1 Definition and basic properties

A nilpotent matrix is a square matrix that becomes the zero matrix after being raised to some positive integer power. This notion captures the idea of repeated action eventually producing no further effect. Nilpotent matrices are central in linear algebra because they provide a canonical example of nonzero operators with strongly constrained long-term behavior.

1.1 Formal definition

A square matrix \(N\) is called nilpotent if there exists an integer \(k \ge 1\) such that \[ N^k = 0, \] where \(0\) denotes the zero matrix of the same size. The smallest such positive integer \(k\) is called the nilpotency index of \(N\).

Nilpotency is a property of the matrix itself, not of a particular basis representation alone. If a linear transformation is represented by a nilpotent matrix in one basis, then every matrix representing that same transformation in another basis is also nilpotent.

1.2 Nilpotency index

The nilpotency index is the least positive integer \(k\) for which \(N^k = 0\). It measures how many repeated multiplications are required before the matrix vanishes. For a zero matrix, the nilpotency index is \(1\).

The nilpotency index is bounded by the size of the matrix. For an \(n \times n\) nilpotent matrix, the index is at most \(n\). This upper bound is sharp, since a single Jordan block of size \(n\) has nilpotency index \(n\).

1.3 Immediate consequences

Nilpotent matrices have several immediate algebraic consequences that follow from the defining equation \(N^k = 0\). These include restrictions on determinant, trace, and eigenvalues, as well as strong implications for the characteristic and minimal polynomials.

1.3.1 Determinant and trace

If \(N\) is nilpotent, then \(\det(N) = 0\). This follows because \(\det(N^k) = (\det N)^k\), and \(N^k = 0\) has determinant \(0\). Thus every nilpotent matrix is singular.

The trace of a nilpotent matrix is also \(0\). In fact, all eigenvalues are zero, and the trace is the sum of the eigenvalues counted with multiplicity.

1.3.2 Eigenvalues

Every eigenvalue of a nilpotent matrix is \(0\). Indeed, if \(Nv = \lambda v\) for a nonzero vector \(v\), then applying \(N^k\) gives \[ 0 = N^k v = \lambda^k v, \] which forces \(\lambda = 0\).

Thus nilpotent matrices have only one possible eigenvalue, and it occurs with full algebraic multiplicity. However, a nilpotent matrix need not be the zero matrix, because it may fail to be diagonalizable.

1.3.3 Characteristic and minimal polynomials

The characteristic polynomial of an \(n \times n\) nilpotent matrix has the form \[ \chi_N(x) = x^n. \] More generally, the minimal polynomial is a power of \(x\), namely \[ m_N(x) = x^r \] for some \(1 \le r \le n\). The exponent \(r\) is exactly the nilpotency index.

These polynomial identities encode the fact that nilpotent matrices are annihilated by some power of the variable \(x\), and they are fundamental in the classification of such matrices.

2 Examples

Nilpotent matrices arise in simple and structured forms. Some examples are immediate from the definition, while others illustrate how nilpotency can appear in small dimensions or through triangular structure.

2.1 Zero matrix

The zero matrix is nilpotent with index \(1\), since \(0^1 = 0\). It is the simplest possible example and serves as the endpoint of repeated multiplication.

2.2 Strictly triangular matrices

Any strictly upper triangular or strictly lower triangular matrix is nilpotent. In an \(n \times n\) strictly triangular matrix, all diagonal entries are zero, and repeated multiplication eventually shifts nonzero entries beyond the matrix boundary. Such a matrix satisfies \(N^n = 0\).

This class provides many standard examples and is especially important because every nilpotent matrix is similar to one of this form over an algebraically closed field.

2.3 Small-dimensional examples

Small matrices make nilpotency easy to visualize and verify directly. They also show how nilpotent behavior can range from immediate annihilation to several steps of persistence before vanishing.

2.3.1 2 × 2 nilpotent matrices

A typical nonzero \(2 \times 2\) nilpotent matrix is \[ \begin{pmatrix} 0 & 1 \\ 0 & 0 \end{pmatrix}. \] Its square is the zero matrix, so its nilpotency index is \(2\). Any nonzero \(2 \times 2\) nilpotent matrix is similar to this one.

2.3.2 3 × 3 nilpotent matrices

A standard \(3 \times 3\) example is \[ \begin{pmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{pmatrix}. \] Its square is nonzero, but its cube is zero, so the nilpotency index is \(3\). This matrix is a single Jordan block of size \(3\).

3 Structural theory

The structure of nilpotent matrices is governed by their Jordan form. This theory explains how nilpotent matrices are organized into blocks and how their kernels and images behave under repeated powers.

3.1 Jordan canonical form

Over an algebraically closed field, every nilpotent matrix is similar to a direct sum of Jordan blocks with eigenvalue \(0\). Since the only eigenvalue is \(0\), the Jordan canonical form consists entirely of blocks whose diagonal entries are zero and whose superdiagonal entries are \(1\).

This form classifies nilpotent matrices up to similarity. The sizes of the blocks determine the matrix completely up to change of basis.

3.2 Jordan blocks

A nilpotent Jordan block of size \(m\) has the form \[ J_m(0)= \begin{pmatrix} 0 & 1 & 0 & \cdots & 0 \\ 0 & 0 & 1 & \cdots & 0 \\ \vdots & \vdots & \ddots & \ddots & \vdots \\ 0 & 0 & \cdots & 0 & 1 \\ 0 & 0 & \cdots & 0 & 0 \end{pmatrix}. \] Such a block satisfies \(J_m(0)^m = 0\) but \(J_m(0)^{m-1} \ne 0\). Therefore its nilpotency index is \(m\).

A general nilpotent matrix is a direct sum of such blocks, and its nilpotency index is the size of the largest block.

3.3 Rank and nullity sequences

The ranks and nullities of powers of a nilpotent matrix form monotone sequences that reveal the block structure. As powers increase, the image shrinks and the kernel grows until the zero matrix is reached.

3.3.1 Kernel stabilization

For a nilpotent matrix \(N\), the sequence \[ \ker N \subseteq \ker N^2 \subseteq \ker N^3 \subseteq \cdots \] eventually stabilizes at the whole space. Each inclusion reflects the fact that if a vector is killed by a smaller power, it is also killed by larger powers.

The stabilization step occurs precisely at the nilpotency index.

3.3.2 Image stabilization

The images form a descending chain: \[ \operatorname{im} N \supseteq \operatorname{im} N^2 \supseteq \operatorname{im} N^3 \supseteq \cdots. \] Since \(N^k = 0\) for some \(k\), these images eventually become zero. The rate at which the image dimensions drop is closely related to the sizes of the Jordan blocks.

4 Criteria for nilpotency

Nilpotency can be recognized through several equivalent or partially equivalent tests, depending on the context. Some criteria are purely polynomial, while others apply only in special settings.

4.1 Polynomial criteria

A matrix \(N\) is nilpotent if and only if its minimal polynomial is a power of \(x\). Equivalently, the matrix is annihilated by some polynomial of the form \(x^r\).

Another useful criterion is that all eigenvalues must be zero. Over an algebraically closed field, this condition together with the characteristic polynomial being \(x^n\) characterizes nilpotency.

4.2 Trace criteria in special cases

In certain special classes of matrices, trace conditions can be informative. For example, for a triangular matrix, if all diagonal entries are zero, then the matrix is nilpotent. Since the trace is the sum of the diagonal entries, a triangular nilpotent matrix automatically has trace zero.

However, trace zero alone does not imply nilpotency in general. It is only a necessary condition, not a sufficient one.

4.3 Similarity invariance

Nilpotency is preserved under similarity. If \(M\) is nilpotent and \(P\) is invertible, then \[ (P^{-1}MP)^k = P^{-1}M^kP = 0, \] so \(P^{-1}MP\) is also nilpotent.

This invariance reflects the fact that nilpotency is a property of the underlying linear transformation rather than of the particular coordinate system used to represent it.

5 Operations and transformations

Nilpotent matrices interact predictably with several standard operations. Their behavior under powers, scalar multiplication, direct sums, and exponentiation is especially useful in applications and theory.

5.1 Powers of nilpotent matrices

If \(N\) is nilpotent, then every sufficiently high power of \(N\) is zero. More precisely, if \(N^k = 0\), then \(N^m = 0\) for all \(m \ge k\).

If \(N\) has nilpotency index \(k\), then the powers \[ N, N^2, \dots, N^{k-1} \] may be nonzero, but \(N^k\) and all higher powers vanish.

5.2 Scalar multiples

If \(N\) is nilpotent and \(c\) is a scalar, then \(cN\) is nilpotent as well. Indeed, \[ (cN)^k = c^k N^k = 0. \] If \(c \ne 0\), the nilpotency index of \(cN\) is the same as that of \(N\).

5.3 Direct sums

A direct sum of nilpotent matrices is nilpotent. If \(N_1^{k_1} = 0\) and \(N_2^{k_2} = 0\), then \[ (N_1 \oplus N_2)^{\max(k_1,k_2)} = 0. \] The nilpotency index of the direct sum is the maximum of the indices of the summands.

This property makes direct sums a natural way to build larger nilpotent matrices from smaller ones.

5.4 Matrix exponentials

For a nilpotent matrix \(N\), the matrix exponential \[ e^N = I + N + \frac{N^2}{2!} + \cdots \] is actually a finite sum, because all sufficiently high powers of \(N\) vanish. Thus \(e^N\) is a polynomial in \(N\).

This fact is useful in linear differential equations and Lie theory, where nilpotent matrices often generate transformations with especially simple exponential behavior.

6 Applications

Nilpotent matrices appear in many settings where repeated application of an operator must eventually disappear. Their finite-step decay makes them useful in analysis, geometry, algebra, and combinatorics.

6.1 Linear transformations

In linear algebra, nilpotent matrices represent linear transformations with no nonzero eigenvalues and no steady-state component. They describe operators that shift vectors along chains until they are annihilated.

This viewpoint is especially helpful in understanding the Jordan decomposition of linear operators, where nilpotent parts capture the non-diagonalizable portion of the transformation.

6.2 Differential equations

Nilpotent matrices simplify systems of linear differential equations of the form \[ x'(t) = Nx(t). \] Because the exponential \(e^{tN}\) is a finite polynomial in \(t\), solutions can be written explicitly. This makes nilpotent systems particularly tractable.

Such systems often exhibit polynomial growth rather than oscillatory or exponential behavior, reflecting the absence of nonzero eigenvalues.

6.3 Lie algebra context

Nilpotent matrices play a major role in Lie algebras and Lie groups. In matrix Lie algebras, nilpotent elements generate flows whose exponential maps are finite polynomials.

They also serve as basic examples of nilpotent operators in the broader algebraic sense, where iterated commutators or adjoint actions may vanish after finitely many steps.

6.4 Combinatorial matrix theory

In combinatorial matrix theory, nilpotent matrices can encode directed acyclic structures and finite-step transitions. The adjacency matrix of a finite directed acyclic graph is nilpotent after a suitable ordering of vertices, since there are no arbitrarily long directed walks.

This connection links nilpotent matrices to path counting, graph structure, and iterative processes with eventual termination.