Loading video...

Video Failed to Load

Go Home

Has anyone ever come across this way of visualizing the convolution of two functions f and g: Look at the graph of the two-variable function f(x)g(y), and consider diagonal slices over the line x + y = k. The area of those slices represents (f * g)(k).

397,338 views • 3 years ago •via X (Twitter)

9 Comments

Grant Sanderson's profile picture
Grant Sanderson3 years ago

I doubt this is original, but I haven't seen it anywhere. I'm searching to see if there's a good source to cite here. If you've come across it, feel free to share links with me!

Nick's profile picture
Nick3 years ago

Yes, any binary operation can be visualized this way. Addition in the integers is a graph over a lattice, and in the continuum, kernels of integral operators like the fourier transform, green's functions, are seen as graphs over the plane rather than matrices. Just order the vars

Smarter Every Day's profile picture
Smarter Every Day3 years ago

It’s pretty. A great way to visualize a hike. I can imagine you doing a slice on a strange meandering path, and then “unrolling” it.

Mathoma's profile picture
Mathoma3 years ago

See 18:00 or so here:

Chris Billington's profile picture
Chris Billington3 years ago

I tried

Craig Gidney's profile picture
Craig Gidney3 years ago

Anecdotally, this is how I've always visualized it and I assumed it was the "typical" way to view it. I do have a couple of old diagrams I drew of it, e.g. at there's

Tanuj Alapati's profile picture
Tanuj Alapati3 years ago

Pretty sure my probability textbook showed this method for when you have two rvs X and Y, and you want to calculate a pdf of Z := X+Y by integrating along a diagonal. Super intuitive

Artur Chakhvadze's profile picture
Artur Chakhvadze3 years ago

I wonder whether this leads to some interesting generalisations of convolution for alternative couplings (e.g. optimal transport)

Luluberlu's profile picture
Luluberlu3 years ago

Je me le suis toujours représenté comme ça jpense pas que c’est nouveau

Related Videos

Do you actually know what convex optimization is in the geometric, guarantee-theoretic sense or have you only met it through solvers and loss curves? Convexity is rare comfort in optimization...there are no spurious local minima, no surprise traps, and inequalities you can use like tools instead of prayers. So, what is this convexity? Let x = (x₁, x₂) and let f(x) be convex. Plot the surface z = f(x). Pick a contact point x₀. The local slope is the gradient p = ∇f(x₀). That p is exactly the data that defines the supporting plane: z = f(x₀) + p · (x − x₀). Thus, f is said to be convex because for every x, f(x) ≥ f(x₀) + p · (x − x₀). So the plane at x₀ can slide under the surface, but it never slices through it. Not near the point...everywhere. Now for here is the interesting part: The slope becomes a coordinate system! Rewrite the same plane as z = p · x − b, where b is the offset. Because the plane passes through (x₀, f(x₀)), the offset is forced to be b = p · x₀ − f(x₀). And that number isn’t just geometry trivia. It’s the convex conjugate: f*(p) = sup over x ( p · x − f(x) ). At a differentiable contact point, the supporting plane touches f tightly enough that the supremum is achieved at x₀, giving the identity f*(p) = p · x₀ − f(x₀) when p = ∇f(x₀). So one moving contact point gives two linked readouts: primal position x₀ dual position (slope) p = ∇f(x₀) dual offset f*(p) One surface. Two worlds. #ConvexOptimization #Optimization #MachineLearning #SignalProcessing #AppliedMath #Engineering

Mathelirium

38,506 views • 6 months ago