Home / Learn / Exam prep / GATE Physics / Lesson 1.8
All lessons

Module 1: Mathematical physics · Lesson 1.8 · about 60 min

The Dirac delta and Green's functions

The Dirac delta as a limit and as a rule under the integral: sifting, scaling, delta of a function, derivatives of the delta, the Fourier representation and the delta in three dimensions; then Green's functions for boundary value problems, jump conditions and when G fails to exist, causal Green's functions for oscillators, and the Poisson, screened and heat kernels.

Key points, meanings and formulas on one page: choose "Save as PDF" in the print box.

Module 1 · about 60 minutes. By the end you will be able to evaluate any integral containing δ(x), δ(g(x)), δ′(x) or δ3(𝐫), build the Green’s function of a second-order boundary value problem from its jump conditions, recognise when it does not exist, and use causal Green’s functions to find the response of an oscillator or a first-order system to any force. It uses the transforms of lesson 1.5 and the modes of lesson 1.7.

Any source can be cut into point sources. If you know the response to one point source, the Green's function, the response to any source is the same sum of point responses. The Dirac delta is the mathematical point source.

1. The delta as a rule under the integral

δ(x−a) is zero away from a, with unit area, so that for every smooth f

∫−∞∞f(x)δ(x−a)dx=f(a)

It is not an ordinary function but a limit of narrow, tall functions of unit area: a Gaussian 1επe−x2/ε2, a Lorentzian ε/πx2+ε2, a box of width 2ε and height 12ε, or sin⁡(x/ε)πx. Check the area of any proposed sequence first: 1εe−|x|/ε tends to 2δ(x), not δ(x). Because its integral over x is a pure number, δ(x) has the units of 1/x.

The rules, each from a substitution or an integration by parts:

  • Range: the spike counts only if it lies inside the limits: ∫12δ(x)dx=0.
  • Products: f(x)δ(x−a)=f(a)δ(x−a); in particular xδ(x)=0.
  • Scaling: δ(ax)=δ(x)|a|, so δ(−x)=δ(x).
  • Delta of a function: for simple real roots xi of g,
δ(g(x))=∑iδ(x−xi)|g′(xi)|
  • Derivatives: ∫fδ′(x−a)dx=−f′(a) and ∫fδ(n)(x−a)dx=(−1)nf(n)(a).
  • Step function: θ′(x)=δ(x).
  • Fourier: δ(x)=12π∫−∞∞eikxdk, so ∫e−ikxdx=2πδ(k).

Worked examples. ∫x2δ(x2−4)dx=44+44=2. ∫02π(1+θ)δ(cos⁡θ)dθ=(1+π2)+(1+3π2)=2+2π, from the two zeros of cos⁡θ, each with |sin⁡θ|=1. ∫e2xδ′(x)dx=−2.

2. The delta in three dimensions

δ3(𝐫−𝐫0)=δ(x−x0)δ(y−y0)δ(z−z0) sifts out values in space with no extra Jacobian: ∫fδ3(𝐫−𝐫0)d3r=f(𝐫0). Its units are length⁻³, so a point charge has density ρ=qδ3(𝐫−𝐫0).

The key identity behind Coulomb’s and Gauss’s laws:

∇⋅𝐫ˆr2=4πδ3(𝐫),∇21r=−4πδ3(𝐫)

Away from the origin both are zero, yet the flux of 𝐫ˆ/r2 through any sphere is 4π, so all of the divergence sits at r=0. Do not confuse δ3(𝐫) with a one-dimensional δ(r−R), which is a thin spherical shell: ∫r2δ(r−2)d3r=4π∫r4δ(r−2)dr=64π.

δ(x − a)
a unit point source at a: zero width, unit area, units of 1/length
Green's function G(x, x′)
the response at x to a unit point source at x′, obeying the problem's boundary conditions
Jump condition
what integrating the equation across the source point forces on G or its derivative
Causal (retarded)
zero before the source acts
Physical meaning
The Coulomb potential, the impulse response of a circuit and the spreading of a drop of heat are all Green's functions. Superposition does the rest.

3. Green’s functions for boundary value problems

To solve Lˆu=f(x) with homogeneous boundary conditions, find G with

LˆG(x,x′)=δ(x−x′),u(x)=∫G(x,x′)f(x′)dx′

For a second-order operator d2dx2+… the recipe is always the same:

Away from x′ the equation is homogeneous. Left of x′ take the solution that satisfies the left boundary condition; right of x′ the one that satisfies the right condition.
G is continuous at x = x′ (otherwise G″ would contain δ′).
Integrating the equation across x′: the slope G′ jumps by 1 (divided by the coefficient of the second derivative, if it is not 1).
Two conditions fix the two remaining constants.

Worked example. G′′=δ(x−x′) on [0,1], G=0 at both ends. Left: Ax; right: B(x−1). Continuity: Ax′=B(x′−1); jump: B−A=1. So A=x′−1, B=x′, and G=x<(x>−1). For u′′=1: u=∫01Gdx′=x(x−1)2, which indeed has u′′=1 and zero ends.

For a string under tension T the deflection obeys −Ty′′=f, so G=x<(L−x>)TL: a tent with its peak under the force. The figure builds spread loads out of these tents.

The Green's function of −(y″ + k²y) on [0, 1] with fixed ends, and the response to a spread load built from it. At k = 0 it is a tent x<(1 − x>); as k approaches π it grows without limit.

When G does not exist. For y′′+k2y on [0,L] with zero ends,

G(x,x′)=sin⁡(kx<)sin⁡(k(x>−L))ksin⁡kL

which blows up when kL=nπ. Then sin⁡kx solves the homogeneous problem with both end conditions: forcing a string exactly at one of its normal-mode frequencies, resonance. A solution of Lˆu=f then exists only if f is orthogonal to that mode. Reciprocity: for self-adjoint operators G(x,x′)=G(x′,x).

4. Causal Green’s functions

In time, we want the response to a kick at t′ to vanish before t′. For a first-order system y˙+ay=f: G(t)=θ(t)e−at, so a step input gives y=1−e−ata.

For the damped oscillator mx¨+mγx˙+mω02x=F(t), the kick changes the velocity by 1/m and leaves the position alone, so for t>0 it is a free oscillation from x=0, x˙=1/m:

G(t)=θ(t)e−γt/2sin⁡ω1tmω1,ω1=ω02−γ2/4,x(t)=∫−∞tG(t−t′)F(t′)dt′

Worked example. An undamped oscillator x¨+4x=F at rest has a constant force 8 switched on at t=0: x=∫0tsin⁡2(t−t′)28dt′=2(1−cos⁡2t). It swings up to 4, twice the static deflection: a suddenly applied load overshoots.

In frequency space, G~(ω)=1m(ω02−ω2−iγω) (with eiωt in the transform) has its poles at ±ω1−iγ/2, below the real axis. Closing the contour above for t<0 encloses nothing, so G=0 there: causality is the position of the poles (lesson 1.4).

5. Green’s functions of the big three operators

operator (source δ) condition Green’s function
∇2 in 3D vanishes at infinity −14π|𝐫−𝐫′|
∇2 in 2D (up to a constant) 12πln⁡|𝐫−𝐫′|
−(∇2−μ2) in 3D vanishes at infinity e−μr4πr (Yukawa, Debye)
−d2dx2+μ2 on a line decays e−μ|x−x′|2μ
∂t−κ∂x2 causal θ(t)4πκte−x2/4κt

The first gives the potential of any charge distribution, V=14πε0∫ρ(𝐫′)|𝐫−𝐫′|d3r′. The heat kernel keeps unit area while spreading, so its height falls as t−1/2 (in d dimensions, t−d/2). The Random walk tool shows the same Gaussian appearing from many small random steps.

Worked example. Solve −y′′+4y=e−|x| on the line. With μ=2, y(0)=∫e−2|x′|4e−|x′|dx′=14⋅23=16.

“δ(g(x)) is just δ(x−x0) at the root.” It carries a weight 1/|g′(x0)| at every simple root, and every root inside the range counts. δ(2x−6) is 12δ(x−3), and δ(x2−4) has two spikes of weight 14. Forgetting the absolute value can even produce a negative weight; a delta always has positive area.

  • Sifting ∫fδ(x−a)=f(a); δ(ax)=δ/|a|; δ(g)=∑δ(x−xi)/|g′(xi)|; ∫fδ(n)=(−1)nf(n); θ′=δ; δ=12π∫eikxdk. Units of 1/x.
  • ∇⋅(rˆ/r2)=4πδ3, ∇2(1/r)=−4πδ3; δ3 needs no Jacobian, δ(r−R) does.
  • Second-order G: homogeneous solutions on each side meeting the boundary conditions, G continuous, G′ jumps by 1. G=x<(x>−1) for d2/dx2 on [0,1].
  • G fails to exist at resonance (a homogeneous solution meets the boundary conditions). Self-adjoint problems are reciprocal.
  • Causal: θ(t)e−at; oscillator θ(t)e−γt/2sin⁡ω1tmω1; poles below the real axis.
  • Kernels: −14πr, 12πln⁡ρ, e−μr4πr, e−μ|x|2μ, heat kernel (4πκt)−1/2e−x2/4κt.

Practise: exam questions on this lesson

From warm-up to top-rank level, with full solutions. More on the delta and Green’s functions practice page.

Before this: lesson 1.7, partial differential equations. Next: lesson 1.9, tensors. Delta potentials, with their jump in slope, return in lesson 4.3, and the oscillator in quantum form in lesson 4.4. Back to the GATE Physics course home.

Check yourself

1What is the integral of x² δ(x² − 9) over the whole line?
2For G″ = δ(x − x′) with G = 0 at both ends of an interval, which is true at x = x′?
3What is the causal Green's function of dy/dt + ay = f(t)?

Sources

  • G. B. Arfken, H. J. Weber and F. E. Harris, Mathematical Methods for Physicists, 7th ed. (2013), chapters 1.11 and 10
  • M. L. Boas, Mathematical Methods in the Physical Sciences, 3rd ed. (2006), chapters 8.11 and 13.8
  • K. F. Riley, M. P. Hobson and S. J. Bence, Mathematical Methods for Physics and Engineering, 3rd ed. (2006), chapters 13 and 21
  • D. J. Griffiths, Introduction to Electrodynamics, 4th ed. (2017), sections 1.5 and 2.3
  • GATE Physics (PH) syllabus, Mathematical Physics section, published by the organising institute each year

Your ticks stay in this browser. Save them as a file on the course home.

Updated 2026-11-22. Not affiliated with the IITs, IISc or the GATE organising institute. The general aptitude part of GATE is not covered here. Check the official GATE website for the current pattern. All questions are original.