You are on page 1of 20

Generalized Coordinate Systems

Grant W. Mason
Department of Physics and Astronomy, Brigham Young University, Provo, Utah 84602

INTRODUCTION
Cartesian coordinates are the familiar rectangular coordinates xi . They have the dimensions of [length] so that speed has the dimensions of [length]/[time] and acceleration has the dimensions of [length]/[time]2. But what if you wish to use plane polar coordinates, r and θ? Theta does not have the dimension of length. How then should we write down an expression for acceleration in terms of polar coordinates? What we attempt below is to deal with so-called generalized coordinates so that we see how to write down correct expressions for positions, velocities and accelerations. We will try to do it in a general way so that we do it once and for all for plane polar coordinates, cylindrical coordinates, spherical coordinates, parabolic coordinates and some coordinate systems you haven’t even made up yet. Unless otherwise indicated, it is assumed that there is a summation on repeated indices.

Deﬁnitions
ˆ ei ˆ a = ai ei qi , dqi ds ∂r bi = ∂qi a = Ai bi bi = ∇qi a = Ai bi gi j = bi · b j gi j = bi · b j ∂b Γkj = bk · ∂q ij i T = 1/2m(ds/dt)2 V Fi = −∇iV L = T −V (unit vectors) (physical components, ai ) (generalized coordinate, differential) (differential line element) (basis vectors, bi ) (contravariant components, Ai ) (reciprocal basis vectors, bi ) (covariant components, Ai ) (covariant metric tensor, gi j ) (contravariant metric tensor, gi j ) (Christoffel symbols, Γkj ) i (kinetic energy) (potential energy) (covariant force component, Fi ) (Lagrangian)

Theorems
ds = dqi bi ds2 = gi j dqi dq j bi · b j = δij √ ai = g(ii) Ai (no sum on (ii)) gi j g jk = δk i Ai = g i j A j Ai = gik Ak bi = gi j b j bi = gi j b j ∂b ∂bi = ∂qij ∂q j 1 Γkj = gkl 2 (∂gil /∂q j + ∂g jl /∂qi − ∂gi j /∂ql ) i ˆ If a = ai ei = Ai bi = Ai bi is acceleration: Ak = (qk + qi q j Γkj ) ¨ ˙ ˙ i 1 d ˙ ˙ ˙ Al = dt (glk qk ) − 2 (∂gi j /∂ql )qi q j √ i ai = g(ii) A √ ai = g(ii) gi j A j d ∂T ∂T Fk = mAk = dt ( ∂qk ) − ∂qk ˙

(Christoffel symbols)

(no sum on (ii)) (no sum on (ii))

As a practical matter, we use this machinery in the following manner. We write down an arbitrary ds from which we form ds2 ≡ ds · ds. We divide the expression for ds2 by dt 2 to form (ds/dt)2 and from this we form T = (1/2)m(ds/dt)2. We then operate on T according to the prescription, Ak = 1 m d ∂T ∂T ( k)− k dt ∂q ˙ ∂q

to obtain Ai . From the Ai we obtain the Ai and from the Ai we obtain the physical components of acceleration, ai . These expressions tell us how to write down correctly the acceleration in our particular coordinate system. Since Fk = mAk , we may also obtain the generalized forces if we know the acceleration a priori. In systems where a potential energy function, V, exists and we can write, Fk = − ∂V ∂qk

we can dispense with most of the machinery by deﬁning L = T −V and operating on L according to the following prescription: d ∂L ∂L ( k)− k = 0 dt ∂q ˙ ∂q

x3 ) are an example. spherical coordinates (r. and it will probably be useful to keep a concrete example in mind. r= x2 + x2 + x2 1 2 3 x2 + x2 1 2 x3 θ = arctan( ) . cylindrical coordinates and plane polar coordinates are alternatives that are sometimes used. Consider. Consider spherical coordinates. Example: Plane polar coordinates ˆ ˆ e1 = r x1 = r cos θ x2 = r sin θ ∂r/∂x1 = x1 / (x2 + x2 ) = cos θ 1 2 ∂θ/∂x1 = −x2 /(x2 + x2 ) = − sin θ/r 2 1 ∂x1 /∂r = cos θ ∂x2 /∂r = sin θ ˆ ˆ e2 = θ r = (x2 + x2 ) 1 2 θ = arctan(x2 /x1 ) ∂r/∂x2 = x2 / (x2 + x2 ) = sin θ 1 2 ∂θ/∂x2 = x1 /(x2 + x2 ) = cos θ/r 2 1 ∂x1 /∂θ = −r sin θ ∂x2 /∂θ = r cos θ An Example of Generalized Coordinates Coordinates are used to specify the position of a particle in space. how should we express the Second Law if we use coordinates that may not even have dimensions of length? In what follows we shall try to solve this problem once and for all for all admissible generalized coordinates. The development is somewhat abstract and general. A is the magnetic vector potential and v is the velocity of the particle.These latter equations are Lagrange’s Equations and they are the differential equations of motion of the system. θ. becomes: L = T −V − eψ + (e/c)(A · v) where e is the charge on the particle. In a threedimensional space. L. In problems with particle motion in electric and magnetic ﬁelds. Newton’s Second Law of Motion can easily be expressed in Cartesian coordinates as Fi = m d 2 xi = mxi ¨ dt 2 where the second derivative of xi with respect to time will have the dimensions of acceleration if Cartesian coordinates are used. φ). only r in the set of spherical coordinates has dimensions of length. we need three such coordinates. ψ is the electric potential. for example. Observe that while the Cartesian coordinates all have dimensions of length. Spherical coordinates. But. x2 . the Lagrangian. The three Cartesian coordinates (x1 .

in the direction of increasing r if θ and φ are held constant to deﬁne the constant surfaces (cone and plane). The velocity of the particle is immediately found in spherical coordinates to be. q3 ). r. x1 Think of each of these as an example of the form ψ(x1 . then a half-plane is deﬁned. If r is a constant. a small displacement resulting from a change in r would be ds(1) = drˆ . we obtain ds2 = dr2 + r2 dθ2 + r2 sin2 θdφ2 . r If we form the dot product. i. If θ is constant. θ. where a double sum is intended. we denote this quantity as ds2 = gi j dqi dq j . θ. Observe the following: ˆ ˆ ˆ 1. dt dt dt dt . 1. we will denote unit vectors ˆ ˆ e associated with generalized coordinates as e1 . A small displacement of the particle is denoted ds. ˆ φ lies along the intersection of sphere and cone in the direction of increasing φ. we will denote generalized coordinates as (q1 . then a spherical surface is deﬁned. 2.e. ˆ ds(2) = rdθθ. See Fig. Finally. These surfaces intersect at a point whose coordinates are (r. (r. v= dθ ˆ dφ ˆ ds dr ˆ = r + r θ + r sin θ φ. Deﬁne unit vectors to lie along the lines of intersection of these surfaces and pointing in the direcˆ tion of increasing coordinate.φ = arctan( x2 ). In general. If θ and φ are held constant in spherical coordinates. q2 . ˆ ds(3) = r sin θdφφ. x3 ) = c discussed in connection with the directional derivative. 5. φ) are an example of generalized coordinates. They have unit length. then a conical surface is deﬁned.ˆ 3 . ˆ ˆ ds = ds(1) + ds(2) + ds(3) = drˆ + rdθθ + r sin θdφφ. e2 . if r and φ are held constant. 4. φ are unit vectors. x2 . In general. In general. r Similarly. θ. ˆ θ lies along the intersection of sphere and plane in the direction of increasing θ. A completely arbitrary displacement in spherical coordinates would be a vector sum of these three. r lies along the line of intersection of the cone and the plane and points away from the origin. ds · ds. 3. For example. φ). Similarly. If r and θ are held constant. If φ is constant.

this latter quantity will be written. ˙ ˙ . The square of the speed is. and half-plane) that deﬁne the spherical coordinates of a point.z r y x FIGURE 1. ˙ ˙ v2 = r2 + r2 θ2 + r2 sin2 θφ2 . v2 = gi j qi q j . cone. ˙ In general. The intersection of constant surfaces (sphere.

Thus deﬁned. but are not necessarily of unit length nor dimensionless. For example. The set of numbers. ˆ ˆ ˆ ˆ a = a1 e1 + a2 e2 + a3 e3 = ai ei . and the surface of constant φ is a half-plane. the surface of constant r is a sphere. the kinetic energy of the particle in spherical coordinates may be written. q3 ). but they do not necessarily have dimensions of length. x2 . 2 The pattern which is illustrated here for ﬁnding the kinetic energy will turn out to be very important. When each is set equal to a constant. the unit vectors must form a linearly independent set for the coordinates system to be “admissible. we will write. For spherical coordinates. qi (x1 . Unlike the unit vectors in a Cartesian system. are (r p. 1 1 ˙ ˙ T = mv2 = m(˙2 + r2 θ2 + r2 sin2 θφ2 ). p p p the surface of constant θ is a cone. the ˆ unit vector r has a different direction for different points in space. The half-plane of constant φ p intersects the circle at a point. P. are said to be the generalized coordinates of a particle in the system. like the unit vectors in spherical coordinates. although. The set of numbers. Generalized Coordinates Imagine a set of generalized coordinates. The generalized coordinates that uniquely deﬁne this point.6. a set of surfaces are deﬁned for which the point of intersection. are said to be the physical components of a. this tangent line varies in space. Finally. the directions of the unit vectors depend on the point at which they are deﬁned. The coordinates must uniquely deﬁne the position of the particle. in spherical coordinates. An arbitrary vector may be expressed as a linear combination of admissible unit vectors. ai . Generalized Basis Vectors We now introduce a new set of basis vectors that have the same direction as the unit vectors. deﬁne a set of unit vectors. P. q2 . r 2 2 In general. qi . θ p . has coordinates (q1 . they often are. The cone of angle θ p intersects the sphere of radius r p in a circle.” The unit vectors are not necessarily orthogonal. Consider a point of intersection of the level surfaces such that q2 and q3 are held ﬁxed and only q1 is p p . which are tangent to the lines of intersection of the various surfaces taken two at a time in the same way that the unit vectors in spherical coordinates were deﬁned. Its direction is given by the tangent line along the intersection of the cone and the half-plane. ei . ˆ Now. x3 ). φ p ). 1 ˙ ˙ T = mgi j qi q j . but as θ or φ varies.

ds = ∂s(i) i dq . are called the contravariant components of the vector a. QED. ∂s(1) ∂r = . denotes summation. ∂qi (Sum on i. bi = ∂s(i)/∂qi.” then we may use them as a basis for expressing an arbitrary vector. 3 ∂q ∂q3 Therefore. q3 + dq3 ). once as a subscript and once as a superscript. bi = ∂r . Thus. ∂q1 ∂q1 ∂s(2) ∂r = . 2 ∂q ∂q2 ∂r ∂s(3) = . By the chain rule. this convention applies henceforth throughout this chapter. Theorem Theorem: ds = bi dqi . ∂s(1) . The faster q1 changes in space along s(1). A tangent vector to this path.allowed to vary. b1 ≡ ∂q1 The partial derivative indicates that q2 and q3 are being held constant. to the position of a particle at position. ∂qi If these new basis vectors are “admissible. by deﬁnition. the partial derivatives imply that. q3) to (q1 + dq1 . Unless otherwise indicated. so ds = bi dqi . ds = ds(1) + ds(2) + ds(3) represents a differential change. Ai . are deﬁned in a similar way. a. q2 + dq2 . Two other generalized basis vectors. . The coefﬁcients. the smaller will be the magnitude of b1 .) But. r. This deﬁnes the line of intersection of the 2-surface and the 3-surface which we can describe as a path s(1). a = b1 A1 + b2 A2 + b3 A3 = bi Ai . Proof: Let ds = ds(1) + ds(2) + ds(3) be the inﬁnitesimal displacement from (q1 . b2 and b3 . They are denoted with a superscript to establish a pattern for the Einstein summation convention: A repeated index. b1 is obtained by differentiation. In this context. dr. we may write. q2 .

we obtain the metric tensor by inspection of the form of ds2 . Thus. For example. we obtain. the tensor property of the metric tensor is not important. . Observe that the matrix is symmetric. These may be grouped together into a 3 × 3 matrix G called the metric tensor with components gi j such that. r Taking the dot product with itself. Proof: Since we have shown that we may write ds = bi dqi . Vectors and tensors are deﬁned by their transformation patterns when one changes from one coordinate system to another. |bi | = bi · bi = g(ii) .The Metric Tensor The inner (dot) product of two arbitrary vectors. any inner product is characterized by the nine products. it follows that. ds · ds = ds2 = (bi · b j )dqi dq j = gi j dqi dq j . The length of one of the basis vectors is obtained from the inner product of the vector with itself. a and c may now be computed using the generalized basis vectors: a · c = (bi Ai ) · (b jC j ) = (bi · b j )AiC j . we have.   1 0 0  0 G =  0 r2 2 sin2 θ 0 0 r We are now in a position to establish the relationship between the unit vectors and our generalized basis vectors. (A tensor is a generalization of a vector. Comparing to the form ds2 = gi j dqi dq j . an arbitrary displacement in spherical coordinates is written. In our present application. ds2 = dr2 + r2 dθ2 + (r sin θ)2 dφ2 .) Theorem Theorem: ds2 = gi j dqi dq j . by inspection. bi · b j . gi j = bi · b j . gi j = g ji . ˆ ˆ ds = drˆ + rdθθ + (r sin θ)φ. As a practical matter. QED.

deﬁnes a different surface. then its reciprocal. indeed. are called reciprocal basis vectors. satisﬁed. (No sum on (ii). ∇qi = bi .(No sum on (ii). j are “reciprocal” if they satisfy the relationship. bi and bi . Two sets of basis vectors. qi (x1 . g(ii) Ai . where each constant. b1 . c. can be written in terms of reciprocal basis vectors just as well as it can be written in terms of unit vectors. Hence. Proof: The proof is a simple application of the chain rule. ∂xk ∂q j ∂q j ∂r j · ∇q j = i k = i = δi . An arbitrary vector..) ˆ ˆ e a = bi Ai = ( g(ii) ei )Ai = ( g(ii) Ai )ˆ i = ai ei . The gradients of these functions of the Cartesian coordinates are vectors that are normal to the respective surfaces at the point (q1 . QED. q2 . has a small magnitude. Reciprocal Basis Vectors Set each of the generalized coordinates. Theorem Theorem: bi · b j = δi . The vectors. bi · b j = QED. ai = √ √ g(ii)Ai . bi . This relationship says that b1 is orthogonal to both b2 and b3 .e. x3 ) = qi (xi ) = c. It also says that if b1 has a large magnitude. x2 . a = A1 b1 + A2 b2 + A3 b3 = Ai bi . we immediately have ˆ ei = bi bi =√ . a. q3 ).) Since the unit vectors are already known to have the same direction as the basis vectors. The generalized coordinates are independent of one another. bi · b j = δi . However. we must show that the reciprocal relationship is. p p p i. i ∂q ∂q ∂x ∂q j . |bi | g(ii) Theorem Theorem: ai = Proof: Thus.

In similar fashion. Ai .The set of numbers. gi j .” The covariant and contravariant forms of the metric tensor are inverses of each other. QED. form the matrix G−1 and are called the contravariant components of the metric tensor to distinguish them from the covariant components of the metric tensor. QED. Proof: Ai = bi · a = bi · (b j A j ) = (bi · b j )A j = gi j A j . The covariant components of a are not the same as the physical components of a nor are they the same as the contravariant components. we maintain a parallelism between the reciprocal basis vectors and the basis vectors. QED. By doing so.” Theorem Theorem: Ai = gi j A j . bi · b j = δij . Theorem Theorem: Ai = bi · a. This important operation is called “lowering the index. bi · a = bi · (A j b j ) = (bi · b j )A j = δij A j = Ai . The covariant and contravariant forms of the metric tensor are useful in relating covariant and contravariant components of vectors. like G. Theorem Theorem: Ai = gi j A j . Proof: We use the relationship. Note that the pattern of notation is consistent: an index repeated once as a subscript and once as a superscript denotes summation. Proof: Ai = bi · a = bi · (b j A j ) = (bi · b j )A j = gi j A j .e. gi j . The reciprocal basis vectors do not necessarily have unit length nor are they necessarily parallel to the unit vectors. The elements. G−1 G = I. i. are called the covariant components of a. but they must be linearly independent if the coordinate system is to be admissible. is symmetric. we can also show Ai = bi · a. This important operation is called “raising the index. Observe that G−1 . Deﬁne gi j ≡ bi · b j . gi j = g ji . .

Hence. In practice. j 3. we have. 2. gki bi = gki (gi j b j ) = δk b j = bk .or. QED. acts exactly as the identity operation. In an admissible generalized coordinate system. gi j g jk = δk . bi = ai j b j . j . bi = gi j b j . gik = bk · (ai j b j ) = ai j (bk · b j ) = ai j δk = aik . Theorem Theorem: bi = gik bk . multiplying both sides of the latter by gki (and forming a sum indicated by the repeated index. i If you apply the combination G−1 G in succession to an arbitrary vector. Proof: 1. 2.” returning you to where you began. Thus. one after the other. gi j gk j = δk i bi = gi j b j . We conclude. this property is used to ﬁnd the contravariant components of the metric tensor after the covariant components are extracted from the form. We have established that. Proof: 1. Theorem Theorem: bi = gi j b j . bi . the reciprocal basis vectors must form a linearly independent set of vectors. the set of basis vectors. ds2 = gi j dqi dq j . We have previously established that. gik = gki = bk · bi so. can be expressed as a linear combination of the reciprocal basis vectors. so that the combination of the two. G will ﬁrst “lower the index” and G−1 will turn around and “raise the index. i).

We choose to express them as a linear combination of basis vectors. 2. reciprocal basis vectors or basis vectors. Theorem Theorem: Γl j = glk i Proof: 1. b j · bk = g jk bi · bk = gik 1 2 ∂gik ∂g jk ∂gi j + i − k ∂q j ∂q ∂q bi · b j = gi j . Theorem Theorem: ∂bi /∂q j = ∂b j /∂qi . We conclude.3. i j ∂q The numbers which form the coefﬁcients of the sum generated by the dummy index k are called Christoffel symbols of the ﬁrst kind. QED. Differentiate each of these expressions in turn to form. by deﬁnition. We have. Proof: This symmetry of pattern merely reﬂects the fact that the order in which partial derivatives are taken does not matter: ∂b j ∂ ∂r ∂ ∂r ∂bi = j i = i j = i. Since each of these nine partial derivatives. bi = gi j b j . ∂gik ∂bk ∂bi = bi · j + j · bk j ∂q ∂q ∂q ∂g jk ∂bk ∂b j = b j · i + i · bk i ∂q ∂q ∂q . j ∂q ∂q ∂q ∂q ∂q ∂q QED. we can express each as some linear combination of either the unit vectors. ∂bi /∂q j are themselves vectors. ∂bi = Γkj bk .

We conclude. thus introducing an additional sum indicated by the dummy index k. 5. We have proved as an earlier theorem that ∂bk ∂b j = k ∂q j ∂q ∂bk ∂bi = k. when we form the combination ∂gik ∂g jk ∂gi j + i − k. 6. It is tempting to solve for Γl j by dividing both sides of this expression by 2glk . ∂gik ∂g jk ∂gi j + i − k = 2[Γl j bl ] · bk = 2Γl j (bl · bk ) = 2Γl j glk . 1 2 2gmk Γl j glk = 2δm Γl j = 2Γm . one for each of the combinations of i. ∂q j ∂q ∂q 3. i but one must remember that the repeated index l indicates a sum of terms. j. ∂gik ∂g jk ∂gi j ∂bi + i − k = 2 j · bk . and k! The correct way to proceed is to multiply both sides by gmk . ∂gik ∂g jk ∂gi j + i − k. l Thus. what we are dealing with here are twenty-seven equations. gmk glk = δm . In fact. i i i ∂q j ∂q ∂q and. We are left with the result. We then use a result that we have already proved. i j ∂q Then. We may now use our assumption that ∂bi = Γl j bl . not an isolated term. ∂q j ∂q ∂q four terms add out and two combine. i ij l i ∂gik ∂g jk ∂gi j + i − k ∂q j ∂q ∂q . Γm = gmk ij QED. . j ∂q ∂q ∂q ∂q 4.∂b j ∂bi ∂gi j = bi · k + k · b j k ∂q ∂q ∂q and form the combination. i ∂q ∂q Thus.

but we must note that as the particle moves. but the outcome is to turn ds into a velocity. but they are completely equivalent ordinary differential equations which describe the motion of a particle. becomes a set of ordinary differential equations. so that the reader gets to choose the most convenient coordinates for a problem without undue concern about how to write down acceleration properly in terms of the coordinates that have been chosen. the acceleration of a particle. and thus to reformulate Newton’s Second Law of Motion. our immediate purpose is to apply the formalism to a particular vector quantity. we must differentiate v with respect to time. The advantage of this new form is that it is done in terms of generalized coordinates. The mathematical formalism can also be used to give general expressions for the divergence. You will be relieved to know that for our purposes this is a formal result essential to the derivation of Lagrange’s equations that follow. We begin with the expression for an arbitrary displacement: ds = dqi bi . 2. the basis vectors change with its position and must therefore be considered functions of time. However. then “lower the index” to ﬁnd the covariant components. APPLICATION TO ACCELERATION The mathematics we have introduced can now be extended in several directions. It was used by Albert Einstein as the original language of his General Theory of Relativity and a simple modiﬁcation of it is also the basis of the most elegant formalism of the Special Theory of Relativity (see Chapter 10). a= dbi dv = qi bi + qi ¨ ˙ . when resolved into components. Newton’s Second Law is a vector equation which. the covariant components of Newton’s Second Law are Lagrange’s equations. but not one that we will use as a practical tool. dt dt . To obtain acceleration. In three dimensional space there are twenty-seven of them. ˙ dt 3. F = ma. Our immediate purpose is to show the connection between Newton’s Second Law and Lagrange’s equations. curl and Laplacian in generalized coordinates. The equations which result from the reformulation of Newton’s Second Law are called Lagrange’s Equations. into a new form. v= ds = qi bi . Hence. We divide both sides by dt. once one knows the covariant and contravariant forms of the metric tensor. we will write the covariant form of Newton’s Second Law and show that for a large class of problems. Finally. the Christoffel symbols can be calculated. We shall ﬁrst ﬁnd the contravariant components of acceleration. Observe that this is a division and not the process of differentiation.In principle. Contravariant Components of Acceleration 1.

˙ dt ∂q dt ∂q From this. a = qi bi + qi q j ¨ ˙ ˙ or. ¨ ˙ ˙ i Covariant Components of Acceleration 1. Al = glk Ak = glk qk + glk Γkj qi q j . we can apply the chain rule. glk gkn = δn . the covariant components of the acceleration can be obtained from this expression once the covariant form of the metric tensor is known. ¨ ˙ ˙ i From this expression. ¨ ˙ ˙ i ∂q j a = [qk + qi q j Γkj ]bk = Ak bk . so that we can write. Al = glk qk + ¨ 1 2 1 2 ∂gil ∂g jl ∂gi j + i − l ∂q j ∂q ∂q . To compute dbi /dt. Again. glk Γkj = glk gkn i 2 ∂q j ∂q ∂q Observe. We may use the expression which we have just derived for the Christoffel symbols to write. we identify the contravariant components of acceleration as. we conclude. ˙ ˙ In principle. ∂bi dq j ∂bi dbi = j = qj j .4. Thus. 1 ∂gin ∂g jn ∂gi j glk Γkj = δn + i − n i l 2 ∂q j ∂q ∂q = 3. ∂bi = qi bi + qi q j Γkj bk . ¨ i ˙ ˙ 2. we lower the index using Al = glk Ak . Ak = qk + qi q j Γkj . To obtain the covariant components of acceleration. ∂gin ∂g jn ∂gi j 1 + i − n . l ∂gil ∂g jl ∂gi j + i − l ∂q j ∂q ∂q qi q j . . this is a formal result that is essential to our derivation. but not one that we will use as a practical tool.

¨ Observe that this term arises in the expression. Al = + 1 2 ∂gl j d (glk qk ) − i qi q j ˙ ˙ ˙ dt ∂q qi q j . are called generalized forces. Newton’s Second Law is the familiar fk = mak . ˙ ¨ ˙ ˙ dt ∂q We may now write. of course. namely. ˙ dt ∂q A renaming of dummy indices (in the second term below) immediately yields. ˙ ¨ ˙ ˙ dt ∂q where we have used the chain rule in the second term on the right to obtain d ∂glk (glk ) = m qm . glk qk . ˙ ˙ ∂qi ∂q ∂qi These equalities may be used to eliminate the middle three terms and to reduce our expression for Al to 1 ∂gi j i j d ˙ qq . ˙ ˙ ∂gil ∂g jl ∂gi j + i − l ∂q j ∂q ∂q Observe that by simply switching dummy indices and using the fact that gil = gli . we arrive at the point of our entire excursion into differential geometry! The covariant components of Newton’s Second Law are Fk = mAk . Fk . the covariant components of force. Consider the ﬁrst term in the expression we have just derived for Al .A Better Form for the Covariant Components of Acceleration We now begin to change the form of the covariant components of acceleration to put them into a form that is a practical tool. we can show that ∂gl j i j ∂gil j i ∂g jl i j qq = jq q = ˙ ˙ ˙ ˙ qq . ∂gl j d (glk qk ) = glk qk + i qi q j . In some texts. In terms of physical components. ˙ ˙ Al = (glk qk ) − dt 2 ∂ql Newton’s Second Law in Covariant Form Finally. ∂glk d (glk qk ) = glk qk + m qm qk . 1. Observe that ds2 = gi j dqi dq j .

and that. Fl = mAl = m 1 ∂gi j ∂T d ∂T d (glk qk ) − m l qi q j = ( l ) − l . qi . We may write. d ∂T d d ( k ) = m (gik qi ) = m (gki qi ) ˙ ˙ dt ∂q ˙ dt dt (The factor of one-half disappears in the latter two expressions because there is a double sum in the expression for T. therefore. then we may simplify the covariant form of Newton’s Second Law. If the system we are describing is “conservative” so that we may deﬁne a potential energy function. V. of a particle as. ˙ ∂T ∂V d ∂T ( l)− l = − l dt ∂q ˙ ∂q ∂q . ˙ ˙ k 2 ∂q ∂q ∂T = mgik qi . written in covariant form is. ˙ ˙ 2 dt 2 2. Fl = d ∂T ∂T ( l)− l. such that Fl = − ∂V ∂ql and V depends only on the generalized coordinates and not on the generalized velocities. If you have trouble seeing this.) 3. you should write out a short double sum so that you see it works. 1 ds 1 T = m( )2 = mgi j qi q j . we may write the kinetic energy. ˙ ˙ ˙ dt 2 ∂q dt ∂q ˙ ∂q Newton’s Second Law. The primary difﬁculty with it is knowing how to write down Fl . We may therefore write. dt ∂q ˙ ∂q This is a perfectly general result. Observe further that if we take the view of generalized coordinates and generalized velocities as independent variables for purposes of taking partial derivatives. ˙ ∂qk ˙ and. T. 1 ∂gi j ∂T = m k qi q j . Lagrange’s Equations If we narrow ourselves to a class of problems that satisﬁes two additional conditions we arrive at a remarkable result.

it may formally be included in the ﬁrst term without making any difference. Solutions of the equations are models of the motion of the system. we do not directly use (for our purposes) bi . i. contravariant and physical components of the acceleration of the particle. θ). Find ds2 = ds · ds in terms of dr and dθ and extract from ds2 the metric tensor gi j . 1. 2. Lagrange’s equations are just the components of Newton’s Second Law for conservative systems written in a different and probably unfamiliar form. Show how these might appear on a diagram. Much of the intervening machinery used to derive the equations disappears and does not return.d ∂(T −V ) ∂(T −V ) − = 0. Using ds2 . Do the following: • Draw a diagram (in two dimensions) of a particle whose position is speciﬁed ˆ ˆ by plane polar coordinates (r. Display the metric tensor as a 2 × 2 matrix. bi . • Determine the covariant. PROBLEMS 1. L ≡ T −V and write ∂L d ∂L ( l ) − l = 0. write down the kinetic energy of the particle in terms of r and θ and their time derivatives.) We now deﬁne the so-called “Lagrangian”.) . it need only indicate directions of the vectors and whether the vectors have unit length or not.e. Show the unit vectors r and θ on your diagram and clearly label them. They are the “equations of motion” of the system. (The diagram is only to be qualitative. dt ∂q ˙ ∂q These are Lagrange’s equations and they are very important in theoretical mechanics. L. • Express an arbitrary differential displacement ds in terms of these coordinates and show this displacement on your diagram. The equations that are obtained by operating on the Lagrangian according to the prescription of Lagrange’s equations are ordinary differential equations. i or gi j . T and V are usually easy to write down correctly and once written down lead to the differential equations of the system in a very methodical way. • Determine gi j and display as a 2 × 2 matrix. These were introduced in order to derive Lagrange’s equations and to show their connection to Newton’s Second Law. dt ∂ql ˙ ∂ql (Since V does not depend on the generalized velocities. Γkj . 4. 3. For example.

aθ = 2v0 ω) 2. the left-hand edge of the block is vertical. 7. Do the following: • Beginning with Newton’s Second Law. ∂qk ∂L d ∂L ( k)− k = 0 dt ∂q ˙ ∂q 4. z = z. 8. Write the equations in terms of r. The length of the tether appears as a constant in the Lagrangian. 5. put another block on the inclined plane so that it slides down the plane (without friction) under the • . r and r by eliminating z. physical velocity and physical acceleration in cylindrical coordinates. 9. we can write Fx = −kx. (Cartesian coordinates) Obtain the differential equations of motion for the earth’s motion about the sun. dt ∂q ˙ ∂q show that if Fk can be expressed in the form. ˙ ¨ Obtain the differential equations of motion for a particle sliding without friction from the top of a hemispherical dome. q .Let a particle move on a circle of constant radius b. The particle does not separate from the dome. (Plane polar coordinates) Obtain the differential equations of motion for a particle moving without friction on the inside of a cone. The system has only one degree of freedom since the radius of the motion appears as a constant in the Lagrangian. ∂V 1 2 3 (q . • ∂T d ∂T ( k ) − k = Fk . The system has only two degrees of freedom and there will only be two differential equations of motion. Find the radial and transverse components of the physical acceleration as functions of time. 6. Calculate the metric tensor.) Obtain the differential equations of motion for a projectile on a ﬂat earth. The cone is parameterized in cylindrical coordinates as z = αr. y = r sin θ. What is the radial component of acceleration (physical)? What is the transverse component? • A bug crawls outward with constant speed v0 along the spoke of a wheel which is rotating with constant angular speed ω. 3. Deﬁne a coordinate X that keeps track of the position of this vertical side. (Spherical coordinates. Now. mx + kx = 0. For the one-dimensional harmonic oscillator. Fk = − then. If the incline slopes toward the right. Observe that the tether line has constant length. ¨ Obtain the differential equations of motion for a tether ball. Deﬁne the function L = T − V and obtain the differential equation of the simple harmonic oscillator.) Consider an inclined plane consisting of a right-triangular shaped block that is free to slide on its bottom along a frictionless horizontal surface. (Plane polar coordinates. (Ans: ar = −v0tω2 . q ). Assume r = 0 at t = 0. Cylindrical coordinates are deﬁned by x = r cos θ. Show 1 that V = 2 kx2 .

inﬂuence of gravity. Keep track of the second block’s position with a coordinate x measured from the top of the inclined plane. X) of each. Here there are two masses. The system has two degrees of freedom. the Lagrangian is written simply as the sum of the the two independent kinetic energies of the two objects minus the sum of the two independent potential energies. nonorthogal coordinates. In such a case. ¨ ¨ (Unusual.) . Obtain the differential equations of the motions of both blocks and solve them directly for the accelerations (x. This problem introduces a problem not found in the previous ones.