Analytical Mechanics This page intentionally left blank
Download 10.87 Mb. Pdf ko'rish
|
V v 0 < b 1
2
√2v < b 3 Fig. 3.6 3.6 One-dimensional motion 107 In the general case, the function F (t) can be expanded in Fourier series (see Appendix 7): F (t) =
∞ n =0 F n cos(n Ω t + γ
n ). (3.46) From the linearity of equation (3.41) it follows that the corresponding particular solution x p (t) is also given by the Fourier series x p (t) = ∞ n =0 B n cos(n Ω t + C
n ), (3.47) where B n and C n can be found by replacing γ by γ n and
Ω by Ω n in equation (3.45). 3.6 Beats
A particular phenomenon known as beats is due to the superposition of harmonic oscillations of frequencies that while different, are close together. More precisely, if ω 1
2 are such frequencies (ω 1 > ω
2 ), then (ω 1 − ω
2 )/ω
1 1. This can happen in various circumstances, but the most important cases occur in acoustics: it can easily be heard by playing the same note on two different instruments, not perfectly tuned to the same pitch. Mathematically, it reduces to the study of sums of the following kind: x(t) = A 1 cos(ω 1 t + α
1 ) + A
2 cos(ω
2 t + α
2 ), (3.48) where we can assume A 1 = A 2 = A, and isolating the excessive amplitude in one of the two vibrations, which henceforth does not contribute to the occurrence of this particular phenomenon. Under this assumption, equation (3.49) is equivalent to x(t) = 2A cos(ωt + α) cos(εt + β), (3.49) where
ω = ω 1 + ω 2 2 , ε =
ω 1 − ω 2 2 , α = α 1 + α 2 2 , β =
α 1 − α 2 2 . The term cos(ωt + α) produces an oscillation with a frequency very close to the frequencies of the single component motions. The amplitude of this oscillation is modulated in a periodic motion by the factor cos(εt + β), whose frequency is much smaller than the previous one. To be able to physically perceive the phenomenon of beats, the base and modulating frequency must be very different. In this case, in a time interval τ much larger than the period 2π/ε there can be found many oscillations of pulse ω and nearly-constant amplitude; one has the impression of a sound of frequency ω with amplitude slowly varying in time. 108 One-dimensional motion 3.7 3.7
Problems 0. Draw the graph of the function T (e) showing the period of the pendulum when e = E/mgl varies in [ −1, +∞].
1. Prove that x = 0 is a point of stable equilibrium for the potential (3.27). Solution
Write the energy (in dimensionless coordinates) in the form E =
1 2 y 2 + V (x).
Clearly V (x) ≤ E. We now distinguish between the cases where E > 0 or E ≤ 0 (Fig. 3.7). If E > 0, we can define x E > 0 such that, if x(0) is initially in the interval ( −x E , x E ), then x(t) must remain in the same interval. Since in the same interval we must have V (x) > −E, it follows that |y| < 2 √ E. Hence the trajectory is confined inside a rectangle; this rectangle is interior to any neighbourhood of the origin if E is chosen sufficiently small. V (x) E > 0 x E x Ј
x x Љ
E < 0 e –1/x 2 Fig. 3.7
3.7 One-dimensional motion 109 If E
≤ 0, we can choose x(0) between two consecutive roots x E , x E of the equation V (x) = E. Once again x(0) ∈ (x E , x E ) ⇒ x(t) ∈ (x E , x
E ), and the interval (x E , x
E ) can be chosen as near to the origin as desired by appropriately selecting |E|. In this interval we have V (x) > −e −1/x 2
, where x M = max( |x E |, |x E |). It follows that 1 2 y 2 < E + e −1/x
2 M and therefore the trajectory can be confined in an arbitrarily small neighbourhood. 2. A point in a horizontal plane is constrained without friction to the curve y = a cos (2π(x/λ)), with a, λ positive constants. The point is attracted to the origin by an elastic force. Discuss the dependence of the equilibrium of the point on the parameters a, λ. 3. Study the motion of a point particle subject to gravity on a cylindrical smooth helix in the following cases: (a) the helix has a vertical axis; (b) the helix has a horizontal axis. Find also the constraint reaction. 4. For a point constrained on a helix as in the previous problem, substitute the gravity force with an elastic force with centre on the axis of the helix. Study the equilibrium of the system. 5. Study the motion of a point particle constrained on a conic section, subject to an attractive elastic force with centre in a focus of the curve. 6. A cylinder of radius R and height h contains a gas subject to the fol- lowing law: pressure × volume = constant. An airtight disc slides without friction inside the cylinder. The system is in equilibrium when the disc is in the middle position. Study the motion of the disc if its initial position at time t = 0 is not the equilibrium position, and the initial velocity is zero. 7. A point particle is constrained to move along a line with friction propor- tional to v p ( v = absolute value of the velocity, p > 0 real). Suppose that no other force acts on the point. Find for which values of p the point comes to a stop in finite time, and what is the stopping time as a function of the initial velocity. 8. A point particle of unit mass is constrained to move along the x-axis, under the action of a conservative force field, with potential V (x) =
x 2 2 + x 3 3 . Determine the equilibrium positions and discuss their stability. Find the equation of the separatrix in the phase plane and draw the phase curves corres- ponding to the energy values 0, 1 8
1 6 , 1 5 , 1 2 . Compute to first order the variation of the frequency of the motion as a function of amplitude for orbits close to the position of stable equilibrium. 110 One-dimensional motion 3.7 9. A point particle of mass m is moving along the x-axis under the action of a conservative force field with potential V (x) = V 0 x
2n , where V 0 and d are positive real constants, and n is an integer, n ≥ 1. Prove that the period of the motion, corresponding to a fixed value E > 0 of the system energy, is T = 2d
2m E E V 0 1/2n 1 0 dy 1 − y
2n . 10. A point particle of mass m is moving along the x-axis under the action of a conservative force field with potential V (x) = V 0 tan 2 (x/d), where V 0 and d
are positive real constants. Prove that the period T of the motions corresponding to a fixed value E > 0 of the system energy is T = 2πmd
2m(E + V 0 ) . 11. A point particle of unit mass is moving along the x-axis under the action of a conservative force with potential V (x) =
⎧ ⎪ ⎨ ⎪ ⎩ (x + 1) 2 , if x ≤ −1, 0, if −1 < x < 1, (x − 1) 2 , if x ≥ 1. Draw the phase curves corresponding to the values E = 0, 1 2
period T of the motion corresponding to a fixed value E > 0 of the system energy is T = 2π 1
2 + 1 π 2 E . 12. A point particle of unit mass is moving along the x-axis under the action of a conservative force field with potential V (x) periodic of period 2π in x and such that V (x) = V
π |x| if x ∈ [−π, π], where V 0
to the values E = V 0 /2, V 0 , 2V
0 of the energy. (Be careful! The potential energy is a function which is continuous but not of class C 1 , therefore . . .) Compute 3.7 One-dimensional motion 111 the period of the motion as a function of the energy corresponding to oscillatory motions. 13. A point particle of mass m is moving along the x-axis under the action of a conservative force field with potential V (x) =
− V 0 cosh 2 (x/d) , where V
0 and d are positive real constants. Determine the equilibrium position, discuss its stability and linearise the equation of motion around it. Draw the phase curves of the system corresponding to the energy values E = −V 0
−V 0 /2, 0, V 0 . Compute the explicit value of T for E = −V 0 /2. 14. A point particle of mass m is moving along the x-axis under the action of a conservative force field with potential V (x) = V 0 (e
− 2e −x/d
), where V 0 and d are two real positive constants. Prove that the motion is bounded only for the values of E in the interval [ −V 0 , 0), and that in this case the period T is given by T = 2πd m
. 15. A point particle of unit mass is moving along the x-axis under the action of a conservative force field with potential V (x) =
⎧ ⎪ ⎨ ⎪ ⎩ (x + 1) 2 , for x < − 1 2 , −x 2 + 1 2 , for
|x| ≤ 1 2 , (x − 1) 2 , for x > 1 2 . Write the equation of the separatrix, draw the phase curves corresponding to values of E = 0, 1 5
1 4 , 1 2 , 1, 3 2 (hint: the potential energy is of class C 1 but not C 2 , therefore . . .) and compute the period T of the motion as a function of the energy. 16. A point particle of unit mass is moving along the x-axis according to the following equation of motion: ¨ x = − Ω 2 (t)x, where
Ω (t) =
ω + ε, if 0 < t < π, ω − ε, if π ≤ t < 2π, 112 One-dimensional motion 3.8 Here ω > 0 is a fixed constant, 0 < ε ω and Ω (t) = Ω (t + 2π) for every t. Prove that, if (x n , ˙ x n ) denotes the position and the velocity of the particle at time t = 2πn, then x n ˙ x n = A n x 0 ˙ x 0 , where A = A + A − is a 2 × 2 real matrix and A ±
⎛ ⎝ cos(ω ± ε) 1 ω
sin(ω ± ε)
−(ω ± ε) sin(ω ± ε) cos(ω ± ε) ⎞ ⎠ . Prove that if ω and ε satisfy the inequality |ω − k| < ε 2
2 + O(ε 2 ), or ω − k −
1 2
ε π k +
1 2 + O(ε), where k is any integer, k ≥ 1; it follows that |TrA| > 2. Deduce that the matrix A has two real distinct eigenvalues λ 1 = 1/λ
2 and prove that in this case the equilibrium position x 0 = ˙ x 0 = 0 is unstable. This instability phenomenon, due to a periodic variation in the frequency of a harmonic motion synchronised with the period of the motion, is called parametric resonance. See the books by Arnol’d (1978a, §25) and Landau and Lifschitz (1976, §27) for a more detailed discussion and applications (such as the swing). 3.8
Additional remarks and bibliographical notes Whittaker’s book (1936, Chapter IV) contains a good discussion of the simple pendulum. More specifically, one can find there the derivation of the double periodicity of the elliptic functions, using the following general result ( §34): in a mechanical system subject only to fixed holonomic constraints and positional forces, the solutions of the equations of motion are still real if the time t is replaced by √ −1t and the initial velocities (v 1 , . . . , v n ) are replaced by ( − √
1 , . . . , − √ −1v
n ). The expressions obtained represent the motion of the same system, with the same initial conditions, but with forces acting in the opposite orientation. Struik (1988, Chapter 1) gives a very detailed description of curves, which one can use as a starting-point for a deeper understanding of the topics considered in this chapter. 3.9 One-dimensional motion 113 3.9
Additional solved problems Problem 1 A point particle of unit mass is moving along a line under the action of a conservative force field with potential energy V (x) = x
α + λx 2 , where α, λ are two given real parameters, not both zero (all variables are dimensionless). (a) Determine for which values of α and of λ the origin x = 0 is a position of stable equilibrium. Linearise the equations around this point and determine the frequency of small oscillations. (b) Consider the motion, with initial conditions x(0) = 0, ˙ x(0) = 1. For which values of α and λ is the motion periodic? For which values of α, λ does the particle go to infinity in finite time? (c) Determine all the periodic motions and the integral of the period. (d) Draw the phase portrait of the system in the case α > 0, λ < 0. Solution
(a) It is immediate to verify that V (0) = 0, for every choice of α and λ. In addition, if α = 0 (λ = 0, respectively) the origin is stable if and only if λ > 0 (α > 0, respectively); in this case, it is also an absolute minimum of the potential energy. If αλ = / 0 we need to distinguish the case αλ > 0, when the potential energy V is defined on the entire line R, and the case αλ < 0 when the two lines x = ± −α/λ are vertical asymptotes of V . In both cases, x = 0 is stable if and only if α > 0. Notice, however, that V (x) = x 4 /α +
O(x 6 ) if α = / 0 and V (x) = x 2 /λ if α = 0. This implies that, if α = / 0, the linearised equation is simply ¨ x = 0 (the system is not linearly stable), while if α = 0 one has ¨ x + (2/λ)x = 0 and the frequency of small oscillations is 2/λ. (b) To ensure that the motion corresponding to the initial condition x(0) = 0, ˙ x(0) = 1, is periodic it is necessary for it to take place between two inversion points x + = −x − > 0, that are solutions of V (x ± ) = E =
1 2 . This is possible in the following cases: α > 0 for any λ; α = 0 and λ > 0; α < 0, λ > 0 and |α| < 1
λ 2 . To ensure that the motion corresponding to the initial condition x(0) = 0, ˙
x(0) = 1 reaches infinity in finite time, T ∞ , we must have λ = 0, α < 0; indeed, in this case we obtain the integral T ∞ = +∞ 0 dx 1 − 2x 4 /α
∞.
114 One-dimensional motion 3.9 (c) For arbitrary initial conditions, the motions are periodic of period T in the following cases. • For α = 0, λ > 0 In this case the period does not depend on the initial conditions (x(0), ˙ x(0))
and is given by T = 2π λ/2.
In all other cases, setting E = 1 2 ˙ x(0)
2 + V (x(0)), we have the following situations. • For λ = 0, α > 0 All motions are periodic, E ≥ 0, x
+ = (αE)
1/4 = −x − , T = 4 (αE) 1/4
0 dx 2[E − x 4 α ] . • For λ > 0, α > 0 All motions are periodic, E ≥ 0,
x + = λE 2 1 + 1 + 4α λ 2 E 1/2 = −x − , T = 4
x + 0 dx 2[E
− x 4 α +λx
2 ] . • For λ > 0, α < 0 Only motions corresponding to E ≥ −4α/λ
2 > 0 are periodic, and take place in the interval ( −x − , −x + ), or (x − , x + ), where x − and x
+ are the
positive roots of V (x ± ) = E: x + = λE 2 1 + 1 + 4α λ 2 E 1/2 , x − = λE 2 1 − 1 +
4α λ 2 E 1/2
, the period is given by T = 2 x
x − dx 2[E − x 4 α +λx 2 ] . • Finally, for λ < 0, α > 0 Only motions corresponding to E > 0 and the initial condition x(0) ∈ − −α/λ, −α/λ are periodic. The inversion points are x + = − λE 2 1 + 4α λ 2 E − 1 1/2 = −x − and the period is given by T = 4 x
0 dx 2[E − x 4 α +λx
2 ] . (d) The phase portrait in the case α > 0, λ < 0 is shown in Fig. 3.8. 3.9 One-dimensional motion 115
1
2
3 = 0 E 4
5
6
a 2a l l = 4a l 2
1
2
3
4
1
2
3
4
5
5
4
3
2
1
4
3
2
1
1
2
3
5
5
6
6
5
5
6
6
6
6
5
5
Fig. 3.8
116 One-dimensional motion 3.9 Problem 2 A point particle of unit mass is moving along a straight line under the action of a field created by two repulsive forces, inversely proportional by the same constant µ to the square of the distance from the respective centre of force. These centres of force are at a distance 2c from each other. Draw the phase portrait and compute the period of the oscillations of bounded motions. Solution
The potential energy is V (x) = µ( |x+c|
−1 + |x−c| −1 ) (where the origin x = 0 is the middle point between the two centres of force). There are two vertical asymptotes of V (x) at x = −c and x = +c and a relative minimum at x = 0 : V (0) = 2µ/c. The phase portrait is shown in Fig. 3.9. The periodic motions are the motions with x(0) ∈ (−c, c), E ≥ 2µ/c. The points of inversion of the motion, x ± , are the roots of µ/ |x ± + c | + µ/|x
± − c| = E. Clearly −c < x − =
+ ≤ 0 ≤ x
+ < c. Setting k = x + /c, one readily finds that k = 1 − (2µ/cE). Hence the period is given by T = 4
kc 0 dx 2[E − V (x)]
= 4 k √ 2E kc 0 1 − x 2 c −2 1 − x
2 k −2 c −2 dx = 2 c 3 (1 − k
2 )µ −1 E(k), where E(k) is the complete elliptic integral of the second kind (see Appendix 2). Problem 3 A point particle of unit mass is moving along a straight line under the action of a conservative force field with potential energy V (x). Suppose that V is a polynomial of degree 4, lim x
V (x) = + ∞ and that there exist values of the energy E for which V (x) − E has four simple zeros −∞ < e 1
2
3
e 4 < + ∞. Prove that in this case the periods of the oscillatory motions between (e 1
2 ) and (e
3 , e
4 ) are equal. Solution Under these assumptions, the periods of the motions are T 12
√ 2 e 2 e 1 dx (x − e 1 )(x
− e 2 )(x − e 3 )(x − e 4 ) , T 34 = √ 2 e 4 e 3 dx (x − e 1 )(x − e 2 )(x − e 3 )(x − e 4 ) , respectively. Since the four points (e 1 , e
2 , e
3 , e
4 ) and (e
3 , e
4 , e
1 , e
2 ) have the same cross-ratio (see Sernesi 1989, p. 325) ((e 1 − e 2 )/(e
2 − e
3 )) · ((e 3 − e
4 )/ (e 4 − e
1 )), there exists a rational transformation ξ = g · x = (Ax + B)/(Cx + D),
3.9 One-dimensional motion 117
3
2
1 2m c c x –c –c c x· g 3
2
2
3
1
3
2
1
Fig. 3.9
118 One-dimensional motion 3.9 where A, B, C, D ∈ R, AD − BC = 1, which maps the quadruple (e 1 , e 2 , e
3 , e
4 ) to (e 3 , e
4 , e
1 , e
2 ). This transformation is easily obtained using the equality of the cross-ratios: e 1 − e 2 e 2 − e
3 · e 3 − x
x − e
1 = e 3 − e
4 e 4 − e 1 · e 1 − ξ ξ − e
3 . This yields ξ = αx + β
γx + δ with α = e 1 − Ge
3 , β = Ge
2 3 − e 2 1 , γ = 1 − G, δ = Ge 3 − e 1 , G = e 4 − e 1 e 2 − e 3 · e 1 − e 2 e 3 − e 4 , and hence the desired transformation can be obtained by normalising the determinant αδ − βγ = −G(e 3 − e 1 ) 2 of the matrix α β γ δ : A = e 1 − Ge 3 √ −G(e 3 − e 1 ) , B = Ge 2 3 − e
2 1 √ −G(e 3 − e 1 ) , C = 1 − G √ −G(e
3 − e
1 ) , D = Ge 3 − e 1 √ −G(e 3 − e 1 ) . The substitution x = g
−1 · ξ =
Dξ − B
A − Cξ
yields dx =
dξ (A − Cξ) 2 , x − e i = g −1 · ξ − e
i = (Ce i + D)ξ
− (Ae i + B) A − Cξ
= ξ − g · e i (A − Cξ)(Ce i + D)
−1 , and hence dx (x − e 1 )(x
− e 2 )(x − e 3 )(x − e 4 ) = dξ (A − Cξ) 2 (ξ − g · e 1 )(ξ − g · e 2 )(ξ − g · e 3 )(ξ − g · e 4 ) (A − Cξ)
4 [(Ce
1 + D)(Ce
2 + D)(Ce
3 + D)(Ce
4 + D)]
−1 = (Ce 1 + D)(Ce
2 + D)(Ce
3 + D)(Ce
4 + D)
(ξ − e
3 )(ξ
− e 4 )(ξ − e 1 )(ξ − e 2 ) dξ, 3.9 One-dimensional motion 119 where we have used the fact that ge i = e
i +2 for i = 1, 2 and ge i = e
i −2 for i = 3, 4. Computing the product (Ce 1 + D) · · · (Ce 4 + D), we find: (Ce 1 + D) · · · (Ce 4 + D) = (1 − G)e
1 + Ge
3 − e
1 √ −G(e 3 − e
1 ) · · · (1 − G)e
4 + Ge
3 − e
1 √ −G(e 3 − e
1 ) = 1 G 2 (e 3 − e 1 ) 4 G(e 3 − e 1 )[e
2 − e
1 + G(e
3 − e
2 )] × (e 3 − e
1 )[e
4 − e
1 + G(e
3 − e
4 )].
Finally, since e 2 − e 1 + G(e 3 − e
2 ) = e
2 − e
1 + e 4 − e
1 e 3 − e 4 (e 2 − e
1 ) = (e
2 − e
1 ) e 3 − e
1 e 3 − e 4 , e 4 − e 1 + G(e
3 − e
4 ) = e
4 − e
1 + e 1 − e
2 e 2 − e 3 (e 4 − e
1 ) = (e
4 − e
1 ) e 3 − e
1 e 3 − e 2 , we arrive at (Ce
1 + D)(Ce
2 + D)(Ce
3 + D)(Ce
4 + D) =
1 G(e
3 − e
1 ) 2 e 2 − e 1 e 3 − e 4 e 4 − e
1 e 3 − e 2 (e 3 − e
1 ) 2 = 1 and the substitution x = Aξ + B
Cξ + D transforms the integral e 2
1 dx (x − e 1 )(x − e 2 )(x − e 3 )(x − e 4 ) into e 4 e 3 dξ (ξ − e
3 )(ξ
− e 4 )(ξ − e 1 )(ξ − e 2 ) , yielding T 12 = T
34 . It is possible to prove in an analogous way that if V is a poly- nomial of degree 3 and V (x) −E has three simple roots −∞ < e 1
2
3
∞, the period of oscillation in the interval [e 1 , e
2 ] is equal to twice the (finite) time needed for the point with energy E to travel the distance [e 3 , + ∞): T 3 = e 2 e 1 dx (x − e
1 )(x
− e 2 )(x − e 3 ) = +∞ e 3 dx (x − e 1 )(x − e 2 )(x − e 3 ) . 120 One-dimensional motion 3.9 The basic idea is to construct the rational transformation of the projective line mapping the quadruple (e 1 , e 2 , e
3 , ∞) to (e 3 , ∞, e 1 , e
2 ): e 1 − e
2 e 2 − e 3 · e 3 − x x − e
1 = − e 1 − ξ ξ − e
3 . These relations have a more general interpretation in the theory of elliptic curves and their periods. This is the natural geometric formulation highlighting the properties of elliptic integrals (see e.g. McKean and Moll 1999). Indeed, both T 12
3 can be reduced to a complete elliptic integral of the first kind by the transformation that maps the quadruples (e 1 , e 2 , e
3 , e
4 ) and (e
1 , e
2 , e
3 , ∞) to (1, −1, 1/k, −1/k), with k determined by the equality of the cross-ratios. Problem 4 Consider the motion with potential energy V ∈ C ∞
0, V (0) > 0. The motion around x = 0 is periodic, with period T (E) given by (3.8) with s 1 , s
2 roots of V (x) = E, for E ∈ (0, E 0
0 . (i) What are the conditions on V that ensure that T (E) is constant? (ii) Study in the general case the behaviour of T (E) for E → 0.
(iii) Using the result of (i) consider the problem of finding a (smooth) curve z = f (x), f ∈ C ∞
−x) = f(x), such that the motion on it due to gravity is isochronous. Solution (i) To answer this we follow Gallavotti (1980, §2.10). Start from the case that V ( −x) = V (x), when equation (3.8) becomes T (E) = 4 m 2
(E) 0 dx E − V (x)
. (3.50)
Introduce the inverse function of V in the interval (0, x(E)): x = ξ(V ) and use it as a change of variable in (3.51), observing that (0, x(E)) → (0, E): T (E) = 4 m 2
0 ξ (V )
√ E − V dV. (3.51)
It is well known (and easily verified) that Abel’s integral equation φ(t) =
t 0 ψ(r)dr √ t − r (3.52) (φ known, φ(0) = 0, ψ unknown) has the unique solution ψ(t) = 1
t 0 φ (r) √ t − r dr. (3.53)
3.9 One-dimensional motion 121 Comparing this with equation (3.52) we conclude 1 π φ = 4 m 2 ξ and T (E) satisfies the equation 4π m 2 ξ(V ) =
V 0 T (z) √ V − z dz. (3.54)
Hence, if we want T (z) = T 0 , we find that 2π m 2 ξ(V ) = √ V T 0 , (3.55) or V =
m 2 ω 2 x 2 , ω =
2π T 0 . (3.56)
It follows that the only symmetric potential that generates isochronous motions is the elastic potential. Considering a generic potential, we need to introduce the right inverse func- tion ξ
+ (V )> 0 and the left inverse function ξ − (V ) < 0. In equation (3.55), 2ξ(V ) is replaced by ξ + (V ) −ξ − (V ); a similar modification appears in (3.56), characterising those perturbations of the elastic potential that preserve isochronicity. (ii) Consider the fourth-order expansion of V : V (x)
mω 2 x 2 2 (1 + c 1 x + c
2 x 2 ), (3.57)
assuming that c = 0. The expansion of ξ ± (V ) to order V 2 is ξ ± (V )
± 2V mω + k ± V + k ± V 3/2 + k ± V 2 , (3.58) where the coefficients need to be determined. Substituting (3.59) into (3.58) and imposing an identity to order V 3/2 , we
find k ± = − c 1 mω 2 , k ± = ± 1 √ 2m 3 ω 3 5 2 c 2 1 − 2c
2 , k ± = 6c 1 c 2 − 4c 3 1 m 2 ω 4 . (3.59)
In the light of the result of (i) (see (3.55) with 2ξ replaced by ξ + − ξ − ) the
linear term in V in the expansion of ξ does not contribute to the difference ξ + − ξ − . Hence the first correction of ξ + − ξ
− is of order V 3/2 :
+ − ξ
− = 2
2V mω 2 + 2 m 3 ω 6 5 2 c 2 1 − 2c 2 V 3/2 . (3.60)
122 One-dimensional motion 3.9 Replacing in equation (3.55) ξ(V ) by 1 2 (ξ + − ξ
− )/2, and writing T (z) = T 0
1 (z) with T 0 = 2π/ω yields for T 1 (z) Abel’s equation 2π 5
c 2 1 − 2c 2 V 3/2 mω 3 = V 0 T 1 (z) √ V − z dz, (3.61)
with solution T 1 (E) = 3 5 2 c 2 1 − 2c 2 1 mω 3 E 0 V E − V dV =
3 2 π E mω 3 5 2 c 2 1 − 2c 2 . (3.62) We conclude that T (E) can be differentiated for E = 0, with T (0) =
3π 2mω
3 5 2 c 2 1 − 2c 2 . (3.63) (iii) We already know one solution: the cycloid (Example 3.1). We want here to start the problem from the equation 1 2 ˙s 2 + gf (x(s)) = E, (3.64) where s = x 0
2(ξ) dξ. Recalling (i), we impose on the function f the condition gf (x) =
1 2 ω 2 s 2 , (3.65)
which produces a harmonic motion of period 2π/ω; it follows that s = A sin ωt. To construct the curve corresponding to (3.66) it is convenient to set ωt = γ/2 and remark that dx dγ 2 = ds dγ 2 − dz dγ 2 , where
z(γ) = ω 2 2g A 2 sin 2 γ 2 . Also set A = αg/ω 2 , which yields dx dγ
= α 2 g 2ω 2 2 cos 2 γ 2 1 − α 2 sin
2 γ 2 . (3.66)
3.9 One-dimensional motion 123 Let us consider first the case that α = 1 (note that 0 < α ≤ 1), when clearly dx dγ = g 2ω 2 cos
2 γ 2 = R(1 + cos γ), R =
g 4ω 2 ; (3.67)
together with the condition x(0) = 0 this yields x = R(γ + sin γ). (3.68) Since A = g/ω 2 we find, for z(γ), z = R(1 − cos γ), (3.69) the cycloid. The choice α = 1 corresponds to the value of the energy allowing the motion to reach the highest possible points (z = 2R) of the cycloid. For α < 1 consider the cycloid x = R(ψ + sin ψ), z = R(1
− cos ψ) (3.70)
and let us verify if the arc defined by (3.67) and by z(γ) = α 2 R(1 − cos γ) lies on it. Write the relation between γ and ψ, expressed by z(ψ) = z(γ), namely α 2 (1 − cos γ) = 1 − cos ψ and compute dx dγ = dx dψ dψ dγ . It is evident that dψ dγ = α 2 sin γ sin ψ and that (3.71) yields dx dγ
2 sin γ
sin ψ . Expressing the right-hand side as a function of γ we find that (dx/dγ) 2 coincides with (3.67). Hence we have proved that the cycloid is the only symmetric curve producing isochronous oscillations under the gravity field. 4 THE DYNAMICS OF DISCRETE SYSTEMS. LAGRANGIAN FORMALISM 4.1 Cardinal equations The mathematical modelling of the dynamics of a constrained system of point particles (P 1 , m
1 ), . . . , (P n , m
n ) is based on the equations of motions for the single points: m
a i = R i , i = 1, . . . , n, (4.1) where R
i denotes the sum of all forces acting on the point (P i , m
i ). In addition to equations (4.1), one has to consider the constraint equations. The forces acting on a single particle can be classified in two different ways: either distinguishing between internal forces and external forces, or using instead the distinction between constraint reactions and the so-called active forces which we used in the study of the dynamics of a single point particle. These two different classifications yield two different mathematical schemes describing the dynamics of systems. In this section we consider the former possibility, distinguishing between internal forces, i.e. the forces due to the interaction of the points of the system among themselves, and external forces, due to the interaction between the points of the system and points outside the system. 1 We note two important facts: (a) internal forces are in equilibrium; (b) the (unknown) constraint reactions may appear among the external as well as among the internal forces. As a consequence of (a) we obtain the cardinal equations of dynamics: ˙ Q = R
(e) , (4.2) ˙ L(O) + v(O) × Q = M (e)
(O), (4.3)
using the standard notation for the linear momentum Q = n i =1 m i v i and the angular momentum L(O) = n i =1 m i (P i − O) × v i (the first can be derived by adding each side of equations (4.1), while the second is obtained by taking the vector product of the two sides of (4.1) with P i − O, with O an arbitrary point, and then adding). Here R (e)
and M (e)
are the resultant and the resultant moment, respectively, of the external force system. 1 To non-inertial observers the so-called apparent forces will also seem external (see Section 6.6 or Chapter 6). 126 The dynamics of discrete systems. Lagrangian formalism 4.1 Define the centre of mass P 0 by m(P 0 − O) =
n i =1 m i (P i − O)
(4.4) (m =
n i =1 m i , O an arbitrary point). Since Q = m ˙ P 0 , (4.5)
equation (4.2) can be interpreted as the equation of motion of the centre of mass (i.e. of the point particle (P 0 , m)):
m ¨ P 0 = R (e)
. Equation (4.3) can be reduced to the form ˙ L(O) = M
(e) (O)
(4.6) if v(O)
× Q = 0, and hence in particular when O is fixed or coincides with the centre of mass. The cardinal equations are valid for any system. On the other hand, in general they contain too many unknowns to yield the solution of the problem. The most fruitful application of such equations is the dynamics of rigid bodies (see Chapters 6 and 7). This is because all reactions due to rigid constraints are internal and hence do not appear in the cardinal equations. In the relevant chapters we will discuss the use of such equations. Here, we only consider the energy balance of the system, which has the form dT dt = W, (4.7)
where T =
1 2 n i =1 m i v 2 i , W = n i =1 R i · v i . Equation (4.7) can be deduced by differentiating T with respect to time and from equations (4.1). In correspondence with the two proposed subdivisions of the forces R i we can isolate the following contributions to the power W : W = W (e)
+ W (i)
(4.8) (W (e) is the power of the external forces, W (i)
is the power of the internal forces), or else W = W (a)
+ W (r)
(4.9) (W (a) is the power of the active forces, W (r)
is the power of the constraint reactions). 4.2 The dynamics of discrete systems. Lagrangian formalism 127 Remark 4.1 Equation (4.7) is not in general a consequence of the cardinal equations (4.2) and (4.3). Indeed, in view of equation (4.8) we can state that it is independent of the cardinal equations whenever W (i)
= / 0. It follows that when the internal forces perform non-vanishing work, the cardinal equations cannot contain all the information on the dynamics of the system. Equation (4.9) suggests that we can expect a considerable simplification of the problem when the constraint reactions have vanishing resulting power. We will examine this case in detail in the next section. 4.2
Holonomic systems with smooth constraints Holonomic systems have been introduced in Chapter 1 (Section 1.10). In analogy with the theory discussed in Chapter 2 (Section 2.4) on the dynamics of the constrained point, we say that a holonomic system has smooth constraints if the only contribution of the constraint reactions to the resulting power W (r)
is due to the possible motion of the constraints. Let Φ
i =1,...,n
φ i (4.10) be the vector representing all constraint reactions. Then the power W (r)
is expressed as W (r)
= Φ · ˙X (4.11) and in view of the decomposition (1.88) for the velocity ˙ X of a representative point, we can write W (r)
= W (r)
+ W (r)∗
, (4.12)
with W (r) = Φ ·V, W (r)∗ = Φ ·V ∗ . (4.13) We call the quantity W (r) the virtual power of the system of constraint reactions. We can now give the precise definition of a holonomic system with smooth constraints. D efinition 4.1 A holonomic system has smooth constraints if the virtual power of the constraint reaction is zero at every time and for any kinematic state of the system.
Equivalently we can say that a holonomic system has smooth constraints if and only if Φ is orthogonal to the configuration space: Φ ∈ (T
X V(t))
⊥ (4.14)
128 The dynamics of discrete systems. Lagrangian formalism 4.3 at every time and for any kinematic state. The latter definition is analogous to that of the orthogonality property for a smooth constraint in the context of the dynamics of a point particle (Definition 2.3). The characterisation of the vector Φ yields the possibility of a unique decomposition in terms of the basis of the orthogonal space: Φ = 3n− j =1 λ j (t) ∇ X f j (X, t),
(4.15) where λ
j (t) are (unknown) multipliers. The property (4.14) yields a system of differential equations characterising the motion of any holonomic system with smooth constraints, from which all constraint reactions can be eliminated. To find this system, consider the vector representing the active forces F (a)
= i =1,...,n F a i (4.16) and the vector representing the momenta Q = i
m i v i . (4.17) Then from equations (4.1) it follows that ˙ Q = F (a) + Φ . (4.18)
Imposing the property (4.14), the two vectors ˙ Q and F
(a) must have the same projection onto the tangent space (T X V(t)). Since the vectors (∂X/∂q k ) k =1,...,
are a basis for the tangent space (T X V(t)) in a fixed system of Lagrangian coordinates (q 1 , . . . , q ), the projection onto (T X V(t))
of a vector Z = (Z 1 , . . . , Z n ) ∈ R 3n is uniquely determined by the components Z Θ
= Z · ∂X ∂q k = n i =1 Z i · ∂P i ∂q k . It follows that F (a)
Θ ,k = n i =1 F (a)
i · ∂P i ∂q k , k = 1, . . . , , (4.19) and the equation of motion can be written as ( ˙ Q) Θ ,k = F
(a) Θ ,k , k = 1, . . . , . (4.20) 4.3
Lagrange’s equations The kinematic term (i.e. the left-hand side) in (4.20) has an interesting connection with the kinetic energy T . To show this connection, we first deduce the expression for T through the Lagrangian coordinates of the phase space. 4.3 The dynamics of discrete systems. Lagrangian formalism 129 By the definition of T and equations (1.95) we easily find T = 1 2 h,k =1 a hk ˙ q h ˙ q k + k =1 b k ˙ q k + c, (4.21)
where a hk (q, t) = n i =1 m i ∂P i ∂q h · ∂P i ∂q k = a kh (q, t), (4.22) b k (q, t) = n i =1 m i ∂P i ∂q k · ∂P i ∂t , (4.23) c(q, t) = 1 2
i =1 m i ∂P i ∂t 2 . (4.24) In the case of fixed constraints, it is possible to choose a Lagrangian coordinate system with respect to which time does not explicitly appear in the expression for X = X(q), and hence in the equations P i = P
i (q). For holonomic systems with fixed constraints we henceforth assume that such a coordinate system has been chosen. In this system all terms in equations (4.21) which are not quadratic vanish. P roposition 4.1 For a holonomic system with fixed constraints, the kinetic energy is a homogeneous quadratic form in the components of the vector ˙q. In the general case, note that T = 1
Q · V (4.25)
(the vector
Q is defined by (4.17)) and that the quadratic term in equations (4.25), i.e. T = 1
h,k =1 a hk ˙ q h ˙ q k , (4.26) is the kinetic energy due to the virtual component of the velocity. The expression (4.26) possesses a fundamental property. T heorem 4.1 T is a positive definite quadratic form. Note that the matrix (a hk ) h,k =1,...,
is the Hessian of T with respect to the variables ˙ q k
a hk = ∂ 2 T ∂ ˙ q h ∂ ˙ q k . (4.27)
Denoting it by H T we can write (4.26) as T = 1 2 ˙q · H
T ˙q.
(4.28) 130 The dynamics of discrete systems. Lagrangian formalism 4.3 As a consequence of Theorem 4.1 we have the following. C orollary 4.1 The matrix H T is positive definite. Proof of Theorem 4.1 Given a holonomic system, and a Lagrangian coordinate system, the coefficients a hk
k , c are uniquely determined. We need to show that T > 0, ∀ ˙q =
/ 0. (4.29)
To this end we express (4.26) in terms of the virtual velocities of the single point particles: T = 1
n i =1 m i v 2 i (4.30) and observe that ˙q = / 0
⇔ V = / 0, and hence not all v i can be zero. We have already remarked how the components of the vector ˙q, for every fixed q and t, can be viewed as parameters, playing the role of kinetic coordinates. We consider the derivatives of T with respect to such parameters: p k = ∂T ∂ ˙ q k , k = 1, . . . , , (4.31)
and hence p k = h =1 a hk ˙ q h + b k , k = 1, . . . , . (4.32) The system (4.32) is linear in p k , ˙
q k and, because of Theorem 4.1, it is invertible. It is easy to recognise that the p k are the Lagrangian components of the vector Q: p k = Q · ∂X ∂q k , k = 1, . . . , . (4.33) For this reason, these are called the kinetic momenta conjugate to the corresponding q k . The variables p k have great importance in mechanics. Remark 4.2 Let us set p k
h =1 a hk ˙ q k . It is useful to note that T = 1 2 p · ˙q. In the case of fixed constraints, this implies T =
1 2 p · ˙q. (4.34)
4.3 The dynamics of discrete systems. Lagrangian formalism 131 Let us return to our original aim of expressing the left-hand side of equation (4.20) as a function of T . Differentiate both sides of (4.33) with respect to time:
˙ p k = ( ˙ Q) Θ ,k + Q · ∂ ˙ X ∂q k (4.35)
and note that Q ·
∂ ˙ X ∂q k = i m i v i · ∂v i ∂q k = ∂T ∂q k , (4.36) which finally yields ( ˙
Q) Θ ,k = d dt ∂T ∂ ˙
q k − ∂T ∂q k . (4.37)
Equations (4.20) can be written in the form d dt ∂T ∂ ˙
q k − ∂T ∂q k = F Θ ,k , k = 1, . . . , (4.38) and are known as Lagrange’s equations. The functions F Θ ,k are defined by (4.19). Equations (4.38) are sufficient to find the solution of the motion problem. T heorem 4.2 The Lagrange equations (4.38) admit a unique solution satisfying the initial conditions q(0) = q
0 , ˙q(0) = w 0 . (4.39) Proof Equations (4.38) are second-order equations, linear with respect to ¨ q k
denoting by ˙q and ¨ q the
× 1 column vectors with components ˙q k and ¨ q k respectively, the system (4.38) can be written as H T ¨ q + ˙ H T ˙q + ˙ b − 1 2 ˙q T ∇ q H T ˙q + ∇ q b T ˙q +
∇ q c = F Θ , (4.40) where b and F Θ denote, respectively, the column vectors with components b k and F
Θ ,k , and c is given by (4.24). Note that the kth component of the column vector 1 2 ˙q T ∇ q H T q, is given by 1 2 i,j =1 (∂a ij /∂q
k ) ˙
q i ˙ q j . Hence Corollary 4.1 yields that the system is solvable with respect to the unknowns ¨ q k
¨ q k = χ k (q, ˙q, t), k = 1, . . . , , (4.41)
132 The dynamics of discrete systems. Lagrangian formalism 4.3 where the functions χ k are easily found. Indeed, χ k is the kth component of the column vector χ = H
−1 T F Θ + 1 2 ˙q T ∇ q H T ˙q +
∇ q b T ˙q +
∇ q c − ˙ H T ˙q − ˙b .
(4.42) The functions χ k contain F Θ ,k , given functions of q, ˙q, t; we assume that these functions are regular. The conclusion of the theorem now follows from the existence and uniqueness theorem for the Cauchy problem for a system of ordinary differential equations (cf. Appendix 1, Theorem A1.1): it is sufficient to set x = (q, ˙q) ∈ R 2
0 , w
0 ) and to write equation (3.21) as ˙x = v(x, t) with v(x, t) = (x +1 , . . . , x 2 , χ
1 (x, t), . . . , χ (x, t)). Remark 4.3 Equations (4.38) imply that the vector F (a) acts on the motion only through its projection onto the tangent space. Remark 4.4 Consider a point particle of mass m constrained to move on a fixed smooth regular surface S ⊆ R 3
1 , q
2 ) is a local para- metrisation of S, it follows that the kinetic energy of the particle can be written as T = m
[E(q 1 , q 2 ) ˙
q 2 1 + 2F (q 1 , q 2 ) ˙
q 1 ˙ q 2 + G(q 1 , q
2 ) ˙
q 2 2 ], (4.43)
where E, F , G are the entries of the first fundamental form of the surface (1.34). Since there are no active forces, F Θ 1
Θ 2 = 0 and Lagrange’s equations (4.38) take the form E ¨
q 1 + F ¨ q 2 + 1 2 ∂E ∂q 1 ˙ q 2 1 + 2 ∂E ∂q 2 ˙ q 1 ˙ q 2 + 2 ∂F ∂q 2 − ∂G ∂q 1 ˙ q 2 2 = 0, F ¨ q 1 + G¨ q 2 + 1 2 2 ∂F ∂q 1 − ∂E ∂q 2 ˙ q 2 1 + 2 ∂G ∂q 1 ˙ q 1 ˙ q 2 + ∂G ∂q 2 ˙ q 2 2 = 0. (4.44)
These can be recognised as the geodesic equations of the surface (1.46). Once again we find that the trajectories of a point particle constrained on a fixed smooth regular surface, with no other forces acting on it, are the geodesics of the surface (Proposition 2.2). Note in addition that this implies that the point acceleration is orthogonal to the surface. Example 4.1 Consider a point particle constrained to move on a surface of rotation without any active force. If x = (u cos v, u sin v, f (u)) is a local parametrisation, the kinetic energy of the point is given by T =
m 2 1 + (f (u)) 2 ˙ u 2 + u
2 ˙v 2 4.3 The dynamics of discrete systems. Lagrangian formalism 133 and Lagrange’s equations are the geodesic’s equations discussed and solved in Example 1.25. A similar conclusion can be reached when considering a holonomic system with fixed smooth constraints, without external forces. In this case the space V of all
possible configurations becomes a Riemannian manifold when endowed with the metric
(ds) 2 = i,j =1 a ij (q 1 , . . . , q ) dq i dq j , (4.45) where a ij are given by equation (4.22); Theorem 4.1 ensures that equation (4.45) defines a Riemannian metric (see Definition 1.30). For this system, Lagrange’s equations become j =1
ij ¨ q j + 1 2 j,k
=1 ∂a ij ∂q k + ∂a ik ∂q j − ∂a jk ∂q i ˙ q j ˙ q k = 0, (4.46)
where i = 1, . . . , . Multiplying by a hi and summing over i (where a hi are
the components of the inverse matrix of A = (a ij )), since i =1 a hi a ij = δ h j , equations (4.46) become ¨ q
+ j,k
=1 Γ h j,k ˙ q j ˙ q k = 0,
h = 1, 2, . . . , , (4.47)
where Γ h j,k are the Christoffel symbols (1.69) associated with the metric (4.45). Equations (4.47) are the geodesic equations (1.68), (note that ˙s 2 = 2T is constant). We have proved the following. T heorem 4.3 The space of configurations of a holonomic system with fixed constraints, endowed with the metric (4.45) induced by the kinetic energy, is a Riemannian manifold. If there are no active forces (and the constraints are smooth) the trajectories of the systems are precisely the geodesics of the Riemannian manifold. Systems of this kind are also called natural Lagrangian systems (see Arnol’d et al. 1988). Example 4.2 Write down Lagrange’s equations for a system of two point particles (P 1 , m
1 ), (P 2 , m
2 ) with P
1 constrained to move on a circle of radius R and centre O, P 2 constrained to move along the line OP 1 , in the presence of the following forces acting in the plane of the circle: F 1 , applied to P 1 , of constant norm and tangent to the circle; F 2 , applied to P 2 , of constant norm, parallel to F 1 but with opposite orientation. 134 The dynamics of discrete systems. Lagrangian formalism 4.3
Download 10.87 Mb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling