JOVANA
Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Signed Volume and Orientation

The abstract form has a face: it is signed volume. We make the [[determinant-as-volume|determinant as signed volume]] precise, explain why the sign encodes [[lin-orientation|orientation]], and connect it to the [[jacobian-determinant|Jacobian determinant]] that rescales volume under change of variables.

The parallelepiped picture

Take the n columns of A as edge vectors emanating from the origin. They span a parallelepiped (a parallelogram in 2D, a slanted box in 3D). The claim of the determinant as signed volume is: |det(A)| equals the n-dimensional volume of that box, and the sign of det(A) records its orientation.

Why must this be true? Signed volume obeys our three rules. Scaling one edge by a scales the volume by a (multilinear). Two equal edges flatten the box to zero volume (alternating). The unit cube has volume 1 (normalized). By the uniqueness theorem of Guide 2, signed volume = det. No separate computation is needed — the rules already proved it.

2D: columns u=(3,0), v=(1,2).
  det([3 1; 0 2]) = 3*2 - 1*0 = 6
  Area of the parallelogram = base 3 * height 2 = 6.  OK.

Sign flip = orientation flip:
  det([1 0; 0 1]) = +1   (e1 then e2: counterclockwise)
  det([0 1; 1 0]) = -1   (swapped: clockwise)
  Same shape (unit square), opposite orientation.
Determinant equals signed area in 2D; the sign tracks orientation.

What the sign means

Volume is never negative, so where does the sign come from? It encodes orientation: whether your ordered list of edges is 'right-handed' like the standard basis (det > 0) or 'left-handed', a mirror image (det < 0). In 2D, positive means the columns turn counterclockwise from first to second; in 3D, positive means right-hand-rule.

This is why a column swap negates det: swapping two edges mirrors the frame, reversing handedness. A linear transformation with det < 0 includes a reflection; one with det = 0 collapses volume to zero, squashing space into a lower dimension (and hence is non-invertible).

The Jacobian: local volume rescaling

The volume picture explains a fact you may have met in calculus. When you change variables in a multiple integral, the volume element dx dy scales by the Jacobian determinant — the determinant of the matrix of partial derivatives of the new coordinates with respect to the old. Locally, a smooth map looks linear; its derivative is a matrix, and det of that matrix is the local volume-stretch factor.

We take its absolute value in the integral because volume is positive, but the sign still carries orientation — it tells you whether the change of variables preserves or reverses orientation. So the determinant you defined abstractly in Guide 1 is literally the conversion rate between coordinate systems' volumes.