Ontological Layer: The Transformer

Systems & LTI

How a System Acts Upon a Signal

The Mechanics of Transformation

A signal in isolation is merely data. It is only when it passes through a System that transformation occurs. Realm 02 categorizes the universe of operators: do they remember the past? Do they anticipate the future? If a system obeys the sacred laws of Linearity and Time Invariance, its entire identity can be captured by a single spark—the Impulse Response. Through Convolution, we predict the output for any conceivable input.

INPUT x(t) → SYSTEM CLASSIFICATION → LINEAR + TIME-INVARIANT → IMPULSE RESPONSE h(t) → CONVOLUTION INTEGRAL → OUTPUT y(t)

The 14 System Chambers

Ontological Mapping: Operators, Classification, and Convolution

CHAMBER 01

What Is a System?

Mathematical operator mapping input to output
Black-box modeling of physical realities
Notation: y(t) = T{x(t)}
SISO vs MIMO concepts

CHAMBER 02

System Classification

The core analytical framework
Linear vs Nonlinear
Time-Invariant vs Time-Varying
Memoryless vs Memory, Causal vs Noncausal

CHAMBER 03

Memory (Dynamic)

Memoryless: y(t) depends only on x(t)
Memory: Depends on past or future inputs
Delays, integrators, accumulators
Static vs Dynamic distinction

CHAMBER 04

Linearity

Homogeneity: T{A·x(t)} = A·T{x(t)}
Additivity: T{x₁ + x₂} = T{x₁} + T{x₂}
Superposition Principle
Zero-input test failure (e.g., y(t) = x(t) + 2)

CHAMBER 05

Time Invariance

System behavior does not change over time
Shift then Transform == Transform then Shift
y(t - t₀) = T{x(t - t₀)}
Systems with x(-t) or x(2t) fail this

CHAMBER 06

Causality

Output depends only on present and past
Cannot anticipate the future
Essential for real-time physical realization
Impulse response h(t) = 0 for t < 0

CHAMBER 07

BIBO Stability

Bounded Input → Bounded Output
Prevents outputs from exploding to infinity
CT: ∫ |h(t)| dt < ∞ (Absolute integrability)
DT: Σ |h[n]| < ∞ (Absolute summability)

CHAMBER 08

Invertibility

Unique mapping from input to output
Ability to exactly recover x(t) from y(t)
Loss of information (e.g., y=x²) is noninvertible
LTI Inverse: h(t) * h_inv(t) = δ(t)

CHAMBER 09

LTI Systems

Linearity + Time-Invariance combined
The Holy Grail of system analysis
Guarantees predictability via Superposition
Fully characterized by Impulse Response h(t)

CHAMBER 10

Impulse Response

System's output when struck by δ(t) or δ[n]
Contains the entire "DNA" of an LTI system
Sifting property foundation
Determines causality and stability directly

CHAMBER 11

Convolution Integral

y(t) = x(t) * h(t) = ∫ x(τ)h(t-τ)dτ
Algorithm: Flip, Shift, Multiply, Integrate
Summation equivalent for Discrete-Time
Sliding overlap area generates output

CHAMBER 12

Convolution Properties

Commutative: x * h = h * x
Associative: Cascaded systems h₁ * h₂
Distributive: Parallel systems h₁ + h₂
System reduction and simplification

CHAMBER 13

Step Response & ODEs

Step Response s(t) is integral of h(t)
LTI systems defined by Differential Equations
Initial conditions / Zero-state response
Difference Equations for DT systems

FINAL PROTOCOL

14 — The Transformer Laboratory

1. SYSTEM: y(t) = x(t) * e^(-t)u(t)

2. h(t) = e^(-t)u(t) -> CAUSAL (h(t)=0 for t<0)

3. ∫ |h(t)| dt = 1 < ∞ -> STABLE (BIBO)

[1] TEST LINEARITY [2] FLIP & SHIFT [3] EVALUATE OVERLAP [4] CHECK INTEGRABILITY [5] DERIVE EQUATION

In Realm 01, we studied signals in isolation. Here, we discovered that if a system is Linear and Time-Invariant, its entire soul is contained within a single waveform: the Impulse Response. Convolution allows us to solve the system perfectly in the time domain. But integration is difficult. To make the math easier, we must transcend time entirely and enter the Frequency Domain—the gateway to Realm 03: Fourier Analysis.

CONVOLUTION DIAGNOSTIC RUNNING