Tensors Unravelled


C. Pozrikidis

Chester & Bennington

420 pages including covers

Download the first version tun_26.01.pdf

Download the accompaning Matlab programs as a tar file: TUNLIB_26.01.tar

Please forgive typographical, conceptual, and other errors. The book should still be readable, mostly accurate, and should fulfil its intended role. Further revisions will correct these errors.

Why now?

I am in poor health and uncertain as to when I will kick the bucket. Remember: It's not the size of the dog in the fight, it's the size of fight in the dog.

What is this?

A vector is a physical entity endowed with magnitude and orientation; examples are the position, the velocity, and the acceleration. A vector is typically described by its components in a chosen frame of reference, Cartesian or non-Cartesian, rectilinear or curvilinear.

A tensor is also a physical entity described by a higher number of components in a chosen frame of reference. In computational practice, the components of a tensor are typically stored in a two- or higher-dimensional array.

Vectors and tensors are distinguished by our ability to deduce their components in a certain frame of reference defined by a base from those in any other frame of reference defined by another base by simple geometrical transformations. To indicate this ability, we say that the physical entity represented by a tensor is objective or frame invariant.

All physical entities should be frame invariant; if they were not, observation and computation would be subjective, that is, the results would depend on the position and orientation of an observer or measuring instrument.

A zeroth-order tensor is a scalar whose value is frame-independent; examples are the temperature of a star, the angle between two vectors, and the distance between two cities. A first-order tensor is a vector that has the same magnitude and points in the same direction independent of the location of an observer.

The stress tensor is a second-order tensor encapsulating the tractions exerted on three small mutually perpendicular faces in a solid or fluid. Higher-order tensors and their components in an arbitrary frame of reference can be defined. Examples are the alternating three-index tensor and the Riemann--Christoffel four-index curvature tensor.

My goal in this book is to present a concise and accessible introduction to vectors and tensors in Cartesian or non-Cartesian, rectilinear or curvilinear coordinates in a way that couples theory and numerical computation.

The notion of uniadic, dyadic, and multiadic bases is emphasized, differential operations on vector and tensor fields inside volumes and over surfaces are derived in terms of the Christoffel symbols and the curvature tensor, and applications in fluid mechanics, membrane theory, and theory of shells are discussed.

Original derivations and novel approaches are presented, including the construction of covariant coordinate fields and the derivation of Green's function of the convection--diffusion equation.

In Chapter 1, the concept of vectors endowed with magnitude and orientation is introduced and the description of a vector in terms of its components in a specified base is discussed. Transformation rules for vector components naturally leads us to the notion of vectors as first-order tensors as opposed to mere one-dimensional numerical arrays. Dyadic bases and the concept of tensors are introduced in a similar way, the description of tensors in terms of their components in a specified base is discussed, and transformation rules are established.

In Chapter 2, vectors and tensor representations in dual biorthogonal bases are discussed, the apparatus of covariant and contravariant bases and associated components is explained, and relevant transformation rules for vectors and tensors are established. The discussion serves as a natural introduction to the subject ot curvilinear coordinates where biorthonormal bases are constructed with reference to contravariant and covariant coordinates defined by families of curved lines in space. This natural introduction serves to emphasize that a tensor is a tensor is a tensor: the concept of covariant and contravariant tensors is not appropriate.

In Chapter 3, basic notions and fundamental concepts underlying the structure, construction, and properties of curvilinear coordinates in two dimensions are discussed. In particular, the dual bases discussed in Chapter 2 are reintroduced with reference to contravariant and covariant coordinates and associated base vectors. Following this introduction, finite-difference methods for solving the Laplace and Poisson equations on structured grids are developed and implemented to demonstrate the practical usefulness of the theoretical apparatus.

In Chapter 4, a comprehensive discussion of tensors in non-Cartesian coordinates is presented from the viewpoint of applied mathematics, physics, and engineering. The Christoffel symbols are defined in terms of derivatives of covariant base vectors with respect to contravariant coordinates, and the notion of covariant derivatives of vector and tensor components is discussed. A covariant derivative is a derivative of a vector or tensor component with respect to a contravariant coordinate.

In Chapter 5, vector and tensor calculus on non-Cartesian coordinates discussed, and expressions for the divergence, the curl, the gradient, the laplacian, and other differential operations are derived using an expedient method that circumvents a great deal of manipulations. Having introduced the necessary framework, applications in mathematical physics are discussed. While stating the contravariant or covariant components of the governing equations in arbitrary curvilinear coordinates is straightforward, subtleties arise in the case of moving coordinates. The notion of convected coordinates is introduced and expressions for Green's functions of the convection--diffusion equation are derived.

In Chapter 6, the apparatus of curvilinear coordinates is specialized to surfaces embedded in space with the introduction of the curvature tensor. Surface calculus is discussed and the Gauss surface divergence theorem is established with applications to force and bending moment equilibria of membranes and thin shells.

A suite of Matlab codes encapsulated in a library named TUNLIB accompany the text. The codes confirm theoretical derivations presented in the book and encode numerical methods for computing solutions of selected partial differential equations.

This book is suitable for self study and as a text in an upper-level undergraduate or graduate level core or elective course. The theoretical discussion and computational developments assume an upper-level undergraduate or entry-level graduate level knowledge of applied mathematics on readily accessible topics.

Random acts of kindness

If you enjoyed reading this book, please consider performing a random act of kindness, saying a prayer for the author, saying another prayer for the pretender, or donating what you can afford to the St. Jude Children's Research Hospital in Memphis, Tennessee. Please be kind to one another, you never know what kind of trauma lies behind the smile. Fly high in Heaven Robin and Chester.