Key Concepts of Basis Functions

Why This Matters

In Representation Theory, basis functions are the building blocks that let you decompose complex functions into manageable pieces—and this decomposition is exactly what exam questions test. You're being asked to understand orthogonality, completeness, weight functions, and domain-specific applications. Whether it's Fourier series breaking down periodic signals or Hermite polynomials solving the quantum harmonic oscillator, each basis function family exists because it's perfectly suited to a particular symmetry or boundary condition.

Don't just memorize which polynomial goes with which equation. Focus on why each basis works where it does: What's the underlying symmetry? What weight function defines orthogonality? What physical or mathematical problem does it solve most naturally? When you understand these connections, you can tackle any comparison question or application problem the exam throws at you.

---

Periodic and Localized Signal Decomposition

These basis functions handle the fundamental trade-off between frequency resolution and spatial/temporal localization.

Fourier Basis Functions

• Decompose periodic functions into sines and cosines—the foundation of frequency analysis across all signal processing applications
• Link time and frequency domains through Fourier series (discrete frequencies) and Fourier transforms (continuous spectra)
• Convergence guarantees ensure accurate approximation of square-integrable functions, making them reliable for both theory and computation

Wavelet Basis Functions

• Capture both frequency and location simultaneously—unlike Fourier methods, wavelets provide multi-resolution analysis
• Ideal for non-stationary signals where frequency content changes over time, enabling localized feature detection
• Power signal compression and denoising applications by isolating relevant components at different scales

---

Spherical and Cylindrical Symmetry

When your problem lives on a sphere or cylinder, these basis functions naturally satisfy the geometry's boundary conditions.

Spherical Harmonics

• Angular solutions to Laplace's equation in spherical coordinates—essential for any problem with spherical symmetry
• Orthogonal on the sphere's surface, enabling expansion of arbitrary angular distributions in quantum mechanics and geophysics
• Indexed by degree $\ell$ and order $m$, with $2\ell + 1$ functions at each degree capturing increasingly fine angular detail

Bessel Functions

• Solutions to Bessel's differential equation $x^2 y'' + xy' + (x^2 - n^2)y = 0$, arising in cylindrical coordinate problems
• Orthogonality with respect to $x$ on finite intervals makes them ideal for series expansions in bounded cylindrical domains
• Appear throughout physics—acoustics (drum vibrations), electromagnetism (waveguides), and heat conduction in cylindrical geometries

---

Classical Orthogonal Polynomials on Intervals

These polynomial families are orthogonal with respect to specific weight functions on bounded intervals, each optimized for different applications.

Legendre Polynomials

• Orthogonal on $[-1, 1]$ with weight $w(x) = 1$—the simplest case, arising naturally in spherical coordinate problems
• Radial component of spherical harmonics through the associated Legendre functions $P_\ell^m(x)$
• Optimal for Gaussian quadrature integration, with roots providing the best node placement for numerical accuracy

Chebyshev Polynomials

• Orthogonal on $[-1, 1]$ with weight $w(x) = (1-x^2)^{-1/2}$—connected to trigonometric functions via $T_n(\cos\theta) = \cos(n\theta)$
• Minimize interpolation error (minimax property), making them the gold standard for polynomial approximation
• Chebyshev nodes prevent Runge's phenomenon—clustering near endpoints stabilizes high-degree polynomial interpolation

---

Weighted Orthogonal Polynomials for Unbounded Domains

When your domain extends to infinity, these polynomials incorporate decay through their weight functions.

Hermite Polynomials

• Orthogonal on $(-\infty, \infty)$ with Gaussian weight $w(x) = e^{-x^2}$—naturally paired with normal distributions
• Eigenfunctions of the quantum harmonic oscillator, with energy levels $E_n = \hbar\omega(n + \frac{1}{2})$
• Foundation of probabilists' and physicists' conventions—watch for the factor of $\sqrt{2}$ difference between definitions

Laguerre Polynomials

• Orthogonal on $[0, \infty)$ with exponential weight $w(x) = e^{-x}$—suited for problems with natural decay
• Solve the radial Schrödinger equation for hydrogen-like atoms, giving the radial wavefunctions $R_{n\ell}(r)$
• Associated Laguerre polynomials $L_n^k(x)$ generalize to different exponential weights, appearing in quantum mechanics with angular momentum

---

Specialized Geometric Domains

Some basis functions are tailored to specific geometric regions where standard polynomials don't naturally apply.

Zernike Polynomials

• Orthogonal on the unit disk $x^2 + y^2 \leq 1$, combining radial polynomials with angular Fourier modes
• Standard for optical aberration analysis—each polynomial corresponds to a specific wavefront distortion (tilt, defocus, astigmatism, coma)
• Separable structure $Z_n^m(\rho, \theta) = R_n^m(\rho) \cdot e^{im\theta}$ allows efficient computation and clear physical interpretation

---

The Unifying Framework: Eigenfunctions

All these basis functions share a common structure—they're eigenfunctions of specific linear operators.

Eigenfunctions of Linear Operators

• Satisfy $\hat{L}f = \lambda f$ where $\hat{L}$ is a linear operator—the function's form is preserved, only scaled by eigenvalue $\lambda$
• Quantum mechanics interpretation: eigenfunctions represent states with definite values of the observable corresponding to $\hat{L}$
• Completeness guarantees that any function in the space can be expanded as $f = \sum_n c_n \phi_n$, with coefficients found via inner products

---

Quick Reference Table

---

Self-Check Questions

1. Which two polynomial families are both orthogonal on $[-1, 1]$ but differ in their weight functions? What applications does each weight function favor?

2. If you're solving a problem with cylindrical symmetry and need a radial basis, which functions should you use? How do they differ from the basis you'd choose for spherical symmetry?

3. Compare and contrast Hermite and Laguerre polynomials: What are their domains, weight functions, and signature physical applications?

4. An FRQ asks you to analyze a non-stationary signal where frequency content changes over time. Why would wavelets outperform Fourier methods, and what property makes this possible?

5. Explain why all the classical orthogonal polynomials (Legendre, Hermite, Laguerre, Chebyshev) can be viewed as eigenfunctions of differential operators. What does this shared structure guarantee about their properties?