Chapter 5 • Theory & Derivations

Regular Singular Points & The Method of Frobenius

Singularity classifications, the indicial polynomial F(r), and the three classical cases of the Frobenius method for regular singular points.

§5.1 Classification of Singular Points (Ordinary, Regular Singular, Irregular Singular)

1. Rigorous Classification of Singular Points

Let $y'' + P(x) y' + Q(x) y = 0$. Suppose $x_0$ is a singular point (either $P(x)$ or $Q(x)$ blows up at $x_0$).

Definition 5.1: The singular point $x_0$ is called a regular singular point if the singularities are at most a simple pole for $P(x)$ and a double pole for $Q(x)$; that is, both functions: $$p(x) = (x - x_0) P(x) \quad \text{and} \quad q(x) = (x - x_0)^2 Q(x)$$ are analytic at $x_0$. If either $p(x)$ or $q(x)$ is not analytic at $x_0$, the point is an irregular singular point (essential singularity).

§5.2 The Method of Frobenius & The Indicial Equation

1. The Frobenius Series Ansatz

About a regular singular point (taken at $x_0 = 0$ without loss of generality), we seek solutions of the generalized power series form:

$$y(x) = x^r \sum_{n=0}^\infty a_n x^n = \sum_{n=0}^\infty a_n x^{n+r}, \quad a_0 \ne 0$$

where the index $r \in \mathbb{C}$ is a parameter to be determined. Expanding $p(x) = \sum p_n x^n$ and $q(x) = \sum q_n x^n$, the lowest-order term in $x$ is $x^r$, whose coefficient gives the indicial equation:

$$F(r) = r(r - 1) + p_0 r + q_0 = 0$$

The two roots $r_1, r_2$ (with $\text{Re}(r_1) \ge \text{Re}(r_2)$) govern the algebraic nature of the solutions near $x = 0$.

§5.3 Case 1: Roots Differing by Non-Integer; Case 2: Equal Roots & Logarithmic Solutions

Case 1: $r_1 - r_2 \notin \mathbb{Z}$ (Roots Not Differing by an Integer)

When the roots do not differ by an integer, two linearly independent Frobenius series solutions exist directly:

$$y_1(x) = x^{r_1} \sum_{n=0}^\infty a_n x^n, \quad y_2(x) = x^{r_2} \sum_{n=0}^\infty b_n x^n \quad (x > 0)$$

Case 2: $r_1 = r_2 = r$ (Equal Indicial Roots)

When the indicial roots coincide, the first solution is $y_1(x) = x^r \sum a_n x^n$. The second linearly independent solution strictly contains a logarithmic branch singularity:

$$y_2(x) = y_1(x) \ln x + x^r \sum_{n=1}^\infty b_n x^n \quad (x > 0)$$

This is derived rigorously by differentiating the parameterized series $y(x, r)$ with respect to the indicial parameter $r$:

$$y_2(x) = \left. \frac{\partial y(x, r)}{\partial r} \right|_{r = r_1}$$

§5.4 Case 3: Roots Differing by a Positive Integer & The Logarithmic Factor Criterion

1. Case 3: $r_1 - r_2 = N \in \mathbb{Z}^+$

When the roots differ by a positive integer $N$, the recurrence relation for the smaller root $r_2$ encounters $F(r_2 + N) = F(r_1) = 0$ in the denominator of $a_N$, which may lead to division by zero.

Theorem 5.1 (Frobenius Case 3 Structure): The second solution takes the general form: $$y_2(x) = C y_1(x) \ln x + x^{r_2} \sum_{n=0}^\infty c_n x^n \quad (c_0 \ne 0)$$ where the constant $C$ is given by: $$C = \lim_{r \to r_2} (r - r_2) a_N(r)$$

If $C = 0$, the logarithmic term vanishes, and $y_2(x)$ is a pure Frobenius series without logarithms.
If $C \ne 0$, the logarithmic term is mandatory.

TIERED UNIVERSITY HONORS PROBLEMS

Step-by-Step Solved Examination Problems

Comprehensive analytical derivations, multi-tier solutions (Foundational, Intermediate Exam, and Honors/Proof Challenge) with complete line-by-line verification.

Tier 1 • Foundational Example 5.1: Frobenius Series Case 1: Roots Differing by Non-Integer

Find the Frobenius series solutions about $x = 0$ for:

$$2x y'' + y' + y = 0$$

Step 1: Identify singularity and indicial equation
Normalized form: $y'' + \frac{1}{2x} y' + \frac{1}{2x} y = 0$. $p(x) = x P(x) = \frac{1}{2} \implies p_0 = \frac{1}{2}$, and $q(x) = x^2 Q(x) = \frac{x}{2} \implies q_0 = 0$. The point $x = 0$ is a regular singular point. Indicial equation:

r(r - 1) + \frac{1}{2} r + 0 = r\left(r - \frac{1}{2}\right) = 0 \implies r_1 = \frac{1}{2}, \quad r_2 = 0

Since $r_1 - r_2 = \frac{1}{2} \notin \mathbb{Z}$, this is Case 1 (two distinct Frobenius series).

Step 2: Recurrence relation for general $r$
Substitute $y = \sum_{n=0}^\infty a_n x^{n+r}$:

2 \sum (n+r)(n+r-1) a_n x^{n+r-1} + \sum (n+r) a_n x^{n+r-1} + \sum a_n x^{n+r} = 0

For $n \ge 1$:

[(n+r)(2n + 2r - 1)] a_n = -a_{n-1} \implies a_n = -\frac{a_{n-1}}{(n+r)(2n + 2r - 1)}

Step 3: Solution for $r_1 = 1/2$
Denominator factor: $(n + 1/2)(2n) = n(2n + 1)$.

a_n = -\frac{a_{n-1}}{n(2n + 1)} \implies y_1(x) = x^{1/2} \left( 1 - \frac{x}{3} + \frac{x^2}{30} - \frac{x^3}{630} + \dots \right)

Step 4: Solution for $r_2 = 0$
Denominator factor: $n(2n - 1)$.

a_n = -\frac{a_{n-1}}{n(2n - 1)} \implies y_2(x) = 1 - x + \frac{x^2}{6} - \frac{x^3}{90} + \dots

Final Answer & Physical Insight

$y(x) = c_1 x^{1/2}\left(1 - \frac{x}{3} + \frac{x^2}{30} - \dots\right) + c_2 \left(1 - x + \frac{x^2}{6} - \dots\right)$

Tier 2 • Intermediate Exam Example 5.2: Frobenius Method Case 2: Equal Indicial Roots & Logarithmic Branch

Find two linearly independent solutions about $x = 0$ for:

$$x y'' + y' - y = 0$$

Step 1: Indicial equation
Multiply by $x$: $x^2 y'' + x y' - x y = 0$. $p(x) = 1 \implies p_0 = 1$, $q(x) = -x \implies q_0 = 0$. Indicial equation: $F(r) = r(r - 1) + r = r^2 = 0 \implies r_1 = r_2 = 0$. Equal roots $\implies$ Case 2 (logarithmic second solution).

Step 2: Recurrence relation for $y(x, r)$
Substitute $y(x, r) = \sum_{n=0}^\infty a_n(r) x^{n+r}$:

F(n + r) a_n = a_{n-1} \implies (n + r)^2 a_n(r) = a_{n-1}(r) \implies a_n(r) = \frac{a_0}{[(1+r)(2+r)\dots(n+r)]^2}

For $r = 0$ and $a_0 = 1$:

a_n(0) = \frac{1}{(n!)^2} \implies y_1(x) = \sum_{n=0}^\infty \frac{x^n}{(n!)^2} = 1 + x + \frac{x^2}{4} + \frac{x^3}{36} + \dots

Step 3: Differentiate with respect to $r$ to obtain $y_2(x)$

y_2(x) = \left. \frac{\partial y(x, r)}{\partial r} \right|_{r=0} = y_1(x) \ln x + \sum_{n=1}^\infty a_n'(0) x^n

Using logarithmic differentiation on $a_n(r)$:

\ln a_n(r) = -2 \sum_{k=1}^n \ln(k + r) \implies \frac{a_n'(r)}{a_n(r)} = -2 \sum_{k=1}^n \frac{1}{k + r}

At $r = 0$: $a_n'(0) = -2 H_n a_n(0) = -2 \frac{H_n}{(n!)^2}$, where $H_n = \sum_{k=1}^n \frac{1}{k}$ is the $n$-th harmonic number.

y_2(x) = y_1(x) \ln x - 2 \sum_{n=1}^\infty \frac{H_n}{(n!)^2} x^n = y_1(x) \ln x - 2\left( x + \frac{3}{8}x^2 + \frac{11}{216}x^3 + \dots \right)

Final Answer & Physical Insight

$y_1(x) = \sum_{n=0}^\infty \frac{x^n}{(n!)^2}, \quad y_2(x) = y_1(x)\ln x - 2\sum_{n=1}^\infty \frac{H_n}{(n!)^2} x^n$

Tier 2 • Intermediate Exam Example 5.3: Hermite Polynomials Generating Function & Orthogonality

Using the generating function $\Phi(x, t) = e^{2xt - t^2} = \sum_{n=0}^\infty \frac{H_n(x)}{n!} t^n$, prove the orthogonality relation:

\int_{-\infty}^\infty e^{-x^2} H_n(x) H_m(x)\,dx = 2^n n! \sqrt{\pi}\, \delta_{nm}

Step 1: Product of generating functions
Consider two generating functions with parameters $t$ and $s$:

\sum_{n=0}^\infty \sum_{m=0}^\infty \frac{t^n s^m}{n!\, m!} \int_{-\infty}^\infty e^{-x^2} H_n(x) H_m(x)\,dx = \int_{-\infty}^\infty e^{-x^2} \Phi(x, t) \Phi(x, s)\,dx

Step 2: Combine exponents

\Phi(x, t) \Phi(x, s) = e^{2xt - t^2} e^{2xs - s^2} = e^{2x(t + s) - (t^2 + s^2)}

The integrand exponent is:

$$-x^2 + 2x(t + s) - (t^2 + s^2) = -[x - (t + s)]^2 + (t + s)^2 - (t^2 + s^2) = -[x - (t + s)]^2 + 2ts$$

Step 3: Evaluate Gaussian integral

\int_{-\infty}^\infty e^{-[x - (t + s)]^2 + 2ts}\,dx = e^{2ts} \int_{-\infty}^\infty e^{-u^2}\,du = \sqrt{\pi} e^{2ts}

Step 4: Taylor series expansion of $e^{2ts}$

\sqrt{\pi} e^{2ts} = \sqrt{\pi} \sum_{n=0}^\infty \frac{(2ts)^n}{n!} = \sqrt{\pi} \sum_{n=0}^\infty \frac{2^n}{n!} t^n s^n

Notice that there are NO terms with unequal powers $t^n s^m$ ($n \ne m$). Hence:

\int_{-\infty}^\infty e^{-x^2} H_n(x) H_m(x)\,dx = 0 \quad \text{for } n \ne m

For $n = m$, equating coefficients of $\frac{t^n s^n}{(n!)^2}$:

\frac{1}{(n!)^2} \int_{-\infty}^\infty e^{-x^2} [H_n(x)]^2\,dx = \sqrt{\pi} \frac{2^n}{n!} \implies \int_{-\infty}^\infty e^{-x^2} [H_n(x)]^2\,dx = 2^n n! \sqrt{\pi}

Final Answer & Physical Insight

$\int_{-\infty}^\infty e^{-x^2} H_n(x) H_m(x)\,dx = 2^n n! \sqrt{\pi}\, \delta_{nm}$