1. Introduction
The complex set is necessary to obtain the roots of all real numbers. It is all too natural to ask whether extensions of complex algebra to higher dimensions introduce more roots: in particular, new roots to the real numbers set. The roots of 
 in a number system 
 reduce to the existence and finding of solutions to the equation 
. If 
, the solutions 
 will be roots of real numbers, whereas if 
, the number and its roots will belong to the 
 set. In quaternion algebra, there is an infinite number of roots for a real number 
 if 
, whereas there are 
m roots for 
 [
1]. The quaternion roots can be economically obtained by writing quaternions in polar form and using De Moivre’s theorem [
2]. The 
 equation has been solved in several other number systems: split quaternions [
3], complexified quaternions [
4], and Clifford algebras 
 with 
 [
5], to mention some of them.
In contrast with the different versions of quaternions where real quantities have infinitely many roots, the number of possible solutions for the 
 root of a real or scator number is 
, where 
n is the number of director dimensions in scator space, as we shall presently see. Not all these roots are necessarily different. In the paper entitled ’
Powers of elliptic scator numbers’, the Victoria equation was established [
6]. This expression can be viewed as a generalization of De Moivre’s formula to scator numbers in 
 dimensions. The equation’s name and its concomitant winged 
 sense, were given in homage to chemist Victoria Guasti; for, in the early days of scator algebra, this result revealed the promising route of the scator’s approach. This manuscript draws heavily on, and should be considered as a follow up to the Victoria equation and other results presented in [
6].
This manuscript is organized as follows: In 
Section 2, the Victoria equation is inverted to obtain the roots of scators. Two versions exist, one in the multiplicative representation, stated in Theorem 1, and another in the additive representation, Theorem 2, that correspond to the rectangular and polar representations of complex numbers in higher dimensions. An important difference between these two representations is that only in the latter can the product be non-associative and eventually lead to zero scator solutions. The conditions that candidate solutions need to satisfy to be roots of non-zero scators are discussed here. In 
Section 3, square roots are considered. No new roots are obtained apart from the complex-like roots present in each scalar-hypercomplex plane. In this and the following sections, a geometrical representation of the roots is presented. Geometric interpretations of hyperbolic scators have been proposed before in terms of higher dimensional spaces [
7,
8,
9] as well as deformed Lorentz metrics [
10]. Elliptic scators and some scator functions of scator variables, in particular the components exponential function, have also been represented geometrically [
11]. Cube roots of one, described in 
Section 4, are the first case where new roots of real numbers with two or more non-vanishing director components exist. The two Abelian groups of these roots lie in orthogonal planes revolving around a point that is not the origin. Fourth roots of one, tackled in 
Section 5, are relevant because they are the lowest order roots that do not satisfy group properties in the additive representation. Fifth roots, described in 
Section 6, are the first case where some Abelian groups of roots no longer lie on planes. Some generalizations for 
 roots of one are discussed in 
Section 7. The conjugation of roots is discussed in this section. Conclusions are drawn in the last section.
  3. Square Roots in the Scator Set
In the multiplicative representation, the number one in 
 is given by a scator with magnitude 
, and 
n multiplicative director coefficients are equal to zero, 
. In the additive representation, one is given by the additive scalar component equal to one, 
, and the 
n additive director coefficients are equal to zero, 
. Recall that real numbers form a subset of the scator set 
. From the Victoria Equation (
5), in 
, the 
 roots of one are
      
      where 
 and the two director axes have been labelled with subindices 
x and 
y instead of numbers; the latter are usually used for dimensions larger than four. The square roots of 1 in 
 are
      
      where the scalar component is 1 if 
 and 
 when they are unequal; the director components are always zero. The usual 
 roots of one are thus obtained.
Spurious roots of 1: However, there seems to be another possibility because the number one in the additive representation in 
 is also attained if 
, 
. The square root from the Victoria Equation (
5) is
      
 However, 
, regardless of the value of the integer 
. A zero scator solution is thus obtained. However, from Lemma 1, this solution is not associative and therefore not a root of 
. It is instructive to work this example explicitly. The multiplicative to additive transformation of the root 
, prior to performing the product of components, is given by Equation (4a) in [
6]:
If this solution were a root of unity, its square should be equal to one. However, the product  is clearly not associative, since , while, recalling that the scator product is commutative, . The fact that this non-associative product can be arranged in such a way that the product is 1 explains why this spurious root occurs. It should be noticed that the  solution cannot be rearranged so that 1 is the product of two equal expressions, consistent with Corollary 2.
Square roots of : In multiplicative variables, minus one is obtained if 
 and either 
, 
, or vice versa. If 
 and 
,
      
      whereas if 
 and 
,
      
 The square roots of minus one in the scator set are the familiar imaginary units of complex algebra, but there is now an imaginary unit for each hypercomplex axis direction. These roots are illustrated in 
Figure 1. The scalar component and the two director components have been depicted in orthogonal directions. The director coefficients in multiplicative variables 
 and 
 are geometrically represented by the angle between the scalar (real) axis and the corresponding hyperimaginary director axis. The product of the roots 
 involves a 
 rotation in the 
 plane, whereas the product 
 produces a 
 rotation. An analogous behaviour occurs in the 
 plane for the 
 roots. These two cases exhaust the possibilities in 
.
Spurious roots of -1: However, in 
, minus one can be written as 
. The solution from (
5) is zero:
 In terms of the components factors, this solution is
      
Evaluation of the square is . Depending on the association, this expression is zero or minus one. Thus, this non-associative solution is a spurious root. Notice that the minus one association, , again cannot be written as the product of two equal expressions.
Square roots can be readily generalized to  dimensions. Consider four subsets of : (i) the set of positive real numbers , (ii) the set of negative real numbers , (iii) the set of scators with at least one non vanishing director component , and (iv) zero.
(i) For elements in the  subset, there are only two different square roots. In additive variables, given a scator with non-zero positive scalar and zero director components, , the two square roots are . In multiplicative variables, given a scator with a non-zero magnitude  and director components equal to zero or a multiple of , the square root is , and the two different square roots are , where the ±sign depends on whether there is an even or odd number of director component arguments equal to . The geometrical interpretation of the roots in this case is the familiar cyclotomic division. For example, the root lying on the negative scalar axis makes a  angle with respect to the positive scalar (or real) axis. The product of the scator root with itself doubles the angle, thereby mapping  onto 1, as in the complex case.
(ii) For elements in the 
 subset, there exist 
 square roots. In additive variables, elements with negative additive scalar and zero director components are considered, 
. In multiplicative variables, all except one of the director components are zero (mod 
). If, for example, the 
 direction has a 
 argument, then
      
The square roots from the Victoria Equation (
5) are
      
      where the only non vanishing term is
      
There are thus two roots, , for each j from 1 to n. In a geometrical representation, the root lying on the  axis makes a  angle with respect to the scalar (or real) axis. The product of the scator root with itself doubles the angle, thereby mapping the imaginary unit onto the negative scalar axis. However, there are now n hyper imaginary axes, each of them mapping positive or negative numbers on the  line onto the negative scalar axis.
(iii) For elements 
, there are only two different square roots in the additive representation, because the possible 
 phase differences in the 
 coefficients can only introduce an overall 
 factor, i.e., 
, and 
. The two square roots are
      
(iv) The square of zero is ; hence, the square root of zero is zero. This is the unique element with only one square root.
The square roots of a scator number 
 are solutions to the polynomial equation 
. In particular, square roots of one satisfy the polynomial 
. The second order polynomial with a non-zero linear term is 
, where 
b is a real coefficient. Surprisingly, this second order polynomial has additional roots that are not present if 
. In 
, depending on the value of the real number 
b, the polynomial 
 has two, six, or eight scator roots [
13].
  4. Cube Roots of 1
The cube roots of 
 in 
 are given by the Victoria equation for 
 and 
:
Labelling the roots 
, the 
 values are shown in 
Table 1. The roots 
 and 
 are the familiar three cyclotomic complex roots, but there are now two circles, one for each scalar-director mutually orthogonal plane, as depicted in 
Figure 2. They account for four roots, in addition, of course, to 
. The remaining four hypercomplex roots have non-zero director coefficients with equal magnitude 
 in both director axes. The roots 
, 
 and 
 form a group since 
 and 
. These three points lie on the parametric curve
      
      depicted in red in 
Figure 2. The two-dimensional surface embedded in three-dimensional space of scators 
 with unit magnitude is
      
      where zero 
 is clearly excluded. This surface is known as a cusphere [
11]. The product of the magnitudes is equal to the magnitude of the products since associativity is satisfied, so that all roots have a magnitude equal to one. Therefore, all the roots lie on the isometric surface as shown in 
Figure 2.
Perform a Euclidean 45° rotation of the cusphere about the 
s axis to bring the hypercomplex roots onto the 
 and 
 planes:
At the 
 plane,
      
      where the positive sign was taken since the magnitude is defined by the positive square root of (
20). Shift 
s by 
 in the negative 
s direction, 
, to obtain 
 that is, an equation of an ellipse with semi-axes 
 and 
. The unshifted ellipse is centred at 
 on the scalar axis. This ellipse and its counterpart, obtained at the 
 plane in the rotated cusphere, are drawn in red in 
Figure 2. The roots 
, 
, and 
, which lie on the ellipse at 
 with respect to the 
 axes, also form a group, 
, 
, and their cubes are equal to 
. The parametric curve for this ellipse is 
. The 45° Euclidean rotation maps the hypercomplex roots to 
 and 
, but these roots no longer have unit scator magnitude. The Euclidean rotation is an auxiliary procedure to exhibit the types of curves that are being obtained. However, the rotated figures must be treated with much care because a Euclidean rotation does not preserve the scator metric, as may be readily seen from (
20).
The projection of the parametric curve (
19) in the 
 plane is 
 = 
. The projections of the unrotated ellipses onto the 
 and 
 planes are thus circles of radius 
, centred at 
. The parameter is doubled, so that a 0 to 
 span describes the complete curve. The hypercomplex root 
 makes an angle 
, measured from the origin (
 is superimposed in this projection). However, if the reference point is 
, this root makes an angle of 
 with respect to the scalar axis, as shown in 
Figure 3. The 
 root (and its superimposed projection 
) exhibit an analogous behaviour. The parametric curve 
, drawn in red in 
Figure 3, is therefore trisected by the hypercomplex roots.
There is a 
 phase increase in each director angle, as the closed curves at 45° with respect to the 
 axes return to an arbitrary departure point. The geometrical descriptions in 
Figure 2, 
Figure 3 and Figure 5, as well as in ([
6], Figure 3), suggest that the equal director roots or powers go around the point 
. However, 
 is not a singularity. For 
, the parametric curve is equal to 
, but this point should not be included in the curve. Recall that zero is not in the cusphere, because the unit magnitude surface cannot include a point that has zero magnitude. Nonetheless, the cusphere points come arbitrarily close to zero. In this sense, the 45° curves come as close as possible to the singularity but jump it, rather than going around it. It is too large a leap to discuss monodromy when many issues, such as giving an unambiguous meaning to ’going around’ in higher dimensional spaces or the topology of the scator space, need to be addressed first. Nonetheless, this part of the discussion aims to show that some of these ideas seem likely to be meaningful in scator algebra.
  5. Fourth Roots of 1
Fourth roots of the multiplicative neutral 
 are a singular case because spurious nilroots are present from the outset. The 16 possible roots in 
, obtained from (
5),
      
      are shown in 
Table 2.
The (four) roots 1 and  are the square roots doubled period. The  and  roots with  and , respectively, correspond to the complex cyclotomic fourth roots, evaluated on the  and  planes. The  and  hypercomplex roots with  and  have minus sign contributions in the products due to the even rs but are otherwise analogous to the previous roots. When  are odd, both director coefficients are odd multiples of . A zero scator is thus obtained in the additive representation. These four odd  are spurious nilroots (Corollary 2). Recall that lack of associativity arises if the additive scalar component of a product vanishes. In this case, the first warning came from the , (or ) roots that square to , but their cube  has a vanishing scalar. Nonetheless, associativity holds because there is only one non-zero director component in either representation. However, when  are both odd, there are two non-zero multiplicative director components, and the products no longer associate in the additive representation.
The set of all fourth roots of one does not satisfy group properties in the additive representation even if nilroots are dismissed, because products of orthogonal roots are zero and associativity between different unit orthogonal roots is not satisfied, i.e., 
. Nonetheless, there are four 
 and three 
 cyclic groups and a Klein four group, represented graphically in 
Figure 4.
In contrast, consider the multiplicative representation where the product satisfies commutative group properties and the exponent distributes over the factors. In 
, the fourth roots of unity are
      
The 16 possible roots in the multiplicative representation are shown in 
Table 3. These 16 elements satisfy Abelian group properties. Its proper subgroups are illustrated in 
Figure 4.
Remark 1. In the multiplicative representation, the product  is irreducible ([6], Definition 2), that is, it cannot be further simplified in this representation since the product has already been performed by having evaluated the sum of arguments. In particular, the multiplicative representation of a scator with multiplicative variables  is , and its magnitude is 1.  The power series of the 
 function is actually the additive representation of this function together with the series expansion of the trigonometric functions. If no reference is made to the power series of the 
 function, as in the previous remark, the question of course arises as to the meaning of this function. The complex exponential can be characterized in several other ways; two of these characterizations have been shown to be also true for the 
 function in 
: (i) the components exponential function 
 is the scator holomorphic function 
 of the scator variable 
, solution to the differential equation [
11]
      
, where 
. (ii) The components exponential function maps scator addition onto scator multiplication ([
6], Lemma 2.1), 
. Therefore, the components exponential function makes sense even if its series representation is not invoked. However, the equivalence between these characterizations in 
 has not been established.
  6. Fifth Roots of 1
The Victoria equation for the fifth root of 
 in 
 is
      
There are 
 possible roots, grouped into six groups with five elements (
), all of them sharing, of course, the neutral element, 
. The roots, labelled 
 with 
 from 0 to 4, are summarized in 
Table 4. Two groups, 
 and 
, correspond to the complex cyclotomic fifth roots, evaluated on the 
 and 
 planes. Two other groups correspond to hypercomplex roots with equal magnitude director coefficients that lie on the 
 planes with respect to the 
 axes. These roots have positive 
s values, 
. The group of roots 
, 
, 
, 
, and 
 has director components with equal signs, both positive or both negative. These roots are shown in red in 
Figure 5, joined by a pentagon with black arrows. The group 
, 
, 
, 
, and 
 has director components with equal magnitudes but opposite signs (purple in 
Figure 5). They exhibit the same structure of the iso-director magnitude groups of the cubic roots. There are now another two sets with a structure that is not present in roots of unity with 
, 
, 
, 
, 
, 
, 
, 
, and 
. All eight roots, depicted in magenta and light blue in 
Figure 5, have the same negative additive scalar coefficient, 
. These eight roots lie on the 
 plane, but 
 does not lie in this plane. Therefore, the five elements of these two groups no longer lie on a plane as in previous cases. The 
 proper Abelian subgroups of these roots are depicted in 
Figure 6.
  7. qth Roots of 1
A scator  raised to the power  has  possible roots. The roots are obtained from the Victoria root Theorem 2 by evaluation of each of the s, from 0 to q, and for j, from 1 to n. In the multiplicative representation, the roots are obtained from Theorem 1. Label the roots for identification with subindices  to , .
The roots of unity, provided that associativity of the factors is ensured, can be grouped in cyclic sets that satisfy Abelian group properties with 
q elements in each set. 
 elements with different 
s belong only to one such set, and the element 
 is the identity element, common to all groups. The 
 roots are grouped into 
g sets; thus, 
, the number of cyclic groups with 
q elements, is then
      
For example, for scators in 
, there are two director components; then, 
. For scators in 
, 
, there are, in particular, 
 cube roots that can be grouped into 
 sets. The elements of a group are obtained by evaluation of the 
p powers of any root in the group except the identity, for 
p from 0 to 
, modulo 
q. Examples for 
q equalling 2, 3, and 5 are shown in 
Figure 6. For example, let 
 be a fifth root of 
: the square is 
, the cube 
, the fourth power 
, and the fifth power 
. For 
, this sequence produces a pentagon, similar to the one depicted with black arrows in 
Figure 5. The sequence 
, 
, 
, 
, and 
 produces a star-like figure, etc. In the multiplicative representation, this scheme is true for any of the 
 roots of unity, since the product is associative. However, in the additive representation, for even 
 roots of unity, the number of roots is reduced, and the set of all roots for a given 
q does not satisfy group properties. There are, nonetheless, cyclic group subsets within the sets of even 
q roots, as shown in 
Figure 4.
  Director Conjugate Roots
The conjugate of a scator leaves the scalar part invariant and changes the sign of the director coefficients, either in the additive or the multiplicative representation. The 
 conjugate or 
director conjugate of a scator 
 is defined by the negative of the 
 director component, while all the remaining components are left unchanged [
14]. In the present context, we refer to the 
 period conjugate when the 
 coefficient of a director is left unaltered, but the fundamental periodicity 
 changes sign, 
.
Lemma 2. If  is a  root of , a  period conjugate scator of this root  is also a root for any  from 1 to n. There are  director period conjugate roots for each ordered set of d non-zero  elements,.
 Proof.  Since 
 is a root, in the multiplicative representation,
so that
 Let 
 in the 
 term be modified to 
, 
, so that 
. Evaluate 
 to the power 
q using Theorem 1,
but 
; thus, 
. Therefore, 
 is also a root. Since any or several 
s can be changed by their negative value independently, given a root 
, 
 roots are obtained from the evaluation of 
 for the ordered sets of 
s different from zero from 1 to 
n. For 
,
For , ;
For ,  □
Director conjugation is applied independently to each director component. Only 
s from 1 to 
, if 
q is odd, or 
, if 
q is even, need to be considered, since 
 is equal to 
. For roots of 
, period director conjugation and director conjugation give the same expressions since 
 for all 
j. The fifth roots of 
 are shown as an example in 
Figure 7, grouped into nine cyclic sets under director conjugation. These sets clearly do not satisfy group properties, except the trivial single element 
 set. Nonetheless, they are cyclic under director conjugation, alternating the conjugation components. For the fifth root, it suffices to consider 
 non-zero values for 
. For 
, 
, there are four possible combinations of 
. Each of these four sets have 
 elements. Take, for instance, the set joined by a horizontal rectangle on the left in 
Figure 7; the 
 element first component conjugation is 
. The superstar followed by a number notation means the conjugation of the corresponding number position. Negative indices are mod 5 mapped onto positive indices for ease of comparison with 
Figure 6. Subsequent elements of these sets upon director conjugation are 
, 
, and 
. This set of four elements is also depicted on the right in 
Figure 7. This clockwise cycle, seen from the positive 
s axis, becomes an anticlockwise cycle if director conjugation is initiated with the second component. Scator conjugation alternates between diagonal elements in these sets, i.e., 
. Since conjugation is a second order involution, if applied twice, it leaves the element invariant. For 
, 
, there are two possible combinations of 
. Each of these two sets have 
 elements. Another two sets of two elements are obtained for 
. The multiplicative identity 
 is invariant under conjugation and is the only element in the remaining set.
To summarize the results in this last section, given a  root of  in , its scator conjugate (changing signs to all director components) is also a root. This result is analogous to complex conjugate roots of one in . Furthermore, the  director conjugate of a  root (changing sign only to the  director component) is also a root. Since the director conjugate can be subsequently evaluated for any another component, say , and so on, all director conjugate possibilities of a  root are also roots. These roots are a higher dimensional generalization of complex conjugate roots in a  scator dimensional space.