From Wikipedia, the free encyclopedia
(Redirected from Sedenions)
Typenonassociative algebra
Unitse0, ..., e15
Multiplicative identitye0
Main propertiespower associativity
Common systems
Less common systems

Octonions () Sedenions ()

In abstract algebra, the sedenions form a 16-dimensional noncommutative and nonassociative algebra over the real numbers; they are obtained by applying the Cayley–Dickson construction to the octonions, and as such the octonions are isomorphic to a subalgebra of the sedenions. Unlike the octonions, the sedenions are not an alternative algebra. Applying the Cayley–Dickson construction to the sedenions yields a 32-dimensional algebra, sometimes called the 32-ions or trigintaduonions.[1] It is possible to continue applying the Cayley–Dickson construction arbitrarily many times.

The term sedenion is also used for other 16-dimensional algebraic structures, such as a tensor product of two copies of the biquaternions, or the algebra of 4 × 4 matrices over the real numbers, or that studied by Smith (1995).


A visualization of a 4D extension to the cubic octonion,[2] showing the 35 triads as hyperplanes through the real vertex of the sedenion example given. Note that the only exception is that the triple , , doesn't form a hyperplane with .

Like octonions, multiplication of sedenions is neither commutative nor associative. But in contrast to the octonions, the sedenions do not even have the property of being alternative. They do, however, have the property of power associativity, which can be stated as that, for any element x of , the power is well defined. They are also flexible.

Every sedenion is a linear combination of the unit sedenions , , , , ..., , which form a basis of the vector space of sedenions. Every sedenion can be represented in the form

Addition and subtraction are defined by the addition and subtraction of corresponding coefficients and multiplication is distributive over addition.

Like other algebras based on the Cayley–Dickson construction, the sedenions contain the algebra they were constructed from. So, they contain the octonions (generated by to in the table below), and therefore also the quaternions (generated by to ), complex numbers (generated by and ) and real numbers (generated by ).

The sedenions have a multiplicative identity element and multiplicative inverses, but they are not a division algebra because they have zero divisors. This means that two non-zero sedenions can be multiplied to obtain zero: an example is . All hypercomplex number systems after sedenions that are based on the Cayley–Dickson construction also contain zero divisors.

A sedenion multiplication table is shown below:

Sedenion properties[edit]

From the above table, we can see that:



The sedenions are not fully anti-associative. Choose any four generators, and . The following 5-cycle shows that these five relations cannot all be anti-associative.

In particular, in the table above, using and the last expression associates.

Quaternionic subalgebras[edit]

The 35 triads that make up this specific sedenion multiplication table with the 7 triads of the octonions used in creating the sedenion through the Cayley–Dickson construction shown in bold:

The binary representations of the indices of these triples bitwise XOR to 0.

{ {1, 2, 3}, {1, 4, 5}, {1, 7, 6}, {1, 8, 9}, {1, 11, 10}, {1, 13, 12}, {1, 14, 15},
{2, 4, 6}, {2, 5, 7}, {2, 8, 10}, {2, 9, 11}, {2, 14, 12}, {2, 15, 13}, {3, 4, 7},
{3, 6, 5}, {3, 8, 11}, {3, 10, 9}, {3, 13, 14}, {3, 15, 12}, {4, 8, 12}, {4, 9, 13},
{4, 10, 14}, {4, 11, 15}, {5, 8, 13}, {5, 10, 15}, {5, 12, 9}, {5, 14, 11}, {6, 8, 14},
{6, 11, 13}, {6, 12, 10}, {6, 15, 9}, {7, 8, 15}, {7, 9, 14}, {7, 12, 11}, {7, 13, 10} }

The list of 84 sets of zero divisors , where :


Moreno (1998) showed that the space of pairs of norm-one sedenions that multiply to zero is homeomorphic to the compact form of the exceptional Lie group G2. (Note that in his paper, a "zero divisor" means a pair of elements that multiply to zero.)

Sedenion neural networks provide[further explanation needed] a means of efficient and compact expression in machine learning applications and have been used in solving multiple time-series and traffic forecasting problems.[3][4]

See also[edit]


  2. ^ (Baez 2002, p. 6)
  3. ^ Saoud, Lyes Saad; Al-Marzouqi, Hasan (2020). "Metacognitive Sedenion-Valued Neural Network and its Learning Algorithm". IEEE Access. 8: 144823–144838. doi:10.1109/ACCESS.2020.3014690. ISSN 2169-3536.
  4. ^ Kopp, Michael; Kreil, David; Neun, Moritz; Jonietz, David; Martin, Henry; Herruzo, Pedro; Gruca, Aleksandra; Soleymani, Ali; Wu, Fanyou; Liu, Yang; Xu, Jingwei (2021-08-07). "Traffic4cast at NeurIPS 2020 – yet more on the unreasonable effectiveness of gridded geo-spatial processes". NeurIPS 2020 Competition and Demonstration Track. PMLR: 325–343.