Show that the fundamental solution to the Laplace equation in 3D is 1/(4π x ).
- User
Show that the fundamental solution to the Laplace equation in 3D is 1/(4π|x|).
Assistant
We need to show that the fundamental solution to the Laplace equation in 3D is 1/(4π|x|). That is, we need to show that Φ(x) = 1/(4π|x|) satisfies ΔΦ = -δ in ℝ^3, where δ is the Dirac delta distribution. Alternatively, we can say that Φ is a solution of ΔΦ = 0 for x ≠ 0 and that it satisfies certain distributional condition.
We need to provide a derivation. The problem likely expects a detailed demonstration: show that for any test function φ, ∫_{ℝ^3} Φ(x) Δφ(x) dx = -φ(0). Or equivalently, show that ΔΦ = -δ. Usually, the approach is to consider the integral over ℝ^3 of Φ Δφ and use Green's identities, or integrate over a domain excluding a ball around the origin and take the limit.
We can present:
Let Φ(x) = 1/(4π|x|). Then for x ≠ 0, compute ΔΦ = 0? Actually, compute Laplacian of 1/r in 3D: In spherical coordinates, for r>0, Δ(1/r) = (1/r^2) ∂/∂r (r^2 ∂/∂r (1/r)) = (1/r^2) ∂/∂r (r^2 * (-1/r^2)) = (1/r^2) ∂/∂r (-1) = 0. So indeed, ΔΦ = 0 for r > 0.
But at the origin there is a singularity, and the distributional Laplacian yields -δ. To show that, we consider a test function φ ∈ C_c^∞(ℝ^3). We want to show ∫_{ℝ^3} Φ Δφ dx = -φ(0). We can do:
∫{ℝ^3} Φ Δφ dx = lim{ε→0} ∫_{|x|>ε} Φ Δφ dx.
Then use Green's second identity on the region Ω_ε = ℝ^3 \ B_ε(0). Since Φ is smooth on Ω_ε and ΔΦ = 0, we have:
∫{Ω_ε} (Φ Δφ - φ ΔΦ) dx = ∫{∂Ω_ε} (Φ ∂_n φ - φ ∂_n Φ) dS,
where ∂_n is the outward normal derivative relative to Ω_ε. The boundary ∂Ω_ε consists of the sphere of radius ε (with normal pointing outward from Ω_ε, i.e., inward into the ball? Let's be careful: Ω_ε = exterior of the ball, so its boundary is the sphere S_ε with outward normal pointing away from Ω_ε, which is inward into the ball (i.e., towards the origin). Usually, we set the outward normal from Ω_ε. For the region outside the ball, the outward normal on the sphere points inward (towards the origin). But it might be easier to consider the region B_R \ B_ε for a large R and then let R → ∞. Since φ has compact support, for large R, the boundary contributions at infinity vanish because φ and its derivatives vanish. So we can consider a bounded domain that includes the support of φ and exclude a small ball around 0. Let's do:
Choose R large enough such that supp(φ) ⊂ B_R(0). Let Ω = B_R(0) \ B_ε(0). Then ∂Ω consists of outer sphere S_R (outward normal pointing outward, i.e., away from origin) and inner sphere S_ε (outward normal relative to Ω, which points away from Ω, i.e., inward toward the origin). Then apply Green's second identity in Ω:
∫Ω (Φ Δφ - φ ΔΦ) dx = ∫{∂Ω} (Φ ∂_n φ - φ ∂_n Φ) dS.
Since ΔΦ = 0 in Ω (as r>ε>0), we have:
∫Ω Φ Δφ dx = ∫{∂Ω} (Φ ∂_n φ - φ ∂_n Φ) dS.
Now, the boundary integral has two parts: over S_R and over S_ε.
On S_R: as R → ∞, φ and ∂_n φ vanish because φ has compact support (and R is large enough to contain supp(φ), actually we choose R fixed such that supp(φ) ⊂ B_R, but then on S_R, φ = 0 and ∂_n φ = 0, so the contribution from S_R is zero). So effectively, we only need to consider the inner sphere S_ε.
On S_ε, the outward normal n points towards the origin, i.e., in the direction of -e_r (since the outward normal relative to Ω points away from Ω, which is into the ball). So ∂_n = -∂/∂r on S_ε. But we need to be consistent with signs.
Let's compute:
On S_ε: x = ε, dS = ε^2 dω (dω solid angle element). Φ = 1/(4π ε). ∂_n Φ = n·∇Φ. Since n points inward, n = -e_r. ∇Φ = ∂/∂r (1/(4π r)) e_r = -1/(4π r^2) e_r. So ∇Φ = -1/(4π ε^2) e_r on the sphere. Then ∂_n Φ = n·∇Φ = (-e_r)·(-1/(4π ε^2) e_r) = +1/(4π ε^2). Alternatively, we can compute: ∂_n = -∂/∂r, so ∂_n Φ = -∂Φ/∂r = -(-1/(4π r^2)) = 1/(4π r^2). So indeed ∂_n Φ = 1/(4π ε^2).
Also ∂_n φ = n·∇φ = (-e_r)·∇φ = - ∂φ/∂r.
Now compute the boundary integral over S_ε:
I_ε = ∫_{S_ε} (Φ ∂_n φ - φ ∂n Φ) dS = ∫{S_ε} [ (1/(4π ε)) (-∂φ/∂r) - φ (1/(4π ε^2)) ] ε^2 dω.
Simplify:
= ∫_{S_ε} [ - (1/(4π ε)) ∂φ/∂r ε^2 - φ (1/(4π ε^2)) ε^2 ] dω? Wait careful:
Term1: Φ ∂_n φ = (1/(4π ε)) * (-∂φ/∂r) = - (1/(4π ε)) ∂φ/∂r. Multiply by dS = ε^2 dω gives: - (1/(4π ε)) ∂φ/∂r * ε^2 dω = - (ε/(4π)) ∂φ/∂r dω.
Term2: - φ ∂_n Φ = - φ * (1/(4π ε^2)). Multiply by dS: - φ * (1/(4π ε^2)) * ε^2 dω = - φ * (1/(4π)) dω.
So I_ε = ∫_{S_ε} [ - (ε/(4π)) ∂φ/∂r - (1/(4π)) φ ] dω.
Thus,
∫_Ω Φ Δφ dx = I_ε.
Now as ε → 0, we want to show that ∫_Ω Φ Δφ dx tends to -φ(0). But note that as ε→0, Ω expands to the whole ball B_R minus the point. Since the integral of Φ Δφ over the whole space (with singularity) is the limit of the integrals over Ω as ε→0. So we have:
∫{ℝ^3} Φ Δφ dx = lim{ε→0} ∫Ω Φ Δφ dx = lim{ε→0} I_ε.
Now evaluate lim_{ε→0} I_ε.
We have I_ε = - (1/(4π)) ∫{S_ε} φ dω - (ε/(4π)) ∫{S_ε} ∂φ/∂r dω.
As ε→0, ∫{S_ε} φ dω → φ(0) * (surface area of unit sphere)?? Wait: dω is the solid angle measure, and ∫{S_ε} dω = 4π (since the total solid angle of a sphere is 4π). But careful: The integral over S_ε with dω is just the solid angle integration, which is over the unit sphere scaled by ε? Actually, we parameterize by angles; dω is the element of solid angle on the unit sphere. For a sphere of radius ε, the surface element dS = ε^2 dω, where dω is the solid angle element (the area element on unit sphere). So our expression I_ε involves integration over dω (not dS). In our derivation we already replaced dS = ε^2 dω and simplified. So ∫{S_ε} φ dω means integrating φ over the unit sphere but with φ evaluated at points on sphere of radius ε. So as ε→0, φ(x) → φ(0) uniformly, so ∫{S_ε} φ dω → φ(0) ∫_{S^2} dω = φ(0) * 4π.
Similarly, ∫_{S_ε} ∂φ/∂r dω: Since ∂φ/∂r is bounded (smooth), the integral is bounded, and multiplied by ε, that term goes to 0.
Thus:
lim_{ε→0} I_ε = - (1/(4π)) * (φ(0) * 4π) - 0 = - φ(0).
Therefore, ∫_{ℝ^3} Φ Δφ dx = -φ(0). This shows that ΔΦ = -δ in the sense of distributions. Hence Φ is the fundamental solution.
We might also want to note that the Laplacian of 1/(4πr) is -δ.
Thus the demonstration is complete.
But the problem statement says "Show that the fundamental solution to the Laplace equation in 3D is 1/(4π|x|)." Possibly they expect a derivation from the equation Δu = δ, using Fourier transform or other methods. But the typical approach is the above using Green's identity.
We should also mention that the fundamental solution is defined as a solution to ΔE = δ, and we have shown that E = 1/(4π|x|) satisfies this.
We may also provide an alternative: using the divergence theorem and properties of the Laplacian.
Thus, answer: show the steps.
Now, as a note, there is also a factor: some define the fundamental solution as a solution to ΔE = δ, others as -ΔE = δ? In many PDE contexts, the fundamental solution of Laplace's equation is the solution to -ΔE = δ? Actually, in Evans, the fundamental solution of Laplace's equation is Φ(x) = -1/(4π|x|) for n=3? Let's check: In Evans' PDE, the fundamental solution of Laplace's equation is defined as the solution to -ΔΦ = δ? Wait, I recall: In Evans, for the Laplace equation, the fundamental solution is defined as the solution of -ΔΦ = δ in ℝ^n. But I think it's a matter of sign convention. In many textbooks, the fundamental solution for the Laplacian is a function Γ such that -ΔΓ = δ. However, some define it as ΔΓ = δ. Which one is more common? In Evans, page 22, for n=3, the fundamental solution is Φ(x) = 1/(4π|x|) and it satisfies -ΔΦ = δ? Let's check: Actually, Evans writes: "We have Δ(1/|x|) = 0 for x ≠ 0, and more precisely Δ(1/|x|) = -c δ_0"? Hmm. I need to recall.
I have a memory: The Newtonian potential in 3D is U = 1/(4πr) and it satisfies Poisson's equation: ΔU = -δ? Let's derive: The potential due to a point charge in electrostatics satisfies ∇^2 φ = -ρ/ε0, and for a point charge at origin, ρ = q δ(r), and φ = q/(4πε0 r). So ∇^2 φ = -q/ε0 δ? Actually, from electrostatics: ∇^2 φ = -ρ/ε0. For a point charge q at origin, ρ = q δ(r), so ∇^2 (q/(4πε0 r)) = - (q/ε0) δ(r). Hence ∇^2 (1/(4π r)) = -δ(r). So that would be Δ(1/(4πr)) = -δ. But in Evans, I think he defines the fundamental solution Φ as the solution to -ΔΦ = δ, so then Φ = 1/(4πr) satisfies -ΔΦ = δ? Let's compute: -Δ(1/(4πr)) = - ( -δ) = δ? Wait, we just computed Δ(1/(4πr)) = -δ? That would give -Δ(1/(4πr)) = δ. So if we want -ΔΦ = δ, then Φ = 1/(4πr) works. In Evans (2nd edition), page 20, he writes: "If we define Φ(x) = -1/(2π) log|x| for n=2, Φ(x) = 1/(n(n-2)α(n) |x|^{n-2}} for n≥3, where α(n) is volume of unit ball. Then -ΔΦ = δ." For n=3, α(3)=4π/3, so n(n-2)α(n) = 31(4π/3) = 4π. So Φ(x) = 1/(4π|x|) indeed, and -ΔΦ = δ. So Evans defines the fundamental solution as solving -ΔΦ = δ. But many other sources define the fundamental solution for the Laplacian as a function E such that ΔE = δ. The difference is just a sign. Which one is correct? Actually, it's a convention. In distribution theory, the fundamental solution for a differential operator L is a distribution E such that L E = δ. So if L = Δ, then ΔE = δ. So then E would be -1/(4πr) because we found Δ(1/(4πr)) = -δ. So if we want ΔE = δ, then E = -1/(4πr). But many authors call 1/(4πr) the fundamental solution (maybe they consider the operator -Δ). For example, in Evans, he calls Φ the fundamental solution of Laplace's equation, but he then writes -ΔΦ = δ. So he's treating the operator -Δ. So it's ambiguous. The problem statement: "Show that the fundamental solution to the Laplace equation in 3D is 1/(4π|x|)." Usually in physics, the fundamental solution is 1/(4π|x|). In mathematics, it's often given as -1/(4π|x|) for ΔE = δ? Let's check: The Laplace equation is Δu = 0. The fundamental solution is a solution to Δu = δ? Actually, a fundamental solution for a linear partial differential operator L is a distribution E such that L E = δ. So for L = Δ, we want ΔE = δ. For L = -Δ, we want -ΔE = δ. So which operator is considered? The "Laplace equation" is Δu = 0. The "fundamental solution" might be defined as a solution to Δu = δ? Not sure. In many PDE texts, the fundamental solution for the Laplacian is defined as the solution to -ΔΓ = δ, because then the solution to Poisson's equation -Δu = f is given by convolution with Γ. Since Poisson's equation often appears as -Δu = f (especially in physics, e.g., electrostatic potential satisfies -Δφ = ρ/ε0). So it's convenient. So the statement likely expects the standard result: the fundamental solution is 1/(4π|x|) and satisfies -ΔΦ = δ. But the phrasing "fundamental solution to the Laplace equation" might be ambiguous. However, the usual representation: The fundamental solution of Laplace's equation in three dimensions is Φ(x) = 1/(4π|x|). And one can show that ΔΦ = -δ. Or sometimes they say Φ is the fundamental solution of the Laplacian, meaning that it satisfies ΔΦ = δ in the sense of distributions? Let's check some sources.
I recall that in Evans, the fundamental solution is defined as the solution to -ΔΦ = δ. He calls it the fundamental solution of Laplace's equation? Actually, on page 20, he says "We next introduce the fundamental solution of Laplace's equation." Then he gives formulas and writes "-ΔΦ = δ". So he defines it as satisfying -ΔΦ = δ. So he calls it the fundamental solution of Laplace's equation (the operator is Laplace). But then the equation is -ΔΦ = δ, not ΔΦ = δ. That seems to be a common convention: the fundamental solution for the Laplacian is the function Γ such that -ΔΓ = δ. I've seen both. Let's check: In distribution theory, if L is a differential operator, a fundamental solution E satisfies L E = δ. If L = Δ, then E should satisfy ΔE = δ. But Evans uses L = -Δ. So it's a matter of definition. Many authors define the fundamental solution for the Laplacian as the solution to ΔE = δ, and then E = -1/(4πr). But then the solution to Poisson's equation Δu = f is u = E * f. But since Δ(E * f) = (ΔE) * f = δ * f = f, that would give Δu = f. So that's consistent: If we want to solve Δu = f, we need E with ΔE = δ. Then u = E * f solves Δu = f. In electrostatics, the potential satisfies Δφ = -ρ/ε0, so they would use E = -1/(4πr) then? Actually, if ΔE = δ, then for f = -ρ/ε0, we get φ = E * f = -1/(4πr) * (-ρ/ε0) = (1/(4πr)) * (ρ/ε0)? That would be (ρ/ε0)(1/(4πr)), which matches the potential for a charge distribution ρ. But the point charge potential is q/(4πε0 r). So if we set f = q δ, then u = E * f = E * (q δ) = q E = q * (-1/(4πr)) = -q/(4πr). That gives negative potential, which is not correct. So to get the correct sign, they would need E = 1/(4πr) and then ΔE = -δ. So for Poisson equation Δu = f, if we want u positive for positive f, we might want E negative. But the sign is just a convention; the physics dictates that the potential for a positive point charge is positive. In electrostatics, the equation is Δφ = -ρ/ε0. So if we define f = -ρ/ε0, then we want a fundamental solution for Δ such that ΔE = δ, then φ = E * f = E * (-ρ/ε0) = -(1/ε0) E * ρ. If E = -1/(4πr), then φ = -(1/ε0)(-1/(4πr))*ρ? Actually, convolution with ρ gives the Newtonian potential: ∫ ρ(y)/|x-y| dy times 1/(4πε0). So the standard potential is (1/(4πε0)) ∫ ρ(y)/|x-y| dy. So the kernel is 1/(4π|x|). So that kernel K satisfies ΔK = -δ. Because Δ (1/(4πr)) = -δ. So the solution to Δu = f is u = K * f? Let's test: If u = K * f, then Δu = Δ(K * f) = (ΔK) * f = (-δ) * f = -f. So Δu = -f. So if we want to solve Δu = -f, then u = K * f solves it. So indeed, if we define Poisson's equation as Δu = -f (like in electrostatics), then the kernel is K = 1/(4πr). So it's all consistent.
Thus, the statement "fundamental solution to the Laplace equation in 3D is 1/(4π|x|)" is ambiguous but common. I'll interpret it as: Show that Φ(x)=1/(4π|x|) is the fundamental solution, i.e., satisfies ΔΦ = -δ in ℝ^3. Or maybe they want to show that it satisfies -ΔΦ = δ. But whichever, the derivation we gave yields ∫ Φ Δφ = -φ(0), which is equivalent to ΔΦ = -δ. If we multiply both sides by -1, we get -ΔΦ = δ. So it's just a sign. So the key is to show the distributional Laplacian yields a Dirac delta with a factor -1.
Given the problem statement exactly: "Show that the fundamental solution to the Laplace equation in 3D is 1/(4π|x|)." Usually in textbooks, they say the fundamental solution of the Laplace equation is the function Φ(x) = 1/(4π|x|) (for n=3) that satisfies -ΔΦ = δ. For example, Evans says: "the fundamental solution of Laplace's equation is Φ(x) = -1/(2π) log|x| (n=2), Φ(x)=1/(n(n-2)α(n)|x|^{n-2}) (n≥3). Then -ΔΦ = δ." So he calls it fundamental solution of Laplace's equation, but the operator is -Δ. So I'll follow that convention: show that Φ(x)=1/(4π|x|) satisfies -ΔΦ = δ. But our derivation gave ∫ Φ Δφ = -φ(0). Multiply by -1: -∫ Φ Δφ = φ(0). But ∫ (-ΔΦ) φ = ∫ Φ (-Δφ) = φ(0)? Wait, careful: If we define T = -ΔΦ, then for test function φ, ⟨T, φ⟩ = ⟨-ΔΦ, φ⟩ = ⟨Φ, -Δφ⟩ = -∫ Φ Δφ dx. Our calculation gave ∫ Φ Δφ dx = -φ(0). So -∫ Φ Δφ dx = φ(0). Thus ⟨-ΔΦ, φ⟩ = φ(0) = ⟨δ, φ⟩. So indeed -ΔΦ = δ. So our derivation shows that -ΔΦ = δ. So we can phrase it as: Φ satisfies -ΔΦ = δ, making it the fundamental solution.
Thus, the answer is correct.
Now we should produce a nicely written solution. Steps:
-
Define Φ(x) = 1/(4π|x|) for x ≠ 0, and as a distribution.
-
For any test function φ ∈ C_c^∞(ℝ^3), consider I = ∫_{ℝ^3} Φ Δφ dx.
-
Because of singularity at 0, we consider the region ℝ^3 \ B_ε(0) and later take ε→0.
-
For large R containing support of φ, consider Ω_ε = B_R(0) \ B_ε(0). Apply Green's second identity: ∫Ω (Φ Δφ - φ ΔΦ) dx = ∫{∂Ω} (Φ ∂_n φ - φ ∂_n Φ) dS.
-
Since ΔΦ = 0 in Ω, we have ∫Ω Φ Δφ dx = ∫{∂Ω} (Φ ∂_n φ - φ ∂_n Φ) dS.
-
On outer sphere S_R, φ and ∂_n φ vanish (since φ=0 on and outside support). So contribution is 0.
-
On inner sphere S_ε, outward normal points toward origin, so ∂_n = -∂_r. Compute:
Φ = 1/(4π ε), ∂_n Φ = -∂_r Φ = -(-1/(4π ε^2)) = 1/(4π ε^2). ∂_n φ = -∂_r φ.
- Then the integral over S_ε yields:
I_ε = ∫_{S_ε} [ (1/(4π ε))(-∂_r φ) - φ (1/(4π ε^2)) ] dS
= ∫_{S^2} [ - (1/(4π ε)) ∂_r φ - (1/(4π ε^2)) φ ] ε^2 dω
= - (1/(4π)) ∫_{S^2} (ε ∂_r φ + φ) dω.
-
As ε→0, ∫{S^2} φ(ε,θ,φ) dω → 4π φ(0), and ∫{S^2} ε ∂_r φ dω → 0.
-
Thus, I = lim_{ε→0} ∫Ω Φ Δφ dx = lim{ε→0} I_ε = -φ(0).
-
Therefore, ∫_{ℝ^3} Φ Δφ dx = -φ(0) for all φ, i.e., ΔΦ = -δ in the sense of distributions.
-
Equivalently, -ΔΦ = δ, so Φ is the fundamental solution.
We can also mention that Φ satisfies ΔΦ = 0 for x≠0.
Thus the solution.
Now, perhaps they also want a derivation using the divergence theorem directly: For any test function, one can write ∫ Φ Δφ = lim_{ε→0} ∫_{|x|>ε} Φ Δφ = ... same.
We'll produce a clear answer.
Now, consider if the problem expects to show that it's a fundamental solution meaning that for any f, the solution to Poisson's equation Δu = f can be expressed as u = Φ * f? But that would require Δ(Φ * f) = f? Actually, if ΔΦ = -δ, then Δ(Φ * f) = (ΔΦ)*f = -δ * f = -f. So that would give Δu = -f. If we want to solve Δu = f, we could use u = -Φ * f. So there is a sign. But the statement likely expects the above.
Anyway, we'll produce the standard proof.
Thus, final answer.
I'll write it in a coherent, step-by-step manner. We show that (\Phi(x)=\frac{1}{4\pi|x|}) satisfies (-\Delta\Phi=\delta) in (\mathbb{R}^{3}) (or equivalently (\Delta\Phi=-\delta)), making it the fundamental solution of Laplace's equation in three dimensions.
1. Definition and regularity
For (x\neq0), (\Phi(x)=\frac{1}{4\pi|x|}) is smooth and satisfies (\Delta\Phi(x)=0).
As a distribution, (\Phi) is locally integrable because (\int_{|x|\le R}\frac{dx}{|x|}=4\pi\int_0^R r,dr<\infty).2. Distributional Laplacian
Let (\varphi\in C_c^{\infty}(\mathbb{R}^{3})) be a test function. We must compute
[ \langle\Delta\Phi,\varphi\rangle = \int_{\mathbb{R}^{3}}\Phi,\Delta\varphi,dx. ]
Because of the singularity at the origin, we excise a ball (B_{\varepsilon}(0)) and later let (\varepsilon\to0).Choose (R>0) so large that (\operatorname{supp}\varphi\subset B_R(0)). For (\varepsilon>0) define
[ \Omega_{\varepsilon}= B_R(0)\setminus B_{\varepsilon}(0). ]
Its boundary (\partial\Omega_{\varepsilon}) consists of the outer sphere (S_R) (radius (R), outward normal (\mathbf{n}) pointing away from the origin) and the inner sphere (S_{\varepsilon}) (radius (\varepsilon), outward normal relative to (\Omega_{\varepsilon}) pointing toward the origin, i.e. (\mathbf{n}=-\mathbf{e}_r)).3. Green's identity
On (\Omega_{\varepsilon}) both (\Phi) and (\varphi) are smooth, and (\Delta\Phi=0) for (r>0). Green's second identity gives
[ \int_{\Omega_{\varepsilon}} (\Phi,\Delta\varphi - \varphi,\Delta\Phi),dx = \int_{\partial\Omega_{\varepsilon}} \left( \Phi,\partial_{\mathbf{n}}\varphi - \varphi,\partial_{\mathbf{n}}\Phi \right)dS. ]
Since (\Delta\Phi=0) in (\Omega_{\varepsilon}),
[ \int_{\Omega_{\varepsilon}} \Phi,\Delta\varphi,dx = \int_{\partial\Omega_{\varepsilon}} \left( \Phi,\partial_{\mathbf{n}}\varphi - \varphi,\partial_{\mathbf{n}}\Phi \right)dS. ]
4. Outer boundary contribution
On (S_R), because (\operatorname{supp}\varphi\subset B_R(0)), we have (\varphi=0) and (\partial_{\mathbf{n}}\varphi=0) (the normal derivative also vanishes on the boundary of the support). Hence the integral over (S_R) is zero.
5. Inner boundary calculation
On (S_{\varepsilon}):
- (\Phi = \dfrac{1}{4\pi\varepsilon}).
- The outward normal derivative (\partial_{\mathbf{n}} = \mathbf{n}\cdot\nabla) with (\mathbf{n}=-\mathbf{e}r), so (\partial{\mathbf{n}} = -\partial_r).
Therefore
[ \partial_{\mathbf{n}}\Phi = -\partial_r\Phi = -\frac{d}{dr}\left(\frac{1}{4\pi r}\right)\Big|{r=\varepsilon} = -\left(-\frac{1}{4\pi r^2}\right)\Big|{r=\varepsilon} = \frac{1}{4\pi\varepsilon^2}. ] - (\partial_{\mathbf{n}}\varphi = -\partial_r\varphi).
- The surface element is (dS = \varepsilon^2 d\omega), where (d\omega) is the solid angle element on the unit sphere.
Thus the inner boundary integral becomes
[ \begin{aligned} I_{\varepsilon}&=\int_{S_{\varepsilon}}\left( \Phi,\partial_{\mathbf{n}}\varphi - \varphi,\partial_{\mathbf{n}}\Phi\right)dS\[4pt] &=\int_{S_{\varepsilon}}\left[ \frac{1}{4\pi\varepsilon},(-\partial_r\varphi) - \varphi,\frac{1}{4\pi\varepsilon^2}\right]\varepsilon^2 d\omega\[4pt] &=\int_{S^2}\left[ -\frac{1}{4\pi},\varepsilon,\partial_r\varphi - \frac{1}{4\pi},\varphi \right] d\omega\[4pt] &= -\frac{1}{4\pi}\int_{S^2}\bigl( \varepsilon,\partial_r\varphi + \varphi \bigr)d\omega. \end{aligned} ]
6. Limit (\varepsilon\to0)
As (\varepsilon\to0):
- Because (\varphi) is smooth, (\varphi(\varepsilon,\omega)\to\varphi(0)) uniformly on the sphere, so
[ \int_{S^2}\varphi(\varepsilon,\omega),d\omega ;\longrightarrow; 4\pi,\varphi(0). ] - (\partial_r\varphi) is bounded, hence
[ \left|\int_{S^2}\varepsilon,\partial_r\varphi,d\omega\right| \le \varepsilon\cdot 4\pi\cdot \sup|\partial_r\varphi| ;\longrightarrow;0. ]
Therefore
[ \lim_{\varepsilon\to0} I_{\varepsilon} = -\frac{1}{4\pi}\cdot 4\pi,\varphi(0) = -\varphi(0). ]
7. Conclusion
We have
[ \int_{\mathbb{R}^{3}}\Phi,\Delta\varphi,dx = \lim_{\varepsilon\to0}\int_{\Omega_{\varepsilon}}\Phi,\Delta\varphi,dx = \lim_{\varepsilon\to0} I_{\varepsilon} = -\varphi(0). ]
Thus, in the sense of distributions,
[ \langle\Delta\Phi,\varphi\rangle = -\varphi(0) = -\langle\delta,\varphi\rangle, ]
i.e., (\Delta\Phi = -\delta). Equivalently, (-\Delta\Phi = \delta).
Hence (\Phi(x)=\frac{1}{4\pi|x|}) is the fundamental solution of Laplace's equation in three dimensions.
-
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.