Analytical Mechanics This page intentionally left blank
Download 10.87 Mb. Pdf ko'rish
|
F
1
2
1
P 2 Fig. 4.1 All constraints are smooth. The system has two degrees of freedom, and we can take as Lagrangian coordinates the angle ϕ between the radius OP 1 and a fixed axis, and the abscissa ξ of P 2 onto the radius OP 1 (Fig. 4.1). The kinetic energy is T =
1 2 m 1 R 2 ˙ ϕ 2 + 1 2 m 2 (ξ 2 ˙ ϕ 2 + ˙
ξ 2 ). (4.48) We need to compute the Lagrangian components F ϕ , F
ξ : F ϕ = F
1 · ∂P 1 ∂ϕ + F 2 · ∂P 2 ∂ϕ , F ξ = F 1 · ∂P 1 ∂ξ + F 2 · ∂P 2 ∂ξ . (4.49) Setting e r = (P
1 − O)/|P
1 − O|, e
ϕ = F
1 / |F 1 |, we find ∂P 1
= Re ϕ , ∂P 2 ∂ϕ = ξe ϕ , ∂P 1 ∂ξ = 0, ∂P 2 ∂ξ = e
r . (4.50) Hence F ϕ = RF 1 − ξF 2 , F ξ = 0.
(4.51) Substituting into equations (4.38), we obtain the desired equations (m 1
2 + m
2 ξ 2 ) ¨ ϕ + 2m
2 ξ ˙
ξ ˙ ψ = RF
1 − ξF
2 , (4.52) ¨ ξ − ξ ˙ϕ 2 = 0.
(4.53) We easily recognise that equation (4.52) is the second cardinal equation written with reference to the point O. Equation (4.53) can instead be interpreted as the equation of motion for P 2 in the reference frame rotating with the line OP 1 , in
which P 2 is subject to the centrifugal acceleration field ξ ˙ ϕ 2 . 4.3 The dynamics of discrete systems. Lagrangian formalism 135
1
2
1
2
2
−P 2
−P 2
Fig. 4.2 We note finally that the equations do not change if F 1 and F
2 are given functions of ξ. Example 4.3 Two point particles P 1 , P 2 of equal mass m move in a plane in which they are subject to the Biot–Savart field (2.19), with the constant C equal in magnitude for the points but with opposite signs. The two points are constrained to preserve a constant distance . Write down the Lagrange equations. The system has three degrees of freedom. We can choose polar coordinates (r, ϕ) for the point P 1 and determine the position of P 2 as a function of the angle ψ between the vectors (P 1 − O) and (P 2 − P
1 ) (Fig. 4.2). The Cartesian coordinates of the two points P 1 , P 2 are given by x 1
y 1 = r sin ϕ, x 2 = r cos ϕ + cos(ϕ − ψ), y
2 = r sin ϕ + sin(ϕ − ψ).
(4.54) Hence
v 2 1 = ˙r 2 + r 2 ˙ ϕ 2 , (4.55) v 2 2 = ˙r 2 + r 2 ˙ ϕ 2 + 2 ( ˙ ϕ − ˙ψ) 2 + 2 [ ˙r sin ψ + r ˙ ϕ cos ψ]( ˙ ϕ − ˙ψ), (4.56) 136 The dynamics of discrete systems. Lagrangian formalism 4.4 and in addition ∂P 1 ∂ϕ = ( −r sin ϕ, r cos ϕ), ∂P 1
= 0, ∂P 1 ∂r = ∂P 2 ∂r = (cos ϕ, sin ϕ), ∂P 2 ∂ϕ = ( −r sin ϕ − sin(ϕ − ψ), r cos ϕ + cos(ϕ − ψ)), ∂P 2
= ( sin(ϕ − ψ), − cos(ϕ − ψ)). (4.57) We still need to determine the Lagrangian components F r , F
ϕ , F
ψ . As in Fig. 4.2 we have F
= F 1 · ∂P 1 ∂r + F 2 · ∂P 2 ∂r = F 2 cos α + π 2 = −F 2 sin α, F ϕ = F 1 · ∂P 1 ∂ϕ + F 2 · ∂P 2 ∂ϕ = F 1 r − F 2 h, F ψ = F
2 · ∂P 2 ∂ψ = F 2 cos(π
− γ) = −F 2 cos γ, (4.58) where
h 2 = r 2 + 2 + 2 r cos ψ (4.59)
and sin α
= sin γ
r = sin ψ h . (4.60) Since T = 1 2 m(v 2 1 + v 2 2 ), we can write down Lagrange’s equations. This is left as an exercise. 4.4 Determination of constraint reactions. Constraints with friction The solution of the initial value problem for the Lagrange system of equations yields the vector q = q(t). Once this vector is known, we can determine the motion of the representative vector X = X(t). Hence the kinematic terms in equations (4.18) and the vector F(X(t), ˙ X(t), t) are known. From this it is easy to find
Φ = Φ (t) and then φ i = φ i (t).
As an example, using expression (4.15), where the multipliers λ j (t) are now known, we can write φ i = 3n−
j =1 λ j (t)
∇ P i f j (X(t), t), (4.61) 4.4 The dynamics of discrete systems. Lagrangian formalism 137 where
∇ P i f j (X, t) denotes the standard gradient vector, obtained by considering only the coordinates of the point P i . Example 4.4 Consider again the system studied in Example 4.2. Suppose the integrals ϕ = ϕ(t) and ξ = ξ(t) of the equations of motion are known; find the constraint reactions φ 1 and φ 2 on the two points of the system. In this particular case, the problem is simple; indeed, once the accelerations a 1 , a 2 are known, it is enough to write φ i = m
i a i − F i . However, it is useful to illustrate the general procedure. We start by computing the unit base vectors of the normal space, writing the constraint equation in the form (see Fig. 4.1) 2f 1
2 1 + y 2 1 − R 2 = 0,
f 2 = y 1 x 2 − x 1 y 2 = 0,
and hence ∇f 1 = (x 1 , y 1 , 0, 0),
∇f 2 = ( −y 2 , x 2 , y
1 , −x 1 ). In addition, ˙ Q = (m
1 ¨ x 1 , m
1 ¨ y 1 , m
2 ¨ x 2 , m
2 ¨ y 2 ) and F (a)
= −F 1 y 1 R , F 1 x 1 R , F 2 y 1 R , −F 2 x 1 R , implying that the vector Φ ∈ R
4 of the constraint reactions can be found by using equation (4.15), and determining the multipliers λ 1 , λ 2 starting from ( ˙ Q − F
(a) ) · ∇f 1 = λ
1 |∇f
1 | 2 , ( ˙
Q − F (a)
) · ∇f
2 = λ
2 |∇f
2 | 2 . In the present case, we have ∇f 1
2 = 0. Using the coordinates ξ, ϕ of Fig. 4.1 one can easily obtain the equations λ 1 = m 1 ˙ ϕ 2 , λ 2 = 1 ξ 2 + R 2 Rξ(m 1 − m
2 cos
2 ϕ) ¨
ϕ − 2m
2 R 2 ˙ ξ sin
2 ϕ ˙
ϕ − F
1 ξ − F 2 R .
For the case of constraints with friction it is necessary to formulate the hypo- thesis linking the constraint reactions φ (a) i
i (strictly speaking, to the virtual velocities v i ) and then include them among the active forces. Using linear links such as 2 φ (a) i = −µ i v i , (4.62) with µ i ≥ 0, one must add the following term to the right-hand side of equation (4.38): Φ (a)
Θ ,k = − n i =1 h=1 µ i ∂P i ∂q h · ∂P i ∂q k ˙ q h . (4.63)
2 The system of coordinates plays an important role. Consider as an example the case of a rotating sphere. Mathematically, the constraint is fixed, as we can represent it by |x| = R,
but in order to take friction into account, the virtual velocity must be computed relative to the constraint, and hence in a coordinate system based on the sphere. 138 The dynamics of discrete systems. Lagrangian formalism 4.5 These Lagrangian components of the friction forces can be deduced from a kind of kinetic variables potential : f D = − 1 2 n i =1 h,k=1 µ i ∂P i ∂q h · ∂P i q k ˙ q h ˙ q k (4.64) in the sense that Φ (a)
Θ ,k = ∂f D ∂ ˙ q k . (4.65) The function f D is called the Rayleigh dissipation function; it is equal to half the power dissipated due to the friction W D = − n i =1 µ i v 2 i . (4.66) 4.5 Conservative systems. Lagrangian function D efinition 4.2 The system of active forces (F (a) i , P i ), i = 1, . . . , n, is conser- vative if there exists a regular function U , called the potential of the system, such that its representative vector F (a) is given by F (a)
= ∇ X U (X). (4.67)
To determine if a system of forces is conservative, and to determine its potential, it is necessary to consider the subdivision into internal and external forces. If each of the external forces is a conservative field with potential given by U (e) i (P i ), the overall potential of the external forces is given by U (e) (X) = n i =1 U (e) i (P i ). (4.68)
As an example, for the gravity field we find U = −mgz
G (m =
i m i , z G is the height of the centre of mass, assuming the z-axis is vertical and oriented upwards). The internal forces are given by interaction pairs. As an example, the interaction between the points P i and P
j is expressed by the pair (F ij , P
i ), (
−F ij , P j ), where
F ij = f ij (P i − P j ) (P i − P j ) |P i − P
j | . (4.69) The conservative interaction pairs are characterised as the conservative central force fields (Example 2.3). P roposition 4.2 The interaction pair (F ij , P
i ), (
−F ij , P j ), where F ij is given
by (4.69), is conservative if and only if f ij depends only on r ij = |P i − P
j |.
4.5 The dynamics of discrete systems. Lagrangian formalism 139 Its potential is given by U (i)
ij (r ij ) = f ij (r ij ) dr ij . (4.70) Proof The proof is a simple extension of Example 2.3. Well-known examples of interaction potentials are the elastic potential U (r) = − 1 2 kr 2 and the gravitational potential U = k/r. The overall potential of the internal forces is the sum of the potentials of the interaction pairs: U (i) (X) = 1≤i U (i)
ij (r ij ). (4.71)
The potential of the system is given by U = U (e)
+ U (i)
. Remark 4.5 The projection of F (a)
onto the tangent space has the structure of a gradient. This means that the Lagrangian components F (a) Θ
are given by F (a) Θ ,k = ∂ U ∂q k , k = 1, . . . , , (4.72) where
U(q, t) = U[X(q, t)]. (4.73)
The time dependence is introduced only through the constraints’ motion (recall that for holonomic systems with fixed constraints the Lagrangian coordinates are by convention such as to yield representations of the form X = X(q)). When the forces are conservative, equations (4.38) can be written more concisely, by introducing the Lagrangian function L(q, ˙q, t) = T + U, (4.74)
where U(q, t) is the function given by (4.73). Remark 4.6 Recall the Lagrangian expression for T , and what we have just mentioned regard- ing the potential. These facts imply that ∂L/∂t = / 0 only if the constraints are in motion. Consider now equations (4.72) and note that all the derivatives ∂ U/∂ ˙q k vanish. We deduce that Lagrange’s equations can be written in the form d dt ∂L ∂ ˙
q k − ∂L ∂q k = 0, k = 1, . . . , . (4.75)
140 The dynamics of discrete systems. Lagrangian formalism 4.5 Remark 4.7 If a Lagrangian coordinate q k does not appear explicitly in the Lagrangian L (in this case we say the coordinate is cyclic), from equation (4.75) it follows that there exists the first integral p k = ∂L/∂ ˙
q k = constant. Example 4.5 Write equations (4.75) for a system of two point particles (P 1 , m
1 ), (P
2 , m
2 ) constrained in a vertical plane as in Fig. 4.3 (smooth constraints). Let ϕ be as shown in the figure and x 2 the x-coordinate of P 2 ; we take these as the Lagrangian coordinates. Then we have L(ϕ, x
2 , ˙
ϕ, ˙ x 2 ) = 1 2 m 1 R 2 ˙ ϕ 2 + 1 2 m 2 ˙ x 2 2 − m
1 gR(1
− cos ϕ) − k 2 (x 2 2 + 2R
2 − 2Rx
2 sin ϕ
− 2R 2 cos ϕ). (4.76) It follows that the required equations are ¨ ϕ +
g R sin ϕ − k m 1 x 2 R cos ϕ
− sin ϕ = 0, (4.77)
¨ x 2 + k m 2 (x 2 − R sin ϕ) = 0. (4.78)
It is possible that some of the forces are conservative and some are not. In this case, writing F (a)
= ∇ X U (X) + G(X, ˙ X, t),
(4.79) y O R P 1
2
1
w Fig. 4.3
4.6 The dynamics of discrete systems. Lagrangian formalism 141 where G cannot be derived from any potential, we can define L = T + U , which yields a mixed form of Lagrange’s equations: d dt ∂L ∂ ˙
q k − ∂L ∂q k = G Θ ,k . (4.80)
Remark 4.8 It can be verified immediately that Lagrange’s equations (4.75) corresponding to the Lagrangians L(q, ˙ q, t), L = cL, L = L + dF/dt, with c a constant and F (q, t) an arbitrary C 2 function, have the same solutions. To clarify this point, we write explicitly dF dt
· ∇ q F (q, t) + ∂F ∂t . 4.6 The equilibrium of holonomic systems with smooth constraints The equilibrium configurations of a holonomic system with fixed, smooth con- straints are given by the constant solutions of equations (4.38). Therefore the equilibrium equations are F (a) Θ ,k (q) = 0, k = 1, . . . , (4.81)
(in this case, it is natural to consider that forces and constraints are time- independent). In the case of conservative fields, using equations (4.73), the equilibrium equations can be written as ∂ U
k = 0,
k = 1, . . . , (4.82)
and express the fact that at equilibrium the restriction of the potential U (X) to the manifold X = X(q) is stationary. Clearly the problem (4.82) can be formulated in R 3n as the problem of finding the stationary points of the potential on the configuration manifold, i.e. ∇ X U (X) + m j =1 λ j f j (X) = 0, (4.83)
f j (X) = 0, j = 1, . . . , m. (4.84)
The Lagrange multipliers λ j are the same as those that give the constraint reactions Φ = m j =1 λ j ∇ X f j (X), (4.85)
142 The dynamics of discrete systems. Lagrangian formalism 4.7 since it must be that, at equilibrium, F (a)
+ Φ = 0. (4.86) Remark 4.9 When the forces are only due to the presence of mass, the equilibrium con- figurations are exactly the configurations for which the height of the centre of mass is extremal. Example 4.6 Study the equilibrium of the system considered in Example 4.2. It is easy to check that of the two equilibrium equations (4.51), F ϕ = 0, F
ξ = 0,
the latter is an identity while the former admits the unique solution ξ = RF 1 /F 2 . For this value of ξ there exist infinitely many equilibrium configurations, obtained by varying ϕ. If F
1 and F
2 are functions of ξ the equation ξ = RF 1 /F
admits either one, many, or no solution. Example 4.7 Study the equilibrium of the system considered in Example 4.3. It is easy to show that there are no equilibrium configurations. Indeed, because of equations (4.58) we should have in particular sin α = cos γ = 0, which is incompatible with equations (4.60), valid for all the possible configurations of the system.
Example 4.8 Study the equilibrium of the system considered in Example 4.5. The equilibrium equations are ∂ U ∂ϕ = 0,
∂ U ∂x 2 = 0,
yielding (set to zero the kinematic terms in (4.77), (4.78)): m 1 g kR sin ϕ − x 2 R cos ϕ + sin ϕ = 0, (4.87) x
− R sin ϕ = 0. (4.88)
If x 2 is eliminated one obtains sin ϕ m 1 g kR − cos ϕ + 1 = 0 ⇔ sin ϕ = 0. (4.89) Hence the only solutions are ϕ = 0, x 2 = 0 and ϕ = π, x 2 = 0.
4.7 Generalised potentials. Lagrangian of an electric charge in an electromagnetic field There are situations when it is possible to define a Lagrangian even if the system of forces depends on velocity. Indeed, note how equations (4.38) imply that, if
4.7 The dynamics of discrete systems. Lagrangian formalism 143 there exists a function U(q, ˙q, t) such that F (a) Θ ,k = ∂ U ∂q k − d dt ∂ U ∂ ˙ q k , k = 1, . . . , , (4.90) the usual definition of the Lagrangian L = T + U still permits us to write the Lagrange equations in the form (4.75). The function U(q, ˙q, t) is called a generalised potential. An important example when it is possible to define a generalised potential is the case of a force applied to a charge e in an electromagnetic field (E, B) (the Lorentz force): F = e E +
1 c v × B . (4.91)
We seek the generalised potential for equation (4.91), starting from Maxwell’s equations div B = 0, (4.92)
curl E + 1 c ∂B ∂t = 0. (4.93) From the first it follows that it is possible to express the field B as B = curl A, (4.94)
where A(x, t) is the so-called vector potential, defined up to an irrotational field (which we assume to be independent of time). Because of equation (4.94), equation (4.93) takes the form curl
E + 1 c ∂A ∂t = 0. (4.95) As a consequence of (4.95) there exists a scalar function ϕ (the usual electrostatic potential when B is independent of time) such that E +
1 c ∂A ∂t = −∇ϕ. (4.96) Substituting equations (4.94) and (4.96) into (4.91) we obtain F = e −∇ϕ −
1 c ∂A ∂t + 1 c v × curl A . (4.97) We now make a second transformation. Note that (v × curl A) i = v
· ∂A ∂x i − v · ∇A
i , i = 1, 2, 3, (4.98) 144 The dynamics of discrete systems. Lagrangian formalism 4.8 and that
v · ∂A ∂x i = ∂ ∂x i (v · A),
(4.99) where, as for equation (4.90), we consider ˙ x 1
x 2 , ˙ x 3 as independent variables. Since finally v · ∇A i = dA i dt − ∂A i ∂t , (4.100)
we can write v × curl A = ∇(v · A) − dA dt + ∂A ∂t , (4.101) and we arrive at the following expression for F: F = e −∇ϕ +
1 c ∇(v · A) − 1 c dA dt . (4.102) Note that F can be expressed in the form 3 F = −∇V + d dt ∇ v V, (4.103) with
V = e ϕ − 1 c v · A (4.104)
and ∇ v = i (∂/∂ ˙ x i ) e i . We have finally obtained the Lagrangian of the charge e in the electromagnetic field (E, B): L = T
− e ϕ − 1 c v · A .
(4.105) 4.8
Motion of a charge in a constant electric or magnetic field For completeness, we examine the motion of a charge e (positive or negative) of mass m in a constant electric field E superimposed on a field with constant induction B. Consider the equation ma = e
E + 1 c v × B .
(4.106) 3 According to definition (4.90) the generalised potential is U = −V . However it is customary to use the potential energy instead of the potential. 4.8 The dynamics of discrete systems. Lagrangian formalism 145
3
1
2
Download 10.87 Mb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling