Hyperdeterminant

In algebra, the hyperdeterminant is a generalisation of the determinant. Whereas a determinant is a scalar valued function defined on an "n"x"n" square matrix, a hyperdeterminant is defined on a multidimensional array of numbers or hypermatrix. Like a determinant, the hyperdeteriminant is a homogeneous polynomial with integer coefficients in the components of the hypermatrix. Many other properties of determinants generalise in some way to hyperdeterminants but unlike a determinant, the hyperdeterminant does not have a simple geometric interpretation in terms of volumes. Instead it is most commonly defined as a discriminant for a singular point on a scalar valued multilinear map.

The notation for deteriminants is extended to hyperdeterminants without change or ambiguity. Hence the hyperdeterminant of a hypermatrix "A" may be written using the vertical bar notation as |"A"| or as "det"("A").

The standard modern textbook on hyperdeterminants is "Discriminants, Resultants and Multidimensional Determinants" by Gel'fand, Kapranov, Zelevinsky ^[2] referred to below as GKZ. Their notation and terminology is followed here.

Cayley's Hyperdeterminant

In the special case of a 2x2x2 hypermatrix the hyperdeteriminant is known as Cayley's Hyperdetermiant after the British mathematician Arthur Cayley who discovered it. The quartic expression for the Cayley's hyperdeterminant of hypermatrix "A" with components "a"_ijk, i,j,k = 0 or 1 is given by

:"det"("A") = "a"²₀₀₀"a"²₁₁₁ + "a"²₀₀₁"a"²₁₁₀ + "a"²₀₁₀"a"²₁₀₁ + "a"²₁₀₀"a"²₀₁₁

::- 2"a"₀₀₀"a"₀₀₁"a"₁₁₀"a"₁₁₁ - 2"a"₀₀₀"a"₀₁₀"a"₁₀₁"a"₁₁₁ - 2"a"₀₀₀"a"₀₁₁"a"₁₀₀"a"₁₁₁ - 2"a"₀₀₁"a"₀₁₀"a"₁₀₁"a"₁₁₀ - 2"a"₀₀₁"a"₀₁₁"a"₁₁₀"a"₁₀₀ - 2"a"₀₁₀"a"₀₁₁"a"₁₀₁"a"₁₀₀

::+ 4"a"₀₀₀"a"₀₁₁"a"₁₀₁"a"₁₁₀ + 4"a"₀₀₁"a"₀₁₀"a"₁₀₀"a"₁₁₁

This expression acts as a discriminant in the sense that it is zero "if and only if" there is a non-zero solution in six unknowns "x"ⁱ, "y"ⁱ, "z"ⁱ, (with superscript i = 0 or 1) of the following system of equations

:"a"₀₀₀"x"⁰"y"⁰ + "a"₀₁₀"x"⁰"y"¹ + "a"₁₀₀"x"¹"y"⁰ + "a"₁₁₀"x"¹"y"¹ = 0

:"a"₀₀₁"x"⁰"y"⁰ + "a"₀₁₁"x"⁰"y"¹ + "a"₁₀₁"x"¹"y"⁰ + "a"₁₁₁"x"¹"y"¹ = 0

:"a"₀₀₀"x"⁰"z"⁰ + "a"₀₀₁"x"⁰"z"¹ + "a"₁₀₀"x"¹"z"⁰ + "a"₁₀₁"x"¹"z"¹ = 0

:"a"₀₁₀"x"⁰"z"⁰ + "a"₀₁₁"x"⁰"z"¹ + "a"₁₁₀"x"¹"z"⁰ + "a"₁₁₁"x"¹"z"¹ = 0

:"a"₀₀₀"y"⁰"z"⁰ + "a"₀₀₁"y"⁰"z"¹ + "a"₀₁₀"y"¹"z"⁰ + "a"₀₁₁"y"¹"z"¹ = 0

:"a"₁₀₀"y"⁰"z"⁰ + "a"₁₀₁"y"⁰"z"¹ + "a"₁₁₀"y"¹"z"⁰ + "a"₁₁₁"y"¹"z"¹ = 0

The hyperdeterminant can be written in a more compact form using the Einstein convention for summing over indices and an alternating tensor density with components ε^ij specified by ε⁰⁰ = ε¹¹ = 0, ε⁰¹ = -ε¹⁰ = 1.

:"b"_kn = (1/2)ε^ilε^jm"a"_ijk"a"_lmn

:"det"("A") = (1/2)ε^ilε^jm"b"_ij"b"_lm

using the same conventions we can define a multilinear form

:"f"(x,y,z) = "a"_ijk"x"ⁱ"y"^j"z"^k

Then the hyperdeterminant is zero if and only if there is a non-trivial point where all partial derivatives of "f" vanish.

General Hyperdeterminants

Definitions

In the general case a hyperdeterminant is defined as a discriminant for a multilinear map "f" from finite-dimensional vector spaces "V"_i to their underlying field "K" which may be $mathbb\{R\}$ or $mathbb\{C\}$.

:$f:\; V\_1\; otimes\; V\_2\; otimes\; cdots\; otimes\; V\_r\; o\; K$

"f" can be identified with a tensor in the tensor product of each dual space "V"^*_i

:$f\; in\; V^*\_1\; otimes\; V^*\_2\; otimes\; cdots\; otimes\; V^*\_r$

By definition a hyperdeterminant "det"("f") is a polynomial in components of the tensor "f" which is zero if and only if the map "f" has a non-trivial point where all partial derivatives with respect to the components of its vector arguments vanish (a non-trivial point means that none of the vector arguments are zero.)

The vector spaces "V"_i need not have the same dimensions and the hyperdeterminant is said to be of format ("k"₁, ... , "k"_r) "k"_i > 0, if the dimension of each space "V"_i is "k"_i+1. It can be shown that the hyperdeterminant exists for a given format and is unique up to a scalar factor, if and only if the largest number in the format is less than or equal to the sum of the other numbers in the format (see GKZ chapter 14 ^[2] )

This definition does not provide a means to construct the hyperdeteriminant and in general this is a difficult task. For hyperdeterminants with formats where "r" ≥ 4 the number of terms is usually too large to write out the hyperdeterminant in full. For larger "r" even the degree of the polynomial increases rapidly and does not have a convenient general formula.

Examples

The case of formats with "r" = 1 deals with vectors of length "k"₁ + 1. In this case the sum of the other format numbers is zero and "k"₁ is always greater than zero so no hyperdeterminants exist.

The case of "r" = 2 deals with ("k"₁ + 1)x("k"₂ + 1) matrices. Each format number must be greater than or equal to the other, therefore only square matrices "S" have hyperdeterminants and they can be identified with the determinant det("S"). Applying the definition of the hyperdeterminant as a discriminant to this case requires that det("S") is zero when there are vectors "X" and "Y" such that the matrix equations "SX" = 0 and "YS" = 0 have solutions for non-zero "X" and "Y".

For "r" > 2 there are hyperdeterminants with different formats satisifying the format inequality. e.g. Cayley's 2x2x2 hyperdeterminant has format (1,1,1) and a 2x2x3 hyperdeterminant of format (1,1,2) also exists. However a 2x2x4 hyperdeterminant would have format (1,1,3) but 3 > 1 + 1 so it does not exist.

Degree

Since the hyperdeterminant is homogeneous in its variables it has a well defined degree that is a function of the format and is written "N"("k"₁, ... , "k"_r). In special cases we can write down an expression for the degree. For example, a hyperdeterminant is said to be of boundary format when the largest format number is the sum of the others and in this case we have (see GKZ p455 ^[2] )

:$N(k\_2\; +\; cdots\; +\; k\_r,\; k\_2,\; cdots,\; k\_r)\; =\; frac\{(k\_2\; +\; cdots\; +\; k\_r\; +\; 1)!\}\{k\_2!\; cdots\; k\_r!\}.$

For hyperdeterminants of dimensions 2^r a convenient generating formula for the degrees N_r is (see GKZ p457 ^[2] )

:$sum\_\{r=0\}^infty\; N\_r\; frac\{z^r\}\{r!\}\; =\; frac\{e^\{-2z\{(1-z)^2\}.$

In particular for r = 2,3,4,5,6 the degree is respectively 2,4,24,128,880 and then grows very rapidly.

Three other special formulae for computing the degree of hyperdeterminants are given in GKZ p477.

for 2 x "m" x "m" use "N"(1,"m"-1,"m"-1) = 2"m"("m"-1)

for 3 x "m" x "m" use "N"(2,"m"-1,"m"-1) = 3"m"("m"-1)²

for 4 x "m" x "m" use "N"(3,"m"-1,"m"-1) = (2/3)"m"("m"-1)("m"-2)(5"m" -3)

A general result is that follows from the hyperdeterminants product rule and invariance properties listed below is that the least common multiple of the dimensions of the vector spaces on which the linear map acts divides the degree of the hyperdeterminant i.e.

:lcm("k"₁+1,...,"k"_r+1) | "N"("k"₁, ... , "k"_r).

Properties of Hyperdeterminants

Hyperdeterminants generalise many of the properties of determinants. The property of being a discrminant is one of them and it is used in the definition above.

Multiplicative properties

One of the most familiar properties of determinants is the multiplication rule which is sometimes known as the Binet-Cauchy formula. For square "n" x "n" matrices "A" and "B" the rule says that

:det("AB") = det("A") det("B")

This is one of the harder rules to generalize from determinants to hyperdeterminants because generalizations of products of hypermatrices can give hypermatrices of different sizes. The full domain of cases in which the product rule can be generalized is still a subject of research. However there are some basic instances that can be stated.

Given a multilinear form "f"(x₁,...,x_r) we can apply a linear transformation on the last argument using an "n" x "n" matrix "B" , y_r = "B" x_r. This generates a new multilinear form of the same format,

:"g"(x₁,...,x_r) = "f"(x₁,...,y_r)

In terms of hypermatrices this defines a product which can be written "g" = "f"."B"

It is then possible to use the definition of the hyperdeterminant to show that

: det("f"."B") = det("f") det("B")^"N"/"n"

Where "n" is the degree the the hyperdeterminant. This generalises the product rule for matrices.

Further generalizations of the product rule have been demonstrated for appropriate products of hypermatrices of boundary format ^[3]

Invariance Properties

A determinants is not usually considered in terms of its properties as an algebraic invariant but when determinants are generalized to hyperdeterminants the invariance is more notable. Using the multiplication rule above on the hyperdeterminant of a hypermatrix "H" times a matrix "S" with determinant equal to one gives

:det("H"."S") = det("H")

In other words the hyperdeterminant is a algebraic invariant under the action of the special linear group "SL"("n") on the hypermatrix. The transformation can be equally well applied to any of the vector spaces on which the multilinear map acts to give another distinct invariance. This leads to the general result,

: The hyperdeterminant of format $(k\_1,cdots,k\_r)$ is an invariant under an action of the group $SL(k\_1+1)\; otimes\; cdots\; otimes\; SL(k\_r+1)$

E.g. the determinant of an "n" x "n" matrix is an "SL"("n")² invariant and Cayley's hyperdeterminant for a 2x2x2 hypermatrix is an "SL"(2)³ invariant.

A more familiar property of a determinant is that if you add a multiple of a row (or column) to a different row (or column) of a square matrix then its determinant is unchanged. This is a special case of its invariance in the case where the special linear transformation matrix is an identity matrix plus a matrix with only one non-zero (off-diagonal) element. This property generalizes immediately to hyperdeterminants implying invariance when you add a multiple of one slice of a hypermatrix to another parallel slice.

A hyperdeterminant is not the only polynomial algebraic invariant for the group acting on the hypermatrix. For example, other algebraic invariants can be formed by adding and multiplying hyperdterminants. In general the invariants form a ring algebra and it follows from Hilbert's basis theorem that the ring is finitely generated. In other words, for a given hypermatrix format, all the polynomial algebraic invariants with integer coefficients can be formed using addition, subtraction and multiplication starting from a finite number of them. In the case of a 2x2x2 hypermatrix, all such invariants can be generated in this way from Cayley's hyperdeteriminant alone, but this is not a typical result for other formats. For example the hyperdeterminant for a hypermatrix of format 2x2x2x2 is an algebraic invariant of degree 24 yet all the invariants can be generated from a set of four simpler invariants of degree 6 and less ^[4] .

History and Applications

The hyperdeterminant was invented and named by Arthur Cayley in 1845 who was able to write down the expression for the 2x2x2 format, but Cayley went on to use the term for any algebraic invariant and later abandoned the concept in favour of a general theory of polynomial forms which he called "quantics" ^[5] . For the next 140 years there were few developments in the subject and hyperdeterminants were largely forgotten until they were rediscovered by Gel'fand, Kapranov and Zelevinsky in the 1980s as an offshoot of their work on generalized hypergeometric functions (see preface to GKZ ^[2] ). This led to them writing their textbook in which the hyperdeterminant is reintroduced as a discriminant.

Since then the hyperdeterminant has found applications over a wide range of disciplines including algebraic geometry, number theory, quantum computing and string theory.

In "algebraic geometry" the hyperdeterminant is studied as a special case of an X-discriminant. A principal result is that there is a correspondance betweem the vertices of the Newton Polytope for hyperdeterminants and the "triangulation" of a cube into simplices (see GKZ) ^[2]

In "quantum computing" the invariants on hypermatrices of format 2^N are used to study the entanglement of N qubits ^[6]

In "string theory" the hyperdeterminant first surfaced in connection with string dualities and black hole entropy. ^[7]

References

: [1] Cayley, A. "On the Theory of Linear Transformations." Cambridge Math. J. 4, 193-209, 1845.

: [2] Gel'fand, I. M.; Kapranov, M. M.; and Zelevinsky, A. V. "Discriminants, Resultants and Multidimensional Determinants" Birkhauser 1994.

: [3] Carla Dionisi, Giorgio Ottaviani, "The Binet-Cauchy Theorem for the Hyperdeterminant of boundary format multidimensional Matrices" arxiv|archive=math|id=0104281

: [4] Luque, J-G, Thibon, J-Y "The polynomial Invariants of Four Qubits" arxiv|archive=quant-ph|id=0212069

: [5] Crilly T, Crilly A.J. "Arthur Cayley: Mathematician Laureate of the Victorian Age", p176, JHU Press 2006.

: [6] Miyake A, "Classification of multipartite entangled states by multidimensional determinants", arxiv|archive=quant-ph|id=0206111

: [7] Duff M., "String triality, black hole entropy and Cayley's hyperdeterminant", arxiv|archive=hep-h|id=0601134