Parameter-dependent integrals and applications

Some integrals refuse to yield to any standard trick — substitution, parts, partial fractions — and yet fall apart the moment you slip an extra constant α\alpha into the integrand and differentiate with respect to it. This is the parameter-differentiation technique: introduce a parameter, differentiate under the integral sign (Leibnitz's rule), integrate the simplified result, and pin down the constant with a known boundary value. The same "sum over a region" logic that makes this work also underlies the standard geometric applications of multiple integrals: computing plane areas by double integrals, volumes by double or triple integrals, and curved surface areas by a two-variable analogue of arc length. This article ties both threads together — the differentiation trick as an evaluation tool, and the region-integral machinery as a measurement tool — exactly as MA10002 pairs them at the close of the integral calculus unit.

The core insight

Differentiating a parameter-dependent integral under the integral sign (Leibnitz's rule) turns hard integrals into easy ones, while the same double/triple-integral machinery — switched to polar, cylindrical, or spherical coordinates when the region is symmetric — computes plane areas, volumes, and curved surface areas.

The core idea

Parameter differentiation as a tool. For fixed limits a,ba,b independent of α\alpha, Leibnitz's rule states

ddα∫abf(x,α) dx=∫ab∂f∂α dx,\frac{d}{d\alpha}\int_a^b f(x,\alpha)\,dx = \int_a^b \frac{\partial f}{\partial \alpha}\,dx,

where: f(x,α)f(x,\alpha) is the integrand depending on both the integration variable xx and the parameter α\alpha, and the right side differentiates inside the integral, treating xx as fixed. The strategy: pick a parameter that makes ∂f/∂α\partial f/\partial\alpha far simpler than ff itself, integrate the simplified expression back with respect to α\alpha, and fix the resulting constant of integration using one special value of α\alpha (often α→∞\alpha\to\infty or α=0\alpha=0, where the original integral is either obviously zero or independently known).

Plane area via a double integral. For a plane region RR,

Area=∬RdA,\text{Area} = \iint_R dA,

with dA=dx dydA = dx\,dy in Cartesian coordinates, or dA=r dr dθdA = r\,dr\,d\theta in polar coordinates (Jacobian J=rJ=r) when RR is circular or sectorial — a disc, an annulus, a region bounded by r=g(θ)r=g(\theta).

Volume via a double or triple integral. For a solid of height z=f(x,y)z=f(x,y) standing over a base region RR,

Volume=∬Rf(x,y) dA,\text{Volume} = \iint_R f(x,y)\,dA,

while a genuinely three-dimensional solid VV uses Volume=∭VdV\text{Volume} = \iiint_V dV, with dV=r dr dθ dzdV = r\,dr\,d\theta\,dz in cylindrical coordinates or dV=ρ2sin⁡ϕ dρ dϕ dθdV = \rho^2\sin\phi\,d\rho\,d\phi\,d\theta in spherical coordinates when the solid has that symmetry (a cylinder, a cone, a sphere or a spherical cap).

Surface area of z=f(x,y)z=f(x,y) over RR.

S=∬R1+(∂f∂x)2+(∂f∂y)2  dA.S = \iint_R \sqrt{1+\left(\frac{\partial f}{\partial x}\right)^2 + \left(\frac{\partial f}{\partial y}\right)^2}\;dA.

where: fx=∂f/∂xf_x=\partial f/\partial x and fy=∂f/∂yf_y=\partial f/\partial y are the surface's two partial slopes at each point, and the square root is the local "tilt factor" that stretches a flat patch dAdA into the corresponding tilted patch of surface. It is the direct two-variable analogue of the arc-length formula s=∫1+(y′)2 dxs=\int\sqrt{1+(y')^2}\,dx: one slope becomes two, and length becomes area.

Figure
A tilted surface z equals f of x y sitting above a flat disc region in the x y plane, with a shaded surface patch above a shaded area patch and slope arrows fx and fy shown at a point
Surface area as a tilted stretch of a flat region region R, patch dA surface z = f(x,y) patch stretches to dS f_x f_y S = double integral over R of the square root of 1 + f_x squared + f_y squared, dA
Derivation AEvaluating a parameter-dependent integral by differentiating under the integral sign

The goal is to evaluate I(α)=∫0∞e−αxsin⁡xx dxI(\alpha) = \displaystyle\int_0^{\infty} \dfrac{e^{-\alpha x}\sin x}{x}\,dx for α>0\alpha>0 — an integral with no elementary antiderivative in xx — by introducing the parameter's derivative as a shortcut, using only the given fact that I(α)→0I(\alpha)\to 0 as α→∞\alpha\to\infty.

Step 1. Differentiate under the integral sign with respect to α\alpha, holding xx as the integration variable and the limits 0,∞0,\infty fixed. Since ∂∂α ⁣(e−αxsin⁡xx)=−x⋅e−αxsin⁡xx=−e−αxsin⁡x\dfrac{\partial}{\partial\alpha}\!\left(\dfrac{e^{-\alpha x}\sin x}{x}\right) = -x\cdot\dfrac{e^{-\alpha x}\sin x}{x} = -e^{-\alpha x}\sin x, the awkward 1/x1/x cancels entirely, leaving

I′(α)=−∫0∞e−αxsin⁡x dx.I'(\alpha) = -\int_0^\infty e^{-\alpha x}\sin x\,dx.

Step 2. Evaluate the remaining integral using the standard Laplace-type result ∫0∞e−αxsin⁡(bx) dx=bα2+b2\int_0^\infty e^{-\alpha x}\sin(bx)\,dx = \dfrac{b}{\alpha^2+b^2} with b=1b=1, giving

I′(α)=−1α2+1.I'(\alpha) = -\frac{1}{\alpha^2+1}.

Step 3. Integrate back with respect to α\alpha: I(α)=−arctan⁡(α)+CI(\alpha) = -\arctan(\alpha) + C. To fix CC, use the given boundary condition I(α)→0I(\alpha)\to 0 as α→∞\alpha\to\infty; since arctan⁡(α)→π/2\arctan(\alpha)\to \pi/2, this forces 0=−π/2+C0 = -\pi/2 + C, so C=π/2C=\pi/2. Hence

I(α)=π2−arctan⁡(α).I(\alpha) = \frac{\pi}{2} - \arctan(\alpha).

Step 4. Read off a bonus result. Setting α=0\alpha=0 gives the classical Dirichlet integral ∫0∞sin⁡xx dx=π2\displaystyle\int_0^\infty \frac{\sin x}{x}\,dx = \frac{\pi}{2} — a value that direct methods cannot easily reach, obtained here purely from differentiating, integrating, and fixing one constant.

Worked example

Problem. Find the surface area of the part of the paraboloid z=x2+y2z=x^2+y^2 that lies above the disc x2+y2≤1x^2+y^2\le 1 (all lengths in metres).

Step 1. Compute the partial slopes: f(x,y)=x2+y2f(x,y)=x^2+y^2 gives fx=2xf_x=2x, fy=2yf_y=2y, so 1+fx2+fy2=1+4x2+4y21+f_x^2+f_y^2 = 1+4x^2+4y^2, and the surface-area integral is

S=∬R1+4x2+4y2  dA,R:x2+y2≤1.S = \iint_R \sqrt{1+4x^2+4y^2}\;dA, \qquad R: x^2+y^2\le 1.

Step 2. Switch to polar coordinates, since RR is a disc and the integrand depends only on x2+y2=r2x^2+y^2=r^2: with dA=r dr dθdA=r\,dr\,d\theta, 0≤r≤10\le r\le1, 0≤θ≤2π0\le\theta\le2\pi,

S=∫02π∫011+4r2  r dr dθ.S = \int_0^{2\pi}\int_0^1 \sqrt{1+4r^2}\;r\,dr\,d\theta.

Step 3. Evaluate the inner integral by substituting u=1+4r2u=1+4r^2, du=8r drdu=8r\,dr (so r=0⇒u=1r=0\Rightarrow u=1, r=1⇒u=5r=1\Rightarrow u=5):

∫011+4r2 r dr=18∫15u du=18[23u3/2]15=112(55−1)≈0.8484.\int_0^1 \sqrt{1+4r^2}\,r\,dr = \frac18\int_1^5\sqrt u\,du = \frac18\left[\frac23u^{3/2}\right]_1^5 = \frac{1}{12}\left(5\sqrt5-1\right)\approx 0.8484.

Step 4. The outer integral over θ\theta just multiplies by 2π2\pi, since the inner result is independent of θ\theta:

S=2π⋅55−112=π(55−1)6≈5.330 m2.S = 2\pi\cdot\frac{5\sqrt5-1}{12} = \frac{\pi(5\sqrt5-1)}{6} \approx 5.330\text{ m}^2.

Answer: S=π(55−1)6≈5.330 m2S = \dfrac{\pi(5\sqrt5-1)}{6}\approx 5.330\text{ m}^2.

Competitive stretch

The same disc also bounds a solid: the volume trapped under the paraboloid and above the disc is V=∬R(x2+y2) dA=∫02π∫01r2⋅r dr dθ=2π[r44]01=π2≈1.571 m3V=\iint_R (x^2+y^2)\,dA = \int_0^{2\pi}\int_0^1 r^2\cdot r\,dr\,d\theta = 2\pi\left[\dfrac{r^4}{4}\right]_0^1 = \dfrac{\pi}{2}\approx1.571\text{ m}^3. Comparing the two results is instructive: the surface area needed the tilt-factor square root, while the volume underneath is a plain height integral — same region, same polar setup, structurally different integrand.

Common traps

  • Forgetting to fix the constant of integration after integrating I′(α)I'(\alpha) back — the parameter-differentiation method is incomplete until CC is pinned down using a known value or limit of I(α)I(\alpha).
  • Applying the surface-area formula with only one partial derivative squared — it needs both fx2f_x^2 and fy2f_y^2 under the square root, unlike single-variable arc length which has just one.
  • Using Cartesian dA=dx dydA=dx\,dy for a circular region when polar dA=r dr dθdA=r\,dr\,d\theta would make the integral far simpler — always look at the region's shape before choosing coordinates.
  • Confusing "volume under a surface" (∬Rf dA\iint_R f\,dA, a double integral) with "volume of a genuine 3D solid" (∭VdV\iiint_V dV, a triple integral) — pick whichever matches how the solid is actually described.

Ask the tutor: if you are unsure whether a given problem wants parameter differentiation, an area/volume setup, or a surface-area setup — or which coordinate system to switch to — ask the AI tutor to walk through a few more mixed examples with you.

Why does differentiating under the integral sign make an integral easier instead of harder?

Because the parameter often appears in a way that, once you differentiate, cancels the exact feature that was blocking a direct evaluation — here the 1/x1/x that made ∫0∞sin⁡xx dx\int_0^\infty \frac{\sin x}{x}\,dx resist elementary methods vanished the moment the extra factor −x-x from differentiating e−αxe^{-\alpha x} multiplied it away. The resulting integral, ∫0∞e−αxsin⁡x dx\int_0^\infty e^{-\alpha x}\sin x\,dx, is a standard exponential-times-trig integral you already know how to do. You then have to undo one derivative (integrate back) and fix one constant, but that trade is usually far cheaper than attacking the original integral head-on.

How do I decide whether a problem needs area, volume, or surface area?

Look at what is being measured. If the answer is a flat two-dimensional region's size, it is ∬RdA\iint_R dA — area. If a solid's capacity or amount of "stuff" is wanted, it is either ∬Rf dA\iint_R f\,dA (volume under one surface, over a flat base) or ∭VdV\iiint_V dV (a fully three-dimensional solid described by inequalities in x,y,zx,y,z) — check whether the solid is described as "under a surface" or as "bounded by these surfaces on all sides." If the question asks for the area of a curved sheet itself, like a dome or a paraboloid's skin, it is the surface-area formula with the square-root tilt factor. Matching the question's wording to one of these three integral types is the first and most important step.

Do I need to memorize the Leibnitz rule formula, or just the technique?

For MA10002, treat differentiation under the integral sign as a black-box tool: you introduce a parameter, differentiate the whole integral with respect to it (which slides the derivative inside, onto the integrand), simplify, integrate back, and fix the constant. You are not asked to re-derive why swapping derivative and integral is valid — that justification belongs to a more advanced real-analysis course. What you must be fluent in is spotting when a parameter makes an integral tractable, executing the differentiate-simplify-integrate-fix-constant chain cleanly, and not dropping sign errors when the integrand contains e−αxe^{-\alpha x}.

Before this

Where this leads