Purdue University

EAS 557

Introduction to Seismology

Robert L. Nowack

Lecture 4

 

Vector and Cartesian Tensor Analysis

 

 

References:

Schey, H.M. (1997)  Div, Grad, Curl and All That, Norton.

Mase, G.E. (1970)  Theory and Problems of Continuum Mechanics, Schaum Outlines, McGraw Hill.

Matthews, P.C. (1998)  Vector Calculus, Springer-Verlag.

 

 

            A vector in Euclidean space is a directed line segment with a given magnitude and direction. 

 

 

 

 

Certain physical quantities, such as force and velocity, which possess both magnitude and direction, may be represented in a three-dimensional space by directed line segments.

 

 

 

                                                                 a couple of equal vectors

 

 

 

 

            Below, I will give the vector notation to the left and the index notation for a Cartesian vector to the right.

 

  ;  xi  ith component of the vector

 

Vectors add according to the parallelogram rule

 

  ;    (components add)

 

 

 

 

Put them head to tail to get the resulting vector addition.

 

Multiplication by a scalar ,      ; 

 

 

 

Unit vectors with a unit length can be written   , where

 

            A vector is usually represented by a column.  For example, here are the components of a vector in R3

 

 

            There are several ways to multiply vectors.

 

1)     The dot product between two vectors results in a scalar.

 

 

        or in index notation

 

 

        In index notation, repeated indices are dummy indices which imply

 

 

This is the so called Einstein sum convection.  Note that . 

 

The geometric interpretation of the dot product is that it “projects” one of the vectors onto the other and multiples the resulting two lengths.  Thus,

 

 

 

 

 

      Ex)  Let  with

 

 

      then,

 

 

      Thus,

 

 

 

Note that for two vectors to be perpendicular to each other, then  and .

 

 

2)   The cross vector product between two vectors results in another vector, where

 

 

      where  is the measure area of parallelogram made from the two vectors  and .

 

 

 

 

 

 

 

 

 

Use the right-hand rule to find orientation of .

 

 

      This is an oriented vector depending on the order of the two vectors and follows a “right hand rule”.  Thus, .  A quick and dirty formula for the cross product can be obtained from the determinant as,

 

    where

 

 

      To express this in index notation, we must introduce the alternating symbol

 

 

      then

 

 

where this has repeated indices on j and k and has an implied sum on them.  Thus,  

 

 

      For example, for the c1 component,

 

 

 

This is the same result as from the other formula.  One of the reasons that a cross product has a complicated index notation form is that one is really trying to represent an area by a vector normal to it.  This can be done in 3-D, but not in higher dimensions where the cross product cannot be represented as a vector, but rather the area itself must be used.  This more general analysis is called the exterior calculus of differential forms.

 

 

3)   The outer product results in a matrix type quantity of second order

 

 

      In index notation, this is very simply written as

 

 

where i is the row index and j is the column index.

 

 

            The volume of a parallel piped in 3-D is given by the scalar  or

 

 

where ai, bj, and ck are summed on i, j, and k.  This results in

 

 

Thus, this determinant measures a volume.

 

            A dyad is a second order tensor with two indices.  A vector is a first order tensor and a scalar is a zero order tensor. 

 

            An identity matrix in R3 can be written as

 

 

This is often written as I or in index notation

 

 

and is called the Kronecker delta.  A useful identity between the Kronecker delta and the alternating symbol is

 

 

A dyad can always be decomposed into its symmetric and an anti-symmetric part

 

 

where DT is the transpose of D.

 

            A linear transformation between two vectors can be represented by a dyad.  Then,

 

 

In index notation, this is written

 

 

where there is an implied sum on j, or

 

 

 

 

Coordinate Systems

 

            In space, any vector can be represented as projections onto three non-zero, non-coplanar vectors.  Let these be .  These give a unique representation of any vector  in 3-D.  Then,

 

 

Most frequently, we will choose the rectangular Cartesian system () where , , and .

 

 

This forms an orthonormal set of basis vectors.

 

            Now, the components of  with respect to () are

 

 

 

 

 

            In addition to rectangular Cartesian coordinates, we could (and will) use cylindrical systems (), spherical systems ().

 

            For cylindrical coordinate systems

 

 

            For spherical coordinate systems

 

 

 

                        Cylindrical coordinates                          Spherical coordinates

 

 

 

 

 

 

 

 

 

 

 

 

 

            These systems provide unique representations, but, in general, do not have fixed directions in space and are functions of position.

 

 

 

Coordinate Transforms

 

            The coordinate transformation from a vector from one coordinate system to another is given by the coordinate transformation equations

 

 

where  are the coordinates of the vector in the xi system and  are the coordinates of the vector in the second system.

 

            For small changes of the coordinates, use a Taylor series expansion as

 

 

(with an implied sum on j).  This is an invertable transformation if  is nonsingular.

 

            Tensors will transform by

 

      vector transformation

 

where  are the components in coordinate system 2 and  are the components in coordinate system 1.  Second order tensors transform as

 

 

(implied sum on r and s).

 

            In Cartesian coordinates, vectors transform from one rotated coordinate system to another as

 

 

(with implied sum on j) where  are the direction cosines between  and  in the two coordinate systems.  Thus,

 

 

where rotation matrices A = aij are orthonormal.  Thus,  and .

 

 

 

 

For example, for the point P, the coordinates in two rotated coordinate systems are related by

 

 

Second order Cartesian tensors transform like

 

 

(with implied sums on p and q).  This in matrix notation is

 

 

            General transforms are done using the Jacobian of the coordinate transformation equations and the introduction of a given metric.  The components or “scale factors” (h1,h2,h3) of the metric tensor come into the index description of the vector operations.  (h1,h2,h3) are the square root of the diagonals of the metric tensor.  For example, for

 

            Cartesian – h1 = 1, h2 = 1, h3 = 1

            spherical – hr = 1, = r,  =

            cylindrical – hr = 1,  = r, hz = 1

 

            Vector and tensor fields assign a vector (or tensor) to every point in space (and time).  Thus,

 

            scalar field –

            vector field –

            tensor field –

 

Coordinate differentiation with respect to xi is expressed by  .

 

            This is given a special symbol,  the “del” or Nabla symbol.  Thus,

 

 

Comma notation is sometimes used

 

 

Several important vector operations with del include,

 

1)     The del operator acting on a scalar field.  This results in a vector field called the gradient or “grad”.

 

  ;  (index notation)

 

        For example, on a surface  is the vector at  that points in the direction of maximum ascent.

 

 

 

 

2)     The del operator acting on a vector using a dot product results in a scalar field which is called the divergence or “div”.

 

 

        In index notation, this can be written as  or  (sum on i).  This measures the net flux into or out of the local region at a given point.

 

 

 

 

3)     The del operator crossed with a vector in 3-D results in a vector field.

 

  ;  

 

        (index notation, sum on j and k) where  is the alternating symbol.  The curl measures the local vorticity or rotation (as in a bathtub drain) at a given point.

 

 

 

                                                                                                                                                                      local paddle wheel

 

 

 

 

 

        The curl points the direction of the rod for clockwise rotation.

 

 

4)     The Laplacian is a second order operation

 

 ;      (index notation, sum on i)

 

 

(sum on i).

 

 

5)     Several identities of interest are

 

 ;     (sum on j and k)

 

 ;      (sum on i, j and k)

 

 

            An essential integral identity is Gauss’ Theorem.  This relates a volume integral to a boundary surface integral for a vector (tensor) field , where  is the outward unit normal of the surface.  Thus,

 

 ;      (sum on i)

 

or for a second order tensor

 

 

            Finally, we give the gradient, divergence, and curl in orthogonal curvilinear coordinates ()

 

 

 

 

 

where the h1, h2, h3 are given for cartesion, cylindrical, and spherical coordinates above.