Some Delta Function Representations

W. A. Horowitz
November 16, 2010

1 Delta Functions
1.1 Principal Value Representation
The Sokhotsky-Weierstrauss Theorem states that, when considered as a distri-
bution,
1 1
lim+ = PV ∓ i π δ(x). (1)
→0 x ± i x
To see this consider
1 x 
f (x) ≡ lim = lim ∓i 2 . (2)
→0+ x ± i →0+ x2 + 2 x + 2
For x 6= 0, f (x) = 1/x. For x = 0, f (x) = ∓ i ∞. So it already looks like the
form of Eq. (1); we really need only check the normalization. If we integrate
f (x) we get, for a < 0 < b,
Z b Z b  
1 b a
dx f (x) = P V dx ∓ i lim+ tan−1 − tan−1 (3)
a a x →0  
Z b
1
= PV dx ∓ i π, (4)
a x

where the Cauchy principal value is defined as the exclusion of a singular value
from a region of integration; e.g. for a < 0 < b
Z b Z − Z b
1 1 1
PV dx ≡ lim+ dx + dx . (5)
a x →0 a x  x

To be specific as to the application of Eq. (1), for a smooth function g(x)

and a < 0 < b
Z b Z b
1 1
lim+ dx g(x) = PV g(x) ∓ i π g(0). (6)
→0 a x ± i a x

1
1.2 Integral Representation
To get the normalization correct for the integral representation of a delta func-
tion note that:
Z ∞
dxeikx f (x) = f˜(k)
−∞
Z ∞
1
f (x) = dke−ikx f˜(k). (7)
2π −∞

Therefore taking f (x) = δ(x), f˜(k) = 1 and, we have that

Z ∞
1
δ(k) = dxeikx . (8)
2π −∞

We also get, for free, another form of the delta function:

Z 0 Z ∞ 
1 i(k−iτ )x i(k+iτ )x
δ(k) = lim+ dxe + dxe
τ →0 2π −∞ 0
 
1 1 1
= lim+ −
τ →0 2π i(k − iτ ) i(k + iτ )
 
1 1 1
= lim+ +
τ →0 2π τ + ik τ − ik
1 τ
= lim+ . (9)
τ →0 π τ 2 + k 2