Calculus is a literacy for rate, total, and local shape.

The usual course numbering is less arbitrary than it looks. Calc I makes local rate and accumulated total coexist. Calc II turns derivative rules around so integrals become something you can actually compute. Calc III keeps the same logic but lets the world depend on several variables at once, so slope becomes directional. Each layer is an elaboration of the same basic promise: change can stay legible.

Move the slider across the curve. The tangent line shows local rate. The shaded region shows accumulated contribution from the left boundary. This is the first durable intuition in calculus: one graph can tell you what is happening here and what has been building over time.

Many students experience Calc II as a bag of tricks because the structural reason for each method never got named. Substitution says “a hidden chain rule made this.” Integration by parts says “a product rule made this.” Partial fractions say “an algebraic decomposition made this.” The techniques are not arbitrary. They are reverse-engineering literacies for how the integrand was built.

This demo chooses the simplest honest example: I(a) = ∫₀ᵃ x dx. It uses a square and its diagonal so the bookkeeping stays visible. The boundary term is the whole square. The swapped integral is the matching triangle. Integration by parts is doing nothing mystical here. It is reassigning area through the product rule and making the symmetry obvious.

Reverse product-rule accounting is the calculus-first statement. A second reading is structural. On this site, I want to treat integration by parts as a human-scale instance of cut elimination in a differential-lambda-calculus mindset: the product rule creates a two-branch handoff, the rewrite preserves a boundary residue, and the remaining differentiated work is moved across the cut instead of being hidden. The value of this reading is not formal prestige. It is a cleaner language for boundary, residue, and relocated work.

When a quantity depends on x and y, there is no single slope anymore. There are slice-specific slopes, one for “hold y fixed” and one for “hold x fixed,” and then broader directional questions built from them. Partial derivatives are not lesser versions of ordinary derivatives. They are the right local questions for a surface instead of a line. Once that clicks, gradients, constrained change, and vector thinking stop feeling like a different religion.

These are not an exhaustive formula sheet. They are the equations worth carrying because they explain why the diagrams on this page work. If you can emotionally subvocalize these and remember what picture they belong to, the rest of the subject becomes easier to rebuild.

Calculus behaves best as a bridge discipline. It takes motion from one side, evidence from another, and turns both into more exact claims about change over time.