Linear Integral Equations Formula Sheet and Short Notes
One-page CSIR NET revision of linear integral equations: types, conversions, separable kernels, characteristic numbers, resolvent kernels, symmetric kernels.
- 7 min
On this page · 9 sections
Every method, kernel and formula from Fredholm and Volterra Equations of the First and Second Kind, Separable Kernels and the Fredholm Alternative and Characteristic Numbers, Eigenfunctions and the Resolvent Kernel, on one page.
The paper in one table
Per the NTA information bulletin for December 2026 (Mathematical Sciences, code 704):
| Part | Given | Attempt | Correct | Wrong | Maximum |
|---|---|---|---|---|---|
| B | 40 | 25 | +3 | −0.75 | 75 |
| C | 60 | 20 | +4.75 | 0 | 95 |
Part C: one or more options may be correct; marks only if you mark all correct options and no wrong one. No partial credit, no negative marking. Details on the Mathematical Sciences overview.
Types
| First kind | Second kind | |
|---|---|---|
| Fredholm | \( \int_a^b K\varphi = f \) | \( \varphi = f + \lambda\int_a^b K\varphi \) |
| Volterra | \( \int_a^x K\varphi = f \) | \( \varphi = f + \lambda\int_a^x K\varphi \) |
Homogeneous: \( f = 0 \). Singular: infinite interval or unbounded kernel. A Volterra equation is a Fredholm equation with \( K = 0 \) for \( t > x \).
Conversions
$$\boxed{\int_a^x\!\!\int_a^s \varphi(t)\,dt\,ds = \int_a^x (x - t)\varphi(t)\,dt} \qquad \boxed{\frac{d}{dx}\int_a^x K\varphi = K(x, x)\varphi(x) + \int_a^x K_x\varphi}$$- IVP \( y'' = F - qy \), \( y(a) = \alpha \), \( y'(a) = \beta \): \( y = \alpha + \beta(x - a) + \int_a^x (x - t)[F - qy]\,dt \).
- IVP with \( y' \) term: take \( u = y'' \); the free term at \( x = a \) equals \( y''(a) \).
- BVP \( y'' + \lambda y = 0 \) on \( [0, 1] \): \( y = \lambda\int_0^1 G y \).
| Boundary conditions | \( G(x, t) \) | \( \lambda_n \) | \( \int G(x, x) \) | \( \iint G^2 \) |
|---|---|---|---|---|
| \( y(0) = y(1) = 0 \) | \( \min(x, t) - xt \) | \( n^2\pi^2 \) | \( \frac{1}{6} \) | \( \frac{1}{90} \) |
| \( y(0) = 0 = y'(1) \) | \( \min(x, t) \) | \( \frac{(2n-1)^2\pi^2}{4} \) | \( \frac{1}{2} \) | \( \frac{1}{6} \) |
| \( y'(0) = 0 = y(1) \) | \( 1 - \max(x, t) \) | \( \frac{(2n-1)^2\pi^2}{4} \) | \( \frac{1}{2} \) | \( \frac{1}{6} \) |
Volterra equations
- Continuous kernel: unique solution for every \( \lambda \); no characteristic numbers.
- Polynomial or exponential kernel: differentiate to an ODE; initial values from \( x = a \).
- Difference kernel: \( \Phi(s) = \dfrac{F(s)}{1 - \lambda\hat k(s)} \).
- First kind: needs \( f(a) = 0 \); differentiate; needs \( K(x, x) \ne 0 \) (else differentiate again).
- Abel: \( \int_0^x \frac{\varphi(t)}{\sqrt{x - t}}dt = f \Rightarrow \varphi = \frac{1}{\pi}\frac{d}{dx}\int_0^x \frac{f(t)}{\sqrt{x - t}}dt \). General \( \alpha \in (0, 1) \): factor \( \frac{\sin\alpha\pi}{\pi} \), inner exponent \( 1 - \alpha \).
| Equation | Solution |
|---|---|
| \( \varphi = 1 - \int_0^x (x - t)\varphi \) | \( \cos x \) |
| \( \varphi = x - \int_0^x (x - t)\varphi \) | \( \sin x \) |
| \( \varphi = 1 + \lambda\int_0^x \varphi \) | \( e^{\lambda x} \) |
| \( \varphi = \sin x + 2\int_0^x \cos(x - t)\varphi \) | \( xe^x \) |
| \( \varphi = e^x + \lambda\int_0^x e^{x - t}\varphi \) | \( e^{(1 + \lambda)x} \) |
| \( \int_0^x e^{x - t}\varphi = \sin x \) | \( \cos x - \sin x \) |
| \( \int_0^x \frac{\varphi(t)}{\sqrt{x - t}}dt = x \) | \( \frac{2}{\pi}\sqrt{x} \) |
Separable kernels
$$K = \sum_{i=1}^n a_i(x)b_i(t), \quad c_i = \int b_i\varphi, \quad A_{ij} = \int b_i a_j, \quad F_i = \int b_i f, \quad (I - \lambda A)c = F$$$$\boxed{\text{characteristic numbers: } \det(I - \lambda A) = 0, \ \lambda = 1/\mu \text{ for eigenvalues } \mu \ne 0 \text{ of } A}$$| Kernel | Interval | Characteristic numbers | Eigenfunctions |
|---|---|---|---|
| \( a(x)b(t) \) | any | \( 1/\int ab \) (none if \( \int ab = 0 \)) | \( a(x) \) |
| \( xt \) | \( [0, 1] \) | 3 | \( x \) |
| \( e^{x + t} \) | \( [0, 1] \) | \( \frac{2}{e^2 - 1} \) | \( e^x \) |
| \( \sin x\cos t \) | \( [0, \pi] \) | none | — |
| \( x + t \) | \( [0, 1] \) | \( -6 \pm 4\sqrt3 \) | \( \sqrt3 x \pm 1 \) |
| \( \sin(x + t) \) | \( [0, 2\pi] \) | \( \pm\frac{1}{\pi} \) | \( \sin x \pm \cos x \) |
| \( \cos(x - t) \) | \( [0, 2\pi] \) | \( \frac{1}{\pi} \) (multiplicity 2) | \( \cos x, \sin x \) |
| \( 1 + 3xt \) | \( [-1, 1] \) | \( \frac{1}{2} \) (multiplicity 2) | \( 1, x \) |
Fredholm alternative. \( \lambda \) not characteristic: unique solution for every \( f \). \( \lambda \) characteristic: solvable iff \( \int f\psi = 0 \) for every solution \( \psi \) of \( \psi(x) = \lambda\int K(t, x)\psi(t)\,dt \) (transposed); then infinitely many solutions. Symmetric kernel: test against the eigenfunctions themselves.
Resolvent kernel
$$K_{n+1}(x, t) = \int K(x, s)K_n(s, t)\,ds \ \ (\text{Volterra: } \textstyle\int_t^x), \qquad \Gamma = \sum_{m \ge 1}\lambda^{m-1}K_m, \qquad \varphi = f + \lambda\int\Gamma f$$| Kernel | \( \Gamma(x, t; \lambda) \) |
|---|---|
| \( a(x)b(t) \), \( \alpha = \int ab \) | \( \frac{a(x)b(t)}{1 - \lambda\alpha} \) |
| \( xt \) on \( [0, 1] \) | \( \frac{3xt}{3 - \lambda} \) |
| \( 1 \) on \( [0, 1] \) (Fredholm) | \( \frac{1}{1 - \lambda} \) |
| \( 1 \) (Volterra) | \( e^{\lambda(x - t)} \) |
| \( e^{x - t} \) (Volterra) | \( e^{(1 + \lambda)(x - t)} \) |
| separable | \( \sum a_i[(I - \lambda A)^{-1}]_{ij}b_j \) |
Neumann series converges for \( \lvert\lambda\rvert < \frac{1}{M(b - a)} \), for \( \lvert\lambda\rvert < \frac{1}{B} \) with \( B^2 = \iint K^2 \), and exactly up to the nearest characteristic number. Volterra: every \( \lambda \). Divergence of the series does not mean the equation has no solution.
Symmetric kernels
| Property | Statement |
|---|---|
| Existence | non-zero symmetric kernel has at least one characteristic number |
| Reality | all characteristic numbers real |
| Orthogonality | eigenfunctions for different \( \lambda \) orthogonal |
| Finiteness | finitely many iff separable; otherwise \( \lvert\lambda_n\rvert \to \infty \) |
| Parseval | \( \sum\lambda_n^{-2} = \iint K^2 \) (with multiplicity); so \( \lvert\lambda_1\rvert \ge 1/B \) |
| Mercer (continuous, positive) | \( \sum\lambda_n^{-1} = \int K(x, x)\,dx \) |
| Solution | \( \varphi = f + \lambda\sum\frac{f_n}{\lambda_n - \lambda}\varphi_n \) |
Master trap list
| False claim | Why false |
|---|---|
| “Every kernel has a characteristic number” | \( \sin x\cos t \) on \( [0, \pi] \) has none |
| “Characteristic numbers are \( \det(A - \lambda I) = 0 \)” | that gives \( 1/\lambda \) |
| “Volterra equations can have characteristic numbers” | not with a continuous kernel |
| “Neumann series diverges, so no solution” | \( x + t \), \( \lambda = 1 \): solution \( -6x - 4 \) |
| “Solvability: \( f \perp \) eigenfunctions of \( K \)” | must use the transposed kernel \( K(t, x) \) |
| “At a characteristic number the solution is unique” | never unique when it exists |
| “Fredholm first kind always solvable” | \( \int_0^1 xt\varphi = x^2 \) has no solution |
| “Volterra first kind: just differentiate” | needs \( f(a) = 0 \) first |
| “\( \min(x, t) - xt \) has \( \lambda_n = n\pi \)” | \( \lambda_n = n^2\pi^2 \) |
Part C checklist
Back to the Linear Integral Equations chapter overview. The finite-dimensional analogues are on Eigenvalues and the Cayley–Hamilton Theorem.