First Year Engineering is coming soon

We're still writing and reviewing the notes for every subject, and we'll open it when every subject is ready. Meanwhile you can practise aptitude, ask Aria about any subject, or see the branches that are ready.

Jacobians

Jacobian determinants: definitions, properties, chain rule, inverse rule, and use in change of variables for multiple integrals, with examples in 2D/3D (polar, cylindrical, spherical).

Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Engineers often change variables to simplify geometry or equations, for example converting Cartesian coordinates to polar or spherical in area/volume integrals or transforming state variables in control and thermodynamics. The Jacobian tells how small areas/volumes scale under such transformations, and it is the key factor in correctly changing variables in multiple integrals.

Key ideas

  • Definition: Jacobian matrix — For a differentiable mapping from ℝ^n to ℝ^n, (u₁, u₂, …, uₙ) = (u₁(x₁,…,xₙ), …, uₙ(x₁,…,xₙ)), the Jacobian matrix is the n×n matrix of first-order partial derivatives: [∂uᵢ/∂xⱼ]. Each row corresponds to one uᵢ and each column to one xⱼ.

  • Definition: Jacobian determinant — The determinant of the Jacobian matrix, denoted as J = ∂(u₁,…,uₙ)/∂(x₁,…,xₙ) = det[∂uᵢ/∂xⱼ]. It measures the local linear scaling (and orientation) of volume (area in 2D) caused by the transformation.

  • Geometric meaning: For a small rectangle/parallelepiped in x-space with side lengths Δx₁,…,Δxₙ, its image in u-space has an n-dimensional volume approximately |J| Δx₁⋯Δxₙ. The sign of J indicates orientation (positive preserves, negative reverses). In integrals of volume or area, we use |J| (the absolute value) because measure is unsigned.

  • Exact statement (Change of Variables Theorem, informal exam form sufficient for engineers): If T maps an open region Rₓ ⊂ ℝⁿ to Rᵤ ⊂ ℝⁿ by a continuously differentiable one-to-one transformation with continuous inverse (a C¹ diffeomorphism) and Jacobian determinant J(x) ≠ 0 on Rₓ, then for an integrable function f on Rᵤ, ∫∫…∫{Rᵤ} f(u) du = ∫∫…∫{Rₓ} f(T(x)) |J(x)| dx. Assumptions commonly stated: T is bijective between regions with piecewise smooth boundaries, u = T(x) is C¹ with nonzero J everywhere, and T⁻¹ exists and is C¹.

  • Statement (Chain rule for Jacobians): If u = u(x, y), v = v(x, y) and x = x(r, s), y = y(r, s), then ∂(u, v)/∂(r, s) = ∂(u, v)/∂(x, y) · ∂(x, y)/∂(r, s). More generally, for composition of mappings, Jacobian determinants multiply.

  • Statement (Inverse function Jacobian rule): If a differentiable transformation u = T(x) has inverse x = T⁻¹(u) and J = ∂u/∂x ≠ 0, then ∂x/∂u = 1 / (∂u/∂x). In n dimensions, det(∂x/∂u) = 1 / det(∂u/∂x).

  • Statement (Functional dependence test): For two functions u(x, y), v(x, y) with continuous first partial derivatives in a region R, if there exists a functional relation F(u, v) = 0 not identically zero in R, then the Jacobian ∂(u, v)/∂(x, y) = 0 at all points where ∇F ≠ 0. Conversely, if ∂(u, v)/∂(x, y) ≠ 0 at a point, then near that point u and v are functionally independent (no nontrivial relation F(u, v) = 0 holds locally).

  • Derivation sketch (area scaling in 2D): Consider u = u(x, y), v = v(x, y) with continuous partials. At a point (x₀, y₀), the linearization gives small changes: Δu ≈ (∂u/∂x)Δx + (∂u/∂y)Δy, Δv ≈ (∂v/∂x)Δx + (∂v/∂y)Δy. The image of a small rectangle spanned by vectors (Δx, 0) and (0, Δy) becomes two vectors in uv-plane: a = ( (∂u/∂x)Δx, (∂v/∂x)Δx ), b = ( (∂u/∂y)Δy, (∂v/∂y)Δy ). The area of the parallelogram spanned by a and b is |det[a b]| = | (∂u/∂x)(∂v/∂y) − (∂u/∂y)(∂v/∂x) | · Δx Δy = |∂(u, v)/∂(x, y)| Δx Δy. This motivates the absolute value in integrals.

  • Standard coordinate transforms and their Jacobians:

    • Polar (2D): x = r cosθ, y = r sinθ, with r ≥ 0, θ in radians. Jacobian |∂(x, y)/∂(r, θ)| = r.
    • Cylindrical (3D): x = r cosθ, y = r sinθ, z = z. Jacobian |∂(x, y, z)/∂(r, θ, z)| = r.
    • Spherical (3D, physics/engineering convention): x = ρ sinφ cosθ, y = ρ sinφ sinθ, z = ρ cosφ, with ρ ≥ 0 (radial distance), 0 ≤ φ ≤ π (polar angle from +z), 0 ≤ θ < 2π (azimuth). Jacobian |∂(x, y, z)/∂(ρ, φ, θ)| = ρ² sinφ.
  • What diagram to draw and label (for an exam answer on meaning of Jacobian): Draw in the x–y plane a small rectangle centered at (x₀, y₀) with sides labeled Δx and Δy aligned with axes. Next to it, in the u–v plane, draw the image as a small parallelogram. Label the vectors a and b as the images of unit steps in x and y, with components (∂u/∂x, ∂v/∂x)Δx and (∂u/∂y, ∂v/∂y)Δy. Mark the area as approximately |∂(u, v)/∂(x, y)| ΔxΔy. Indicate orientation with an arrow showing that a sign change flips orientation.

  • When assumptions fail: If J = 0 at a point, the mapping locally flattens/folds space; inverse may not exist, and change of variables may be invalid there. If T is not one-to-one over the integration region, you must split the region into subregions where it is one-to-one and sum integrals, each with |J|.

Formulas

∂(u, v)/∂(x, y) = (∂u/∂x)(∂v/∂y) − (∂u/∂y)(∂v/∂x) — 2D Jacobian determinant. Symbols u, v, x, y are variables (dimensionless unless variables carry units). Applies for differentiable u(x, y), v(x, y).

∂(u₁, u₂, …, uₙ)/∂(x₁, x₂, …, xₙ) = det[∂uᵢ/∂xⱼ] — nD Jacobian determinant as a determinant. All variables are scalars (dimensionless unless specified). Applies for C¹ mappings.

∫∫_{Rᵤ} f(u, v) du dv = ∫∫_{Rₓ} f(u(x, y), v(x, y)) |∂(u, v)/∂(x, y)| dx dy — 2D change of variables formula. Units: area measure du dv and dx dy are in square units consistent with u, v or x, y. Applies when mapping is one-to-one C¹ with nonzero Jacobian.

∭_{Rᵤ} f(u, v, w) du dv dw = ∭_{Rₓ} f(u(x, y, z), v(x, y, z), w(x, y, z)) |∂(u, v, w)/∂(x, y, z)| dx dy dz — 3D change of variables. Volume elements in cubic units.

∂(u, v)/∂(r, s) = ∂(u, v)/∂(x, y) · ∂(x, y)/∂(r, s) — Chain rule for Jacobians in 2D. Variables are dimensionless unless given. Applies to composed C¹ mappings.

∂(x, y)/∂(u, v) = 1 / ( ∂(u, v)/∂(x, y) ) — Inverse rule in 2D. Nonzero denominator required. Applies when inverse exists and is C¹.

|∂(x, y)/∂(r, θ)| = r — Polar coordinate Jacobian. r in metres (if x, y in metres), θ in radians. Applies to x = r cosθ, y = r sinθ.

|∂(x, y, z)/∂(r, θ, z)| = r — Cylindrical coordinate Jacobian. r in metres, θ in radians, z in metres. Applies to x = r cosθ, y = r sinθ, z = z.

|∂(x, y, z)/∂(ρ, φ, θ)| = ρ² sinφ — Spherical coordinate Jacobian. ρ in metres, φ and θ in radians. Applies to x = ρ sinφ cosθ, y = ρ sinφ sinθ, z = ρ cosφ.

If ∂(u, v)/∂(x, y) ≠ 0 at (x₀, y₀), then u, v are locally independent near (x₀, y₀) — Functional independence criterion (qualitative). Variables dimensionless. Applies for C¹ functions.

Worked examples

Example 1: Jacobian for a given 2D transformation and its inverse Given u = x² − y² and v = 2xy.

  1. Compute J₁ = ∂(u, v)/∂(x, y).
  2. Find ∂(x, y)/∂(u, v) at a nonzero point.
  3. Verify the chain rule with a secondary transform x = r cosθ, y = r sinθ.

Step 1: Compute partial derivatives.

  • ∂u/∂x = 2x; ∂u/∂y = −2y
  • ∂v/∂x = 2y; ∂v/∂y = 2x So, J₁ = (2x)(2x) − (−2y)(2y) = 4x² + 4y² = 4(x² + y²).

Step 2: Inverse Jacobian at a point with x² + y² ≠ 0.

  • ∂(x, y)/∂(u, v) = 1 / J₁ = 1 / [4(x² + y²)]. At, for example, (x, y) = (1, 2): J₁ = 4(1² + 2²) = 4(5) = 20, so ∂(x, y)/∂(u, v) = 1/20.

Step 3: Introduce polar: x = r cosθ, y = r sinθ. Compute J₂ = ∂(x, y)/∂(r, θ) = r. By chain rule, ∂(u, v)/∂(r, θ) = ∂(u, v)/∂(x, y) · ∂(x, y)/∂(r, θ) = 4(x² + y²) · r. Since x² + y² = r², we get ∂(u, v)/∂(r, θ) = 4r² · r = 4r³. Direct check in polar: u = x² − y² = r²(cos²θ − sin²θ) = r² cos2θ; v = 2xy = 2r² sinθ cosθ = r² sin2θ. Then

  • ∂u/∂r = 2r cos2θ; ∂u/∂θ = −2r² sin2θ
  • ∂v/∂r = 2r sin2θ; ∂v/∂θ = 2r² cos2θ So det = (2r cos2θ)(2r² cos2θ) − (−2r² sin2θ)(2r sin2θ) = 4r³(cos²2θ + sin²2θ) = 4r³, which matches the chain rule result. Hence verified. Final answers:
  • J₁ = 4(x² + y²)
  • ∂(x, y)/∂(u, v) = 1/[4(x² + y²)] (defined where x² + y² ≠ 0)
  • ∂(u, v)/∂(r, θ) = 4r³

Example 2: Change of variables in a double integral using polar coordinates Evaluate I = ∬_{D} (x² + y²) dA, where D is the disk x² + y² ≤ a², a > 0.

Step 1: Choose transformation: x = r cosθ, y = r sinθ with r ∈ [0, a], θ ∈ [0, 2π]. Jacobian |∂(x, y)/∂(r, θ)| = r.

Step 2: Write integrand and area element.

  • x² + y² = r².
  • dA = dx dy = |J| dr dθ = r dr dθ.

Step 3: Set limits and integrate. I = ∫{θ=0}^{2π} ∫{r=0}^{a} r² · r dr dθ = ∫{0}^{2π} ∫{0}^{a} r³ dr dθ. Compute inner integral: ∫{0}^{a} r³ dr = (1/4) a⁴. Then outer integral: ∫{0}^{2π} (1/4) a⁴ dθ = (1/4) a⁴ (2π) = (π/2) a⁴.

Final answer: I = (π/2) a⁴.

Example 3: Triple integral in spherical coordinates (volume of a sphere and a check) Compute V = ∭_{x²+y²+z²≤R²} 1 dV using spherical coordinates.

Step 1: Transformation and limits.

  • x = ρ sinφ cosθ, y = ρ sinφ sinθ, z = ρ cosφ.
  • Jacobian: |∂(x, y, z)/∂(ρ, φ, θ)| = ρ² sinφ.
  • Region: ρ from 0 to R, φ from 0 to π, θ from 0 to 2π (entire sphere).

Step 2: Set up the integral. V = ∫{θ=0}^{2π} ∫{φ=0}^{π} ∫_{ρ=0}^{R} 1 · (ρ² sinφ) dρ dφ dθ.

Step 3: Integrate step-by-step.

  • Inner: ∫_{0}^{R} ρ² dρ = (1/3) R³.
  • Middle: ∫_{0}^{π} sinφ dφ = 2.
  • Outer: ∫_{0}^{2π} dθ = 2π. Multiply: V = (1/3) R³ × 2 × 2π = (4/3) π R³.

Final answer: V = (4/3) π R³.

Common mistakes

  • Wrong order of variables in the Jacobian: ∂(u, v)/∂(x, y) ≠ ∂(v, u)/∂(x, y); swapping rows or columns changes sign.
  • Dropping factors in standard transforms: Missing the factor r in polar or ρ² sinφ in spherical, leading to major errors in integrals.
  • Applying change of variables when the mapping is not one-to-one over the region: E.g., using θ from 0 to 2π and also doubling a symmetric sector, causing double counting.
  • Using degrees instead of radians for trigonometric derivatives: In calculus, θ is in radians; otherwise derivatives like d/dθ (sinθ) change by factors of π/180.
  • Assuming nonzero Jacobian guarantees global invertibility: It ensures local invertibility only; for global, you must check injectivity and the domain.

For your exam

  • Derivations: Derive the 2D Jacobian formula from linearization and area of a parallelogram; prove the chain rule for Jacobians using matrix determinants; show Jacobian of inverse is reciprocal (use det of matrix inverse); derive polar/cylindrical/spherical Jacobians from partial derivatives.
  • Numericals: Evaluate double/triple integrals after a coordinate transform; compute Jacobians for given transformations; check independence using nonzero Jacobian.
  • Diagrams: Small rectangle in x–y and its image parallelogram in u–v, with labels Δx, Δy, vectors mapping, and area scaling; coordinate grids (polar: circles and rays) with mapping of a ring sector to a rectangle in (r, θ) space.
  • Full-marks answers include: Clear statement of assumptions (C¹ mapping, nonzero Jacobian, one-to-one over region), correct computation of partial derivatives, determinant shown step-by-step, correct absolute value in the integral, correct limits in new variables, and units/radians where relevant.

Quick check

  1. State the Jacobian determinant for u = u(x, y), v = v(x, y).
  2. What is |∂(x, y)/∂(r, θ)| for x = r cosθ, y = r sinθ?
  3. If ∂(u, v)/∂(x, y) ≠ 0 at a point, what does it imply about u and v near that point?

Answers: (1) ∂(u, v)/∂(x, y) = (∂u/∂x)(∂v/∂y) − (∂u/∂y)(∂v/∂x). (2) r. (3) u and v are locally functionally independent and the mapping is locally invertible there.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?