Tensors Explained: An Introductory Guide for Beginners

A tensor is a mathematical object that generalizes scalars, vectors, and matrices. A scalar is a rank-0 tensor (one number), a vector is a rank-1 tensor (numbers with one index), a matrix can be a rank-2 tensor (numbers with two indices), and higher ranks continue the pattern. What makes a tensor a tensor is not the array of numbers but how those numbers change when you switch coordinate systems.

Tensors matter because physics must not depend on the coordinates we choose to describe it. Stress inside a bridge, the curvature of spacetime in relativity, and the inertia of a spinning top are all tensors: their components reshuffle under rotations, but the physical object stays the same.

Key Concepts

• Rank (order): the number of indices; a rank-2 tensor in 3D has 3 times 3 = 9 components.
• Transformation law: components in new coordinates are built from old components using the Jacobian matrices of the coordinate change; any array failing this law is just an array, not a tensor.
• Einstein summation convention: a repeated index in a product means "sum over it", so a(i) b(i) means the sum of a1 b1 + a2 b2 + a3 b3.
• Contraction: setting two indices equal and summing reduces the rank by 2, like the trace of a matrix.

Formulas

• Rank-2 transformation: T prime(ij) = (partial x^k / partial x prime^i) (partial x^l / partial x prime^j) T(kl), with sums over k and l.
• Vector transformation (rank 1): v prime^i = (partial x prime^i / partial x^j) v^j.
• Stress-force relation: the traction vector t(i) = sigma(ij) n(j), where sigma is the stress tensor and n is the surface normal.
• Rotation of vector components in 2D by angle theta: v prime = (v1 cos theta + v2 sin theta, minus v1 sin theta + v2 cos theta).

Worked Example

Rotate the vector v = (1, 0) by theta = 30 degrees and verify its length is unchanged, showing the tensor character of a vector.

1. Apply the rotation: v1 prime = 1 times cos 30 degrees + 0 times sin 30 degrees = 0.8660.
2. v2 prime = minus 1 times sin 30 degrees + 0 times cos 30 degrees = minus 0.5.
3. The components changed from (1, 0) to (0.8660, minus 0.5), exactly as the transformation law demands.
4. Compute the squared length in the new frame: 0.8660 squared + 0.5 squared = 0.75 + 0.25 = 1.00.
5. The length is still 1, confirming that while components depend on coordinates, the geometric object does not.

Common Misconceptions

• "Every matrix is a tensor." False. A tensor's components must transform by the tensor law; a random table of numbers with no transformation rule is not a tensor.
• "Rank means the size of the matrix." No. Rank is the number of indices; a 3-by-3 matrix and a 4-by-4 matrix can both be rank-2 tensors in different dimensions.
• "Tensors are only for relativity." Stress, strain, inertia, conductivity, and piezoelectricity are everyday engineering tensors.

Real-World Applications

• General relativity: the metric tensor and Einstein tensor describe gravity as spacetime curvature.
• Structural engineering: the stress tensor predicts where bridges and aircraft wings will fail.

Tensors are the coordinate-free language of physics: the components dance when coordinates change, but the object they describe stands still. Learning to think in tensors is the gateway to relativity, continuum mechanics, and modern geometry.