Mathematics
Nilpotent Matrices and the Jordan Canonical Form
Quick fact
Every square matrix over the complex numbers can be written in Jordan canonical form, which is almost diagonal but with 1's just above the diagonal in certain blocks. These extra 1's are exactly the footprint of nilpotent 'shift' operations that encode how repeated eigenvalues behave.
Why this is interesting
You may know that some matrices can be diagonalized, but what about those that stubbornly refuse? What if a matrix behaves like a 'shift' that eventually wipes everything out—could that be the key to a universal simplification?
Read the full explanation
Understanding Nilpotent Matrices and the Jordan Canonical Form
Imagine a matrix as a machine that moves vectors around. Sometimes it stretches or rotates, and you can find special directions (eigenvectors) that it only scales. But when you have repeated eigenvalues, sometimes there aren't enough independent eigenvectors—the machine also 'shears' or 'shifts' vectors. A nilpotent matrix is like a pure shift: apply it enough times and everything collapses to zero. For example, take a matrix that maps standard basis vectors forward: it sends e1 to e2, e2 to e3, ... and the last to zero. After three applications, everything is zero. That's nilpotent. Now, any matrix can be broken into diagonal scaling plus a nilpotent shift, and the Jordan form organizes this decomposition into neat blocks. Each block corresponds to one eigenvalue and shows exactly how much shifting happens.
A deeper explanation
The Jordan canonical form arises from decomposing a linear transformation over the complex numbers. Any matrix A can be written as A = PJP⁻¹, where J is block diagonal with Jordan blocks. A Jordan block for an eigenvalue λ has λ on the diagonal, 1s on the superdiagonal, and zeros elsewhere. Such a block can be written as λI + N, where N is nilpotent—it shifts basis vectors along a chain. The key mechanism is that the generalized eigenspaces decompose the vector space into invariant subspaces. Within each generalized eigenspace, you choose a basis that is a chain of generalized eigenvectors: (A - λI)v₁ = v₂, (A - λI)v₂ = v₃, etc., and (A - λI)vₖ = 0. This chain directly gives the Jordan block structure. The sizes of the blocks are determined by the dimensions of the kernels of (A - λI)^m. The Jordan form is fundamental because it simplifies many computations, such as matrix powers and exponentials, making it crucial in solving systems of differential equations and in theoretical applications.