Open access peer-reviewed chapter

Fractal Pyramid: A New Math Tool to Reorient and Accelerate a Spacecraft

Written By

Alexander P. Yefremov

Submitted: August 1st, 2017 Reviewed: October 18th, 2017 Published: December 20th, 2017

DOI: 10.5772/intechopen.71751

Chapter metrics overview

1,238 Chapter Downloads

View Full Metrics


An original mathematical instrument matching two different operational procedures aimed to change orientation and velocity of a spacecraft is suggested and described in detail. The tool’s basements, quaternion algebra with its square-root (pregeometric) image, and fractal surface are represented in a parenthetical but in a sufficient format, indicating their principle properties providing solution to the operational task. A supplementary notion of vector-quaternion version of relativity theory is introduced since the spacecraft-observer mechanical system appears congenitally relativistic. The new tool is shown to have a simple pregeometric image of a fractal pyramid whose tilt and distortion evoke needed changes in the spacecraft’s motion parameters, and the respective math procedures proved to be simplified compared with the traditionally used math methods.


  • spacecraft motion
  • operation
  • quaternion
  • fractal surface

1. Introduction

In classical mechanics, rotation of a rigid body (in particular, a spacecraft) and its translational motion are normally regarded as drastically different actions leading to changes in its position and are respectively described by different groups. Relativistic mechanics, in its turn, deals with these two types of motions “more homogeneously” since rotation and linear motion are described in this case by 4 × 4 matrices from the Lorentz group SO13. However, it is well known that the special relativity limits itself by inertial motions of the involved frames of reference while use of general relativity comprising any types of motion but demanding math methods of tensor calculus seems unapproved sophisticated. Happily, there exists a simpler vector version of the relativity theory admitting arbitrary accelerated motion of the frames. A brief formulation of the theory is made with the help of quaternion vector units, each set of the units representing a Cartesian-type frame of reference. In this case, the rotation-and-translation operator is given by 3×3 matrix belonging to the group SO3С known to be 1:1 isomorphic to the group SO13. However, the calculations of the body’s complex motions even within the framework of the vector-quaternion relativity remain prolonged and cumbersome, a simpler method is desired. Such a method is found due to existence of 1:2 isomorphism of the groups SO3С and SL2С, the last being a spinor group operating in fractal two-dimensional complex-number valued space (a fractal surface). It is necessary to mention that the subgroup of SL2С, rotational group SU2, is normally used in space-flight practice, providing comparatively simple mathematical computations for a spacecraft reorientation tasks [1, 2]. This method is based upon similarity-type transformations of the initial quaternion triad, in fact assuming nontrivial multiplication of at least three different quaternions, though it straightforwardly gives the data describing the axis of single rotation and value of the respective angle. However, this method provides no translational motion.

In this study, we suggest an essential development of the last (single rotation) method leading, first, to noticeable simplification of computations, and second, to possibility of introduction of additional parameters responsible for the spacecraft acceleration. This development is fully based on fundamental properties of subgeometric dyad forming the fractal space in a way underlying the 3D physical space. Moreover, we suggest subgeometric images (fractal joystick and fractal pyramid) of the math tools realizing the spacecraft’s reorientation and acceleration tasks. As well, we give a brief comparative analysis of simplicity (or complexity) of conventional and new methods.

The study is composed as following. In Sections 2–4 we offer a detailed mathematical introduction. In Section 2, we renew our knowledge of quaternion algebra giving traditional (Hamiltonian) and more compact (tensor) notions and correlations. In Section 3, we briefly reproduce the quaternion version of the relativity theory. In Section 4, we consider main notions and properties of the 2D fractal space and show how to build a 3D frame out of a dyad element.

Sections 5–7 are devoted to new math methods making operations of a spacecraft simpler and more functional. Section 5 is devoted to presentation of three methods to reorient a spacecraft with accent on convenience of the single rotation method involving a fractal joystick model. In Section 6, we suggest a very simple way to introduce (apart from space rotation) an acceleration of the spacecraft and demonstrate a subgeometric image of the respective math tool having a shape of fractal pyramid. Finally, in Section 7, we give a sketch of a technological map previewing necessary steps to simultaneously reorient and accelerate the spacecraft followed by a series of relevant pictures.


2. Basic notions and relations of quaternion algebra

Quaternion (Q-) numbers were discovered by Hamilton in 1843 [3]. A quaternion is a math object of the type q=a1+bi+cj+dk (in Hamilton’s notation), where a,b,c,d are real coefficients at the real unit 1 (the symbol is normally omitted in the number) and at three imaginary units i,j,k forming the postulated multiplication table (16 equalities).


Q-numbers and the multiplication law (1) can be more compactly rewritten in the vector (and tensor) notations i,j,kq1,q2,q3qk, j,k,l,m,n=1,2,3; then, a quaternion is a sum of scalar a and vector bkqk parts qa+bkqk, where a,bkR, and the multiplication table (1) has the form


Summation in repeated indices is implied, and δkl and εklj are the 3D Kronecker and Levi-Chivita symbols (see e.g., [4]).

Quaternions admit the same operations as real and complex numbers. Comparison of Q-numbers is reduced to their equality: two Q-numbers are equal if coefficients at respective units are equal. Commutative addition (subtraction) of Q-numbers is made by components. Q-numbers are multiplied as polynomials; the rules (1, 2) state that multiplication is noncommutative (left and right products are defined), but still associative. A quaternion qa+bkqk has its conjugate q¯abkqk, the norm q2qq¯=q¯q, and the modulus (positive square root from the norm) qqq¯=a2+bkbk. Inverse number is q1=q¯/q2; so, for two quaternions q1 and q2, division (left and right) is defined as q1/q2left=q2q1/q22 and q1/q2right=q1q2/q22. If q is a product of two multipliers q1=a+bkqk and q2=c+dnqn, then from definition of the norm one finds


Written in components, Eq. (3) becomes the famous identity of four squares


Identities of the type (4) exist only in four algebras: of real numbers (trivial identity), of complex numbers (two squares), of quaternions (four squares), and of octonions (the last exclusive algebra with one real and seven imaginary units admits identity of eight squares; multiplication in this algebra is no more associative).

Geometrically, the imaginary Q-units are associated with three unit vectors initiating a Cartesian coordinate system (Q-triad, Q-frame). This image, in particular, follows from the fact that, according to Eq. (2), each imaginary unit appears as ordered product of the two others: q1=q2q3, q2=q3q1, q3=q1q2 (vector products in Gibbs-Heaviside algebra). One can easily construct a set of such units. To demonstrate this, we consider a couple of 2×2-matrices, A=abca, B=defd, traceless: TrA=TrB=0, and not degenerate: detA0, detB0. We use the matrices to build two different imaginary units as


We form the product of the two units and demand that its trace vanishes that is given as


then Eq. (6) gives expression for the third imaginary Q-unit q1q2=q3, and as a whole, we get the Q-triad qk, the real unit always remaining the unit matrix 11001. One readily checks up that the triad given by Eqs. (5) and (6) identically satisfies the multiplication law (2). Built in a similar way, the simplest representation of Q-units qk˜ is given by the Pauli matrices pk˜ with factor –i: qk˜=ipk˜


and the imaginary Q-triad given as Eq. (7) describes a constant Q-vector frame.

However, a Q-frame may be variable, rotating, and moving. There are two types of transformations changing the frame but retaining the form of the multiplication law (2). The first is rotational-type transformation


where Okn is a 3×3-matrix (its components are in general complex numbers) having orthogonal properties OknOmn=δkm, hence this matrix belongs to the special orthogonal group of 3D rotations over field of complex numbers OknSO3C. The matrix Onk can be always represented as a product of plane (or simple) rotations, irreducible representations of SO3C. For such matrices, a special notation will be used, e.g., OnΘ, where the lower index indicates the rotation axis (the frame’s unit vector) and upper index shows the rotation angle. Depending on the math nature of the angle Θ, we distinguish two types of simple rotations. If Θ=αR, then we have a real simple rotation OnΘRnα; if the angle is imaginary Θ=ηiR, then we have a simple hyperbolic rotation OnΘHnη; for example Eq. (9)


Superposition of any number (N) of real rotations (product of relevant matrices) gives a (nonplane) real rotation


Product of multiple hyperbolic rotations is physically sensible if accompanied by real rotations in the framework of vector version of theory of relativity (see Section 3); so in general, the matrices of the type


are used in applications.

The second type of transformations is performed by an operator U and its inverse U1 is given as


It is evident that the transformation (12) keeps the form of the basic law (2). The operators U are known to form the (spinor) group USL2C of special linear 2D transformations over field of complex numbers; this group is 2:1 isomorphic to SO3C and similarly to the Lorentz group. A special case of the transformation (12) is a real rotation made by means of the subgroup SU2SL2C, and this spinor subgroup is 2:1 isomorphic to vector group SO3R. It is necessary to note that the transformation of the type (12) with USU2 is most frequently used for solution of a spacecraft orientation problem (see Section 5.2).

As well, in formulation of quaternion relativity (see Section 3), we shall need notion of a biquaternion (BQ-) number. Such a number has the form b=x+ykqk, where x,ykC while 1, qk are Q-units. BQ-numbers admit addition, multiplication, and conjugation b=xykqk. But the norm is not well defined since the product bb¯=x2+ykyk in general is not a real (and positive) number. A real number “norm” exists in the subset of vector biquaternions


whose real and imaginary parts are mutually orthogonal


There are evidently zero dividers in Eq. (14), hence division is not well defined, but the subset (13 and 14) comprises basic formulas describing relative motion of arbitrary accelerated frames of reference.


3. Vector-quaternion version of the relativity theory

According to Eqs. (13) and (14), the interval of Einstein’s relativity theory1


admits a BQ-square root


where displacement of observed object dxk is orthogonal to a unit vectorek directing change in time dt:ekdxk=0. Under these conditions, square of Eq. (16) yields Eq. (15) dsds=ds2. It is convenient to explicitly relate displacement dxk to a plane orthogonal to time-directing vector ek with the help of metric-projector bknδkneken dxk=dxnbnk,

then the orthogonality condition is fulfilled automatically ekdxk=ekdxnbkn=0.

The interval (15) is invariant under Lorentz transformations of coordinate system dxα=Lλαdxλ, LλαSO13, while the Q-frame can be subject to SO3C rotations qk=Oklql; simultaneous application of the transformations, together with demand that the BQ-vector (16) form be conserved, leads to correlation between components of matrices Okl and Lαλ2 [5, 6]


Eqs. (17) and (18) in particular mean that within the group SO3C a set of ordered simple rotations of the type (11) are distinguished, real and hyperbolic, each performed about one-unit vector of Q-triad. If for instance, direction No. 1 of Lαλ is not involved in the transformation (ek=ek=δ1k), then Eqs. (17) and (18) represent the matrix O as function of components of Lorentz matrix L


The matrix (19) may describe a series of simple rotations, but real rotations should be always performed about vector q1 (initial or transformed), while hyperbolic rotations are allowed about vectors q2 and q3. It is easily checked up that all matrices O of the type (19) constitute a subgroup SO12SO3C of the ordered rotations of Q-triads.

Main idea of Q-version of relativity is to replace line element of Einstein’s relativity (15) and its invariance under Lorentz group by adequate BQ-vector (16) invariant under rotational group represented by matrices OSO12. Then, instead of quadratic form of four-dimensional coordinates, an observer has at his disposal a movable Q-triad with time and distances measured along its unit vectors and dealt with the vector basement as with the Newtonian mechanics or general relativity in tetrad formulation. However, on this way, an essential peculiarity arises. Eq. (16) implies that the constructed space-time model has six dimensions, and it is a symmetric sum of two three-dimensional (3D) spaces Q6=R3T3, where R3 is the usual 3D space where coordinate and velocity change, whereas T3 is also a 3D space but imaginary with respect to R3. In this model, the observer works only with some sections of the 6D space; but since the objects of the observations are found in real 3D space, and imaginary time axis is distinguished, an illusion of four dimensions emerges.

Physical measurements in the Q-model are made with the help of three spatial rulers qk and built-in geometric clock represented by “imaginary time rulers” (Pauli-type matrices) pkiqk, the two triads being obviously co-aligned. The tool-set Σpkqk with an observer in the initial point represents full physical frame of reference, Eq. (16) can be rewritten as


Now, the principal statement of the Q-version of relativity follows: all physically sustainable frames of reference are interconnected by “rotational equations”


The sustainability means form-invariance of BQ-vector (16) or (20) under transformations (21). Kinematic effects of special relativity are straightforwardly found in the Q-version; here, we demonstrate only one effect important for fractal pyramid technology accelerating a spacecraft (see Section 6).

Boost. Σ-observer always can align one of his spatial vectors (e.g., q2) with velocity of moving body, so basic BQ-vector can be written in the form


Let the frame Σ be a result of a hyperbolic rotation of a constant frame Σ


with the matrix H3ηfrom Eq. (9b) (rotation about q3 by angle η). This simple rotation, physically a boost, obviously keeping BQ-vector (20) form-invariant


yields familiar coordinate transformations


with respective effects of length and time segments contraction. If observed particle is the body of reference of the frame Σ′, then dr=0, and one finds that the frame Σ′ is moving with the velocity


Specific features of the Q-vector version of relativity will be effectively used below in the fractal-pyramid math method to operate a spacecraft. Now, we turn to notions of a fractal space.


4. Fractal space underlying physical space

In this section, we show that a 3D space (e.g., physical space) may be endowed with a pregeometry [7] mathematically described by a complex-numbered surface, a 2D fractal space, each its vector having dimensionality half compared to that of the 3D space. We start with 2D space and construct out of its basic elements a basis of 3D space.

Let there exist a smooth 2D space (surface) endowed with a metric gAB (and inverse: gBCgBCgABgBC=δAC) and with a system of coordinates xA=x1x2; here A,B,C=1,2, δAC is a 2D Kronecker symbol, summation in repeated indices is also implied. The line element of the surface is


the surface may be curved, so covariant and contravariant metric components differ. In a point, we choose a couple of unit orthogonal vectors aA, bB (a dyad)


A domain of the surface in vicinity of the dyad’s initial point (together with respective part of tangent plane having the metric δMN=δMN=δmN) will be called a “2D-cell.”

Considering direct (tensor) products of the dyad vectors with mixed components [8], we can construct only four such products (2×2 matrices): two idempotent matrices


and two nilpotent matrices


Next, we built sum and difference of the idempotent matrices


and sum and difference of the nilpotent matrices


If the units Eqs. (31b) and (31c) are slightly corrected so that their product is the third unit (31d), then we obtain the basis of quaternion (and biquaternion) numbers


Now, we recall the spectral theorem (of the matrix theory) stating that any invertible matrix with distinct eigenvalues can be represented as a sum of idempotent projectors with the eigenvalues as coefficients, the projectors being direct products of vectors of a biorthogonal basis. The unit q3 defined in Eqs. (32), (31b) is the characteristic example


Right and left eigenfunctions of q3 are vectors aA, bB and covectors aA, bB of the dyad, respectively; the eigenvalues are +i (for a) and i (for b), and GBA, HBA are the projectors.

As mentioned above, the similarity transformation of the units


preserves the form of algebras’ multiplication law (2). Therefore, vector units from Eq. (32) can be obtained from a single unit, say, q3 by a transformation (34). Then, all vector units have same eigenvalues ±i, and the eigenfunctions of the derived units are linear combinations of the eigenfunctions of the initial unit [9]. This also means that the mapping (34) is a secondary one, but the primary one is SL2C transformation of dyad vectors, thus forming a set of spinors from the viewpoint of the 3D space described by the triad vectors qk.

Hereinafter, we introduce shorter 2D-index-free matrix notations for the dyad: a vector is a column, a co-vector is a row, and a parity indicator + or marks the sign of the eigenvalue ±i


this helps to rewrite the above expressions more compactly. The dyad orthonormality conditions (28, 29) acquire the form


the idempotent projectors are denoted as C+G=ψ+φ+,CH=ψφ,

and the units (32) are expressed through the single dyad vectors (co-vectors) as


Eq. (37) obviously demonstrates that the dyad elements are in a way “square roots” from 3D vector units. So, if we put dimensionality of any 3D line to be a unity, then dimensionality of a line on the 2D space (e.g., dimensionality of a dyad vector) must be ½; hence from the viewpoint of the 3D space, the surface determined by a dyad is fractal. The next important observation concerns transformations. The transformation (34) clearly results from the SL2C transformations of the dyad vectors (covectors)


So, apart from vector-type (8) and spinor-type (12) transformations of a Q-triad (an element of 3D space), there exists a possibility to deal with more fundamental math elements, vectors, and covectors describing “pregeometric” 2D cell of a fractal surface. These simpler math objects are subject to evidently simpler mapping (38); moreover, in the following sections, we will show that the operators of the transformations, being themselves BQ-numbers, suggest simpler and less numerous equations to solve, thus reducing degree of math load and probability of mistakes.


5. Three methods to reorient a spacecraft and fractal joystick

The orientation tasks are relevant with computations over 3D flat space modeling a local domain of the physical space. Two types of the orientation problem solutions are traditional: (i) a series of subsequent several angles rotation and (ii) a one-angle rotation about an instant axis. Mixed variants exist, but are less productive, and they are not normally considered.

If magnitudes involved in calculations are generically measured in real numbers, then both techniques (i) and (ii) are based on the vector rotation group SO3R. Math content of the technique (i) implies a multiple set of plane rotations [of type of Eq. (9a)] by Euler (or Krylov, or others) angles about selected axes. The technique (ii) in its turn represents a nontrivial problem of determining the instant axis of a single rotation.

Quaternions are widely known to fit better than real numbers for the orientation tasks mostly due to the fact that three vector units represent models of three mutually orthogonal gyroscope axes. As well, use of the Q-algebra formalism essentially simplifies calculations, especially for the technique (ii), since both the vector rotation group SO3R and its spinor “equivalent” SU2 reflection group can be used whatever enigmatic were formulas describing spinor rotations. However, the quaternion algebra reveals its unique property to split axial 3D vectors into dyad sets belonging to a fractal subspace as in Eq. (37), see also the basic work [10]. The above-described fractalization procedure, mathematically nontrivial and much less known, on the one hand clarifies “mysterious” two-side SU2 quaternion vector multiplication and on the other hand endows all algebraic objects and actions with distinct geometric sense; moreover, the calculations become most primitive. Solution of a spacecraft reorientation task as transformation of a fractal dyad represents the third math method (iii) suggested here. However, all three math methods are described in detail in this section.

5.1. Quaternion SO(3,R) approach to the reorientation problem: Technique (i)

Orientation of a spacecraft in 3D space is determined by three angles between axes of some global coordinate system and unit vectors of a frame attached to the moving body taking into account its physical symmetry. The global coordinates, e.g., are represented by a spherical system, and its local initiating vectors pointing: q1 to the north along the Earth’s meridian), q2 along a parallel, and q3 to zenith direction. The directing vectors qk are considered constant. Then, the orientation of a spacecraft bearing a frame qk, (with q1 along the body, q2 a transverse one, and q3 along gravity) is determined by three angles: “yaw” ψ, the angle between q1 and q1 (rotation about q3); “roll” φ, angle q2q2 (rotation about q1); and “pitch” θ, angle q3q3 (rotation about q2). Within these notations, the spacecraft’s orientation in the space is described by the matrix equation


Outlined above technique (i) demands that the matrix Rnk be represented as a product of simple rotations, irreducible representations of SO3R [a special notation for such matrix is Rnα, see Section 2, Eqs. (9, 10)], each performed about a frame’s unit vector. Simple rotations with the above parameters of the probe’s orientations are given by the matrices


Direct reorientation problem, i.e., reaching object’s assigned orientation, can be solved by a sequence of plane rotations mathematically described by a sequent multiplication of matrices [see Eq. (10)]. This problem has no unique solution since the group SO3R is not commutative; i.e., different multiplication order of the matrices (40) with the same parameters (angles) generally gives different result; e.g., the products R=R3ψR2φR1θ and R=R1θR2φR3ψ are, in general, different RR. Vice versa, different orders of the matrix product with other parameters may yield the same result, e.g., products R=R3ψR2φR1θ and R=R1θR2φR3ψ may represent equivalent rotational result R=R. The possibility to represent an arbitrary SO3R matrix as a product of its irreducible representations given in different order in particular entails uncertainty in solution of the inverse problem when one has to determine values of angles securing an assigned reorientation of the spacecraft. Therefore, the technique (i) does not provide single-valued results.

Even with more difficult, we meet trying to use matrices from the group SO3Rin the technique (ii). As is known from the theory of matrices (see e.g., [11]) in this case, we have to solve the characteristic equation RX=X searching for the matrix operator R an eigenvector X with unit eigenvalue, the vector X pointing direction of the instant rotation axis. This tough algebraic task then followed by sophisticated calculations aimed to find the instant rotation angle. The use of hypercomplex numbers essentially helped to avoid these math troubles, and about half of a century ago, quaternion algebra became a common tool serving for engineering goals of navigation and orientation. Indeed, the similarity transformation UqU1 of a quaternion q performed with the help of auxiliary quaternion Ua+bq geometrically leads to conical rotation of the vector part of q about an axis whose direction is determined by the unit Q-vector q (e.g., [2]); the value of the instant rotation angle is computed as 2arctanb/a. Below, we suggest a detailed analysis of this type of description of rotations.

5.2. Reorientation by a single rotation of the quaternion frame: Technique (ii)

Consider a 2×2matrix (with complex-number components) Uxzwy, belonging to a special linear group USL2,C, detU=xywz=1. The multiplication law (6) is obviously form invariant under the similarity-type transformation


One readily demonstrates that the matrix U is a biquaternion with the definable norm; indeed,




and q is a Q-vector unit


The unit vector (44) represented through the constant basis (7) has the form; q=lkqk=bk/bqk where lk=bk/b are components of a unit vector pointing in 3D space a vector with components bk, then the condition detU=xywz=1 takes the form a2+b2=1, b2=bkbk. This general biquaternion case will be used in subsequent studies when combined rotation-plus-translational motion is regarded (see Section 6). In this section, we consider only quaternion case: a,bkR, so the matrix U is unimodular if




with lk representing cosines of angles between Q-vectors qk and the direction determined by q. With the help of Eqs. (46) and (2), we reproduce the transformation (42) in the developed form


Eq. (47) in fact interlinks the SO3R rotation matrix components and the parameters of SU2 transformations of a Q-frame [compare with (39)]. As well, Eq. (47) helps to make the following geometric analysis.

Multiplied by lk (with summation in index k), Eq. (47) yields the equality lkqk=lnqn, meaning that vectors of the transformed frame qk have the same projections onto vector lk as the initial frame qk; i.e., the transformation may be represented as a conical rotation about lk, Φ2α, which is angle of the rotation in the orthogonal plane with the metric pkn=δknlkln [see the second term in Eq. (47)]. Let two unit vectors ek,nk form this plane pkn=eken+nknn, then lmεmkn=eknnennk, and the SO3R-matrix comprised in Eq. (47) acquires the form


Introducing now two artificial unit vectors with complex number components skek+ink/2 and skekink/2, we get the final (canonical) expression


Eq. (49) is just an explicit formulation of the spectral theorem applied on a 3D orthogonal matrix. Since its determinant differs from zero, this matrix is nonsingular, all its eigenvalues λi are different, so it is simple; therefore, it can be expanded into a series of projectors Ci with λi as coefficients


Here, λ1=1, λ2=eiΦ, λ3=e, C1kn=lkln, C(2)kn=sksn*, C3kn=sksn; the projectors are idempotents CiN=Ci, N being a natural number, TrCi=1, detCi=0. It is important to note that the decomposition of a matrix R into the series (49, 50) necessarily leads to appearance of the complex-numbered 2D basis sk sk; we will indicate similar features in the fractal technique (iii) below.

The value of the single rotation angle follows from computation of the trace of the matrix (49)


antisymmetric part of the matrix yields the components of unit vector directing the rotation axis


Eqs. (51) and (52) represent parameters of the single rotations, the angle Φ. and components lj of the vector pointing the rotation axis, as functions of an arbitrary SO3C rotation angles, e.g., yaw, roll, and pitch ψφθ, and parameters of an equivalent single rotation, the value of the angle Φ and components (in the initial frame) lj of the vector pointing the rotation axis.

5.3. Reorientation as transformation of a fractal surface, technique (iii)

In Section 4, we demonstrated that each vector of any Q-triad qk is a linear combination of vector-covector direct products of its proper biorthogonal basis ψ±φ± belonging to a domain of complex-number valued 2D fractal space [see Eqs. (37)]. Then, rotation (reorientation) of the frame qk by the technique (ii) on the base of the transformation (42) induces specific type of the “interior” rotation on the fractal surface level [see Eq. (38)]


Further on, we use for the dyad the eigenvectors ψ± [and eigencovectors as Hermitian conjugation of the vectors φ±=ψ±T] of q3 of any Q-triad, where respective eigenvalues being ±i. In the simplest case of q3 from Eq. (7), the constant dyad is


Normalization and orthogonality conditions are identically satisfied. The matrix U, as a quaternion (46), is expressible in terms of the fractal basis


Therefore, Eq. (53) takes the form


Eq. (56) shows that the nonlinear problem formulated within the technique (ii), on the fractal surface level, is reduced to a linear task of the 2D basis rotation.

To get technological formulas convenient for fast numerical computation, we denote the final values of the new 2D basis as


Then, we notice that only one new dyad vector is to be computed,


The second vector ψ and the co-vectors are simply expressed through the factors (57) and their complex conjugation


This helps to represent the 3D reorientation processes “subgeometrically”, on the 2D fractal level, as a displacement of a “joystick” tool (see [12] and Figure 1).

Figure 1.

Fractal “joystick tool”.

2D complex-numbered space can be imaged as a pyramid (with no base) consisting of one real, one imaginary, and two mixed real-imaginary joined surfaces. The joystick has one of its end matched with the pyramid’s top by a hinge; a certain shift of the stick gives components of a new dyad vectors and co-vectors. From these fractal elements, a new Q-frame providing the assigned reorientation of the spacecraft is straightforwardly built.

All reorientation parameters providing operations in the fractal space are in fact the components of the matrix USU2; therefore, the unit vector directing the axis of instant rotation is given by Eq. (52); the fractal rotation angle is


Eqs. (59), (52), (56), and (37) suggest a very simple algorithm for computation of all parameters of a single rotation and resulting matrices of a reoriented Q-triad describing new orientation of a spacecraft.

The technological scheme of the reorientation procedure can be briefly outlined as the following steps:

  • A spacecraft reorientation is assigned by a series of simple rotations [Eq. (40)].

  • Components of the rotation axis vector are computed [Eq. (52)].

  • The angle of fractal rotation is computed [Eq. (59)].

  • The dyad and resulting Q-triad are computed [Eqs. (56), (37), much simpler than in Eq. (49)].

  • If the computed and assigned frames match, then the rotation parameters are sent to the operational systems realizing the reorientation.

The study suggested in Section 5 gives detailed analysis of math mechanisms linking two different approaches to solution of an object’s reorientation task, a consequent 3D rotations described by matrices and a single rotation about an instant axis described by matrices. We like to emphasize importance (and original form) of Eqs. (48) and (49) explicitly demonstrating the projector-eigenvalue decomposition of any SO3R matrix, so immediately giving technological values of the single rotation. Another novel math feature of the problem is its connection with subgeometric properties of a fractal complex number surface.

However, thorough analysis of the Q-math reveals its additional, and important, option quite helpful in operational tasks. Namely, extension of the groups SO3R and SU2 to the rotations with complex parameters, SO3С and SL2C, respectively, with the vector-quaternion version of relativity theory taken into account, may open a possibility not only reorient but as well simultaneously endow a spacecraft with velocity assigned in value and direction. Apparently, this math tool matching rotations and accelerations, if possible in 3D space, should exist as fractal mechanism. Designing of such original (and exotic) operational instrument is a challenging task; it is in detail analyzed in the next section.


6. Hyperbolic rotations and a fractal pyramid

In this section, we essentially extend the methods briefly described above. The crucial point of the extension is introduction of an imaginary parameter of rotation, thus involving hyperbolic functions. We assume that this action will result in possibility to control not only orientation, but as well dynamics of the spacecraft. We will prove the assumption within extended formulation of the technique (iii).

But at first, to make the picture more clear, we show it in framework of 3D serial rotations [technique (i)], and for simplicity, we implement just one supplement plane hyperbolic rotation about one axis


so that hyperbolic functions are introduced. Then, complete rotational operator is


We rewrite the operator (61) in the spinor-type form where the tilde denotes some initial basis


and the components of the instant rotation axis vector given by Eq. (52). It is important to note that in the computation procedure, we have to deal with vectors belonging to the same frame. Therefore, we express q3=R3nqn and make multiplication in Eq. (62) to obtain


This expression is again a quaternion and we denote it as




parameter Θ being a complex number. One straightforwardly verifies fulfilling the identity

cos2Θ+sin2Θq2=coshη2cosΦ2isinhηsinΦ2R3n˜ln˜coshη2cosΦ2isinhηsinΦ2R3p˜lp˜++coshη2sinΦ2ln˜isinhη2cosΦ2R3n˜+sinΦ2R3j˜lm˜εjmn×             ×coshη2sinΦ2ln˜isinhη2coshΦ2R3n˜+sinΦ2R3l˜lp˜εlpsqn˜qs˜=1.E67

Expression for the vector-directing axis of the single rotation is found from Eqs. (65) and (66)


Eq. (61) represents an operator performing the serial rotation, and Eqs. (65), (68) give parameters of a single rotation. Physical content of this rotation is easily revealed when the mapping is made in the fractal surface format, and then returned into 3D space. Despite seeming complexity of the given expressions, the final calculation is shown to be very simple.

So, following the ideology of geometrization of the algebraic actions, we plunge into the fractal medium, and we consider the technique (iii). We rewrite fractal mapping with the operator (62) in the form


where the intermediate dyad is a result of the real rotation (similar with the covectors)


We also stress that all dyad elements used in the computations are always the eigenvectors (eigencovectors) of the quaternion unit q3


hence, Eq. (69) produces a new fractal basis simply multiplying the intermediate dyad by an exponent


By other words, one dyad vector and one co-vector (here ψ+and φ) become longer, and the others (ψ and φ+) become shorter, all of them though preserving unit length, i.e., rescaled.

This primitive mapping has clear physical sense concerning kinematic of a spacecraft. To reveal it, we, using Eq. (75), build an “imaginary constituent” of the 3D frame vector q1 as in Eq. (37b).


However from Eqs. (37b, c), we find


substitution of the Eq. (74) into Eq. (73) yields


Eq. (75a) rewritten in terms of the Pauli-type matrices [as in Eqs. (20), (22)] piq has the form


Using results of Section 3, we associate the hyperbolic functions with the time ratio


(linking time dt of an immobile frame and proper time dt of moving spacecraft) and with the relative velocity ratio (c is speed of light).


Then, Eq. (75b) takes the form of “vector interval” of quaternion version of relativity theory (23)


when squared, it gives the spacecraft’s special relativistic space-time interval linked with the frame at rest by the Lorentz (hyperbolic) transformation


describing kinematics of a frame moving along q2 with velocity V, while the vectors p1(or p1) play the role of direction of time in the immobile (or moving) spacecraft. It is always possible to choose the direction q2 as pointing the “yaw” of a spacecraft. In particular, the velocity can be small sufficiently to reduce the calculations into classical format


besides, the velocity modulus may be variable in time; hence, the spacecraft is accelerated.

So, introducing imaginary rotation angles, we obtain a possibility to control an arbitrary space reorientation of a spacecraft with variation of its velocity in the direction that can be as well changing with time (In this sample, the vector q2 is in fact permanently rotating.)

This math tool has two important properties. First, a spacecraft endowed by the tool with a velocity is initially described as a relativistic system; one comes to the classical mechanics considering the hyperbolic parameter small. Second, the tool accelerates the spacecraft always in the direction of the frame vector appointed to indicate “yaw”; if this vector rotates, changing the yaw, the acceleration arrow changes with it; i.e., the spacecraft is accelerated along a curve line. These properties can be useful in real motion control.

On the 2D fractal level, the spacecraft’s more complex 3D motion comprising reorientation and acceleration is accompanied by respective rotation and deformation of the mentioned above fractal pyramid. Here, this subgeometric image of the math instrument necessarily enriches a simpler model of the joystick, and moreover, to make the picture symmetric, we show positive and negative directions of the pyramid (see Figure 2).

Figure 2.

Case (a): The spacecraft performs a 3D rotation, the pyramid is tilted by respective halfangle. Rotations and displacements of a spacecraft (Pioneer-10) accompanied by respective 2D rotations and deformations of the fractal pyramid. Case (b): The reoriented spacecraft rectilinearly moves with some velocity, and the tilted pyramid is distorted: Two its edges become shorter, and the other two edges become longer. Case (c): The spacecraft (?frees-framed?) is reoriented by another angle, and the distorted pyramid as tilted by respective halfangle. Case (d): The spacecraft moves along a curve trajectory with changing velocity (accelerated), and the pyramid is subject to permanent respective tilt and distortion.

Computations providing the spacecraft’s reorientation and acceleration are performed on the fractal level by Eq. (58) with the functions A,B generalized as


with hyperbolic conjugation (:e±η/2eη/2), similar to the complex conjugation, introduced, e.g.,


where vector lk directs axis of the single space rotation by angle Φ. Then (as in Section 5), only one equation is to be solved, e.g., that determining the dyad vector


and rest of the dyad elements is found by primitive math actions


Eqs. (82), (37) immediately give expressions of all spacecraft’s frame vectors, thus solving the reorientation and acceleration problem in explicit form.

One straightforwardly finds that use of the fractal technique (iii) essentially simplifies computation procedures. In paper [13], we compare math difficulty of the discussed three techniques in solution of the simple problem of the spacecraft’s one-plane space rotation and acceleration. It is demonstrated there that the techniques (i) and (ii) demand solution of at least seven equations, among them are matrix equations, while the fractal technique (iii) suggests solution of only four relatively simple algebraic equations.


7. Technological scheme and concluding remarks

A sketch of technological scheme aimed to realize mixed rotation-acceleration maneuver of a spacecraft can be suggested as the following consequence of actions fit for any mentioned above approach.

  • The initial and final parameters of reorientation and acceleration are assigned and memorized.

  • Parameters as functions of time must be determined and input.

  • Time intervals are divided into standard steps (quantized), the standard input.

  • Process of computation of quantum steps starts resulting in obtaining of a series of related parameter values describing the orientation and velocity of the spacecraft’s frame.

  • The data of each step are transmitted to the systems changing the spacecraft orientation and velocity until the assigned values are achieved.

And we emphasize two most important results of this study.

First, we succeeded to show that an extrarotation by an imaginary angle entails endowing a spacecraft with a (relativistic) velocity, hence in addition to reorientation, to accelerate it. This math observation seems to be a novel one since no similar information is met in related literature.

Second, we show that the most mathematically economical way to compute operational parameters needed for realization of the maneuver is to utilize the “fractal pyramid” technique (definitely a new tool) comprising minimal number of math actions, where major of them are simple algorithms, other approaches having no such advantages.


  1. 1. Halijak Ch.A. Quaternions Applied to Missle Systems, US Army Missle Research and Development Command, AD A 052232, Redstone Arsenal, Alabama 35809, Apr, 18, 1978 (Approved for public release; distribution unlimited)
  2. 2. Branets VN, Shmyglevskiy IP. Application of Quaternions to Rigid Body Rotation Problems, (translated from the Russian in 1974). Washington D.C.: Scientific Translation Service, National Aeronautics and Space Administrationp. 352, (1973. Report No.NASA-TT-F-15414)
  3. 3. Hamilton WR The Mathematical Papers of William Rowan Hamilton. Vol. 3. Cambridge: Cambridge University Press; 1967
  4. 4. Yefremov AP. Quaternions and biquaternions: Algebra, geometry and physical theories. Hypercomplex Numbers in Geometry and Physics. 2004;1:104-119, arXiv: math-ph/0501055
  5. 5. Yefremov AP. Bi-quaternion square roots, rotational relativity, and dual space-time intervals. Gravitation & Cosmology. 2007;133(51):178-184
  6. 6. Yefremov AP. Quaternion model of relativity: Solutions for non-inertial motions and new effects. Advanced Science Letters. 2008;1:179-186
  7. 7. Wheeler JA. Pregeometry: Motivations and prospects. In: Marlov AR, editor. Quantum Theory and Gravitation. New York: Academic Press; 1080. pp. 1-11
  8. 8. Yefremov AP. Splitting of 3D quaternion dimensions into 2D-cells and a “world screen technology”. Advanced Science Letters. 2012;5(1):288-293
  9. 9. Yefremov AP. Fundamental properties of quaternion Spinors. Gravitation and Cosmology. 2010;16(2):137-139
  10. 10. Yefremov AP. The conic-gearing image of a complex number and a spinor-born surface geometry. Gravitation and Cosmology. 2011;17(1)
  11. 11. Lancaster P, Tismenetsky M. The Theory of Matrices with Applications. 2nd ed. San Diego, London: Academic Press; 1985. p. 154
  12. 12. Yefremov AP. Fractal Surface as the Simplest Tool to Control Orientation of a Spacecraft. Acta Astronautica. 2016;129:174
  13. 13. Yefremov AP. New fractal math tool providing simultaneous reorientation and acceleration of spacecraft. Acta Astronautica. 2017;139:481-485


  • Standard interval of special relativity is regarded for simplicity; similarly, interval of general relativity can be considered in tangent space ds2=θ02−θkθk with θα=gαλdyλ being basic one-form and Greek indices in brackets enumerating tangent space tetrad, and those without brackets are related to curved manifold holonomic coordinatesηαβ=diag1−1−1−1
  • ds2=θ02−θkθkθα=gαλdyλ: four-dimensional indices are raised and lowered by Minkowski metric ηαβ=diag1−1−1−1.

Written By

Alexander P. Yefremov

Submitted: August 1st, 2017 Reviewed: October 18th, 2017 Published: December 20th, 2017