Convolution

Prime #

751

Origin domain

Mathematics

Subdomain

signal processing → Mathematics

Core Idea

Convolution slides one fixed kernel across an input, taking a weighted local sum at every position so each output value is a localized, time-shifted mixture of input values produced by a single mixing pattern applied identically everywhere.

How would you explain it like I'm…

Slide and Smear

Imagine smearing each dot of a drawing a little into its neighbours to make it blurry. To find the new colour of one spot, you mix in a bit of all the spots around it. You slide that same little mixing recipe across the whole picture. That sliding-and-mixing is convolution.

The Sliding Mixer

Convolution takes two signals and lets one of them reshape the other by sliding, weighting, and adding. You take a small pattern of weights, called a kernel, and lay it over each spot of the input, multiply the nearby values by those weights, and add them up to get the new value at that spot. Then you slide the kernel one step over and do it again, everywhere. A blur kernel averages neighbours together to smooth a photo; a different kernel can sharpen edges instead. The trick is that the same little recipe is applied identically at every position.

Sliding Weighted Sum

Convolution is a sliding weighted sum: at every position, you replace a value with a weighted mixture of the input values around it, using a fixed pattern called the kernel. Formally you flip the kernel and slide it across the input, multiplying and summing at each offset. The single idea covers things that look unrelated: smoothing and edge detection (each point becomes a weighted average of its neighbourhood), fading memory (the present output is a decaying mixture of past inputs, as in an echo), and combining randomness (the distribution of a sum of two independent random variables is the convolution of their distributions). What stays the same is the mixing pattern; only its meaning changes from blur to memory to probability.

 

Convolution is the operation in which one signal acts on another by sliding, weighting, and summing: each output value is a localized, time-shifted mixture of input values produced by a single fixed kernel applied identically everywhere. In continuous form (f*g)(t) = integral of f(tau)g(t-tau) dtau, and the discrete sum mirrors it; in both, the second function is reflected and slid across the first. The same shape captures three families that look distinct until the kernel is named. Filtering and smoothing replace each point with a weighted neighbourhood average (moving averages, Gaussian blur, edge detection). Memory and persistence make the present a fading mixture of the past (impulse-response systems, RC circuits). And spread of independent contributions appears because the density of a sum of independent random variables is the convolution of their densities. Convolution is unusually well-behaved algebraically: it is commutative, associative, and bilinear, has the delta function as identity, and diagonalizes under the Fourier transform, which turns convolution into pointwise multiplication and makes the cross-domain transfer load-bearing rather than cosmetic.

Broad Use

• Signal and image processing: low-pass filters, Gaussian blur, edge detection, and reverb are all convolution with the right kernel.
• Probability: the density of a sum of independent variables is the convolution of their densities, and the central limit theorem is iterated convolution.
• Neural computation: convolutional networks share one kernel across positions to encode translation invariance, modeling visual receptive fields.
• Physics: a linear time-invariant system's response is the convolution of input with its impulse response; Green's functions solve linear PDEs the same way.
• Economics: distributed-lag models convolve past shocks with a response weight; queueing waiting-times convolve service times.
• Social diffusion: cascade models compute influence at each node as a convolution of upstream activations with an influence kernel.

Clarity

Naming a process as convolution forces the analyst to write down the kernel — the mixing pattern itself — so disputes about "what causes the smoothing" become questions about a function one can plot.

Manages Complexity

A potentially infinite history of dependence collapses into one small, transparent, sliceable object: identify the kernel and reason against it rather than against the full input-output behaviour.

Abstract Reasoning

Convolution's algebraic richness — commutativity, associativity, a delta-function identity, and Fourier diagonalization into pointwise multiplication — lets filters cascade, reorder, and be analysed in whichever domain is easier.

Knowledge Transfer

• Policy design: a central banker's choice of how to respond to past shocks is a kernel-design problem importing signal-processing intuition about smoothing and leakage.
• Diagnostics: the impulse response is a universal probe — engineers inject an impulse, macroeconomists compute impulse-response functions, both reading off one kernel.
• Inference: deconvolution recovers an unknown cause from an observed effect given a propagation kernel, identical in astronomy, medical imaging, and policy evaluation.

Example

A structural engineer computes a building's earthquake response as the convolution of ground acceleration with the building's impulse response, collapsing an infinite-dimensional dynamic mapping into one interpretable function and adding dampers to reshape it.

Relationships to Other Primes

Parents (1) — more general patterns this builds on

• Convolution is a kind of Function (Mapping) — The file: convolution is 'the UNIQUE linear, translation-invariant' function mapping — one fixed kernel applied identically everywhere. A sharply constrained specialization of function_mapping (an arbitrary input-output rule).

Path to root: Convolution → Function (Mapping)

Not to Be Confused With

• Convolution is not Correlation because convolution flips the kernel and computes what a system produces (commutative, cascading), whereas cross-correlation slides without flipping and measures similarity at each lag.
• Convolution is not Function Mapping because convolution is the unique linear, translation-invariant mapping — one fixed kernel applied identically everywhere — whereas a general mapping can have a different rule at every position.
• Convolution is not Attention because convolution's weights are a fixed kernel applied at every position, whereas attention computes input-dependent, position-varying weights from the data at run time.