Newton's Laws, Phase Space and Stability
CSIR NET classical mechanics: Newton's laws, variable mass, 1D energy method, phase portraits, fixed points, linear stability and Liouville's theorem.
- 22 min
On this page · 15 sections
- Prerequisites
- The Hook: The Pendulum That Will Not Stand Up
- Newton’s Laws, Stated Exactly
- Variable Mass: The Rocket
- One-Dimensional Motion: The Energy Method
- Dynamical Systems and Phase Space
- Fixed Points and Linear Stability
- Liouville’s Theorem
- Memory Tricks
- Don’t Confuse
- Common Mistakes
- Practice
- Quick Revision Table
- Key Takeaways
- What’s Next
Prerequisites
- Physical Sciences overview: where classical mechanics sits in the paper and how it is marked.
- Linear ODEs and Variation of Parameters: constant-coefficient second-order equations, which every linearised system reduces to.
- Eigenvalues and Cayley–Hamilton: the trace and determinant of a \( 2\times2 \) matrix decide the type of every fixed point on this page.
The Hook: The Pendulum That Will Not Stand Up
Newton’s Laws, Stated Exactly
First law. There exist frames (inertial frames) in which a body subject to no net force moves with constant velocity. The first law is a definition of the frames in which the second law holds, not a special case of it.
Second law. In an inertial frame,
$$\boxed{\mathbf{F} = \frac{d\mathbf{p}}{dt}, \qquad \mathbf{p} = m\mathbf{v}}$$For a constant mass this is \( \mathbf{F} = m\mathbf{a} \). For a body whose mass changes, \( \mathbf{F} = d\mathbf{p}/dt \) applied blindly to the body alone is wrong; you must apply it to a closed system (body plus the mass it gains or loses).
Third law. Forces come in equal and opposite pairs, \( \mathbf{F}_{12} = -\mathbf{F}_{21} \) (weak form). The strong form adds that the pair acts along the line joining the particles. The weak form gives conservation of linear momentum for an isolated system; the strong form is what you need for conservation of angular momentum. Magnetic forces between moving charges violate the strong form, and the “missing” momentum is carried by the field.
Consequences for a system of particles
For \( N \) particles with total mass \( M \) and centre of mass \( \mathbf{R} = \sum m_i\mathbf{r}_i/M \):
| Quantity | Equation | Conserved when |
|---|---|---|
| Linear momentum \( \mathbf{P} = M\dot{\mathbf{R}} \) | \( \dot{\mathbf{P}} = \mathbf{F}_{\text{ext}} \) | \( \mathbf{F}_{\text{ext}} = 0 \) |
| Angular momentum \( \mathbf{L} = \sum\mathbf{r}_i\times\mathbf{p}_i \) | \( \dot{\mathbf{L}} = \boldsymbol{\tau}_{\text{ext}} \) | \( \boldsymbol{\tau}_{\text{ext}} = 0 \) and strong third law |
| Energy \( T + V \) | \( dE/dt = \) power of non-conservative forces | all forces conservative |
Two decompositions are used constantly: \( \mathbf{L} = \mathbf{R}\times M\dot{\mathbf{R}} + \mathbf{L}_{\text{cm}} \) and \( T = \tfrac{1}{2}M\dot R^2 + T_{\text{cm}} \). The motion of the centre of mass and the motion about it separate.
Conservative forces
A force field is conservative if any one of these holds (they are equivalent on a simply connected region):
- \( \nabla\times\mathbf{F} = 0 \);
- \( \mathbf{F} = -\nabla V \) for some scalar \( V \);
- the work \( \oint\mathbf{F}\cdot d\mathbf{r} \) around every closed loop is zero.
A quick test: \( \mathbf{F} = (y, x, 0) \) has zero curl and \( V = -xy \); \( \mathbf{F} = (y, -x, 0) \) has curl \( (0, 0, -2) \) and is not conservative. Every central force \( \mathbf{F} = f(r)\hat{\mathbf{r}} \) is conservative.
Variable Mass: The Rocket
A rocket of mass \( m \) ejects exhaust at speed \( u \) relative to the rocket. Applying momentum conservation to rocket plus ejected mass over a time \( dt \):
$$m\frac{dv}{dt} = -u\frac{dm}{dt} + F_{\text{ext}}$$The term \( -u\,dm/dt \) is the thrust (positive, since \( dm/dt < 0 \)). In free space (\( F_{\text{ext}} = 0 \)), integrating gives the Tsiolkovsky rocket equation:
$$\boxed{\Delta v = u\ln\frac{m_0}{m_f}}$$The logarithm is the whole story of rocketry: doubling the speed gain needs the mass ratio squared, not doubled. In a uniform gravitational field fired vertically, add \( -gt \) to the right side.
One-Dimensional Motion: The Energy Method
For a particle in a 1D potential, \( E = \tfrac{1}{2}m\dot x^2 + V(x) \) is a first integral, so the motion reduces to a quadrature:
$$\dot x = \pm\sqrt{\frac{2}{m}\bigl(E - V(x)\bigr)}, \qquad t = \int\frac{dx}{\sqrt{\frac{2}{m}\bigl(E - V(x)\bigr)}}$$- Motion is allowed only where \( V(x) \leq E \). Points with \( V(x) = E \) are turning points.
- Between two turning points \( x_1 < x_2 \) the motion is periodic with period $$T(E) = \sqrt{2m}\int_{x_1}^{x_2}\frac{dx}{\sqrt{E - V(x)}}.$$
- Near a minimum \( x_0 \), \( V \approx V(x_0) + \tfrac{1}{2}V''(x_0)(x - x_0)^2 \), so small oscillations have $$\boxed{\omega = \sqrt{\frac{V''(x_0)}{m}}}$$ independent of amplitude. Larger amplitudes probe the anharmonic terms and the period changes (for the pendulum, it grows).
Dynamical Systems and Phase Space
A dynamical system is a set of first-order equations \( \dot{\mathbf{z}} = \mathbf{f}(\mathbf{z}, t) \). It is autonomous if \( \mathbf{f} \) does not depend on \( t \). Every Newtonian problem is one: \( m\ddot x = F(x, \dot x) \) becomes
$$\dot x = \frac{p}{m}, \qquad \dot p = F\!\left(x, \frac{p}{m}\right)$$The space of \( (x, p) \) (or \( (x, \dot x) \)) is phase space, and each solution is a curve in it, a trajectory. The full picture of trajectories is the phase portrait.
Three structural facts are tested repeatedly:
- Trajectories of an autonomous system never cross. Through each point passes exactly one trajectory, by uniqueness of solutions. A drawing with crossing curves is wrong, except at fixed points, which trajectories approach only asymptotically.
- Conservative 1D systems have trajectories that are level curves of \( E(x, p) = p^2/2m + V(x) \). They are traversed clockwise when \( x \) is horizontal and \( p \) vertical (for \( p > 0 \), \( x \) increases).
- Dimension limits the behaviour. In a 2D autonomous phase space trajectories can only approach fixed points, closed orbits or limit cycles (the Poincaré–Bendixson theorem). Chaos needs at least three phase-space dimensions, which a driven 1D oscillator supplies (time acts as the third).
Standard phase portraits
| System | \( H \) or equation | Phase trajectories |
|---|---|---|
| Free particle | \( p^2/2m \) | horizontal lines \( p = \) const |
| Constant force \( F \) | \( p^2/2m - Fx \) | parabolas \( x = p^2/2mF + \) const |
| Harmonic oscillator | \( p^2/2m + \tfrac{1}{2}m\omega^2x^2 \) | ellipses, area \( 2\pi E/\omega \) |
| Inverted oscillator | \( p^2/2m - \tfrac{1}{2}m\omega^2x^2 \) | hyperbolas, saddle at origin |
| Damped oscillator (underdamped) | \( \ddot x + \gamma\dot x + \omega_0^2x = 0 \) | spirals into the origin |
| Pendulum | \( p_\theta^2/2ml^2 - mgl\cos\theta \) | loops (libration), wavy lines (rotation), separatrix between |
The ellipse area deserves a line of algebra: semi-axes \( x_{\max} = \sqrt{2E/m\omega^2} \) and \( p_{\max} = \sqrt{2mE} \), so area \( = \pi x_{\max}p_{\max} = 2\pi E/\omega \). That area divided by \( 2\pi \) is the action \( J = E/\omega \), the quantity that stays constant when \( \omega \) is changed slowly and that Bohr–Sommerfeld quantisation sets equal to \( nh \) (giving \( E = nh\nu \)).
Fixed Points and Linear Stability
A fixed point \( \mathbf{z}^* \) satisfies \( \mathbf{f}(\mathbf{z}^*) = 0 \). Write \( \mathbf{z} = \mathbf{z}^* + \boldsymbol{\eta} \) and keep first-order terms:
$$\dot{\boldsymbol{\eta}} = J\boldsymbol{\eta}, \qquad J_{ij} = \left.\frac{\partial f_i}{\partial z_j}\right\rvert_{\mathbf{z}^*}$$The eigenvalues \( \lambda \) of the Jacobian \( J \) decide everything.
One dimension. \( \dot x = f(x) \): the fixed point is stable if \( f'(x^*) < 0 \), unstable if \( f'(x^*) > 0 \), and undecided by linearisation if \( f'(x^*) = 0 \).
Two dimensions. With \( \tau = \operatorname{tr}J = \lambda_1 + \lambda_2 \) and \( \Delta = \det J = \lambda_1\lambda_2 \), the eigenvalues are \( \lambda = \tfrac{1}{2}\left(\tau \pm \sqrt{\tau^2 - 4\Delta}\right) \):
| Condition | Eigenvalues | Fixed point |
|---|---|---|
| \( \Delta < 0 \) | real, opposite signs | saddle (always unstable) |
| \( \Delta > 0,\ \tau^2 > 4\Delta,\ \tau < 0 \) | real, both negative | stable node |
| \( \Delta > 0,\ \tau^2 > 4\Delta,\ \tau > 0 \) | real, both positive | unstable node |
| \( \Delta > 0,\ \tau^2 < 4\Delta,\ \tau < 0 \) | complex, negative real part | stable spiral (focus) |
| \( \Delta > 0,\ \tau^2 < 4\Delta,\ \tau > 0 \) | complex, positive real part | unstable spiral |
| \( \Delta > 0,\ \tau = 0 \) | purely imaginary | centre (linear) |
| \( \tau^2 = 4\Delta \) | repeated | degenerate (star) node |
graph TD
A[Fixed point: compute J, then trace and det] --> B{det J negative?}
B -->|yes| S[Saddle: unstable]
B -->|no| C{trace J zero?}
C -->|yes| D[Centre in the linear approximation]
C -->|no| E{trace squared exceeds 4 det?}
E -->|yes| F[Node: stable if trace negative]
E -->|no| G[Spiral: stable if trace negative]Conservative systems: read stability off the potential
For \( m\ddot x = -V'(x) \) the Jacobian at an equilibrium is \( \begin{pmatrix}0 & 1/m\\ -V''(x^*) & 0\end{pmatrix} \), with \( \tau = 0 \) and \( \Delta = V''(x^*)/m \). Therefore:
- minimum of \( V \) (\( V'' > 0 \)): centre, with \( \omega^2 = V''/m \);
- maximum of \( V \) (\( V'' < 0 \)): saddle, with growth rate \( \sqrt{-V''/m} \).
Hamiltonian flows can never have stable nodes or spirals: they preserve phase-space area (see Liouville below), and a node would squeeze a region to a point.
The pendulum in full
For \( \ddot\theta = -(g/l)\sin\theta \), with \( \omega_0^2 = g/l \):
- \( \theta = 0 \): \( V'' > 0 \), centre. Small oscillations at \( \omega_0 \).
- \( \theta = \pi \): \( V'' < 0 \), saddle. Departures grow as \( e^{\omega_0 t} \).
- The separatrix has the energy of the inverted pendulum at rest, \( E_s = 2mgl \) (measured from the bottom). Below it the pendulum librates; above it the pendulum rotates; on it the pendulum takes infinite time to reach the top.
On the separatrix \( \dot\theta = \pm2\omega_0\cos(\theta/2) \), which is why the period diverges logarithmically as \( E \to E_s \).
Damping turns centres into spirals
For \( \ddot x + \gamma\dot x + \omega_0^2x = 0 \), \( J = \begin{pmatrix}0 & 1\\ -\omega_0^2 & -\gamma\end{pmatrix} \), \( \tau = -\gamma \), \( \Delta = \omega_0^2 \):
| Damping | \( \tau^2 - 4\Delta \) | Origin is |
|---|---|---|
| Underdamped, \( \gamma < 2\omega_0 \) | \( < 0 \) | stable spiral |
| Critically damped, \( \gamma = 2\omega_0 \) | \( = 0 \) | degenerate stable node |
| Overdamped, \( \gamma > 2\omega_0 \) | \( > 0 \) | stable node |
| Negative damping, \( \gamma < 0 \) | unstable spiral or node |
Limit cycles and bifurcations (the minimum you need)
A limit cycle is an isolated closed orbit that neighbouring trajectories spiral towards (stable) or away from (unstable). It cannot occur in a conservative system, where closed orbits come in continuous families. The standard example is the van der Pol oscillator \( \ddot x - \mu(1 - x^2)\dot x + x = 0 \) with \( \mu > 0 \): negative damping for small \( \lvert x\rvert \) pumps energy in, positive damping for large \( \lvert x\rvert \) takes it out, and the balance gives a self-sustained oscillation.
A bifurcation is a qualitative change in the fixed points as a parameter is varied. For \( \dot x = r + x^2 \): two fixed points \( x^* = \pm\sqrt{-r} \) for \( r < 0 \) (the negative one stable), one at \( r = 0 \), none for \( r > 0 \). This is a saddle-node bifurcation. The bead on a rotating hoop in Rigid Bodies and Non-Inertial Frames shows a pitchfork bifurcation: the bottom of the hoop loses stability above a critical spin and two new equilibria appear.
Liouville’s Theorem
For a Hamiltonian system, the flow in phase space is incompressible:
$$\nabla\cdot\dot{\mathbf{z}} = \sum_i\left(\frac{\partial\dot q_i}{\partial q_i} + \frac{\partial\dot p_i}{\partial p_i}\right) = \sum_i\left(\frac{\partial^2H}{\partial q_i\partial p_i} - \frac{\partial^2H}{\partial p_i\partial q_i}\right) = 0$$so the phase-space volume occupied by an ensemble of systems is constant in time, though its shape can distort without limit. Equivalently, the density \( \rho(q, p, t) \) is constant along trajectories: \( d\rho/dt = 0 \).
With damping \( \gamma \), \( \nabla\cdot\dot{\mathbf{z}} = -\gamma \), and areas shrink as \( e^{-\gamma t} \). That is why dissipative systems have attractors and Hamiltonian ones do not.
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Weak third law (equal, opposite): momentum conserved | Strong third law (also central): angular momentum conserved | |
| Exhaust speed relative to the rocket (\( u \)) | Exhaust speed relative to the ground | |
| Centre (conservative minimum): closed orbits | Stable spiral (damped): orbits shrink to the point | |
| Saddle: unstable, \( \Delta < 0 \) | Unstable node: \( \Delta > 0 \), \( \tau > 0 \) | |
| Limit cycle: isolated, attracts neighbours | Conservative closed orbit: one of a continuous family | |
| Liouville: volume conserved | Shape conserved (false: it stretches and folds) |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper.
Level 1: Foundation
- Is \( \mathbf{F} = (2xy, x^2, 0) \) conservative? If so, give \( V \).
- For \( \dot x = x(1 - x) \), which fixed point is stable?
- What is the phase-space trajectory of a free particle in 1D?
Answers
- Yes. \( \nabla\times\mathbf{F} = (0, 0, 2x - 2x) = 0 \), and \( V = -x^2y \) (check: \( -\partial V/\partial x = 2xy \), \( -\partial V/\partial y = x^2 \)).
- \( x = 1 \). \( f'(x) = 1 - 2x \): \( f'(0) = 1 > 0 \) unstable, \( f'(1) = -1 < 0 \) stable.
- A horizontal straight line \( p = \) constant, traversed in the direction of \( p \).
Level 2: Part B Style
Practice Q1. The fixed point at the origin of the system \( \dot x = x - y \), \( \dot y = x + y \) is:
(a) a stable node (b) an unstable spiral (c) a saddle (d) a centre
Answer: (b) an unstable spiral
\( J = \begin{pmatrix}1 & -1\\ 1 & 1\end{pmatrix} \): \( \tau = 2 \), \( \Delta = 2 \), \( \tau^2 - 4\Delta = -4 < 0 \). The eigenvalues are \( 1 \pm i \): complex with positive real part, an unstable spiral. Not a saddle (\( \Delta > 0 \)); not a centre (\( \tau \neq 0 \)).
Practice Q2. A particle of mass \( m \) moves under a constant force \( F > 0 \) along \( x \). Its trajectories in the \( (x, p) \) plane are:
(a) straight lines (b) ellipses (c) parabolas (d) hyperbolas
Answer: (c) parabolas
\( E = p^2/2m - Fx \) is constant, so \( x = p^2/2mF - E/F \): parabolas opening towards \( +x \). Ellipses belong to the oscillator, hyperbolas to the inverted oscillator.
Practice Q3. A rocket in free space has exhaust speed 2 km/s relative to itself. When its mass has fallen to one quarter of the initial value, its gain in speed is closest to:
(a) 1.5 km/s (b) 2.8 km/s (c) 6.0 km/s (d) 8.0 km/s
Answer: (b) 2.8 km/s
\( \Delta v = 2\ln4 \approx 2.77 \) km/s. Option (c) comes from treating the gain as \( u(m_0/m_f - 1) \), and (d) from \( u\,m_0/m_f \): neither follows from the momentum balance, which integrates to a logarithm.
Practice Q4. A particle of mass \( m \) moves in \( V(x) = \alpha x^4 - \beta x^2 \) with \( \alpha, \beta > 0 \). The angular frequency of small oscillations about a stable equilibrium is:
(a) \( \sqrt{\beta/m} \) (b) \( 2\sqrt{\beta/m} \) (c) \( \sqrt{2\beta/m} \) (d) \( \sqrt{\beta/2m} \)
Answer: (b)
Minima at \( x^2 = \beta/2\alpha \), where \( V'' = 12\alpha x^2 - 2\beta = 4\beta \). Then \( \omega = \sqrt{4\beta/m} \). Option (c) uses \( V''(0) \) in magnitude, which belongs to the unstable maximum.
Level 3: Part C Style
Practice Q5. A unit-mass particle obeys \( \ddot x = -x + x^3 \). Consider the statements:
- The origin is a centre.
- The points \( x = \pm1 \), \( \dot x = 0 \) are saddles.
- The separatrix has energy \( E = 1/4 \), with \( E = \tfrac{1}{2}\dot x^2 + V(x) \) and \( V(0) = 0 \).
- A particle released from \( x = 0 \) with speed 0.8 stays bounded for all time.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 1, 2, 3 and 4 (d) 2 and 3 only
Answer: (b) 1, 2 and 3 only
The force is \( -V'(x) \) with \( V = \tfrac{1}{2}x^2 - \tfrac{1}{4}x^4 \). Equilibria: \( x - x^3 = 0 \), so \( x = 0, \pm1 \).
\( V''(x) = 1 - 3x^2 \): \( V''(0) = 1 > 0 \), a minimum, so the origin is a centre (genuinely, since energy is conserved). Statement 1 true. \( V''(\pm1) = -2 < 0 \), maxima, so saddles. Statement 2 true.
\( V(\pm1) = \tfrac{1}{2} - \tfrac{1}{4} = \tfrac{1}{4} \). The separatrix is the energy level through the saddles. Statement 3 true.
Starting at \( x = 0 \) with speed 0.8: \( E = 0.32 > 0.25 \), so the particle clears the hump at \( x = 1 \) and then \( V \to -\infty \): it escapes. Statement 4 false. (Speed below \( 1/\sqrt{2} \approx 0.707 \) would keep it bounded.)
Practice Q6. A 1 kg mass on a spring of constant 16 N/m moves in a medium with damping force \( -b\dot x \), \( b = 10 \) kg/s. Consider:
- The origin of the \( (x, \dot x) \) phase plane is a stable spiral.
- The eigenvalues of the linear system are \( -2 \) and \( -8 \) s\( ^{-1} \).
- At late times the displacement decays as \( e^{-2t} \) (with \( t \) in seconds).
- Reducing \( b \) to 8 kg/s makes the motion critically damped.
Which are correct?
(a) 1 and 2 only (b) 2, 3 and 4 only (c) 2 and 3 only (d) 1, 3 and 4 only
Answer: (b) 2, 3 and 4 only
The equation is \( \ddot x + 10\dot x + 16x = 0 \); characteristic roots \( \lambda^2 + 10\lambda + 16 = 0 \), so \( \lambda = -2, -8 \). Statement 2 true.
Both roots are real and negative: a stable node, not a spiral (\( \tau^2 - 4\Delta = 100 - 64 = 36 > 0 \)). Statement 1 false.
The slower mode \( e^{-2t} \) dominates at late times. Statement 3 true.
Critical damping needs \( b^2 = 4mk = 64 \), \( b = 8 \) kg/s. Statement 4 true.
Practice Q7. An ensemble of free particles of mass \( m \) occupies, at \( t = 0 \), the region \( 0 \leq x \leq a \), \( p_0 \leq p \leq p_0 + \Delta p \) of phase space. At time \( t \), which of the following is correct?
(a) The region is a rectangle of area \( a\,\Delta p \). (b) The region is a parallelogram of area \( a\,\Delta p \) and spatial extent \( a + \Delta p\,t/m \). (c) The region is a parallelogram of area \( (a + \Delta p\,t/m)\Delta p \). (d) The region shrinks in area as \( t \) increases because the beam spreads.
Answer: (b)
Each particle moves as \( x \to x + pt/m \) at fixed \( p \): the faster edge pulls ahead by \( \Delta p\,t/m \), so the rectangle shears into a parallelogram. The base (along \( x \) at fixed \( p \)) is still \( a \) and the height \( \Delta p \), so area \( a\,\Delta p \), as Liouville’s theorem demands. The total spatial extent is \( a + \Delta p\,t/m \). Option (c) multiplies the full extent by the height, which overcounts; (d) contradicts Liouville.
Quick Revision Table
| Idea | Result |
|---|---|
| Second law | \( \mathbf{F} = d\mathbf{p}/dt \) in an inertial frame |
| Conservative | \( \nabla\times\mathbf{F} = 0 \iff \mathbf{F} = -\nabla V \) |
| Rocket | \( \Delta v = u\ln(m_0/m_f) \) |
| Small oscillation | \( \omega^2 = V''(x_0)/m \) |
| Period in a well | \( T = \sqrt{2m}\int dx/\sqrt{E - V} \) |
| Fixed-point type | \( \Delta < 0 \) saddle; \( \tau < 0 \) stable; \( \tau^2 < 4\Delta \) spiral; \( \tau = 0 \) centre |
| Conservative system | minima centres, maxima saddles; no nodes or spirals |
| Pendulum separatrix | \( E_s = 2mgl \) above the bottom |
| Oscillator orbit | ellipse, area \( 2\pi E/\omega \) |
| Liouville | phase-space volume conserved for Hamiltonian flow |
| Damped flow | area shrinks as \( e^{-\gamma t} \) |
Key Takeaways
What’s Next
- Central Forces, Scattering and Collisions: the energy method applied to the effective radial potential.
- Lagrangian and Hamiltonian Mechanics, Conservation Laws: Hamilton’s equations as the dynamical system behind Liouville’s theorem.
- Small Oscillations and Normal Modes: linearisation about a minimum with many degrees of freedom.
- Classical Mechanics: Quick Revision.