Geometric and Wave Optics: /20 Summary

Context and scope

the supplied physics reference, Applications of Classical Physics — Chapters 7–10 Geometric Optics · Diffraction · Interferometry & Coherence · Nonlinear Optics

/20 Summary

• Three regimes of wave propagation, distinguished by the ratio of wavelength $\lambda$ to relevant length scale $L$:
  • $\lambda \ll L$ ⇒ Geometric (ray) optics (eikonal limit, WKB)
  • $\lambda \sim L$ ⇒ Diffraction / wave optics (Helmholtz, Fresnel-Kirchhoff)
  • $\lambda \gg L$ ⇒ Lumped/long-wavelength limit (not central to optics)
• Three categories of phenomena the chapter develops:
  1. Propagation (geometric optics, eikonal, paraxial)
  2. Coherent superposition (diffraction, interference, holography)
  3. Statistical superposition (coherence theory, V-C-Z theorem, intensity interferometry)
• Nonlinear optics = breakdown of linear superposition when $E \sim E_{\text{atomic}}$; generates harmonics, mixes frequencies, enables solitons and parametric amplification.
• Universal toolkit:
  • Eikonal $|\nabla\psi|^2 = n^2$ for rays
  • Fresnel-Kirchhoff integral for diffraction
  • Wiener-Khinchin / Van Cittert-Zernike for coherence
  • $\chi^{(n)}$ susceptibility expansion for nonlinearity
• Practical payoff: explains imaging resolution, telescopes & microscopes, lasers, fiber optics, holography, LIGO, parametric oscillators, frequency combs.

Master Map

mindmap
  root((Optics))
    Geometric Optics (Ch.7)
      Eikonal equation
      Ray equation
      Fermat principle
      Snell's law
      Hamiltonian rays
      Polarization transport
      Paraxial / Gaussian beams
      Caustics
      WKB
    Diffraction (Ch.8)
      Helmholtz equation
      Huygens-Fresnel
      Fresnel-Kirchhoff
      Fraunhofer
      Fresnel zones
      Airy, Rayleigh
      Fourier optics
      Babinet
    Coherence (Ch.9)
      Temporal coherence
      Spatial coherence
      Mutual coherence Γ
      Van Cittert-Zernike
      Michelson stellar
      Fabry-Perot
      HBT intensity interferometry
      Photon statistics
      Holography
    Nonlinear (Ch.10)
      χ⁽ⁿ⁾ expansion
      SHG, sum/diff freq
      Phase matching
      Parametric amp
      Four-wave mixing
      Kerr / Self-focusing
      Solitons
      Raman, Brillouin

Geometric Optics & the Eikonal Approximation

From Wave Equation to Eikonal

Start from scalar wave equation (or component of Maxwell) in slowly-varying medium $n(\mathbf{x})$:

$$\nabla^2 \psi - \frac{n^2}{c^2}\frac{\partial^2 \psi}{\partial t^2} = 0$$

Eikonal ansatz (WKB):

$$\psi(\mathbf{x}, t) = A(\mathbf{x})\, e^{i[k_0 S(\mathbf{x}) - \omega t]}, \quad k_0 = \omega/c$$

Substitute, separate large-$k_0$ and finite terms:

Validity: $|\nabla n|/n \ll k_0 n$, i.e., $n$ varies over scales $\gg \lambda$.

Rays as Trajectories

Definition: ray = curve everywhere parallel to $\nabla S$ (gradient of phase).

Tangent vector $\hat{\mathbf{t}} = \nabla S/|\nabla S| = \nabla S/n$.

Ray equation (from $d/ds$ of eikonal):

$$\boxed{\frac{d}{ds}\!\left(n\,\hat{\mathbf{t}}\right) = \nabla n}$$

Equivalent vector form: $\frac{d}{ds}\!\left(n\frac{d\mathbf{x}}{ds}\right) = \nabla n$.

Special cases:

• Homogeneous medium ($\nabla n = 0$): straight lines.
• Snell's law (interface, $n$ jump): tangential component of $n\hat{\mathbf{t}}$ continuous → $n_1\sin\theta_1 = n_2\sin\theta_2$.
• Gradient-index (GRIN): ray bends toward higher $n$ (toward denser/cooler air → mirages, atmospheric refraction).

Fermat's Principle

The eikonal equation is equivalent to:

$$\delta \int n\, ds = 0$$

— rays extremize the optical path length between two points.

flowchart LR
    A["Fermat\nδ∫n ds = 0"] <--> B["Eikonal\n∣∇S∣² = n²"]
    A <--> C["Ray eq\n(d/ds)(n t̂) = ∇n"]
    B <--> D["Snell's law\nn₁ sinθ₁ = n₂ sinθ₂"]
    A <--> D

Analogous to Maupertuis principle in mechanics ($\delta\int p\,dq = 0$). Provides the bridge "geometric optics ↔︎ classical mechanics," with $n \leftrightarrow p$.

Hamiltonian Formulation

Treat phase $S$ as Hamilton's action; rays as trajectories in phase space $(\mathbf{x}, \mathbf{k})$ with $\mathbf{k} = k_0\nabla S$ playing role of momentum.

Optical Hamiltonian:

$$H(\mathbf{x}, \mathbf{k}) = \frac{1}{2k_0}\!\left(\frac{|\mathbf{k}|^2}{n^2} - n^2\right) k_0 \;\;\text{(one form)}\quad \text{or}\quad H = \tfrac{1}{2}(k^2 - k_0^2 n^2)$$

Hamilton's equations:

$$\frac{d\mathbf{x}}{d\sigma} = \partial_{\mathbf{k}} H, \qquad \frac{d\mathbf{k}}{d\sigma} = -\partial_{\mathbf{x}} H$$

with $\sigma$ a parameter along the ray.

Liouville's theorem holds: phase-space volume in $(\mathbf{x},\mathbf{k})$ space is conserved along rays. This is the basis of conservation of étendue (area × solid angle × $n^2$) in optical systems.

Polarization Transport (Parallel Transport along a Ray)

For Maxwell (not scalar) waves, polarization vector $\hat{\mathbf{e}}$ obeys:

$$\frac{d\hat{\mathbf{e}}}{ds} = -(\hat{\mathbf{e}}\cdot\nabla\ln n)\,\hat{\mathbf{t}}$$

i.e., the projection onto the wavefront is parallel-transported.

Consequence: in a twisted ray (e.g., optical fiber with torsion), polarization picks up a geometric (Berry) phase = solid angle subtended by $\hat{\mathbf{t}}$ on the unit sphere. Verified in helical fibers (Tomita & Chiao 1986).

Caustics

A caustic is where neighboring rays converge → amplitude diverges → eikonal breaks down. Locally, two rays merge; the geometric-optics field has a square-root singularity.

Examples: focal point of a lens; bright curve on bottom of a swimming pool; rainbow caustic (Airy function structure across it).

Topology: Arnold's classification — fold, cusp, swallowtail, butterfly, umbilic. Locally diffraction smooths the singularity into Airy- and Pearcey-function patterns.

Paraxial Approximation & Gaussian Beams

Paraxial = rays make small angles with optical axis.

Paraxial wave equation: write $\psi = u(\mathbf{r}_\perp, z)\,e^{ikz}$, $u$ slowly varying:

$$\boxed{2ik\frac{\partial u}{\partial z} + \nabla_\perp^2 u = 0}$$

— a Schrödinger equation in $(\mathbf{r}_\perp, z)$ with $z\leftrightarrow t$.

Fundamental Gaussian beam (TEM₀₀ mode):

$$u(r, z) = \frac{w_0}{w(z)}\exp\!\left[-\frac{r^2}{w(z)^2}\right]\exp\!\left[ikz + ik\frac{r^2}{2R(z)} - i\eta(z)\right]$$

Key invariant (uncertainty-style):

$$w_0 \times \theta = \lambda/\pi \quad \Rightarrow \quad \text{small waist ⇔ large divergence}$$

ABCD Matrix Optics (Paraxial Ray Tracing)

Track a ray by $(y, y')$ where $y$ = height from axis, $y' = dy/dz$:

$$\begin{pmatrix} y_{\text{out}} \\ y'_{\text{out}}\end{pmatrix} = \begin{pmatrix} A & B \\ C & D \end{pmatrix}\begin{pmatrix} y_{\text{in}} \\ y'_{\text{in}}\end{pmatrix}$$

System matrix = product of element matrices in reverse order of propagation. Always $\det = n_{\text{in}}/n_{\text{out}}$.

For Gaussian beam propagation: complex parameter $q$ satisfies $q' = (A q + B)/(C q + D)$, where $1/q = 1/R - i\lambda/(\pi w^2)$.

Diffraction

Helmholtz Equation & Green's Function

For monochromatic field $\psi(\mathbf{x},t) = U(\mathbf{x}) e^{-i\omega t}$:

$$(\nabla^2 + k^2) U = 0 \quad\text{(Helmholtz)}$$

Spherical-wave Green's function:

$$G(r) = \frac{e^{ikr}}{r}$$

— solution to $(\nabla^2 + k^2)G = -4\pi\delta^3(\mathbf{x})$.

Huygens-Fresnel Principle (Modernized)

Every point on a wavefront acts as a source of a secondary spherical wave; the total field is the superposition.

Kirchhoff's integral theorem:

$$U(P) = \frac{1}{4\pi}\oint_{\partial V}\!\left[U\,\partial_n G - G\,\partial_n U\right]\, dS$$

For diffraction from an aperture $\Sigma$ (illuminated by spherical/plane wave), Fresnel-Kirchhoff formula:

$$\boxed{U(P) = \frac{-i}{\lambda}\int_\Sigma U_0(P_0)\,\frac{e^{ikr}}{r}\,K(\theta)\, dA}$$

with obliquity factor $K(\theta) = \tfrac{1}{2}(1 + \cos\theta)$ (Kirchhoff) or $\cos\theta$ (Rayleigh-Sommerfeld).

Fresnel Number — Regime Classification

For aperture of size $a$, viewing at distance $z$, wavelength $\lambda$:

$$\boxed{N_F = \frac{a^2}{\lambda z}}$$

Fraunhofer Diffraction = Fourier Optics

For an aperture in plane $z=0$ with field $U_0(\mathbf{r}_\perp)$, observed at distance $z \gg a^2/\lambda$:

$$\boxed{U(\boldsymbol\rho) = \frac{e^{ikz}}{i\lambda z} e^{ik\rho^2/2z} \int U_0(\mathbf{r}_\perp)\, e^{-i\mathbf{k}_\perp\cdot\mathbf{r}_\perp}\, d^2 r_\perp}$$

with $\mathbf{k}_\perp = k\boldsymbol\rho/z$. The far field is the 2-D spatial Fourier transform of the aperture amplitude.

Standard apertures:

Airy disk:

• First zero at $ka\sin\theta = 3.8317 \Rightarrow \sin\theta = 1.22\lambda/(2a) = 1.22\lambda/D$ for diameter $D$.
• 84% of power within first zero.

Rayleigh criterion: two point sources resolved when one's center sits on other's first dark ring:

$$\boxed{\theta_{\min} = 1.22\,\lambda/D \quad \text{(circular aperture)}}$$

Diffraction Grating

$N$ slits, period $d$, width $a$:

$$I(\theta) = I_0\,\underbrace{\text{sinc}^2\!\left(\tfrac{\pi a \sin\theta}{\lambda}\right)}_{\text{single-slit envelope}}\cdot\underbrace{\left[\frac{\sin(N\pi d\sin\theta/\lambda)}{\sin(\pi d\sin\theta/\lambda)}\right]^2}_{\text{principal maxima at } d\sin\theta = m\lambda}$$

Fresnel Diffraction & Zones

For near-field, must keep quadratic phase. The aperture is divided into Fresnel zones — annular regions whose contributions to $U(P)$ alternate in sign.

Fresnel zone radii (axis-to-zone-edge distance, for source/observation at distance $z$): $$r_n = \sqrt{n\lambda z}$$

Field at axis = approximate sum of alternating contributions:

$$U(\text{axis}) \approx \tfrac{1}{2}(U_1 + U_{\text{last}})$$

Zone plate = alternating opaque/clear zones → focuses like a lens (without refraction). Focal length $f = a^2/(\lambda)$ for first-zone radius $a$.

Fresnel integrals (used for slit edge, knife edge):

$$C(u) = \int_0^u \cos(\pi t^2/2)\, dt, \quad S(u) = \int_0^u \sin(\pi t^2/2)\, dt$$

→ Cornu spiral.

Babinet's Principle

Field of an aperture + field of its complement = field of unobstructed wave:

$$U_{\text{aperture}}(P) + U_{\text{complement}}(P) = U_{\text{unobstructed}}(P)$$

Consequence: away from the geometric shadow (where $U_{\text{unobstructed}} = 0$), an opaque disk and a circular hole produce the same diffraction pattern in intensity.

Diffraction from the Information-Theory View

Optical imaging = bandlimited spatial Fourier reconstruction.

Abbe / Rayleigh resolution limit: maximum spatial frequency the lens admits is $k_{\max} = (2\pi/\lambda)\sin\alpha = 2\pi\,\text{NA}/\lambda$. The image-plane resolution is

$$\Delta x \approx \lambda/(2\,\text{NA})$$

NA = numerical aperture = $n\sin\alpha$ at object side.

Super-resolution techniques (STED, PALM, structured illumination, near-field) break this classical limit by exploiting nonlinearity or proximity.

Interferometry & Coherence

Two-Beam Interference (Coherent Case)

Two fields $E_1, E_2$ with intensities $I_1, I_2$ and phase difference $\delta$:

$$I_{\text{tot}} = I_1 + I_2 + 2\sqrt{I_1 I_2}\cos\delta$$

Visibility (fringe contrast):

$$\boxed{V = \frac{I_{\max} - I_{\min}}{I_{\max} + I_{\min}} = \frac{2\sqrt{I_1 I_2}}{I_1 + I_2} \quad \text{(coherent)}}$$

Equal intensities $I_1=I_2$ ⇒ $V=1$. Unequal or partially coherent ⇒ $V<1$.

Multi-Beam Interference — Fabry-Perot

Two parallel mirrors, reflectivity $R$, separation $L$:

$$\frac{I_t}{I_0} = \frac{1}{1 + F\sin^2(\delta/2)}, \quad F = \frac{4R}{(1-R)^2}, \quad \delta = 2kL\cos\theta$$

— the Airy function (transmission, not Airy disk).

Used in: laser cavities, optical filters, gravitational wave detectors (LIGO arm cavities, $\mathcal{F}\sim 450$), atomic clocks.

Coherence — Statistical Description of Light

Real light has finite spectral width and angular extent. Coherence theory quantifies "how well-defined is the phase" in time + space.

Mutual coherence function (two points $\mathbf{r}_1, \mathbf{r}_2$, time lag $\tau$):

$$\Gamma_{12}(\tau) = \langle E^*(\mathbf{r}_1, t)\, E(\mathbf{r}_2, t+\tau)\rangle$$

Normalized:

$$\gamma_{12}(\tau) = \frac{\Gamma_{12}(\tau)}{\sqrt{\Gamma_{11}(0)\,\Gamma_{22}(0)}}, \quad |\gamma_{12}| \le 1$$

General visibility:

$$V = \frac{2\sqrt{I_1 I_2}}{I_1 + I_2}\, |\gamma_{12}|$$

Temporal Coherence (Self-Coherence)

For single point ($\mathbf{r}_1 = \mathbf{r}_2$), $\gamma(\tau)$ measures how the field correlates with delayed copy of itself.

Coherence time: $\tau_c \sim 1/\Delta\nu$ (width of $|\gamma(\tau)|$). Coherence length: $L_c = c\tau_c$.

Wiener-Khinchin (again): $\Gamma_{11}(\tau)$ ↔︎ power spectrum $S(\omega)$. So:

• Spectrally narrow source ⇒ long coherence time/length.
• White light: $\tau_c \sim 1/\Delta\omega \sim$ optical cycle.

Typical numbers:

Michelson interferometer sweeps $\tau$; envelope of fringes = $|\gamma_{11}(\tau)|$ → Fourier-transform spectroscopy (FTIR).

Spatial Coherence

For lag $\tau = 0$ but two separated points $\mathbf{r}_1, \mathbf{r}_2$: $\gamma_{12}(0)$ measures spatial coherence.

Van Cittert-Zernike theorem: for a distant, incoherent, quasi-monochromatic source with intensity distribution $I_s(\boldsymbol\sigma)$ (angular):

$$\boxed{\gamma_{12}(0) = \frac{\int I_s(\boldsymbol\sigma)\, e^{-i\mathbf{k}\cdot(\mathbf{r}_2-\mathbf{r}_1)\cdot\boldsymbol\sigma/R}\, d^2\sigma}{\int I_s\, d^2\sigma}}$$

— the spatial coherence function is the Fourier transform of the source brightness distribution.

Implications:

• Coherence area on Earth ≈ $\lambda R/d$ where $d$ = source size.
• For Sun ($d\approx 1.4\times 10^9$ m at $R\approx 1.5\times 10^{11}$ m, $\lambda = 500$ nm): $\sim 50\,\mu$m. Why pinhole experiments need pinholes.
• For star (apparent diameter $\theta_*$): coherence baseline = $\lambda/\theta_*$.

Stellar Interferometry

Michelson stellar interferometer (1920, measured Betelgeuse): two-aperture interference; baseline $D$ → $V = |J_1(\pi\theta_*D/\lambda)|$ for uniform disk. First null gives $\theta_* = 1.22\lambda/D$.

Modern long-baseline interferometers (CHARA, VLTI, GRAVITY): kilometer baselines → milli-arcsec resolution. Used for stellar diameters, exoplanet imaging, AGN black holes (Event Horizon Telescope = VLBI at radio wavelengths).

Hanbury Brown–Twiss (Intensity Interferometry)

Correlate intensities at two detectors instead of amplitudes:

$$g^{(2)}(\tau) = \frac{\langle I(t)I(t+\tau)\rangle}{\langle I\rangle^2}$$

For thermal/chaotic light: $g^{(2)}(0) = 2$ ("bunching"), decaying to 1 at $\tau\gg\tau_c$. For coherent (laser): $g^{(2)}(\tau) = 1$ (uncorrelated arrivals). For single-photon source: $g^{(2)}(0) = 0$ ("antibunching" — purely quantum, no classical wave can do this).

HBT-Siegert relation (for chaotic Gaussian light): $g^{(2)}(\tau) = 1 + |g^{(1)}(\tau)|^2$, with $g^{(1)} = \gamma_{11}(\tau)$.

Advantage of intensity interferometry: insensitive to atmospheric phase fluctuations → very long baselines. HBT measured Sirius stellar diameter (1956). Modern revivals (VERITAS, etc.) probe stellar surfaces.

Photon Statistics

For monochromatic light, intensity ⟶ photon counts $\to$:

Mandel $Q = (\langle\Delta n^2\rangle - \langle n\rangle)/\langle n\rangle$.

Standard quantum limit (SQL) for phase: $\Delta\phi_{SQL} = 1/\sqrt{\bar n}$ — Poisson counting noise. Squeezed light beats SQL (used in upgraded LIGO).

Holography

Record interference pattern between object wave $U_O$ and reference wave $U_R$ on a photographic plate; intensity recorded:

$$I = |U_O + U_R|^2 = |U_O|^2 + |U_R|^2 + U_R^* U_O + U_R U_O^*$$

Readout: illuminate hologram with reference $U_R$ → third term gives $|U_R|^2 U_O$ → reconstructs object wave (virtual 3-D image).

Key feature: stores both amplitude and phase of object wave. Originally Gabor (1948, Nobel 1971) for electron microscopy; revolutionized with lasers (Leith-Upatnieks 1962). Modern variants: rainbow holograms, digital holography, holographic optical elements.

Nonlinear Optics

Nonlinear Polarization

In linear media: $\mathbf{P} = \epsilon_0\chi^{(1)}\mathbf{E}$.

In nonlinear: expand in powers of $\mathbf{E}$:

$$\boxed{P_i = \epsilon_0\!\left[\chi^{(1)}_{ij} E_j + \chi^{(2)}_{ijk} E_j E_k + \chi^{(3)}_{ijkl} E_j E_k E_l + \cdots\right]}$$

Symmetries:

• $\chi^{(2)} \ne 0$ only in non-centrosymmetric media (BBO, KDP, LiNbO₃, GaAs).
• $\chi^{(3)}$ allowed in all media (incl. glass, air).

Magnitudes: $\chi^{(2)} \sim 10^{-12}$ m/V, $\chi^{(3)} \sim 10^{-22}$ m²/V². Nonlinear effects need intense fields → lasers, especially pulsed.

Three-Wave Mixing (from $\chi^{(2)}$)

Two input waves at $\omega_1, \omega_2$ generate output components at $\omega_1 \pm \omega_2$ (and harmonics $2\omega_1, 2\omega_2$, plus DC). Energy and momentum conservation:

$$\omega_3 = \omega_1 + \omega_2 \quad\text{(or differences)}$$ $$\mathbf{k}_3 = \mathbf{k}_1 + \mathbf{k}_2 \quad \text{(phase matching)}$$

Coupled-wave equations (slowly varying envelope approximation):

$$\frac{dA_3}{dz} = \frac{i\omega_3}{n_3 c} d_{\text{eff}} A_1 A_2\, e^{i\Delta k z}, \quad \Delta k = k_1 + k_2 - k_3$$

Conversion efficiency $\propto \text{sinc}^2(\Delta k\, L/2)$ — phase mismatch kills it.

Phase Matching

Phase matching = ensuring $\Delta k = 0$ over the interaction length. Hard because of dispersion ($n(\omega) \ne $ const).

Methods:

1. Birefringent phase matching: use ordinary/extraordinary rays in anisotropic crystal. Tune $n_e(\theta)$ via crystal orientation.
2. Temperature tuning: $n(T)$ varies; lock to phase-matching condition.
3. Quasi-phase matching (QPM): periodically poled crystal (e.g., PPLN) flips sign of $\chi^{(2)}$ every coherence length $L_c = \pi/\Delta k$. Equivalent to grating momentum kick.

Second-Harmonic Generation (SHG)

Special case of three-wave mixing with $\omega_1 = \omega_2 = \omega$, $\omega_3 = 2\omega$.

Conversion efficiency (low-depletion, phase-matched):

$$\eta = \frac{P_{2\omega}}{P_\omega} = \frac{8\pi^2 d_{\text{eff}}^2 L^2 I_\omega}{n^3\epsilon_0 c\lambda^2}$$

— quadratic in length, linear in input intensity. Doubles 1064 nm Nd:YAG to 532 nm green; the basis of green laser pointers.

Parametric Processes

Parametric amplification: pump at $\omega_p$, signal at $\omega_s$ → idler at $\omega_i = \omega_p - \omega_s$ (with $\omega_s + \omega_i = \omega_p$, $\mathbf{k}_s + \mathbf{k}_i = \mathbf{k}_p$). Signal grows exponentially when phase-matched:

$$A_s(z) = A_s(0)\cosh(\gamma z), \quad \gamma \propto \sqrt{I_p}$$

Optical parametric oscillator (OPO): place in cavity → coherent tunable source. Widely used for mid-IR / UV generation.

Spontaneous parametric down-conversion (SPDC): quantum process where pump photon spontaneously splits into entangled signal-idler pair. Foundation of quantum optics (Bell tests, quantum key distribution, two-photon interferometry).

Four-Wave Mixing & Kerr Effect (from $\chi^{(3)}$)

Generic: $\omega_4 = \omega_1 + \omega_2 - \omega_3$.

Kerr effect (degenerate FWM, $\omega_1 = \omega_2 = \omega_3 = \omega$):

$$n(I) = n_0 + n_2 I, \quad n_2 = \frac{3\chi^{(3)}}{4\epsilon_0 c n_0^2}$$

— refractive index depends on intensity.

Consequences:

Solitons

Nonlinear Schrödinger equation (envelope $A$ in fiber):

$$i\frac{\partial A}{\partial z} - \frac{\beta_2}{2}\frac{\partial^2 A}{\partial t^2} + \gamma |A|^2 A = 0$$

with $\beta_2 < 0$ (anomalous dispersion) and $\gamma > 0$ (Kerr).

Fundamental soliton: $A(t, z) = A_0\,\text{sech}(t/T_0)\,e^{i\gamma A_0^2 z/2}$. Propagates without distortion — dispersion and nonlinearity cancel.

Used in fiber-optic communications, mode-locked lasers (passive soliton mode-locking).

Stimulated Raman & Brillouin Scattering

SRS: photon scatters off molecular vibration; emits Stokes photon at $\omega - \Omega_v$ (with $\Omega_v \sim$ THz for molecules). Gain $\propto I_p$.

SBS: photon scatters off acoustic phonon; backward-propagating Stokes at $\omega - \Omega_B$ with $\Omega_B \sim$ GHz. Limits power in long fibers (couples forward to backward).

Frequency Combs (a Modern Triumph)

Mode-locked laser → train of pulses with rep rate $f_r$ → spectrum is a comb of teeth at $f_n = n f_r + f_0$ ($f_0$ = carrier-envelope offset).

Self-referencing (f-2f): broaden spectrum to octave; SHG of low-frequency tooth interferes with high-frequency tooth → measure $f_0$. Locking both $f_r$ and $f_0$ to atomic clocks ⇒ all comb teeth become frequency standards.

Impact: revolutionized optical frequency metrology, optical clocks, exoplanet RV spectroscopy, attosecond pulse generation. Hänsch & Hall Nobel 2005.

Workflow / Process

flowchart TD
    A[Optical problem] --> B{Length scales}
    B -->|λ ≪ L| C[Geometric optics]
    B -->|λ ~ L| D[Wave optics]
    B -->|λ ≫ L| E[Lumped/quasi-static]
    C --> F[Eikonal / ray equation]
    C --> G[Paraxial / Gaussian beam]
    D --> H{Aperture vs distance}
    H -->|N_F ≫ 1| I[Fresnel diffraction]
    H -->|N_F ≪ 1| J[Fraunhofer = Fourier transform]
    D --> K{Coherent or partially coherent?}
    K -->|Coherent| L[Add amplitudes, then intensity]
    K -->|Partial| M[Mutual coherence Γ]
    M --> N[Van Cittert-Zernike]
    A --> O{Intensity high?}
    O -->|Yes, |E| ~ |E_atomic|| P[Nonlinear χⁿ]
    P --> Q[Phase matching condition]

Comparison Tables

Wave/Ray Regimes

Imaging Resolution

Coherence Concepts

Nonlinear Processes ($\chi^{(2)}$)

Photon Statistics Summary

Common Mistakes

• ❌ Using geometric optics where $\lambda \sim L$. Always check $N_F$ or eikonal validity.
• ❌ Confusing Fresnel and Fraunhofer regimes. Plug in numbers for $N_F$ before choosing approximation.
• ❌ Adding intensities instead of amplitudes for coherent superposition. Coherent: $|U_1 + U_2|^2$. Incoherent: $|U_1|^2 + |U_2|^2$.
• ❌ Forgetting the obliquity factor / Kirchhoff sign. Different texts use different forms; verify before plugging into a problem.
• ❌ Using paraxial Gaussian beam formulas at large angles. Breaks down when $w_0 \sim \lambda$.
• ❌ Treating Airy disk first zero as the resolution by Sparrow criterion. Sparrow uses where pattern flattens, $\sim 0.95 \lambda/D$. Rayleigh = $1.22\lambda/D$. Pick one consistently.
• ❌ Confusing visibility $V$ with mutual coherence $|\gamma|$. They coincide only for equal intensities.
• ❌ Forgetting that VCZ gives Fourier transform of source intensity, not amplitude. Source is incoherent; coherence is built up by propagation distance.
• ❌ Using $g^{(2)}(0) > 1$ as proof of quantum effect. That's classical bunching (thermal light). Only $g^{(2)}(0) < 1$ is non-classical.
• ❌ Ignoring polarization in $\chi^{(2)}$ processes. It's a 3-tensor; ordinary/extraordinary mixing essential for phase matching.
• ❌ Computing SHG efficiency without phase matching condition. A factor of $10^4$–$10^6$ comes from $\sin^2(\Delta k L/2)/(\Delta k L/2)^2$.
• ❌ Confusing Kerr nonlinearity $n_2$ with electro-optic Pockels effect. Pockels is $\chi^{(2)}$ (linear in $E$); Kerr is $\chi^{(3)}$ (quadratic in $E$).
• ❌ Treating coherence as binary (coherent/incoherent). It's continuous: $0 \le |\gamma| \le 1$.
• ❌ Assuming all light is monochromatic. Real light has finite $\Delta\nu$; affects coherence, interferometer fringe envelope.

Expert Insights

Geometric optics is the classical limit of wave optics, exactly analogous to classical mechanics being the limit of QM. Eikonal ↔︎ Hamilton-Jacobi; ray equation ↔︎ Newton's equation; $\hbar \leftrightarrow \lambda$.

Caustics are where geometric optics commits suicide. Diffraction integrals there yield Airy / Pearcey / higher catastrophe functions. The colors of rainbows live exactly in this regime.

The lens does a Fourier transform. This is the central insight of Fourier optics — image plane = FT of object plane. Spatial filtering in the Fourier plane = phase contrast, dark field, edge enhancement, modern phase microscopy.

Resolution ≠ sampling. A telescope's $1.22\lambda/D$ is the angular resolution; a CCD pixel array can be coarser (under-sampling, aliasing) or finer (over-sampling). Match them at Nyquist.

Coherence is geometric. It's not a property of "the light" alone but of "the light + the experimental setup." A pinhole increases spatial coherence (at cost of flux). A monochromator increases temporal coherence (same trade-off).

The fluctuation-dissipation theorem hides in HBT. The bunching of thermal photons is a stat-mech statement about $\Delta n$ for Bose distribution — directly inherited from Part II.

Holography is a metaphor for many things. Storing phase = storing 3-D info. Quantum holography, gravitational holography (AdS/CFT) borrow the name.

The Van Cittert-Zernike theorem and the Wiener-Khinchin theorem are sibling FTs. One in space (incoherent source), one in time (random process). Both unmask randomness as Fourier dual of correlations.

Quasi-phase matching is the right answer. Most useful crystals can't be birefringently phase-matched at desired wavelengths. PPLN (periodically poled LiNbO₃) made nonlinear optics industrial.

Squeezed light is not "more photons in less space." It's a redistribution of the quantum uncertainty between amplitude and phase quadratures. Anti-squeezing the unmeasured quadrature is the cost.

Solitons are particles to fluid dynamicists, but pulses to optical engineers. Their stability comes from balance of two opposing effects (dispersion vs nonlinearity in optics; depth-dispersion vs amplitude in shallow water).

Frequency combs make light into a ruler. Phase-locking a comb to a microwave standard transfers atomic-clock precision to optical frequencies — billion-fold gain in precision.

Atmospheric phase fluctuations kill amplitude interferometry but not intensity interferometry. That's the deep reason HBT survives at intercontinental baselines while Michelson interferometers struggle past 100 m.

Numerical aperture is the same as $\sin\theta$ for the marginal ray. Higher NA = wider acceptance angle = better resolution AND smaller depth of focus. Trade-off baked in.

Diffraction limit is not fundamental. It's the linear far-field limit. Nonlinear / near-field / statistical tricks (STED, PALM, NSOM) all break it routinely now.