{t}\delta_{k}^{l},\tag{16.166}
we have
detAAlu=13!δrstijk(δurδilAjsAkt+δusδjlAirAkt+δutδklAirAjs).(16.167)\frac{\partial\det A}{\partial A_{l}^{u}}=\frac{1}{3!}\delta_{rst} ^{ijk}\left( \delta_{u}^{r}\delta_{i}^{l}A_{j}^{s}A_{k}^{t}+\delta_{u} ^{s}\delta_{j}^{l}A_{i}^{r}A_{k}^{t}+\delta_{u}^{t}\delta_{k}^{l}A_{i} ^{r}A_{j}^{s}\right) .\tag{16.167}
Absorbing the Kronecker deltas into the delta systems, we find that
detAAlu=13!(δustljkAjsAkt+δrutilkAirAkt+δrsuijlAirAjs).(16.168)\frac{\partial\det A}{\partial A_{l}^{u}}=\frac{1}{3!}\left( \delta _{ust}^{ljk}A_{j}^{s}A_{k}^{t}+\delta_{rut}^{ilk}A_{i}^{r}A_{k}^{t} +\delta_{rsu}^{ijl}A_{i}^{r}A_{j}^{s}\right) .\tag{16.168}
Now, recall the combination corresponding to the cofactor Aˉri\bar{A}_{r}^{i}, i.e.
Aˉri=12!δrstijkAjsAkt,(16.158)\bar{A}_{r}^{i}=\frac{1}{2!}\delta_{rst}^{ijk}A_{j}^{s}A_{k}^{t}, \tag{16.158}
and observe that each term in parentheses in the preceding identity equals 2!Aˉul2!\bar{A}_{u}^{l}. Thus, the sum in parentheses equals 3!Aˉul3!\bar{A}_{u}^{l} and we can conclude that
detAAlu=Aˉul,(16.159)\frac{\partial\det A}{\partial A_{l}^{u}}=\bar{A}_{u}^{l}, \tag{16.159}
as we set out to prove. It is left as an exercise to generalize this identity to lower and higher dimensions.
Let us now turn our attention to the second property of the cofactor Aˉri\bar {A}_{r}^{i}. Namely, that it equals the product of detA\det A and the matrix inverse of AirA_{i}^{r}. In the tensor notation, this relationship is captured by the equation
AˉriAlr=detA δli.(16.169)\bar{A}_{r}^{i}A_{l}^{r}=\det A~\delta_{l}^{i}.\tag{16.169}
To prove this identity, denote the product AˉriAlr\bar{A}_{r}^{i}A_{l}^{r} by DliD_{l}^{i}, i.e.
Dli=AˉriAlr.(16.170)D_{l}^{i}=\bar{A}_{r}^{i}A_{l}^{r}.\tag{16.170}
Since the cofactor Aˉri\bar{A}_{r}^{i} is given by
Aˉri=12!δrstijkAjsAkt,(16.158)\bar{A}_{r}^{i}=\frac{1}{2!}\delta_{rst}^{ijk}A_{j}^{s}A_{k}^{t}, \tag{16.158}
the product DliD_{l}^{i} is given by
Dli=12!δrstijkAlrAjsAkt(16.171)D_{l}^{i}=\frac{1}{2!}\delta_{rst}^{ijk}A_{l}^{r}A_{j}^{s}A_{k}^{t}\tag{16.171}
and we must show that
Dli=detA δli.(16.172)D_{l}^{i}=\det A~\delta_{l}^{i}.\tag{16.172}
We will demonstrate this identity by considering two specific examples -- one where ili\neq l and another where i=li=l. These two examples will convince us of the general validity of the above identity.
First, consider the case where i=1i=1 and l=2l=2, i.e.
D21=12!δrst1jkA2rAjsAkt.(16.173)D_{2}^{1}=\frac{1}{2!}\delta_{rst}^{1jk}A_{2}^{r}A_{j}^{s}A_{k}^{t}.\tag{16.173}
The nonzero contributions to the sum on the right come from the combinations for which j,k=2,3j,k=2,3 and j,k=3,2j,k=3,2, i.e.
D21=12!(δrst123A2rA2sA3t+δrst132A2rA3sA2t).(16.174)D_{2}^{1}=\frac{1}{2!}\left( \delta_{rst}^{123}A_{2}^{r}A_{2}^{s}A_{3} ^{t}+\delta_{rst}^{132}A_{2}^{r}A_{3}^{s}A_{2}^{t}\right) .\tag{16.174}
However, each of these terms is zero since A2rA2sA_{2}^{r}A_{2}^{s} is symmetric in rr and ss and A2rA2tA_{2}^{r}A_{2}^{t} is symmetric in rr and tt, and, as we demonstrated earlier, a double contraction of a skew-symmetric system and a symmetric system vanishes. Thus, D21=0D_{2}^{1}=0 and, more generally,
Dli=0 when il.(16.175)D_{l}^{i}=0\text{ when }i\neq l.\tag{16.175}
Next, consider the case where i=i= l=1l=1, i.e.
D11=12!δrst1jkA1rAjsAkt.(16.176)D_{1}^{1}=\frac{1}{2!}\delta_{rst}^{1jk}A_{1}^{r}A_{j}^{s}A_{k}^{t}.\tag{16.176}
Once again, the nonzero contributions to the sum on the right come from the combinations for which j,k=2,3j,k=2,3 and j,k=3,2j,k=3,2, i.e.
D11=12!(δrst123A1rA2sA3t+δrst132A1rA3sA2t).(16.177)D_{1}^{1}=\frac{1}{2!}\left( \delta_{rst}^{123}A_{1}^{r}A_{2}^{s}A_{3} ^{t}+\delta_{rst}^{132}A_{1}^{r}A_{3}^{s}A_{2}^{t}\right) .\tag{16.177}
It is left as an exercise to show that, this time, each term in parentheses equals detA\det A. Thus, D11=detAD_{1}^{1}=\det A and, more generally,
Dli=detA when i=l.(16.178)D_{l}^{i}=\det A\text{ when }i=l.\tag{16.178}
In summary,
Dli={0 when ildetAwhen i=l,(16.179)D_{l}^{i}=\left\{ \begin{array} {rr} 0 & \text{ when }i\neq l\\ \det A & \text{when }i=l \end{array} \right. ,\tag{16.179}
or
Dli=detAδli,(16.172)D_{l}^{i}=\det A\delta_{l}^{i}, \tag{16.172}
as we set out to show.
In this Section, we will show that the partial derivative
ZZk(16.180)\frac{\partial\sqrt{Z}}{\partial Z^{k}}\tag{16.180}
of the volume element Z\sqrt{Z}, where ZZ is the determinant of the covariant metric tensor ZijZ_{ij} is given by the beautiful formula
ZZk=ZΓiki.(16.181)\frac{\partial\sqrt{Z}}{\partial Z^{k}}=\sqrt{Z}\Gamma_{ik}^{i}.\tag{16.181}
This formula is of undeniable intrinsic interest. Furthermore, it will be used on a number of occasions in our narrative, including the proof of the metrinilic property of the covariant derivative with respect to the Levi-Civita symbols introduced in the next Chapter, the proof of the Voss-Weyl formula in Chapter 18, and the proof of Gauss's theorem, also known as the divergence theorem, in a future book.
To begin our analysis, note that by the chain rule,
ZZk=12ZZZk.(16.182)\frac{\partial\sqrt{Z}}{\partial Z^{k}}=\frac{1}{2\sqrt{Z}}\frac{\partial Z}{\partial Z^{k}}.\tag{16.182}
Thus, we can focus our attention on ZZ itself.
Treat ZZ as a function of the elements ZijZ_{ij}. This is similar to the way we treated detA\det A as a function of the elements of AriA_{r}^{i} in the previous Section. By the chain rule,
ZZk=ZZijZijZk.(16.183)\frac{\partial Z}{\partial Z^{k}}=\frac{\partial Z}{\partial Z_{ij}} \frac{\partial Z_{ij}}{\partial Z^{k}}.\tag{16.183}
As we showed in the previous Section,
ZZij=ZZij.(16.184)\frac{\partial Z}{\partial Z_{ij}}=ZZ^{ij}.\tag{16.184}
Recall that the derivative Zij/Zk\partial Z_{ij}/\partial Z^{k} of the metric tensor was calculated in Chapter 12, where we found that it is given by the identity
ZijZk=Γi,jk+Γj,ik.(12.40)\frac{\partial Z_{ij}}{\partial Z^{k}}=\Gamma_{i,jk}+\Gamma_{j,ik}. \tag{12.40}
Combining the last two identities, we find
ZZk=ZZij(Γi,jk+Γj,ik).(16.185)\frac{\partial Z}{\partial Z^{k}}=ZZ^{ij}\left( \Gamma_{i,jk}+\Gamma _{j,ik}\right) .\tag{16.185}
Absorbing ZijZ^{ij} into the Christoffel symbols yields
ZZk=Z(Γjkj+Γiki),(16.186)\frac{\partial Z}{\partial Z^{k}}=Z\left( \Gamma_{jk}^{j}+\Gamma_{ik} ^{i}\right) ,\tag{16.186}
or, equivalently,
ZZk=2ZΓiki.(16.187)\frac{\partial Z}{\partial Z^{k}}=2Z\Gamma_{ik}^{i}.\tag{16.187}
Since, as we noted earlier,
ZZk=12ZZZk(16.182)\frac{\partial\sqrt{Z}}{\partial Z^{k}}=\frac{1}{2\sqrt{Z}}\frac{\partial Z}{\partial Z^{k}} \tag{16.182}
we have arrived at the identity
ZZk=ZΓiki,(16.181)\frac{\partial\sqrt{Z}}{\partial Z^{k}}=\sqrt{Z}\Gamma_{ik}^{i}, \tag{16.181}
as we set out to show.
For illustration purposes, let us confirm this identity in cylindrical coordinates, where
Z=r.(9.61)\sqrt{Z}=r. \tag{9.61}
Since Z1Z^{1}, Z2Z^{2}, and Z3Z^{3} correspond to rr, θ\theta, and zz, we have
ZZ1=1          (16.188)ZZ2=0          (16.189)ZZ3=0.          (16.190)\begin{aligned}\frac{\partial\sqrt{Z}}{\partial Z^{1}} & =1\ \ \ \ \ \ \ \ \ \ \left(16.188\right)\\\frac{\partial\sqrt{Z}}{\partial Z^{2}} & =0\ \ \ \ \ \ \ \ \ \ \left(16.189\right)\\\frac{\partial\sqrt{Z}}{\partial Z^{3}} & =0.\ \ \ \ \ \ \ \ \ \ \left(16.190\right)\end{aligned}
Recall that the nonzero elements of the Christoffel symbol in the cylindrical coordinates are given by
Γ221=r    and    Γ122=Γ212=1r.(12.22)\Gamma_{22}^{1}=-r\text{\ \ \ \ and \ \ \ }\Gamma_{12}^{2}=\Gamma_{21} ^{2}=\frac{1}{r}. \tag{12.22}
Therefore, we have
ZΓi1i=r(Γ111+Γ212+Γ313)=r(0+1r+0)=1          (16.191)ZΓi2i=r(Γ121+Γ222+Γ323)=r(0+0+0)=0          (16.192)ZΓi3i=r(Γ131+Γ232+Γ333)=r(0+0+0)=0,          (16.193)\begin{aligned}\sqrt{Z}\Gamma_{i1}^{i} & =r\left( \Gamma_{11}^{1}+\Gamma_{21}^{2} +\Gamma_{31}^{3}\right) =r\left( 0+\frac{1}{r}+0\right) =1\ \ \ \ \ \ \ \ \ \ \left(16.191\right)\\\sqrt{Z}\Gamma_{i2}^{i} & =r\left( \Gamma_{12}^{1}+\Gamma_{22}^{2} +\Gamma_{32}^{3}\right) =r\left( 0+0+0\right) =0\ \ \ \ \ \ \ \ \ \ \left(16.192\right)\\\sqrt{Z}\Gamma_{i3}^{i} & =r\left( \Gamma_{13}^{1}+\Gamma_{23}^{2} +\Gamma_{33}^{3}\right) =r\left( 0+0+0\right) =0,\ \ \ \ \ \ \ \ \ \ \left(16.193\right)\end{aligned}
as we set out to show.
Exercise 16.1Show that any of the equations
Aijk=Ajik (switch 1st and 2nd)          (16.194)Aijk=Aikj (switch 2nd and 3rd)          (16.195)Aijk=Akji (switch 1st and 3rd),          (16.196)\begin{aligned}A_{ijk} & =-A_{jik}\text{ (switch }1^{\text{st}}\text{ and }2^{\text{nd} }\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.194\right)\\A_{ijk} & =-A_{ikj}\text{ (switch }2^{\text{nd}}\text{ and }3^{\text{rd} }\text{)}\ \ \ \ \ \ \ \ \ \ \left(16.195\right)\\A_{ijk} & =-A_{kji}\text{ (switch }1^{\text{st}}\text{ and }3^{\text{rd} }\text{),}\ \ \ \ \ \ \ \ \ \ \left(16.196\right)\end{aligned}
follows from the other two.
Exercise 16.2Show that
eijk=ejki.(16.197)e_{ijk}=e_{jki}.\tag{16.197}
Exercise 16.3Show that
eijkeijk=3!(16.198)e^{ijk}e_{ijk}=3!\tag{16.198}
Exercise 16.4Show that
eijkekji=3!(16.199)e^{ijk}e_{kji}=-3!\tag{16.199}
Exercise 16.5Show that
eijkeijl=2!δlk.(16.200)e^{ijk}e_{ijl}=2!\delta_{l}^{k}.\tag{16.200}
Exercise 16.6Describe the values of all the 363^{6} elements of the system
eijkerst.(16.201)e^{ijk}e_{rst}.\tag{16.201}
Exercise 16.7Similarly, show that
eijkTiZj(16.202)e^{ijk}\frac{\partial T_{i}}{\partial Z^{j}}\tag{16.202}
is a tensor for a tensor TiT_{i}.
Exercise 16.8Show that for any system AiA_{i},
eijkAiAj=0.(16.203)e^{ijk}A_{i}A_{j}=0.\tag{16.203}
Exercise 16.9Show that if UijmU_{ij}^{m} is symmetric in its subscripts, i.e. Uijm=UjimU_{ij} ^{m}=U_{ji}^{m}, then
eijkUijm=0.(16.204)e^{ijk}U_{ij}^{m}=0.\tag{16.204}
In particular,
eijkΓijm=0.(16.205)e^{ijk}\Gamma_{ij}^{m}=0.\tag{16.205}
Exercise 16.10We have shown that if AijA_{ij} is skew-symmetric, then AijUiUj=0A_{ij}U^{i}U^{j}=0 for any UiU^{i}. Show the converse, i.e. if AijUiUj=0A_{ij}U^{i}U^{j}=0 for any UiU^{i}, then AijA_{ij} is skew-symmetric.
Exercise 16.11Show that
δijkijk=3!(16.206)\delta_{ijk}^{ijk}=3!\tag{16.206}
and, more generally,
δi1ini1in=n!.(16.85)\delta_{i_{1}\cdots i_{n}}^{i_{1}\cdots i_{n}}=n!. \tag{16.85}
Exercise 16.12Evaluate
δjkiijk.(16.207)\delta_{jki}^{ijk}.\tag{16.207}
Exercise 16.13Show that
δlmnijkδrstlmn=3!δrstijk.(16.208)\delta_{lmn}^{ijk}\delta_{rst}^{lmn}=3!\delta_{rst}^{ijk}.\tag{16.208}
Exercise 16.14Evaluate
δrstijkδijkrst.(16.209)\delta_{rst}^{ijk}\delta_{ijk}^{rst}.\tag{16.209}
Exercise 16.15Evaluate
δlmnijkδrstlmnδijkrst.(16.210)\delta_{lmn}^{ijk}\delta_{rst}^{lmn}\delta_{ijk}^{rst}.\tag{16.210}
Exercise 16.16Show that
δlmnijkAlAm=0.(16.211)\delta_{lmn}^{ijk}A^{l}A^{m}=0.\tag{16.211}
Exercise 16.17Confirm that in three dimensions,
δrsij=δrskijk(16.212)\delta_{rs}^{ij}=\delta_{rsk}^{ijk}\tag{16.212}
and
2δri=δrjij.(16.213)2\delta_{r}^{i}=\delta_{rj}^{ij}.\tag{16.213}
Exercise 16.18Show that
δrstijk=δriδstjkδrjδstik+δrkδstij.(16.214)\delta_{rst}^{ijk}=\delta_{r}^{i}\delta_{st}^{jk}-\delta_{r}^{j}\delta _{st}^{ik}+\delta_{r}^{k}\delta_{st}^{ij}.\tag{16.214}
Generalize this formula to higher dimensions.
Exercise 16.19In three dimensions, consider a first-order system TiT_{i} and let
Tjk=eijkTi.(16.215)T^{jk}=e^{ijk}T_{i}.\tag{16.215}
Show that TjkT^{jk} is anti-symmetric and that TiT_{i} can be recovered from TjkT^{jk} by contraction with eijke_{ijk}, i.e.
Ti=12!eijkTjk.(16.216)T_{i}=\frac{1}{2!}e_{ijk}T^{jk}.\tag{16.216}
Exercise 16.20Similarly, consider an anti-symmetric second-order system TijT_{ij} and let
Tk=eijkTij.(16.217)T^{k}=e^{ijk}T_{ij}.\tag{16.217}
Show that TijT_{ij} can be recovered from TkT^{k} by contraction with eijke_{ijk} , i.e.
Tij=eijkTk.(16.218)T_{ij}=e_{ijk}T^{k}.\tag{16.218}
Exercise 16.21If TrsT_{rs} is a tensor, show that
δijkrstTrsZt(16.219)\delta_{ijk}^{rst}\frac{\partial T_{rs}}{\partial Z^{t}}\tag{16.219}
is also a tensor.
Exercise 16.22Use the formula
detA=13!δrstijkAirAjsAkt(16.117)\det A=\frac{1}{3!}\delta_{rst}^{ijk}A_{i}^{r}A_{j}^{s}A_{k}^{t} \tag{16.117}
to show that the determinant of δji\delta_{j}^{i} is 11.
Exercise 16.23Confirm the identity
detA=12(AiiAjjAjiAij)(16.151)\det A=\frac{1}{2}\left( A_{i}^{i}A_{j}^{j}-A_{j}^{i}A_{i}^{j}\right) \tag{16.151}
by evaluating the quantities AiiA_{i}^{i} (which equals AjjA_{j}^{j}) and AjiAijA_{j}^{i}A_{i}^{j} for a general 2×22\times2 matrix
[abcd].(16.220)\left[ \begin{array} {cc} a & b\\ c & d \end{array} \right] .\tag{16.220}
Exercise 16.24Derive the formula for the determinant of a 3×33\times3 matrix AA in terms of the traces of AA, A2A^{2}, and A3A^{3}.
Exercise 16.25In three dimensions, show that the determinant AA of a skew-symmetric system AijA_{ij} is zero. Show that this result extends to any odd dimension.
Exercise 16.26Demonstrate the identity
detAAlu=Aˉul(16.159)\frac{\partial\det A}{\partial A_{l}^{u}}=\bar{A}_{u}^{l} \tag{16.159}
in two, as well as nn, dimensions.
Exercise 16.27Confirm the identity
ZZk=ZΓiki(16.181)\frac{\partial\sqrt{Z}}{\partial Z^{k}}=\sqrt{Z}\Gamma_{ik}^{i} \tag{16.181}
in spherical coordinates.
Exercise 16.28Demonstrate the identity
1ZZk=1ZΓiki(16.221)\frac{\partial\frac{1}{\sqrt{Z}}}{\partial Z^{k}}=-\frac{1}{\sqrt{Z}} \Gamma_{ik}^{i}\tag{16.221}
and confirm it in cylindrical and spherical components.
Send feedback to Pavel