Open access peer-reviewed chapter

Some Applications of Clifford Algebra in Geometry

Written By

Ying-Qiu Gu

Submitted: November 1st, 2019 Reviewed: July 21st, 2020 Published: September 3rd, 2020

DOI: 10.5772/intechopen.93444

Chapter metrics overview

807 Chapter Downloads

View Full Metrics


In this chapter, we provide some enlightening examples of the application of Clifford algebra in geometry, which show the concise representation, simple calculation, and profound insight of this algebra. The definition of Clifford algebra implies geometric concepts such as vector, length, angle, area, and volume and unifies the calculus of scalar, spinor, vector, and tensor, so that it is able to naturally describe all variables and calculus in geometry and physics. Clifford algebra unifies and generalizes real number, complex, quaternion, and vector algebra and converts complicated relations and operations into intuitive matrix algebra independent of coordinate systems. By localizing the basis or frame of space-time and introducing differential and connection operators, Clifford algebra also contains Riemann geometry. Clifford algebra provides a unified, standard, elegant, and open language and tools for numerous complicated mathematical and physical theories. Clifford algebra calculus is an arithmetic-like operation that can be well understood by everyone. This feature is very useful for teaching purposes, and popularizing Clifford algebra in high schools and universities will greatly improve the efficiency of students to learn fundamental knowledge of mathematics and physics. So, Clifford algebra can be expected to complete a new big synthesis of scientific knowledge.


  • Clifford algebra
  • geometric algebra
  • gamma matrix
  • multi-inner product
  • connection operator
  • Keller connection
  • spin group
  • cross ratio
  • conformal geometric algebra

1. A brief historical review

It is well known that a rotational transformation in the complex plane is equivalent to multiplying the complex number by a factor eθi. How to generalize this simple and elegant operation to three-dimensional space is a difficult problem for many outstanding mathematicians in the early nineteenth century. William Rowan Hamilton (1805–1865) spent much of his later years studying the issue and eventually invented quaternion [1]. This generalization requires four elements 1ijk, and the spatial basis should satisfy the multiplying rules i2=j2=k2=1and jk=kj=i, ki=ik=j, and ij=ji=k. Although a quaternion is still a vector, it constitutes an associative algebra according to the above rules. However, the commutativity of multiplication is violated. Quaternion can solve the rotational transformation in three-dimensional space very well and simplify the representation of Maxwell equation system of electromagnetic field.

When Hamilton introduced his quaternion algebra, German high school teacher Hermann Gunther Grassmann (1809–1877) was constructing his exterior algebra [2]. He defined the exterior product or outer product abof two vectors aand b, which satisfies anti-commutative law ab=baand associativity abc=abc. The exterior product is a generalization of cross product in three-dimensional Euclidian space. Its geometrical meaning is the oriented volume of a parallel polyhedron. Exterior product is now the basic tool of modern differential geometry, but Grassmann’s work was largely neglected in his lifetime.

British mathematician William Kingdon Clifford (1845–1879) was one of the few mathematicians who read and understood Grassmann’s work. In 1878, he combined the algebraic rules of Hamilton and Grassmann to define a new algebraic system, which he himself called geometric algebra [3]. In this algebra, both the inner and exterior products of vectors can be uniquely represented by a linear combination of geometric product. In addition, geometric algebra is always isomorphic to some special matrix algebra.

Clifford algebra combines all the advantages of quaternion with the advantages of vector algebra and uniformly and succinctly describes the contents of geometry and physics. However, the vector calculus introduced by Gibbs had also successfully described the mathematical physics problem in three-dimensional space [4]. Clifford died prematurely at the age of 34, so that the theory of geometric algebra was not deeply researched and fully developed, and people still could not see the superiority of this algebra at that time. Thus, the important insights of Grassmann and Clifford were lost in the late nineteenth century papers. Mathematicians abstracted Clifford algebra from its geometric origins, and, for the most part of a century, it languished as a minor subdiscipline of mathematics and became one more algebra among so many others.

With the establishment of relativity, especially the introduction of Pauli and Dirac’s matrix algebra for spin and the successful application in quantum theory [5], it was felt that there is an urgent need for a mathematical system to deal with problems in high-dimensional space-time. In the 1920s, Clifford algebra re-entered the field of vision and was paid attention and researched by some of the famous mathematicians and physicists such as R. Lipschitz, T. Vahlen, E. Cartan, E. Witt, C. Chevalley, and M. Riesz [6, 7, 8]. When only formal algebra is involved, we usually use the term “Clifford algebra,” but more often use the “geometric algebra” named by Clifford himself if applied to geometric problems.

The first person who realized that Clifford algebra is a unified language in geometry and physics should be David Hestenes. By the 1960s, Hestenes began to restore the geometric meaning behind Pauli and Dirac algebra. Although his initial motivation was to gain insight into the nature of quantum mechanics, he quickly realized that Clifford algebra was a unified language and tool for mathematics, physics, and engineering. He published “space-time algebra” in 1966 and has been working on the promotion of Clifford algebra in teaching and research [9, 10, 11, 12]. Because representation and algorithm in geometric algebra are seemingly as ordinary as arithmetic, his work has been neglected by the scientific community for more than 20 years. Only with the joint impetus of computer-aided design, computer vision and robotics, protein folding, neural networks, modern differential geometry, mathematical physics [13, 14, 15, 16, 17], and especially the Journal “Advances in Applied Clifford algebras” founded by Professor Jaime Keller, geometric algebra began to move towards popularity and prosperity.

As a unified and universal language of natural science, Clifford algebra is developed by many mathematicians, physicists, and engineers according to their different requirements and knowledge background. Such situation leads to “There are a thousand Hamlets in a thousand people’s eyes.” In this chapter, by introducing typical application of Clifford algebra in geometry, we show some special feature and elegance of the algebra.


2. Application of Clifford algebra in differential geometry

In Euclidean space, we have several important concepts such as vector, length, angle, area, volume, and tensor. The study of relationship between these concepts constitutes the whole content of Euclidean geometry. The mathematical tools previously used to discuss these contents are vector algebra and geometrical method, which are complex and require much fundamental knowledge. Clifford algebra exactly and faithfully describes the intrinsic properties of vector space by introducing concepts such as inner, exterior, and geometric products of vectors and thus becomes a unified language and standard tool for dealing with geometric and physical problems. Clifford algebra has the characteristics of simple concept, standard operation, completeness in conclusion, and easy understanding.

Definition 1For Minkowski spaceMnover number fieldF, if the multiplication rule of vectors satisfies


the algebra is calledGrassmann algebraandxyexterior product.

The Grassmann is also called exterior algebra. The geometrical meaning of xyis oriented area of a parallelogram constructed by xand y, and the geometrical meaning of xyzis the oriented volume of the parallelohedron constructed by the vectors (see Figure 1). We call xytwo-vector, xyzthree-vector, and so on. For k-vector xΛkand l-vector yΛl, we have

Figure 1.

Geometric meaning of exterior products of vectors.


By the definition, we can easily check:

Theorem 1For exterior algebra defined inV=Mn, we have


The dimension of the algebra is


Under the orthonormal basise1e2en, the exterior algebra takes the following form:


in whichwjklF, ejkl=ejekel, andejkl=1.

The exterior product of vectors contains alternating combinations of basis, for example:


Definition 2For any vectorsx,y,zMn, Clifford productof vectors is denoted by


Clifford product is also calledgeometric product.

Similarly, we can define Clifford algebra for many vectors as xyz. In (6), xy=ηabxaybis the scalar productor inner productin Mn. By xy=yx, we find Clifford product is not commutative. By (6), we have


Definition 3For Minkowski spaceMp,qwith metricηab=diagIpIq, if the Clifford product of vectors satisfies


then the algebra


is called asClifford algebraorgeometric algebra, which is denoted asCp,q.

There are several definitions for Clifford algebra [18, 19]. The above definition is the original definition of Clifford. Clifford algebra has also 2ndimensions. Comparing (11) with (4), we find the two algebras are isomorphic in sense of linear algebra, but their definitions of multiplication rules are different. The Grassmann products have clear geometrical meaning, but the Clifford product is isomorphic to matrix algebra and the multiplication of physical variables is Clifford product. Therefore, representing geometrical and physical variables in the form of (4) will bring great convenience [20, 21]. In this case, the relations among three products such as (6)(9) are important.

In physics, we often use curvilinear coordinate system or consider problems in curved space-time. In this case, we must discuss problems in ndimensional pseudo Riemann manifold. At each point xin the manifold, the tangent space TMxis a ndimensional Minkowski space-time. The Clifford algebra can be also defined on the tangent space and then smoothly generalized on the whole manifold as follows.

Definition 4Inn=p+qdimensional manifoldTMp,qoverR, theelementis defined by


whereγais the local orthogonal frame andγathe coframe. Thedistanceds=dxandoriented volumesdVkis defined by


in whichηab=diagIpIqisMinkowski metricandgμνisRiemann metric.


isGrassmann basis. The following Clifford-Grassmann number with basis


defines real universalClifford algebraCp,qon the manifold.

The definitions and treatments in this chapter make the corresponding subtle and fallible concepts in differential geometry much simpler. For example, in spherical coordinate system of R3, we have element dxand the area element dsin sphere dr=0as


We have the total area of the sphere


The above definition involves a number of concepts, some more explanations are given in the following:

  1. 1. The geometrical meanings of elements dx,dy,dxdyare shown in Figure 2. The relation between metric and vector basis is given by:

Figure 2.

Geometric meaning of vectorsdx,dyanddxdy.


which is the most important relation in Clifford algebra. Since Clifford algebra is isomorphic to some matrix algebra, by (17)γais equivalent to some special matrices [20]. In practical calculation, we need not distinguish the vector basis from its representation matrix. The relation between the local frame coefficientfaμfμaand metric is given by:


  1. 2. Assume γaa=12nto be the basis of the space-time, then their exterior product is defined by [22]:


In which σa1a2akb1b2bkis permutation function, if b1b2bkis the even permutation of a1a2ak, it equals 1. Otherwise, it equals −1. The above formula is a summation for all permutations, that is, it is antisymmetrization with respect to all indices. The geometric meaning of the exterior product is oriented volume of a higher dimensional parallel polyhedron. Exterior algebra is also called Grassmann algebra, which is associative.

  1. 3. By (12) and (13) we find that, using Clifford algebra to deal with the problems on a manifold or in the tangent space, the method is the same. Unless especially mentioned, we always use the Greek alphabet to stand for the index in curved space-time, and the Latin alphabet for the index in tangent space. We use Einstein summation convention.

  2. 4. In Eq. (15), each grade-kterm is a tensor. For example, c0IΛ0is a scalar, cμγμΛ1is a true vector, and cμνγμνΛ2is an antisymmetric tensor of rank-2, which is also called a bivector, and so on. In practical calculation, coefficient and basis should be written together, because they are one entity, such as (12) and (15). In this form, the variables become coordinate free. The coefficient is the value of tensor, which is just a number table, but the geometric meaning and transformation law of the tensor is carried by basis.

The real difficulty in learning modern mathematics is that in order to get a little result, we need a long list of subtle concepts. Mathematicians are used to defining concepts over concepts, but if the chain of concepts breaks down, the subsequent contents will not be understandable. Except for the professionals, the common readers impossibly have so much time to check and understand all concepts carefully. Fortunately, the Clifford algebra can avoid this problem, because Clifford algebra depends only on a few simple concepts, such as numbers, vectors, derivatives, and so on. The only somewhat new concept is the Clifford product of the vector bases, which is isomorphic to some special matrix algebra; and the rules of Clifford algebra are also standardized and suitable for brainless operations, which can be well mastered by high school students.

Definition 5For vectorx=γμxμΛ1and multivectorm=γθ1θ2θkmθ1θ2θkΛk, their inner product is defined as


in which


Theorem 2For basis of Clifford algebra, we have the following relations


Proof.Clearly γμγθ1θ2θkΛk1Λk+1, so we have


Permuting the indices θ1and θ2, we find a2=a1. Let μ=θ1, we get a1=1. Check the monomial in exterior product, we get A=1. Thus, we prove (21). In like manner, we prove (22). For orthonormal basis γa, by (22) we have:


Again by γanγan=1(not summation), we prove (23). Other equations can be proved by antisymmetrization of indices. The proof is finished.

Likewise, we can define multi-inner productAkBbetween multivectors as follows:


We use AkBrather AkB, because the symbol “” is too small to express exponential power. Then for the case γμ1μ2μfγθ1θ2θk, we have similar results. For example, we have


In C1,3, denote the Pauli matrices by


We use k,f,jstanding for spatial indices. Define Dirac γmatrix by:


γaforms the grade-1 basis of Clifford algebra C1,3. In equivalent sense, the representation (33) is unique. By γ-matrix (33), we have the complete bases of C1,3as follows [21]:


Based on the above preliminaries, we can display some enlightening examples of application, which show how geometric algebra works efficiently. For a skew-symmetrical torsion TμνωgμβTνωβin M1,3, by Clifford calculus, we have:


and then


where g=detgμν. So we get:


So, the skew-symmetrical torsion is equivalent to a pseudo vector in M1,3. This example shows the advantages to combine variable with basis together.

The following example discusses the absolute differential of tensors. The definition of vector, tensor, and spinor in differential geometry involving a number of refined concepts such as vector bundle and dual bundle, which are too complicated for readers in other specialty. Here, we inherit the traditional definitions based on the bases γaand γμ. In physics, basis of tensors is defined by direct products of grade-1 bases γμ. For metric, we have [23]:


For simplicity, we denote tensor basis by:


In general, a tensor of rank nis given by:


The geometrical information of the tensor such as transformation law and differential connection are all recorded by basis γμ, and all representations of rank rstensor denote the same one practical entity Tx. Tμνis just a quantity table similar to cμνin (15), but the physical and geometrical meanings of the tensor Tare represented by basis γμ. Clifford algebra is a special kind of tensor with exterior product. Its algebraic calculus exactly reflects the intrinsic property of space-time and makes physical calculation simple and clear.

For the absolute differential of vector field A=γμAμ, we have


We call dαconnection operator[23]. According to its geometrical meanings, connection operator should satisfy the following conditions:

  1. It is a real linear transformation of basis γμ,

  2. It satisfies metric consistent condition dg=0.

Thus, the differential connection can be generally expressed as:


For metric g=gμνγμγν, by metric consistent condition we have:


By (43), we have:


Since dxαδXais an arbitrary vector in tangent space, (44) is equivalent to:


(45) is a linear nonhomogeneous algebraic equation of ΠαβμTαβμ.

Solving (45), we get the symmetrical particular solution “Christoffel symbols” as follows;


in which Γμναis called Levi-Civita connection determined by metric, πμνα=πνμαis a symmetrical post-metric part of connection. In this chapter, the “post-metric connection” means the parts of connection cannot be determined by metric, i.e., the components πμναand Tμναdifferent from Levi-Civita connection Γμνα. Denote


where Kμναis called contortion with total n3components [24]. Substituting (46) and (47) into metric compatible condition (45), we get 12n+1n2constraints for Kμνα,


By (48), Kμναhas only 12n1n2independent components. Noticing torsion Tμναhas just 12n1n2independent components, so Kμναor πμναcan be represented by Tμνα.

Theorem 3For post-metric connections we have the following relations


and consistent condition


T˜=T˜μνωγμνωΛ3is an arbitrary skew-symmetrical tensor.

ProofIf we represent πμναby Tμνα, by (48) and symmetry we have solution as (49). By (49), we get consistent condition (52). By (49) and (47), we get (50).

If we represent Tμναby πμνα, we generally have linear relation


in which kis a constant to be determined, T˜μναis particular solution as πμνα0. T˜μναsatisfies


So this part of torsion is a skew-symmetrical tensor T˜=T˜μνωγμνωΛ3, which has Cn3=16n2n1nindependent components. Substituting (53) into (48), we get


Calculating the summation of (55) for circulation of μνα, we also get consistent condition (52). Substituting (52) into (55) we get k=13. Again by (53), we get solution (51). It is easy to check, (49) and (51) are the inverse representation under condition (52). The proof is finished.

Substituting (42) into


we get


To understand the meaning of πμναand Tμνα, we examine the influence on geodesic.


The term Tμναvμvν=0due to Tμνα=Tνμα. So the symmetrical part πμναinfluences the geodesic, but the antisymmetrical part Tμναonly influences spin of a particle. This means πμνα0violates Einstein’s equivalent principle. In what follows, we take πμνα=0.

By (42) and (57), we get:

Theorem 4In the caseπμνα0, the absolute differential of vectorAis given by


in whichαdenotes the absolute derivatives of vector defined as follows:


whereA;αμand Aμ;αare usual covariant derivatives of vector without torsion. TorsionTμνωΛ3is an antisymmetrical tensor ofCn3independent components.

Similarly, we can calculate the absolute differential for any tensor. The example also shows the advantages to combine variable with basis.

Now we take spinor connection as example to show the power of Clifford algebra. For Dirac equation in curved space-time without torsion, we have [23, 25, 26]:


Γμis called spinor connection. Representing γμΓμΛ1Λ3in the form of (15), we get:


where αμis current operator, p̂μis momentum operator, and sμspin operator. They are defined respectively by:


where σμ=faμσaand σ˜μ=faμσ˜aare the Pauli matrices in curved space-time. ϒμΛ1is called Keller connection, and ΩμΛ3is called Gu-Nester potential, which is a pseudo vector [23, 26, 27]. They are calculated by:


where Sabμνfa{μfbν}signabfor LUdecomposition of metric. In the Hamiltonian of a spinor, we get a spin-gravity coupling potential sμΩμ. If the metric of the space-time can be orthogonalized, we have Ωμ0.

If the gravitational field is generated by a rotating ball, the corresponding metric, like the Kerr one, cannot be diagonalized. In this case, the spin-gravity coupling term has nonzero coupling effect. In asymptotically flat space-time, we have the line element in quasi-spherical coordinate system [28]:


in which UVWis just functions of rθ. As rwe have:


where mLare mass and angular momentum of the star, respectively. For common stars and planets, we always have rmL. For example, we have m=̇3km for the sun. The nonzero tetrad coefficients of metric (66) are given by:


Substituting it into (65) we get


By (70), we find that the intensity of Ωαis proportional to the angular momentum of the star, and its force line is given by:


(71) shows that the force lines of Ωαis just the magnetic lines of a magnetic dipole. According to the above results, we know that the spin-gravity coupling potential of charged particles will certainly induce a macroscopic dipolar magnetic field for a star, and it should be approximately in accordance with the Schuster-Wilson-Blackett relation [29, 30, 31].


3. Representation of Clifford algebra

The matrix representation of Clifford algebra is an old problem with a long history. As early as in 1908, Cartan got the following periodicity of 8 [18, 19].

Theorem 5For real universal Clifford algebraCp,q, we have the following isomorphism


For C0,2, we have C=tI+xγ1+yγ2+zγ12with


By (73), we find Cis equivalent to a quaternion, that is, we have isomorphic relation C0,2H.

Similarly, for C2,0, we have C=tI+xγ1+yγ2+zγ12with


By (74), the basis is equivalent to


Thus, (75) means C2,0Mat2R.

In geometry and physics, the matrix representation of generators of Clifford algebra is more important and fundamental than the representation of whole algebra. Define γμby


which forms the generator or grade-1 basis of Clifford algebra C1,3. To denote γμby Γμmis for the convenience of representation of high dimensional Clifford algebra. For any matrices Cμsatisfying C1,3Clifford algebra, we have [20, 32]:

Theorem 6Assuming the matricesCμsatisfy anti-commutative relation ofC1,3


then there is a natural numbermand an invertible matrixK, such thatK1CμK=Γμm.

This means in equivalent sense, we have unique representation (76) for generator of C1,3. In [20], we derived complex representation of generators of Cp,qbased on Theorem 6 and real representations according to the complex representations as follows.

Theorem 7Let


Otherγμ,μ3are given by(76). Then the generators of Clifford algebraC1,4are equivalent toγμ,μ=0,1,2,3,5.

In order to express the general representation of generators, we introduce some simple notations. Imstands for m×munit matrix. For any matrix A=Aab, denote block matrix


in which the direct product of matrix is Kronecker product. Obviously, we have I2I2=I4, I2I2I2=I8, and so on. In what follows, we use Γμmdefined in (76). For μ0,1,2,3, Γμmis 4m×4mmatrix, which constitute the generator of C1,3. Similar to the above proofs, we can check the following theorem by method of induction.

Theorem 8

  1. 1. In equivalent sense, forC4m, the matrix representation of generators is uniquely given by


in whichn=2m1N, where N is any given positive integer. All matrices are2m+1N×2m+1Ntype.

  1. 2. ForC4m+1, besides(80)we have another real generator


If and only ifk=l, this representation can be uniquely expanded as generators ofC4m+4.

  1. 3. For anyCp,q, pqp+q4mmodp+q41, the combination ofp+qlinear independent generatorsγμiγνtaking from(80)constitutes the complete set of generators. In the casepqp+q4mmodp+q4=1, besides the combination ofγμiγν, we have another normal representation of generator taking the form(81)withkl.

  2. 4. ForCm,m<4, we have another2×2Pauli matrix representation for its generatorsσ1σ2σ3.

Then, we get all complex matrix representations for generators of real Cp,qexplicitly.

The real representation of Cp,qcan be easily constructed from the above complex representation. In order to get the real representation, we should classify the generators derived above. Let Gcnstand for any one set of all complex generators of Cngiven in Theorem 8, and set the coefficients before all σμand σ˜μas 1or i. Denote Gc+stands for the set of complex generators of Cn,0and Gcfor the set of complex generators of C0,n. Then, we have:


By the construction of generators, we have only two kinds of γμmatrices. One is the matrix with real nonzero elements and the other is that with imaginary nonzero elements. This is because all nonzero elements of σ2are imaginary but all other σμμ2are real. Again assume

Gc+=GrGi,Gr=γrμγrμis real,Gi=γiμγiμis imaginary.E83

Denote J2=iσ2, we have J22=I2. J2becomes the real matrix representation for imaginary unit i. Using the direct products of complex generators with I2J2, we can easily construct the real representation of all generators for Cp,qfrom Gc+as follows.

Theorem 9

  1. ForCn,0, we have real matrix representation of generators as


  2. ForC0,n, we have real matrix representation of generators as


  3. ForCp,q, we have real matrix representation of generators as


Obviously we haveCnpCnq=Cnp2choices for the real generators ofCp,qfrom each complex representation.

Proof.By calculating rules of block matrix, it is easy to check the following relations:


By these relations, Theorem 9 becomes a direct result of Theorem 8.

For example, we have 4×4real matrix representation for generators of C0,3as follows:


It is easy to check


4. Transformation of Clifford algebra

Assume Vis the base vector space of Cp,q, then Clifford algebra has the following global properties [22, 33, 34]:


Cp,qis a 2-graded superalgebra, and Cp,q+is a subalgebra of Cp,q. We have:


Definition 6Theconjugationof element inCp,qis defined by


Themain involutionof element is defined by


Thenormandinverseof element are defined by


By the definition, it is easy to check


Definition 7ThePin groupandSpin groupofCp,qare defined by


The transformation xαgxgis called sandwich operator. Pin or Spin group consists of two connected components with Ng=1or Ng=1,


For gPinp,q,xV, the sandwich operator is a linear transformation for vector in V,


In all transformations of vector, the reflection and rotation transformations are important in geometry. Here, we discuss the transformation in detail. Let mΛ1be a unit vector in V, then the reflection transformationof vector XΛ1with respect to n1dimensional mirror perpendicular to mis defined by [35]:


Let m=γama,X=γaXa, substituting it into (106) and using (21), we have:


Eq. (107) clearly shows the geometrical meaning of reflection. By (106), we learn reflection transformation belongs to Pinp,qgroup (Figure 3).

Figure 3.

Reflection transformationX=XX.

The rotation transformationRSpinp,q,


The group elements of elementary transformation in Λ2are given by [22, 36]:


The total transformation can be expressed as multiplication of elementary transformations as follows:


(111) has 12n1ngenerating elements like SOn. In (111), we have commutative relation as follows:


in which


If abcd, the right hand terms vanish, and then two elementary transformations commute with each other.

Rforms a Lie Group of 12n1nparaments. In the case Cn,0or C0,n, Ris compact group isomorphic to SOn. Otherwise, Ris noncompact one similar to Lorentz transformation. The infinitesimal generators of the corresponding Lie group is γab, and the Lie algebra is given by:


Thus, Λ2Mp,qis just the Lie algebra of proper Lorentz transformation of the space-time Mp,q.


5. Application in classical geometry

Suppose the basic space of projective geometry is n-dimensional Euclidean space π(see Figure 4), and the basis is γaa=12n. The coordinate of point xis given by x=γaxa. The projective polar is P, and its height from the basic space πis h. The total projective space is n+1dimensional, and an auxiliary basis γn+1=γpis introduced. The coordinate of the polar Pis p=γμpμ. In this section, we use Greek characters for n+1indices. Assume the unit directional vector of the projective ray is t=γμtμ, the unit normal vector of the image space πis n=γμnμ, coordinate in πis y=γμyμ, and the intercept of πwith the n+1coordinate axis is a. Then, we have:

Figure 4.

Diagram of parameter setting for projective geometry.


The equation of projective ray is given by:


where λis parameter coordinate of the line. In the basic space π, we have sn+1=0and λ=h/tp, so the coordinate of the line in πreads


Let s=yand substitute (117) into (116) we get image equation as follows:


In the above equation tn0, which means tcannot be perpendicular to n; otherwise, the projection cannot be realized. Eliminating coordinate tin (118) and (119), we find the projective transformation yxis nonlinear. In (119), only the parameters anare related to image space π; so, all geometric variables independent of two parameters anare projective invariants. In what follows we prove the fundamental theorems of projective geometry by Clifford algebra.

Theorem 10For 4 different pointsy1y2y3y4on a straight lineL, the following cross ratio is a projective invariant


ProofSubstituting (119) into (120) we get


By (19) and (20), we get


where mis the unit normal vector of the plane spanned by tatb, which is independent of the image space π. Substituting it into (121), we get


(123) is independent of an; so, it is a projective invariant. Likewise, 1324and 1423are also projective invariants. The proof is finished.

Now we examine affine transformation. In this case, the polar Pat infinity and the directional vector tof rays becomes constant vector. The equation of rays is given by y=x+λt. Substituting it into (116), we get the coordinate transformation from basic space πto image space π,


Since tand nare constant vectors for all rays, the affine transformation yxis linear. A variable independent of anis an affine invariant.

Theorem 11Assumex1x2x3are 3 points on a straight lineLin basic spaceπ, andy1y2y3are respectively their projective images on lineLinπ. Then the simple ratio


is an affine invariant.

ProofBy equation of transformation (124) we get


In (126), only the parameters anare related to image space π. Substituting (126) into (125), we have:


Denote the unit directional vector of line Lby k, then we have


Substituting them into (127) we get:


This proves the simple ratio 1213is an affine invariant. Likewise, we can prove 1223and 1323are also affine invariants. The proof is finished.

The treatment of image information by computer requires concise and general algebraic representation for geometric modeling as well as fast and robust algebraic algorithm for geometric calculation. Conformal geometry algebra was introduced in this context. By establishing unified covariant algebra representation of classical geometry, the efficient calculation of invariant algebra is realized [13, 14, 15]. It provides a unified and concise homogeneous algebraic framework for classical geometry and algorithms, which can thus be used for complicated symbolic geometric calculations. This technology is currently widely applied in high-tech fields such as computer graphics, vision calculation, geometric design, and robots.

The algebraic representation of a geometric object is homogeneous, which means that any two algebraic expressions representing this object differ by only one nonzero factor and any such algebraic expressions with different nonzero multiple represent the same geometric object. The embedding space provided by conformal geometric algebra for ndimensional Euclidean space is n+2dimensional Minkowski space. Since the orthonormal transformation group of the embedding space is exactly double coverage of the conformal transformation group of the Euclidean space, this model is also called the conformal model. The following is a brief introduction to the basic concepts and representation for geometric objects of conformal geometric algebra. The materials mainly come from literature [13].

In conformal geometry algebra, an additional Minkowski plane M1,1is attached to ndimensional Euclidean space Rn, M1,1has an orthonormal basis e+e, which has the following properties:


In practical application, e+eis replaced by null basis e0e


They satisfy


A unit pseudo-scalar Efor M1,1is defined by:


In conformal geometric algebra, we work with Mn+1,1=RnM1,1.

Define the horosphereof Rnby:


Nenis a homogeneous model of Rn. The powerful applications of conformal geometry come from this model. By calculation, for xRnwe have:


which is a bijective mapping xRnxNen, we have NenRn. xis referred to as the homogeneous pointof x. Clearly, 0Rne0Nenand RneNenare in homogeneous coordinate.

Now we examine how conformal geometric algebra represents geometric objects. For a line passing through points aand b, we have


Since ab=abais the moment for a line through point awith tangent ab, eabcharacterizes the line completely.

Again by using (135) and (136), we get


We recognize abcas the moment of a plane with tangent baca. Thus, eabcrepresents a plane through points {a, b, c}, or, more specifically, the triangle (2-simplex) with these points as vertices.

For a sphere with radius ρand center pRn, we have xp2=ρ2. By (135), the equation in terms of homogeneous points becomes


Using xe=1, we get:




From these properties, the form (139) and center pcan be recovered. Therefore, every sphere in Rnis completely characterized by a unique vector sMn+1,1. According to (140), slies outside the null cone. Analysis shows that every such vector determines a sphere.


6. Discussion and conclusion

The examples given above are only applications of Clifford algebra in geometry, but we have seen the power of Clifford algebra in solving geometrical problems. In fact, Clifford algebra is more widely used in physics. Why does Clifford algebra work so well? As have been seen from the above examples, the power of Clifford algebra comes from the following features:

  1. In the geometry of flat space, the basic concepts are only length, angle, area, and volume, which are already implicitly included in the definition of Clifford algebra. So, Clifford algebra summarizes these contents of classical geometry and algebraize them all. By introducing the concepts of inner, exterior, and direct products of vector, Clifford algebra summarizes the operations of scalars, vectors, and tensors and then can represent all the physical variables in classical physics, because only these variables are included in classical physics.

  2. By localizing the basis or frame of space-time, Clifford algebra is naturally suitable for the tangent space in a manifold. If the differential μand connection operator dμγνare introduced, Clifford algebra can be used for the whole manifold, so it contains Riemann geometry. Furthermore, Clifford algebra can express all contents of classical physics, including physical variables, differential equations, and algebraic operations. Clifford algebra transforms complicated theories and relations into a unified and standard calculus with no more or less contents, and all representations are neat and elegant [23, 36].

  3. If the above contents seem to be very natural, Clifford algebra still has another unusual advantage, that is, it includes the theory of spinor. So, Clifford algebra also contains quantum theory and spinor connection. These things are far beyond the human intuition and have some surprising properties.

  4. There are many reasons to make Clifford algebra become a unified and efficient language and tool for mathematics, physics, and engineering, such as Clifford algebra generalizes real number, complex number, quaternion, and vector algebra; Clifford algebra is isomorphic to matrix algebra; the derivative operator γμμcontains grad, div, curl, etc. However, the most important feature of Clifford algebra should be taking the physical variable and the basis as one entity, such as g=gμνγμγνand T=Tμνωγμνω. In this representation, the basis is an operator without ambiguity. Clifford algebra calculus is an arithmetic-like operation which can be well understood by everyone.

    “But, if geometric algebra is so good, why is it not more widely used?” As Hestenes replied in [11]: “Its time will come!” The published geometric algebra literature is more than sufficient to support instruction with geometric algebra at intermediate and advanced levels in physics, mathematics, engineering, and computer science. Though few faculty are conversant with geometric algebra now, most could easily learn what they need while teaching. At the introductory level, geometric algebra textbooks and teacher training will be necessary before geometric algebra can be widely taught in the schools. There is steady progress in this direction, but funding is needed to accelerate it. Malcolm Gladwell has discussed social conditions for a “tipping point” when the spread of an idea suddenly goes viral. Place your bets now on a Tipping Point for Geometric Algebra!



I would like to thank Dr. Min Lei for her kind invitation and help. The discussion on torsion is completed under the inspiration and guidance of Prof. James M. Nester. The content of conformal geometric algebra is added according to the suggestion of Dr. Isiah Zaplana. The chapter has been improved according to the comments of a referee.


  1. 1. Hamilton W. On quaternions, or on a new system of imaginaries in algebra. Philosophical Magazine. 1844
  2. 2. Grassmann H. Die Lineale Ausdehnungslehre, ein neuer Zweig der Mathematik [The Theory of Linear Extension, a New Branch of Mathematics]. O. Wigand, 1844, New York: Cambridge University Press; 2013
  3. 3. Clifford W. Application of Grassmann’s extensive algebra. American Journal of Mathematics. 1878;1:350-358
  4. 4. Gibbs W. The Scientific Papers of Willard Gibbs. Vol. 3. London: Longmas, Green and Company; 1906
  5. 5. Dirac P. The quantum theory of the electron. Proceedings of the Royal Society of London. 1928;A117:610-624
  6. 6. Cartan E. The Theory of Spinors. Cambridge, MA: The M.T.I. Press; 1966
  7. 7. Chevalley C. The Algebraic Theory of Spinors and Clifford Algebras. Berlin: Springer; 1996
  8. 8. Riesz M. In: Bolinder EF, Lounesto P, editors. Clifford Numbers and Spinors. Netherlands: Springer; 1993
  9. 9. Hestenes D. Space Time Algebra. New York: Gordon and Breach; 1966
  10. 10. Hestenes D. Oersted medal lecture 2002: Reforming the mathematical language of physics. American Journal of Physics. 2003;71:104-121
  11. 11. Hestenes D. The genesis of geometric algebra: A personal retrospective. Advances in Applied Clifford Algebras. 2017;27(1):351-379
  12. 12. Sobczyk G. Clifford geometric algebras in multilinear algebra and non-Euclidean geometries. In: Byrnes J, editor. Computational Noncommutative Algebra and Applications. Netherlands: Springer; 2004
  13. 13. Li HB, Hestenes D, Rockwood A. Generalized homogeneous coordinates for computational geometry. In: Sommer G, editor. Geometric Computing with Clifford Algebras. Heidelberg: Springer-Verlag; 2001. pp. 27-60
  14. 14. Doran C, Lasenby A. Geometric Algebra for Physicists. Cambridge: Cambridge University Press; 2003
  15. 15. Lasenby A, Lasenby J, Wareham R. A Covariant Approach to Geometry Using Geometric Algebra, Technical Report. Cambridge, UK: University of Cambridge Department of Engineering; 2004
  16. 16. Doran CJL, Lasenby AN, Gull SF, Somaroo S, Challinor A. Spacetime algebra and electron physics. Advances in Electronics and Electron Physics. 1996;95:272-383
  17. 17. Sommer G, editor. Applications of Geometric Algebra in Engineering. 1st ed. Springer Verlag; 1999
  18. 18. Lounesto P. Clifford Algebras and Spinors. Cambridge: Cambridge University Press; 2001
  19. 19. Shirokov DS. Clifford algebras and their applications to Lie groups and spinors. In: Mladenov IM, Yoshioka A, editors. Proceedings of the Nineteenth International Conference on Geometry, Integrability and Quantization. Sofia: Avangard Prima; 2018. pp. 11-53. arXiv:1709.06608
  20. 20. Gu YQ. A Note on the Representation of Clifford Algebra. Preprints; 2020. p. 2020020466. [Preprint]. DOI: 10.20944/preprints202002.0466.v1
  21. 21. Pavsic M. Clifford algebra, geometry and physics. NATO Science Series II. 2003;95:165-174. arXiv:gr-qc/0210060
  22. 22. Lachiéze-Rey M. Spin and Clifford algebras, an introduction. Advances in Applied Clifford Algebras. 2009;19(3–4):687-720. arXiv:1007.2481
  23. 23. Gu YQ. Space-time geometry and some applications of Clifford algebra in physics. Advances in Applied Clifford Algebras. 2018;28(4):79
  24. 24. Hehl FW, von der Heyde P, Kerlick GD, Nester JM. General relativity with spin and torsion: Foundation and prospects. Reviews of Modern Physics. 1976;48:393-416
  25. 25. Bergmann PG. Two-component Spinors in general relativity. Physics Review. 1957;107(2):624-629
  26. 26. Gu YQ. The Simplification of Spinor Connection and Classical Approximation. [Preprint] arXiv:gr-qc/0610001
  27. 27. Nester JM. Special orthonormal frames. Journal of Mathematical Physics. 1992;33:910
  28. 28. Gu YQ. The series solution to the metric of stationary vacuum with axisymmetry. Chinese Physics B. 2008;19(3):90-100
  29. 29. Blackett PMS. The magnetic field of massive rotating bodies. Nature. 1947;159:658-666
  30. 30. Dolginov A. Electromagnetic field created by rotation of celestial bodies. Journal of Modern Physics. 2016;7(16):2418-2425
  31. 31. Gu YQ. A new explanation for the origin of magnetic field of celestial bodies. Natural Science (in Chinese). 2019;7(6):464-470
  32. 32. Gu YQ. A canonical form for relativistic dynamic equation. Advances in Applied Clifford Algebras. 1997;V7(1):13-24
  33. 33. Garling DJH. Clifford Algebras: An Introduction. Cambridge: Cambridge University Press; 2011
  34. 34. Lawson HB, Michelsohn ML. Spin Geometry. Princeton: Princeton University Press; 1989
  35. 35. Gunn CG. Doing Euclidean plane geometry using projective geometric algebra. Advances in Applied Clifford Algebras. 2017;27:1203-1232
  36. 36. Gu YQ. Clifford algebra, Lorentz transformation and unified field theory. Advances in Applied Clifford Algebras. 2018;28(2):37

Written By

Ying-Qiu Gu

Submitted: November 1st, 2019 Reviewed: July 21st, 2020 Published: September 3rd, 2020