State Transition Matrix

The state transition matrix is one of the most fundamental tools in modern control theory. It describes how the state of a linear time-invariant system evolves over time from any given initial condition. Understanding this matrix is essential for analyzing the free response of a system, solving state-space equations, and preparing for GATE questions on time-domain analysis.

Core Concept Explanation

The state transition matrix (STM), denoted Φ(t) or sometimes Φ(t, 0), answers a fundamental question in system analysis: if a system starts at a known state x(0) with no external input applied, what will its state be at any future time t? The answer is given by the simple product x(t) = Φ(t) · x(0). The matrix Φ(t) completely encodes the natural dynamics of the system.

For a linear time-invariant system described by the state equation dx/dt = Ax, the state transition matrix is the matrix exponential e^(At). This is not a scalar exponential but a matrix function defined through its Taylor series expansion. The matrix A is the system matrix that characterizes all the interconnections between state variables.

Physically, each column of Φ(t) tells you what happens to the system if it starts with all its initial energy concentrated entirely in one state variable. The matrix exponential continuously transforms the initial state vector as time progresses, rotating and scaling it according to the eigenstructure of A.

Mathematical Expression

The matrix exponential is defined by the convergent series:

e^(At) = I + At + (At)²/2! + (At)³/3! + ... = Σ (Aᵏtᵏ)/k! for k from 0 to infinity

For practical computation, the Laplace transform method is most commonly used in examinations. Taking the Laplace transform of dx/dt = Ax gives sX(s) - x(0) = AX(s), which rearranges to X(s) = (sI - A)⁻¹ x(0). Comparing with the time-domain solution X(s) = Φ(s)·x(0), we get:

Φ(t) = L⁻¹ [(sI - A)⁻¹]

This is the most testable formula in GATE. The matrix (sI - A) is called the resolvent matrix, and its inverse is a rational matrix of s whose inverse Laplace transform gives Φ(t). For a 2x2 system, this involves computing a 2x2 matrix inverse with polynomial entries.

The Cayley-Hamilton theorem provides another route: every matrix satisfies its own characteristic equation, so higher powers of A can be reduced to lower-order polynomial expressions. This reduces the infinite series to a finite sum of n terms for an n-dimensional system.

Practical Understanding

The complete solution of the state-space equation with input u(t) uses the STM in the variation of parameters formula: x(t) = Φ(t)·x(0) + integral from 0 to t of Φ(t-τ)·B·u(τ)dτ. The first term is the zero-input response and the second is the zero-state response driven by the input. The STM thus forms the backbone of both free and forced response analysis.

The eigenvalues of A determine the poles of the system and are directly visible in the STM. If A has eigenvalue λᵢ, then e^(At) contains terms of the form e^(λᵢt). Stable systems have all eigenvalues with negative real parts, causing Φ(t) to decay to zero as t increases. This connects state-space analysis directly to classical pole-zero analysis.

Numerical Example

Consider a system with A = [[0, 1], [-2, -3]]. The Laplace method finds Φ(t) by computing (sI - A)⁻¹. First, sI - A = [[s, -1], [2, s+3]]. The determinant is s(s+3) + 2 = s² + 3s + 2 = (s+1)(s+2). The inverse is (1/det)·[[s+3, 1], [-2, s]]. Each element is partial-fraction expanded and inverse Laplace transformed to obtain Φ(t).

Given:
A = [[0, 1], [-2, -3]]
x(0) = [1, 0]ᵀ

Why this formula applies:
For zero-input LTI system, x(t) = e^(At) x(0) = Φ(t) x(0)
Use Φ(t) = L⁻¹[(sI - A)⁻¹]

Formula:
Φ(t) = L⁻¹[(sI - A)⁻¹]

Substitution:
sI - A = [[s, -1], [2, s+3]]
det = s² + 3s + 2 = (s+1)(s+2)
(sI-A)⁻¹ = [[（s+3)/((s+1)(s+2)),  1/((s+1)(s+2))],
             [-2/((s+1)(s+2)),     s/((s+1)(s+2))]]

Calculation:
(s+3)/((s+1)(s+2)) = 2/(s+1) - 1/(s+2)  → 2e⁻ᵗ - e⁻²ᵗ
1/((s+1)(s+2))     = 1/(s+1) - 1/(s+2)  → e⁻ᵗ - e⁻²ᵗ
-2/((s+1)(s+2))    = -2/(s+1) + 2/(s+2) → -2e⁻ᵗ + 2e⁻²ᵗ
s/((s+1)(s+2))     = -1/(s+1) + 2/(s+2) → -e⁻ᵗ + 2e⁻²ᵗ

Final Answer:
Φ(t) = [[2e⁻ᵗ - e⁻²ᵗ,   e⁻ᵗ - e⁻²ᵗ ],
         [-2e⁻ᵗ + 2e⁻²ᵗ, -e⁻ᵗ + 2e⁻²ᵗ]]
x(t) = Φ(t)·[1,0]ᵀ = [2e⁻ᵗ - e⁻²ᵗ, -2e⁻ᵗ + 2e⁻²ᵗ]ᵀ

Exam Tip: In GATE, always use Φ(t) = L⁻¹[(sI - A)⁻¹] for computation. Verify your answer by checking Φ(0) = I. If your result does not give the identity matrix at t = 0, there is a calculation error.

Key Mechanism Points

• The STM Φ(t) = e^(At) is computed most efficiently by the Laplace method: Φ(t) = L⁻¹[(sI-A)⁻¹]. This approach is standard in GATE solutions.
• The denominator of (sI-A)⁻¹ is always the characteristic polynomial det(sI-A), whose roots are the eigenvalues of A and also the poles of the system.
• Stability is directly visible from Φ(t): if all eigenvalues have negative real parts, every entry of Φ(t) decays to zero, confirming asymptotic stability.
• The semigroup property Φ(t₁+t₂) = Φ(t₁)·Φ(t₂) enables breaking the response into sequential intervals, which is useful for piecewise analysis.
• Always verify Φ(0) = I after computing the STM. This is a fast self-check and is frequently tested as a short conceptual question in GATE.

Quick Revision

• State transition matrix Φ(t) = e^(At) maps x(0) to x(t) for zero-input LTI systems.
• Computation: Φ(t) = L⁻¹[(sI - A)⁻¹]. The resolvent (sI - A)⁻¹ is partial-fraction expanded element-wise and inverse Laplace transformed.
• Properties: Φ(0) = I, Φ⁻¹(t) = Φ(-t), d/dt[Φ(t)] = A·Φ(t), and semigroup property.
• Eigenvalues of A appear as exponents in Φ(t). Negative real parts guarantee stable decay.
• Complete response: x(t) = Φ(t)x(0) + ∫₀ᵗ Φ(t-τ)Bu(τ)dτ.
• Exam trap: Do not confuse scalar exponential e^(at) with matrix exponential e^(At). They are structurally different objects.
• Series method e^(At) = I + At + A²t²/2! + ... converges always but is impractical for hand calculation. Use Laplace or Cayley-Hamilton in exams.