Coherent States of the Harmonic Oscillator

The energy eigenstates of the quantum harmonic oscillator are magnificent and useless for one particular job: describing anything that actually swings. Every ∣n⟩|n\rangle is stationary — its probability cloud is frozen, its average position and momentum pinned at zero forever. A pendulum in a number state does not go back and forth. Schrödinger noticed the scandal as early as 1926 and asked for the opposite: a quantum state whose wave packet oscillates like a classical particle, without dissolving. The answer he built — rediscovered, generalized, and named by Roy Glauber in 1963 — is the coherent state. The recipe is almost insolent in its simplicity: take the ground state's Gaussian blob, pick it up, and set it down off-center. Because the oscillator's energy levels are perfectly evenly spaced, every Fock component of this displaced blob rotates in phase, and the packet swings through the well rigidly — same shape, same minimum uncertainty Δx Δp=ℏ/2\Delta x\,\Delta p = \hbar/2, at every instant, forever. It is the most classical object quantum mechanics permits. The chapter's oscillator articles — the analytic solution and Ladder Operators and the Algebraic Oscillator — built the operators a^,a^†\hat a, \hat a^\dagger and the states ∣n⟩|n\rangle; this closing article puts them to work. You will meet the coherent state three ways — eigenstate of a^\hat a, displaced vacuum, minimum-uncertainty packet — prove they are the same thing, watch the classical limit emerge as ∣α∣|\alpha| grows, and discover why the light from every laser you have ever seen is precisely this state.

Key takeaways
  • A coherent state ∣α⟩|\alpha\rangle is the eigenstate of the annihilation operator, a^∣α⟩=α∣α⟩\hat a|\alpha\rangle = \alpha|\alpha\rangle, with α\alpha any complex number encoding classical amplitude and phase.
  • In the number basis it is a Poissonian superposition: mean quantum number nˉ=∣α∣2\bar n = |\alpha|^2 and variance (Δn)2=∣α∣2(\Delta n)^2 = |\alpha|^2 — the statistics of ideal laser light and the origin of shot noise.
  • Equivalently, it is the vacuum displaced in phase space: ∣α⟩=D^(α)∣0⟩|\alpha\rangle = \hat D(\alpha)|0\rangle with D^(α)=eαa^†−α∗a^\hat D(\alpha) = e^{\alpha\hat a^\dagger - \alpha^*\hat a}.
  • It saturates Δx Δp=ℏ/2\Delta x\,\Delta p = \hbar/2 with the ground state's widths — and keeps saturating it forever: α(t)=α0e−iωt\alpha(t) = \alpha_0 e^{-i\omega t}, the centroid riding the exact classical trajectory, no spreading.
  • The family is overcomplete — never orthogonal (∣⟨β∣α⟩∣2=e−∣α−β∣2|\langle\beta|\alpha\rangle|^2 = e^{-|\alpha-\beta|^2}), yet resolving the identity: 1π∫d2α ∣α⟩⟨α∣=1^\tfrac1\pi\int d^2\alpha\,|\alpha\rangle\langle\alpha| = \hat 1.
  • As ∣α∣→∞|\alpha| \to \infty the relative fluctuations Δn/nˉ=1/∣α∣\Delta n/\bar n = 1/|\alpha| vanish — the classical oscillator, and the classical light wave, emerge smoothly.
  • Key idea: a coherent state is the ground-state blob displaced in phase space; equal level spacing makes it rotate rigidly, so it rides the classical trajectory with permanently minimal uncertainty.
The core insight

A coherent state is the vacuum's ground-state Gaussian pushed off-center in phase space by the displacement operator; because the oscillator's levels are equally spaced (ℏω\hbar\omega apart), this displaced blob rotates rigidly about the origin, reproducing classical motion with permanently minimal, non-spreading uncertainty.

Where this lives today

The coherent state is the reference point of modern quantum technology — the "classical" against which every quantum advantage is measured. Gravitational-wave detectors are the grandest example: LIGO's sensitivity is capped by the shot noise of the coherent laser field in its arms, Δn=nˉ\Delta n = \sqrt{\bar n} from equation (3), and since 2019 the observatories inject squeezed vacuum — a coherent-state blob deliberately deformed — to slip below that Poisson floor. Superconducting quantum computers speak coherent states natively: every microwave readout pulse is one, and "cat qubits" — superpositions ∣α⟩+∣−α⟩|\alpha\rangle + |-\alpha\rangle of well-separated blobs — are engineered as error-protected logical qubits precisely because equation (10) makes the components as distinguishable as the overlap allows. In ion traps, coherent states of motion like Example 2's are prepared and interfered daily — quantum-limited force sensing, where an unknown push reveals itself as a measurable displacement α\alpha.

The physical picture

Draw phase space: ⟨x^⟩\langle\hat x\rangle across, ⟨p^⟩\langle\hat p\rangle up. A classical oscillator is a point circling the origin at angular frequency ω\omega. The quantum ground state ∣0⟩|0\rangle is not a point but a fuzzy blob of area ∼ℏ\sim\hbar parked at the origin — minimum uncertainty (Zero-Point Energy proved it the unique such Gaussian), going nowhere. The coherent state is that same blob picked up and placed at the classical point (x0,p0)(x_0, p_0). Nothing about its shape changes; only its address. Now let it evolve. Each Fock component ∣n⟩|n\rangle acquires phase e−i(n+1/2)ωte^{-i(n+1/2)\omega t}, and because the levels are equally spaced these phases advance in lockstep — the superposition never dephases against itself. The blob simply circles the origin at ω\omega, rigid as a carousel horse. Its center obeys Newton exactly; its fuzz stays the ground state's fuzz. Contrast the cruelty of a free particle's Gaussian, which spreads irreversibly (Wave Packet Spreading and Dispersion quantified the σx(t)\sigma_x(t) law): the oscillator's perfect level ladder is what holds the coherent packet together.

Figure
Phase space: the vacuum blob, displaced, orbits like a classical point
⟨x⟩ ⟨p⟩ vacuum |0⟩: blob at rest D(α) |α⟩: same blob, new address ω classical phase point rides the same dashed orbit rigid rotation: the blob never deforms — Δx·Δp = ℏ/2 at every instant

The same rigidity, seen in position space rather than phase space: the probability cloud is the ground state's Gaussian, sliding back and forth in the well with its center riding the classical trajectory. Half a period later it is the identical blob at the opposite turning point — nothing has spread.

Figure
A packet that swings without spreading
V(x) = ½mω²x² classical energy t = 0 t = T/2 same Gaussian, swinging at ω — |ψ|² never spreads

Why Poisson statistics? Because ∣α⟩|\alpha\rangle is an eigenstate of the annihilation operator: removing one quantum leaves the state unchanged. Detection events are therefore memoryless — each photon click carries no news about the next — and memoryless counting is, by definition, Poissonian. A laser beam's graininess (shot noise) is this arithmetic heard on a photodiode.

Three definitions, one state

Start with the face Glauber made fundamental. A classical oscillation is specified by two data — amplitude and phase — which pack naturally into a single complex number; the quantum object carrying the same information is the annihilation operator a^\hat a, the oscillator's complex combination of x^\hat x and p^\hat p inherited from Ladder Operators and the Algebraic Oscillator. Demanding a state that a^\hat a merely rescales, rather than degrades, gives the defining eigenvalue equation:

a^ ∣α⟩=α ∣α⟩,α∈C(1)\hat a\,|\alpha\rangle = \alpha\,|\alpha\rangle, \qquad \alpha \in \mathbb{C} \qquad (1)

where: a^\hat a — the (non-Hermitian) annihilation operator; α\alpha — a complex eigenvalue whose modulus sets the amplitude and whose argument sets the oscillation phase. Non-Hermitian operators may have complex eigenvalues, and a^\hat a obligingly has one for every complex number (a^†\hat a^\dagger, by contrast, has none). What does this state look like on the energy ladder? Project the eigenvalue condition onto the number basis and a one-step recurrence appears — each coefficient is α/n\alpha/\sqrt{n} times the one below it — which telescopes all the way down to the ground level. Normalizing the resulting series gives the explicit recipe:

∣α⟩=e−∣α∣2/2∑n=0∞αnn! ∣n⟩(2)|\alpha\rangle = e^{-|\alpha|^2/2}\sum_{n=0}^{\infty}\frac{\alpha^n}{\sqrt{n!}}\,|n\rangle \qquad (2)

where: ∣n⟩|n\rangle — the oscillator's energy eigenstates. Every rung of the ladder participates — this is emphatically not an energy eigenstate — with weights that climb to a peak near n=∣α∣2n = |\alpha|^2 and then die off as the factorial takes over. Squaring those coefficients answers the question a photodetector actually asks: how many quanta will I count in this state? The answer lands on the most famous distribution in counting statistics:

P(n)=∣⟨n∣α⟩∣2=e−∣α∣2∣α∣2nn!,nˉ=∣α∣2,(Δn)2=∣α∣2(3)P(n) = |\langle n|\alpha\rangle|^2 = e^{-|\alpha|^2}\frac{|\alpha|^{2n}}{n!}, \qquad \bar n = |\alpha|^2, \quad (\Delta n)^2 = |\alpha|^2 \qquad (3)

where: nˉ\bar n — mean quantum number; the variance equals the mean, the Poisson fingerprint. This is the memoryless counting promised in the physical picture, and its nˉ\sqrt{\bar n} width is the shot noise on every photodiode trace. The third face explains where such a state comes from in a laboratory. Nothing exotic is required: take the vacuum and push it — apply the unitary that translates a state across phase space, the quantum version of giving a resting pendulum one clean shove. The pushed vacuum is the coherent state:

∣α⟩=D^(α) ∣0⟩,D^(α)=eαa^†−α∗a^(4)|\alpha\rangle = \hat D(\alpha)\,|0\rangle, \qquad \hat D(\alpha) = e^{\alpha\hat a^\dagger - \alpha^*\hat a} \qquad (4)

where: D^(α)\hat D(\alpha) — the unitary displacement operator, the oscillator's version of a phase-space translation ei(p0x^−x0p^)/ℏe^{i(p_0\hat x - x_0\hat p)/\hbar}; it moves ⟨x^⟩\langle\hat x\rangle and ⟨p^⟩\langle\hat p\rangle but leaves every variance untouched. That is the experimental recipe in one line: a resonant laser drive, an RF push on a trapped ion, a microwave pulse on a cavity — each is an implementation of D^(α)\hat D(\alpha). The Baker–Campbell–Hausdorff identity (using only [a^,a^†]=1[\hat a, \hat a^\dagger] = 1) splits the push into normal order, D^(α)=e−∣α∣2/2 eαa^†e−α∗a^\hat D(\alpha) = e^{-|\alpha|^2/2}\,e^{\alpha\hat a^\dagger}e^{-\alpha^*\hat a} — act on the vacuum and expansion (2) drops straight out: faces (1), (2), and (4) are one state. Finally, set the state swinging and ask whether it survives its own evolution. Run the Schrödinger equation through the expansion (2): each ∣n⟩|n\rangle acquires the phase e−i(n+1/2)ωte^{-i(n+1/2)\omega t}, and because the levels are equally spaced every factor can be absorbed into a redefinition of the label:

∣α(t)⟩=e−iωt/2 ∣α0e−iωt⟩(5)|\alpha(t)\rangle = e^{-i\omega t/2}\,\big|\alpha_0 e^{-i\omega t}\big\rangle \qquad (5)

where: the overall phase e−iωt/2e^{-i\omega t/2} is unobservable; the physics is that a coherent state stays coherent, with α\alpha circling the phase-space origin at ω\omega — the rigid carousel rotation of the hero diagram, now proved rather than pictured. One state, three costumes — and each earns its keep on different problems:

faceequationreach for it when
eigenstate of a^\hat a(1)computing moments — every a^\hat a acts sideways as a number
Poisson ladder sum(2)–(3)anything about energy or photon counting
displaced vacuum(4)–(5)state preparation, phase-space pictures, time evolution
Derivation AMinimum uncertainty for every alpha and for all time

We compute the position and momentum spreads of an arbitrary coherent state and show Δx Δp=ℏ/2\Delta x\,\Delta p = \hbar/2 exactly — the Heisenberg floor — independent of α\alpha, and hence (via the rotating label of equation 5) at every instant of the motion. The strategy is pure ladder bookkeeping: write x^,p^\hat x, \hat p in terms of a^,a^†\hat a, \hat a^\dagger, and let the eigenvalue equation turn every operator average into arithmetic with the number α\alpha. The step to watch is the single +1+1 that refuses to cancel — the commutator's irreducible contribution, which is the entire quantum content of the result.

Step 1. Express the observables in ladder language: x^=ℏ2mω(a^+a^†)\hat x = \sqrt{\tfrac{\hbar}{2m\omega}}(\hat a + \hat a^\dagger) and p^=iℏmω2(a^†−a^)\hat p = i\sqrt{\tfrac{\hbar m\omega}{2}}(\hat a^\dagger - \hat a). Why: every oscillator moment reduces to bookkeeping with a^\hat a and a^†\hat a^\dagger — the algebra owned by the ladder-operator article.

Step 2. Take the means. Using (1) and its adjoint ⟨α∣a^†=α∗⟨α∣\langle\alpha|\hat a^\dagger = \alpha^*\langle\alpha|: ⟨x^⟩=ℏ2mω(α+α∗)\langle\hat x\rangle = \sqrt{\tfrac{\hbar}{2m\omega}}(\alpha + \alpha^*), ⟨p^⟩=iℏmω2(α∗−α)\langle\hat p\rangle = i\sqrt{\tfrac{\hbar m\omega}{2}}(\alpha^* - \alpha). Why: the eigenvalue equation lets every a^\hat a act sideways as a number.

Step 3. Take the second moment of x^\hat x, respecting operator order:

⟨x^2⟩=ℏ2mω⟨a^2+a^†2+a^†a^+a^a^†⟩=ℏ2mω[(α+α∗)2+1](6)\langle\hat x^2\rangle = \frac{\hbar}{2m\omega}\big\langle \hat a^2 + \hat a^{\dagger 2} + \hat a^\dagger\hat a + \hat a\hat a^\dagger \big\rangle = \frac{\hbar}{2m\omega}\left[(\alpha + \alpha^*)^2 + 1\right] \qquad (6)

Why the crucial +1+1: it is the commutator at work — a^a^†=a^†a^+1\hat a\hat a^\dagger = \hat a^\dagger\hat a + 1 — the irreducible quantum fuzz that no displacement can remove.

Step 4. Subtract the squared mean. (Δx)2=⟨x^2⟩−⟨x^⟩2=ℏ2mω(\Delta x)^2 = \langle\hat x^2\rangle - \langle\hat x\rangle^2 = \dfrac{\hbar}{2m\omega}. Every trace of α\alpha cancels: the width is exactly the ground state's width. Why: displacement moves the blob; only the commutator sets its size.

Step 5. Repeat for momentum and multiply. (Δp)2=ℏmω2(\Delta p)^2 = \tfrac{\hbar m\omega}{2}, so

Δx Δp=ℏ2(7)\Delta x\,\Delta p = \frac{\hbar}{2} \qquad (7)

Why this holds forever: equation (5) keeps the state coherent with ∣α(t)∣|\alpha(t)| merely rotating, and nothing in Steps 3–5 cared about α\alpha's value. The coherent state is a minimum-uncertainty state that stays one. *Verified: numerically in a 200-level Fock basis: Δx Δp=0.500000 ℏ\Delta x\,\Delta p = 0.500000\,\hbar at α=1.3−0.7i\alpha = 1.3-0.7i; matches Sakurai §2.3 and Cohen-Tannoudji G_V.*

Go deeper · Derivation B · 1 minThe centroid obeys Newton exactly
Derivation BThe centroid obeys Newton exactly

We show that the coherent state's mean position executes the classical trajectory xcl(t)x_{\rm cl}(t) — not approximately, but exactly — so that together with the non-spreading width of the previous card, the state is the literal quantum shadow of a swinging pendulum. The strategy: insert the rotating label of equation (5) into the mean-value formulas and differentiate. Watch how the complex number α0\alpha_0 turns out to be precisely the classical amplitude-and-phase, packaged.

Step 1. Insert the rotating label (5) into the mean position from the previous card: ⟨x^⟩(t)=ℏ2mω(α0e−iωt+α0∗eiωt)\langle\hat x\rangle(t) = \sqrt{\tfrac{\hbar}{2m\omega}}\big(\alpha_0 e^{-i\omega t} + \alpha_0^* e^{i\omega t}\big). Why: time evolution costs nothing but α→αe−iωt\alpha \to \alpha e^{-i\omega t}.

Step 2. Write α0=∣α0∣eiϕ\alpha_0 = |\alpha_0|e^{i\phi} and combine the conjugates:

⟨x^⟩(t)=2ℏmω ∣α0∣cos⁡(ωt−ϕ)(8)\langle\hat x\rangle(t) = \sqrt{\frac{2\hbar}{m\omega}}\,|\alpha_0|\cos(\omega t - \phi) \qquad (8)

Why: a pure cosine — amplitude set by ∣α0∣|\alpha_0|, phase by arg⁡α0\arg\alpha_0.

Step 3. Differentiate for the momentum: ⟨p^⟩(t)=−2ℏmω ∣α0∣sin⁡(ωt−ϕ)=m d⟨x^⟩/dt\langle\hat p\rangle(t) = -\sqrt{2\hbar m\omega}\,|\alpha_0|\sin(\omega t - \phi) = m\,\mathrm{d}\langle\hat x\rangle/\mathrm{d}t. Why: Ehrenfest's Theorem holds exactly here — for quadratic potentials the quantum averages obey the classical equations with no correction at all.

Step 4. Differentiate once more: d2⟨x^⟩/dt2=−ω2⟨x^⟩\mathrm{d}^2\langle\hat x\rangle/\mathrm{d}t^2 = -\omega^2\langle\hat x\rangle. Why: the centroid solves the classical oscillator equation; with the never-spreading width, the correspondence principle is realized as an identity rather than a limit. *Verified: numerically — evolving ∣α⟩|\alpha\rangle with the exact e−iHte^{-iHt} in the truncated Fock basis reproduces e−iωt/2∣α0e−iωt⟩e^{-i\omega t/2}|\alpha_0e^{-i\omega t}\rangle to 5×10−165\times10^{-16} and ⟨x^⟩(t)=2 ∣α0∣cos⁡(ϕ−ωt)\langle\hat x\rangle(t) = \sqrt2\,|\alpha_0|\cos(\phi - \omega t) (units ℏ=m=ω=1\hbar{=}m{=}\omega{=}1) to 10−610^{-6}.*

The number statistics come straight from expansion (2): projecting onto ∣n⟩|n\rangle and squaring gives the Poisson law of equation (3) — and forming the relative fluctuation delivers the classical limit as a one-line dial:

Δnnˉ=∣α∣∣α∣2=1∣α∣=1nˉ(9)\frac{\Delta n}{\bar n} = \frac{|\alpha|}{|\alpha|^2} = \frac{1}{|\alpha|} = \frac{1}{\sqrt{\bar n}} \qquad (9)

where: brightness buys classicality — at nˉ=1016\bar n = 10^{16} photons per second in a laser pointer the graininess is one part in 10810^8. The two ends of the dial: α→0\alpha \to 0 is the vacuum itself; ∣α∣→∞|\alpha| \to \infty has fluctuations vanishing relative to amplitude, the definite classical oscillation emerging smoothly. (Numerically verified: the Fock-basis coherent vector at α=1.3−0.7i\alpha = 1.3-0.7i has nˉ=(Δn)2=2.1800\bar n = (\Delta n)^2 = 2.1800 to 10−610^{-6}; ⟨E⟩=(∣α∣2+12)ℏω\langle E\rangle = (|\alpha|^2{+}\tfrac12)\hbar\omega and ΔE=ℏω∣α∣\Delta E = \hbar\omega|\alpha| exactly.)

Overcomplete on purpose — the family as a (too-rich) basis

Coherent states refuse to behave like a textbook basis, productively. First, no two are ever orthogonal — expansion (2) gives the exact overlap

∣⟨β∣α⟩∣2=e−∣α−β∣2(10)|\langle\beta|\alpha\rangle|^2 = e^{-|\alpha-\beta|^2} \qquad (10)

where: the overlap decays with the phase-space distance between the blobs — never zero, but below 2%2\% once ∣α−β∣=2|\alpha - \beta| = 2: distinguishable in practice, never in principle. Second — despite the redundancy, or rather because of it — the family spans everything:

1π∫d2α  ∣α⟩⟨α∣=1^(11)\frac{1}{\pi}\int d^2\alpha\;|\alpha\rangle\langle\alpha| = \hat{\mathbb 1} \qquad (11)

where: the integral runs over the complex plane and the 1/π1/\pi is fixed by Gaussian integrals (diagonal elements check exactly: 1π∫d2α ∣⟨n∣α⟩∣2=1\tfrac1\pi\int d^2\alpha\,|\langle n|\alpha\rangle|^2 = 1 for every nn). Equation (11) is an identity resolution built from non-orthogonal states — an "overcomplete" frame — and it licenses the phase-space representations of quantum optics: any field state expands over coherent states (the Glauber–Sudarshan PP-representation), and how badly that expansion misbehaves is the modern diagnostic line between classical-like and genuinely quantum light.

Limiting cases

CaseResultMeaning
α→0\alpha \to 0∣α=0⟩=∣0⟩\lvert\alpha=0\rangle = \lvert 0\ranglezero displacement is the vacuum itself; nˉ=0\bar n = 0
∣α∣→∞\lvert\alpha\rvert \to \inftyΔn/nˉ=1/∣α∣→0\Delta n/\bar n = 1/\lvert\alpha\rvert \to 0fluctuations vanish relative to amplitude — the classical wave emerges
Fock state at the same nˉ\bar n(Δn)2=0(\Delta n)^2 = 0 but ⟨x^⟩=0\langle\hat x\rangle = 0 alwayssharp energy, no motion — the anti-coherent extreme
two coherent states∣⟨β∣α⟩∣2=e−∣α−β∣2\lvert\langle\beta\vert\alpha\rangle\rvert^2 = e^{-\lvert\alpha-\beta\rvert^2}never exactly orthogonal, effectively distinguishable once the blobs separate by ≳1\gtrsim 1
free particle instead of oscillatorpacket spreadsno equal level spacing, no rigidity — the ladder is what preserves the shape

Worked examples

Example 1 — the laser pointer, counted photon by photon (confidence builder). A red laser pointer emits P=1.0 mWP = 1.0\ \mathrm{mW} at λ=650 nm\lambda = 650\ \mathrm{nm}. For a 1.0 ns1.0\ \mathrm{ns} counting window, find the mean photon number, the coherent-state amplitude ∣α∣|\alpha|, and the shot-noise fluctuation.

Step 1. Energy per photon: E=hc/λ=(6.63×10−34)(3.0×108)/(650×10−9)=3.06×10−19 JE = hc/\lambda = (6.63\times10^{-34})(3.0\times10^{8})/(650\times10^{-9}) = 3.06\times10^{-19}\ \mathrm{J}.

Step 2. Photon rate: R=P/E=(1.0×10−3)/(3.06×10−19)=3.3×1015R = P/E = (1.0\times10^{-3})/(3.06\times10^{-19}) = 3.3\times10^{15} photons per second.

Step 3. Mean number in the window: nˉ=Rτ=(3.3×1015)(1.0×10−9)=3.3×106\bar n = R\tau = (3.3\times10^{15})(1.0\times10^{-9}) = 3.3\times10^{6}, so ∣α∣=nˉ≈1.8×103|\alpha| = \sqrt{\bar n} \approx \boxed{1.8\times10^{3}}.

Step 4. Shot noise from (3): Δn=nˉ≈1.8×103\Delta n = \sqrt{\bar n} \approx 1.8\times10^{3} photons — an absolute jitter of thousands, but a relative jitter of Δn/nˉ=1/∣α∣≈5.5×10−4\Delta n/\bar n = 1/|\alpha| \approx \boxed{5.5\times10^{-4}}. A photodiode fast enough sees exactly this Poissonian graininess; slower electronics average it into the steady beam you perceive.

Example 2 — a trapped ion swinging coherently (exam level). A single 40^{40}Ca+^+ ion (m=6.6×10−26 kgm = 6.6\times10^{-26}\ \mathrm{kg}) sits in a trap of frequency ω=2π×1.0 MHz\omega = 2\pi\times1.0\ \mathrm{MHz}. A resonant push prepares the coherent state with ∣α∣=10|\alpha| = 10. Find the ground-state width, the oscillation amplitude, the mean phonon number, and the energy spread.

Step 1. Ground-state (and coherent-state) width from (7): Δx=ℏ2mω=1.055×10−342(6.6×10−26)(6.28×106)=11 nm\Delta x = \sqrt{\dfrac{\hbar}{2m\omega}} = \sqrt{\dfrac{1.055\times10^{-34}}{2(6.6\times10^{-26})(6.28\times10^{6})}} = \boxed{11\ \mathrm{nm}}.

Step 2. Oscillation amplitude from (8): xmax⁡=2ℏmω ∣α∣=2 Δx ∣α∣=2(1.1×10−8)(10)=0.23 μmx_{\max} = \sqrt{\dfrac{2\hbar}{m\omega}}\,|\alpha| = 2\,\Delta x\,|\alpha| = 2(1.1\times10^{-8})(10) = \boxed{0.23\ \mathrm{\mu m}} — the packet swings through twenty times its own width: visibly "classical" motion on the trap's scale.

Step 3. Phonon statistics: nˉ=∣α∣2=100\bar n = |\alpha|^2 = 100, with Δn=∣α∣=10\Delta n = |\alpha| = 10 — each experimental run counts a slightly different number of phonons, Poisson-distributed about 100.

Step 4. Energy: ⟨E⟩=ℏω(nˉ+12)=100.5 ℏω\langle E\rangle = \hbar\omega(\bar n + \tfrac12) = 100.5\,\hbar\omega, with spread ΔE=ℏω∣α∣=10 ℏω\Delta E = \hbar\omega|\alpha| = 10\,\hbar\omega. The 10% energy uncertainty is the price of having a moving packet — sharp energy and motion are incompatible, exactly as the warning box insists.

Step 5. Cross-check: amplitude over width =2∣α∣=20= 2|\alpha| = 20 ✓, independent of every trap parameter — a signature a lab verifies by imaging the ion's smeared trajectory.

Origins · 1 minRoy J. Glauber and how this was discovered
RJ
Roy J. Glauber
1925–2018 · Harvard

When the laser arrived in 1960, optics faced an embarrassment: its theory of coherence was entirely classical, and nobody could say what distinguished laser light from starlight at the quantum level. Glauber — who as an 18-year-old had computed critical masses at Los Alamos — answered in 1963 with the quantum theory of optical coherence. He resurrected Schrödinger's non-spreading oscillator packets, named them coherent states, showed they are the eigenstates of the photon annihilation operator (hence the states a photodetector, which destroys photons, naturally resolves), and proved that an ideal laser emits precisely them. His PP-representation expands any field state over coherent states, drawing the sharpest available line between classical-like and genuinely quantum light. The 2005 Nobel Prize arrived four decades later — with the annual duty, which he performed for years, of sweeping paper airplanes off the Ig Nobel stage.

History
1926
— Schrödinger constructs non-spreading Gaussian packets of the harmonic oscillator, seeking states that obey the correspondence principle exactly
1960
— Maiman's ruby laser produces the first coherent optical field, making the quantum theory of coherence an urgent question
1963
— Glauber (and, in parallel, Sudarshan) builds the quantum theory of optical coherence on coherent states; the modern name and formalism date here
1985
— Slusher's group at Bell Labs generates squeezed light, deforming the coherent state's circular blob and beating shot noise in one quadrature
2005
— Glauber shares the Nobel Prize in Physics for the quantum theory of optical coherence

Going further

References. Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., §2.3 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. I, Complement G_V · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 7 & 21 · Glauber, Phys. Rev. 131, 2766 (1963) · Klauder & Skagerstam, Coherent States (World Scientific, 1985).

Provenance. All equations and numbers are standard textbook results. A verifier should: (1) confirm the BCH disentangling of D^(α)\hat D(\alpha) requires [a^,a^†]=1[\hat a,\hat a^\dagger] = 1 (Sakurai §2.3); (2) reproduce Δx Δp=ℏ/2\Delta x\,\Delta p = \hbar/2 via normal ordering (Cohen-Tannoudji Complement G_V); (3) verify the Poisson statistics nˉ=(Δn)2=∣α∣2\bar n = (\Delta n)^2 = |\alpha|^2 and the overlap ∣⟨β∣α⟩∣2=e−∣α−β∣2|\langle\beta|\alpha\rangle|^2 = e^{-|\alpha-\beta|^2} (Glauber 1963); (4) recompute both worked examples (nˉ=3.3×106\bar n = 3.3\times10^{6}, ∣α∣≈1.8×103|\alpha| \approx 1.8\times10^{3}; Δx≈11 nm\Delta x \approx 11\ \mathrm{nm}, xmax⁡≈0.23 μmx_{\max} \approx 0.23\ \mathrm{\mu m}, nˉ=100\bar n = 100, ΔE=10 ℏω\Delta E = 10\,\hbar\omega). For this edition, all four items plus the grafted additions were re-executed numerically and symbolically (numpy 200-level Fock matrices + sympy): eigenstate relation, Poisson moments, ΔxΔp=ℏ/2\Delta x\Delta p = \hbar/2 at complex α\alpha, D(α)∣0⟩=∣α⟩D(\alpha)|0\rangle = |\alpha\rangle and its BCH split, exact rigid rotation under e−iHte^{-iHt}, the overlap law, the resolution-of-identity diagonals n=0…3n = 0\ldots3, ΔE=ℏω∣α∣\Delta E = \hbar\omega|\alpha|, and every laser-pointer and trapped-ion number — the v3 spine's figures all reverified clean, no corrections required. This article folds the v3 pilot spine with the AllQFT base (overcompleteness, BCH, exact overlap grafted).

Builds on: Quantum Harmonic Oscillator: Analytic Solution · Ladder Operators and the Algebraic Oscillator · Zero-Point Energy · Wave Packets and Their Construction

Leads to: Wave Packet Spreading and Dispersion · Hermite Polynomials and Oscillator Eigenfunctions

Is laser light literally in a coherent state, or is that an idealization?

An idealization in the same sense a plane wave is — and an excellent one. An ideal single-mode laser far above threshold emits photon statistics and phase behaviour matching a coherent state (with a slowly diffusing overall phase); real lasers add technical noise on top. The claim with teeth is the floor: nothing done to a laser's classical controls takes its intensity noise below the Poisson nˉ\sqrt{\bar n} — only genuinely nonclassical (squeezed) light does that.

Why does the annihilation operator have eigenstates but the creation operator none?

Run the recurrence of expansion (2) for a^†∣ψ⟩=β∣ψ⟩\hat a^\dagger|\psi\rangle = \beta|\psi\rangle: the coefficients link upward with no ground floor to anchor on, and the resulting series has divergent norm — no normalizable solution exists. Physically: removing a quantum from a Poisson pile leaves it statistically unchanged, but adding one always changes it (P(0)P(0) becomes exactly zero, a detectable fingerprint). Memorylessness works in one direction only.

If coherent states aren't orthogonal, how can equation (11) be a completeness relation?

Orthogonality is sufficient for a resolution of identity, not necessary. The family is a frame: too many states, each overlapping its neighbours, whose weighted outer products still sum to the identity — like reconstructing a photo from overlapping blurry tiles. The price is non-uniqueness of expansions, whose coefficient functions can be wilder than any wavefunction (the P-representation of squeezed light is a genuine distribution). That wildness is diagnostic gold: "no well-behaved P-function" is a standard definition of nonclassical light.

What exactly is squeezed light, in the blob picture?

Take the coherent state's circular blob and deform it into an ellipse: quieter in one quadrature, noisier in the conjugate one — Heisenberg's product respected, the fuzz's distribution engineered. Detection aligned with the thin axis then beats shot noise, which is precisely how LIGO gains sensitivity. Squeezing comes from nonlinear optics (parametric down-conversion); the coherent state is the boundary case — the round, unsqueezed, "most classical" member of the minimum-uncertainty family.

Does the coherent state stay coherent if the oscillator is damped or driven?

Remarkably, yes — their deepest stability property. A classical drive just adds to α\alpha, and even damping (photon loss) maps ∣α⟩→∣αe−γt/2⟩|\alpha\rangle \to |\alpha e^{-\gamma t/2}\rangle — still coherent, just dimmer, with no entanglement generated with the environment. That is why coherent states are the "pointer states" of the electromagnetic field: decoherence selects them. Fock and cat states degrade dramatically under the same loss — the fragility that makes bosonic error correction hard.

Is a coherent state of a single photon possible? What does α = 1 mean?

∣α=1⟩|\alpha = 1\rangle has mean photon number one, but it is not "a photon": P(0)=37%P(0) = 37\%, P(1)=37%P(1) = 37\%, P(2)=18%P(2) = 18\%… A true single-photon state is the Fock state ∣1⟩|1\rangle — definite number, zero coherent amplitude, maximally nonclassical by the P-function test. Different design briefs: coherent states carry phase (communications, interferometry); Fock states carry countability (cryptography, metrology). Confusing them is the most common quantum-optics error in journalism.

Before this

Where this leads