Determinants and Jacobians

Interpret volume changes, orientation, and changes of variables. Theory, formulas, worked examples and interactive graphs.

What it is

The determinant of a square matrix A is a single number, det(A), that measures how much a linear transformation scales the volume (or area, in two dimensions) of any region of space, and whether it also reverses orientation. A nonzero determinant indicates that the transformation is invertible; a zero determinant indicates that it collapses space into a lower dimension.

The Jacobian extends this idea to nonlinear transformations between coordinates: it is the determinant of the matrix of partial derivatives of a change of variables, and it measures locally how much the infinitesimal volume is stretched or compressed when moving from one coordinate system to another, at each point in space.

Why it appears in physics

It appears because many physical calculations require changing coordinate systems: switching from Cartesian to polar or spherical coordinates to exploit a symmetry, or changing variables in an integral to simplify it. In each change, the volume element transforms by multiplying by the absolute value of the Jacobian, and forgetting that factor produces results that are wrong by a scale factor.

It also appears when deciding whether a system of linear equations has a unique solution (nonzero determinant), when calculating the density of states in statistical mechanics, and when determining whether a coordinate transformation preserves or reverses orientation, which matters in electromagnetism with cross products and in mechanics with left-handed or right-handed reference frames.

Visual or geometric idea

The visual idea is this: take a unit square (or a unit cube in three dimensions) and apply the linear transformation A. The area (or volume) of the resulting figure is exactly |det(A)| times the original area. If det(A) is negative, the figure has also been reflected, reversing its orientation.

For the Jacobian, instead of a fixed square, imagine a grid of curved coordinates overlaid on space. At each point, the Jacobian says how much a small cell of that grid has been locally deformed relative to an equivalent Cartesian cell, which is why it appears inside an integral when changing variables.

Minimal symbolic language

For a 2×2 matrix, det([[a, b], [c, d]]) = ad − bc. For larger matrices, it is calculated by cofactor expansion or through properties such as swapping rows (which changes the sign) or adding a multiple of one row to another (which does not change the determinant).

The Jacobian of a change of variables (x, y) → (u, v) is J = det([[∂x/∂u, ∂x/∂v], [∂y/∂u, ∂y/∂v]]), and the area element transforms as dx dy = |J| du dv. In polar coordinates, this calculation gives the familiar factor r in dx dy = r dr dθ.

What the result means

The determinant's result means a signed volume scale factor: its absolute value is how much the volume is multiplied by, and its sign indicates whether orientation is preserved (positive) or reversed (negative). A zero determinant means the transformation is not invertible because it destroys information, flattening space into a lower dimension.

The Jacobian's result means the same thing, but locally and generally varying from point to point: at each position, it says how much the infinitesimal volume is deformed, and that factor must multiply the differential element when changing variables in an integral.

When not to use it

Do not use the determinant to invert a matrix if you already know it is zero: in that case no inverse exists, and you must look for another strategy, such as solving the homogeneous system directly or using decompositions that handle singular matrices. Also do not confuse the determinant of a matrix with its trace or with the sum of its entries: these are different quantities with different meanings.

For the Jacobian, never omit it when changing variables in a multiple integral, and do not assume its value is constant if the transformation is not linear: in general it depends on the point and must be evaluated inside the integral, not outside it.

Operational summary

Calculate the determinant by the direct formula for small matrices or by cofactor expansion for larger ones, interpret its absolute value as a volume scale factor and its sign as preservation or reversal of orientation. When changing variables in an integral, calculate the Jacobian of the partial derivatives, take its absolute value, and multiply it by the differential element in the new coordinates.

Procedure

Before calculating

Identify whether you need the determinant of a constant matrix (for invertibility, orientation, or volume scale) or the Jacobian of a change of variables (to transform a differential element inside an integral). Both use the same algebraic calculation, but the physical context differs.

Note the size of the matrix or the number of variables in the coordinate change, since it determines the most efficient calculation method.

Step 1: identify the mathematical object

For a 2×2 matrix, the determinant is calculated directly as ad − bc. For 3×3, Sarrus's rule or cofactor expansion along a row or column with many zeros is convenient. For a Jacobian, first identify the old and new variables and write the corresponding matrix of partial derivatives.

If the matrix has a row or column with obvious zeros, choose to expand along that row or column to simplify the calculation.

Step 2: choose the operation

If the goal is to decide invertibility, calculate the determinant and compare it with zero. If the goal is to transform an integral, calculate the full Jacobian of partial derivatives and take its absolute value.

If the matrix is large, consider using properties such as row reduction (Gaussian elimination), remembering that swapping two rows changes the sign of the determinant and that multiplying a row by a constant multiplies the determinant by that constant.

Step 3: preserve units and signs

Keep the sign of the determinant until the end: it is relevant physical information about orientation, not just a technical detail. For the Jacobian, take the absolute value only at the end, after correctly calculating the determinant with its sign.

If the variables carry different physical units, check that the Jacobian has the correct units so that the new differential element is dimensionally consistent with the original one.

Step 4: calculate

Expand the determinant step by step, either by the direct formula, by cofactors, or by row reduction. For the Jacobian, calculate each partial derivative separately before assembling the matrix and its determinant.

Verify the result in a simple particular case, such as evaluating the Jacobian at a point where the coordinate change is well known (for example, the origin in polar coordinates, taking care with singularities there).

Step 5: interpret the result

If you calculated a determinant, interpret its absolute value as a volume scale factor and its sign as preservation or reversal of orientation. If it is zero, explain which direction or combination of vectors collapses.

If you calculated a Jacobian, interpret its absolute value as the factor that corrects the differential element in the new coordinate system, and check that the result matches known cases, such as the factor r in polar coordinates or r² sin θ in spherical coordinates.

Self-check checklist

Check that you used the calculation method suited to the size of the matrix, that you kept the sign until the correct final step, that you correctly interpreted a zero determinant, and that in a change of variables the Jacobian multiplies rather than divides the differential element (or vice versa, depending on the direction of the change).

Guided example

Physical situation

The area of a circular region needs to be calculated by integrating in polar coordinates instead of Cartesian ones, because the symmetry of the problem (a radially symmetric charge or mass distribution) makes it much simpler to describe the boundary in terms of r and θ.

Mathematical translation

The change of variables is x = r cos θ, y = r sin θ. The area element dx dy must be transformed by multiplying by the absolute value of the Jacobian of this transformation, which needs to be calculated before rewriting the integral.

Step-by-step calculation

The partial derivatives are ∂x/∂r = cos θ, ∂x/∂θ = −r sin θ, ∂y/∂r = sin θ, ∂y/∂θ = r cos θ. The Jacobian is J = cos θ (r cos θ) − (−r sin θ)(sin θ) = r cos²θ + r sin²θ = r. Therefore, dx dy = r dr dθ.

Reading the result

The factor r is not an arbitrary convention: it means that, farther from the origin, the same angular variation dθ corresponds to a larger area. This factor prevents underestimating the contribution of regions far from the origin when integrating in polar coordinates.

Common error avoided

This avoids writing dx dy = dr dθ directly, forgetting the factor r, an error that systematically underestimates any radial integral and produces incorrect results by a factor that depends on the integration region.

Inverse example

Given result

A 2×2 matrix has a determinant equal to zero.

What physical question it can answer

This allows one to conclude that the associated transformation collapses the plane onto a line (or a point), that the corresponding system of linear equations has no unique solution, and that the two columns of the matrix are parallel vectors, that is, linearly dependent. Physically, it can indicate that two equations of a system actually describe the same constraint, without contributing independent information.

Common errors

Error 1: forgetting the Jacobian when changing variables in an integral

It seems reasonable to substitute the new variables directly without adjusting the differential element. It fails because the area or volume element deforms differently at each point, and that factor must be included explicitly. Detect it by comparing the result with a known case, such as the area of a circle calculated in polar coordinates without the factor r. Correct it by always calculating the full Jacobian before rewriting the integral. Mini-example: calculating the area of a circle of radius R as ∫∫ dr dθ gives a wrong result; one must use ∫∫ r dr dθ.

Error 2: confusing the determinant with the trace

It seems reasonable to think both summarize the "size" of a matrix in some similar way. It fails because the trace is the sum of the diagonal entries and has no direct relation to volume scaling, while the determinant does. Detect it by comparing both values in a simple numerical example. Correct it by remembering that the determinant measures volume scale and orientation, while the trace appears in other contexts, such as the sum of eigenvalues. Mini-example: a pure rotation matrix has determinant 1 but a trace generally different from 1.

Error 3: assuming the Jacobian is always a constant

It seems reasonable to treat it as a simple fixed conversion factor, like converting meters to centimeters. It fails because, except in linear transformations, the Jacobian depends on the point where it is evaluated. Detect it by checking whether the partial derivatives of the transformation depend on the variables. Correct it by evaluating the Jacobian as a function of the point and leaving it inside the integral. Mini-example: in polar coordinates, the Jacobian r is not constant but grows with distance from the origin.

Error 4: ignoring the sign of the determinant

It seems reasonable to keep only the absolute value because what matters is "how much" the volume scales. It fails when the physical problem depends on orientation, as in cross products or in left-handed and right-handed reference frames. Detect it by asking whether the problem distinguishes between a transformation and its mirror image. Correct it by keeping the sign of the determinant until checking whether orientation matters in that context. Mini-example: a reflection has determinant −1, the same volume scaling as a rotation but with reversed orientation.

Error 5: calculating the determinant by expanding along the wrong row or column

It seems reasonable to always expand along the first row. It fails because the calculation becomes unnecessarily complicated if that row has no zeros while another one does. Detect it by comparing the calculation effort across different rows or columns before starting. Correct it by always choosing the row or column with the most zeros to expand. Mini-example: a 4×4 matrix with a row of zeros except for one entry reduces the calculation to a single cofactor.

Quick detection rule

If the result of a changed-variable integral does not match a known case, if the sign of a determinant relevant to orientation has been lost, or if the Jacobian has been treated as constant in a nonlinear transformation, the error is probably in the setup of the change of variables, not in the final arithmetic.

Physical applications

1. Coordinate changes in integrals

It appears when switching from Cartesian to polar, cylindrical, or spherical coordinates to simplify an integral with symmetry. The Jacobian of the transformation is calculated and multiplied by the differential element. The result gives the correct volume or area element in the new coordinates. The dangerous error is omitting the Jacobian factor or using a constant value when it actually depends on the point. As a final control, the learner must explain why an extra factor appears when changing coordinate systems.

2. Systems of linear equations

It appears when deciding whether a physical system of linear equations (for example, a circuit or a static structure) has a unique solution. The determinant of the coefficient matrix is calculated. The result indicates whether the system is well posed or has redundant equations. The dangerous error is trying to solve a system with a zero determinant as if it had a unique solution. As a final control, the learner must explain what it physically means for two equations of the system to be linearly dependent.

3. Solid mechanics and orientation

It appears when checking whether a coordinate transformation preserves the orientation of a reference frame (for example, whether it remains right-handed). The sign of the determinant of the basis-change matrix is calculated. The result determines whether cross products and angular momenta must adjust their sign. The dangerous error is ignoring a change of orientation when combining different reference frames. As a final control, the learner must explain what physical consequence would follow from unknowingly reversing the orientation of a coordinate system.

4. Statistical mechanics

It appears when calculating volumes in phase space or densities of states, where a change of variables between coordinates and momenta requires a Jacobian. The result directly affects the calculation of entropies and partition functions. The dangerous error is omitting the Jacobian when changing from microscopic to macroscopic variables. As a final control, the learner must explain why the volume of phase space must be transformed carefully when changing coordinates.

5. Optics and beam deformation

It appears when describing how an optical system or a lens deforms a light beam in position-angle space. The determinant of the system's transfer matrix is calculated to check whether it preserves the beam area in that space. The result predicts whether the system is focusing, defocusing, or neutral in terms of brightness conservation. The dangerous error is ignoring the determinant of the optical matrix when combining several elements. As a final control, the learner must explain what it physically means for the determinant of an ideal optical system to equal one.

Practice

Level 1: recognition

Identify, for a given problem, whether a determinant is needed (invertibility, orientation, constant volume scale) or a Jacobian (change of variables in an integral).

Exercise: for three different situations (solving a linear system, changing to spherical coordinates, checking axis orientation), classify which tool corresponds to each one.

Level 2: direct application

Calculate the determinant of a simple 2×2 or 3×3 matrix and verify the result with the direct formula or by cofactors.

Exercise: calculate the Jacobian of the change to polar coordinates and check that it matches the known factor r.

Level 3: choosing the tool

Decide whether it is best to expand a determinant by cofactors, use row reduction, or apply the direct 2×2 or 3×3 formula, depending on the size and structure of the matrix.

Exercise: given a 3×3 matrix with a row that is almost all zeros, decide which row to expand along and justify the choice.

Level 4: physical interpretation

Interpret the sign and absolute value of a determinant in a specific physical context: volume scale, axis orientation, or invertibility of a system.

Exercise: given a coordinate change with a negative Jacobian in a certain region, explain what that sign change physically means.

Level 5: intentional error

Analyze a changed-variable integral where the Jacobian was omitted or incorrectly treated as constant. Detect the error, correct it, and explain the physical consequence.

Exercise: invent a plausible change of variables, deliberately miscalculate the Jacobian, show why the error seems reasonable, correct it, and state the incorrect result it would have produced.