Calculus's Only Trick: It's About Change

Don't Be Intimidated

Calculus has one gimmick. It is the math of change. Anything that varies — a position over time, a loss over weight values, a price over interest rates — has calculus inside it.

The two halves:

Differentiation — given a function, what's its slope at a point? How fast is it changing right now?
Integration — given a rate of change, what's the accumulated total? How much of "change" has piled up?

That's it. That's all of calculus, distilled. The rest is technique for computing the slopes and totals of specific kinds of functions.

Why AI Doesn't Care About Most of Calculus

Most calculus textbooks teach you how to find derivatives by hand for specific function families (polynomials, trig, exponentials). AI doesn't need this. PyTorch's autograd computes derivatives automatically by tracking operations. What you actually need is the intuition: what does a derivative mean, when do gradients vanish, why does the chain rule matter?

Calculus is about change. Differentiation reads the rate of change; integration sums it up. Modern AI cares about the first far more than the second, and never makes you compute it by hand.

import torch # Autograd computes derivatives automatically x = torch.tensor(3.0, requires_grad=True) y = x ** 2 + 2 * x + 1 # y = (x + 1)^2 y.backward() print(f"dy/dx at x=3: {x.grad.item()}") # 8.0 (= 2x + 2 evaluated at x=3)

Calculus's Only Trick: It's About Change

Don't Be Intimidated

Why AI Doesn't Care About Most of Calculus

Code

External links

Exercise

Progress

Comments 0