Rank-2 Tensor · Interactive Module

Trace & Determinant
Invariance

What stays constant when everything appears to change

Theory Page
← Back to Demo

Trace Invariance

The trace of a tensor is the sum of its diagonal entries equivalently, the sum of its eigenvalues. For your tensor with principal values a and b:

// In demo you can see the definition of the Tensor;
Trace(T) = T11 + T22 = a + b

// After rotating by angle t=theta:
Trace(T') = T'11 + T'22 = a + b  - identical, always

// Numerical example: a = 3, b = 1, c=0
At t= theta = 0:   T11 + T22 = 3 + 1 = 4
 - check in demo by varying slider.
At t= theta = 45: T'11 + T'22 = 2 + 2 = 4  - confirmed
At t= theta = 90: T''11+ T''22= 1 + 3 = 4  - confirmed
Car Analogy - Trace

The Fixed Inertia Budget

Imagine your parked car has a total rotational stubbornness a fixed amount of resistance it possesses against being pushed or spun. Push it from behind (easy), push from the side (hard), push at 45 degrees (somewhere in between).

The trace is exactly this total budget of resistance. When you walk around the car and measure from a new angle, you redistribute the resistance between your x-axis and y-axis measurements but the total never changes.

At 45 degrees, both axes feel an equal resistance of 2 instead of 3 and 1. But 2 + 2 = 3 + 1 = 4. The car has not changed only your ruler has rotated.

At t=theta either 0 or 180 degrees, means you are pushing car either from front side or back side, it is easy compare to pushing by the sides, an angle that is in between 0 to 180 degree (or 180 to 360 degree).

What the Demo Shows

As you drag the theta slider in demo, watch the diagonal entries a' and b' change one rises as the other falls but their sum stays locked at a + b throughout the full 360 degree rotation.

In deeper mathematics, the trace equals the sum of all eigenvalues . This connects directly to the characteristic polynomial of the tensor the trace is its second coefficient, and the characteristic polynomial belongs to the matrix itself, not to any coordinate system used to write it.

Determinant Invariance

The determinant of a symmetric 2x2 tensor is the product of its eigenvalues. For the tensor T = [[a, c], [c, b]], set it to in demo and look for all the results:

det(T) = a*b - c^2

// In the principal frame where c = 0:
det(T) = a x b  = lambda_1 x lambda_2

// Numerical example: a = 3, b = 1, c = 0
At theta = 0:   det = 3 x 1 - 0 = 3
At theta = 45: det = 2 x 2 - 1^2 = 4 - 1 = 3  -- confirmed
At theta = 60: det = lambda_1 x lambda_2 = 3 x 1 = 3  -- confirmed

// Area of the ellipse is proportional to:
Area ~ pi / sqrt( det(T) )  -- fixed, since det is invariant
Car Analogy - Determinant

The Shape Signature of Asymmetry

The determinant measures how different the two principal resistances are from each other, it is the car's shape of asymmetry.

A car that is 3 times harder to push sideways than forward has a determinant of 3. This ratio of difficulty is a physical fact about the car's geometry. No matter which angle you stand at to measure, the car remains equally asymmetric. You cannot make it more or less car-shaped by changing your coordinate axes.

The determinant is directly linked to the area of the ellipse in our demo. Since the ellipse shape never changes as you rotate theta, its area never changes and since area is proportional to 1/sqrt(det), the determinant never changes either. This calculation you can directly observe in to the demo by rotating the theta i.e. t.

Special case: if det = 0, one eigenvalue is zero the car offers no resistance in some direction, like a car on a perfectly frictionless rail. This fundamental physical fact cannot be hidden by any rotation of axes.

What the Demo Shows

As t=theta changes, the off-diagonal term c' grows and shrinks. But a'*b' - c'^2 always returns to exactly a*b. The ellipse shape its elongation and orientation never changes. The determinant is the mathematical guardian of that shape.

The Two Invariants at a Glance

Invariant Formula Car Analogy Physical Meaning
Trace lambda_1 + lambda_2 Total inertia budget sum of all resistances Average response strength; related to mean eigenvalue
Determinant lambda_1 x lambda_2 Shape signature how asymmetric the car is Degree of anisotropy; governs ellipse area
Eigenvalues lambda_1, lambda_2 The actual easy and hard push resistances Principal physical values in the natural frame
Eigenvectors e_1, e_2 The car's actual forward and sideways directions Principal axes frame where tensor is diagonal

Why Invariants Matter in Physics

The Core Principle

Coordinate Independence of Physical Reality

Physical reality cannot depend on the arbitrary choice of coordinate system. When you rotate your axes by t=theta, you have changed nothing about the physical world only your description of it.

So any quantity that is genuinely physical must be invariant under rotation. This is why physicists search for invariants they are nature's way of separating physical truth from human bookkeeping.

The components T_11, T_12, T_21, T_22 are bookkeeping they depend on how you drew your axes. The trace and determinant are truth they belong to the physical object itself, independent of any observer.

Here are four domains where tensor invariants play an essential role:

🔩

Stress Tensor

The trace of the stress tensor equals 3 times the hydrostatic pressure average compression in all directions. An engineer measuring stress in a steel beam gets the same hydrostatic pressure regardless of how they orient their measurement frame.

🌀

Inertia Tensor

The trace of a rigid body's inertia tensor equals the sum of all three principal moments Ixx + Iyy + Izz the average rotational kinetic energy over all spin axes. This is independent of measurement orientation.

Conductivity Tensor

In anisotropic crystals, the trace of the conductivity tensor gives the mean electrical conductivity over all directions. Rotating the crystal in your measurement setup cannot change this mean it is a property of the crystal itself.

🌍

General Relativity

Einstein built his entire framework on tensor equations precisely because they are automatically invariant under coordinate changes. The curvature scalars tell you whether spacetime is genuinely curved near a black hole, regardless of coordinates.

Quantum Mechanics Connection

The energy levels of the hydrogen atom are eigenvalues of the Hamiltonian operator a rank-2 tensor in function space. They are invariants: the spectral lines you observe in a telescope are identical regardless of how Earth is oriented in space. Nature's invariants appear directly as measurable spectral lines, resonant frequencies, and energy levels.

What to Watch in Demo

Observation Guide

Verifying Invariance Live in the Demo

1 Set a = 3, b = 1, c = 0 Trace = 4, Det = 3
2 Drag theta to any angle Trace stays 4, Det stays 3
3 Watch a' rise as b' falls they always sum to 4
4 The ellipse shape never changes its area is pinned by det = 3
5 Set a = b = 2 ellipse becomes a circle, det = 4, trace = 4
6 Drag theta now nothing changes at all this is the isotropic tensor
The Key Realisation

Dragging theta changes all four components T_11, T_12, T_21, T_22 continuously. But the trace and determinant sit perfectly still because the car, the ellipse, and the physics have not moved at all. Only your ruler rotated.

The Single Most Important Insight

"Invariants are nature's way of separating physical truth from human bookkeeping."

The tensor components are bookkeeping they depend on how you drew your axes. The trace, determinant, and eigenvalues are truth they belong to the physical object itself, independent of any observer, any coordinate system, any angle of measurement.

In our demo, drag theta freely. All four numbers in the matrix change. But the sum and the product of eigenvalues never move. That is not a coincidence that is physics.

Next: To Quadratic Form → Back to Demo