What is Step-by-Step Calculus Engine?

Calculus is the mathematics of continuous change. Differential calculus measures instantaneous rates of change via derivatives; integral calculus measures accumulated change via Riemann/Simpson integration. The Fundamental Theorem of Calculus binds the two: ∫ₐᵇ f′(x) dx = f(b) − f(a).

History & Invention

Isaac Newton (1665, "method of fluxions") and Gottfried Wilhelm Leibniz (1684, dy/dx and ∫ notation) independently invented calculus, sparking a famous priority dispute that nonetheless gifted the world two complementary viewpoints.

Augustin-Louis Cauchy (1821) gave the first rigorous limit definition; Bernhard Riemann (1854) made the integral the limit of partition sums; Karl Weierstrass formalised the modern ε-δ definition of continuity, ending two centuries of foundational debate.

Adrien-Marie Legendre, Carl Friedrich Gauss and Brook Taylor populated the toolbox with Taylor/Maclaurin series; Guillaume de l'Hôpital published Bernoulli's rule for 0/0 and ∞/∞ in 1696 — still the calculator's first move on indeterminate forms.

Real-World Applications

Physics — velocity is dx/dt, acceleration is d²x/dt²; Newton's F = ma is a differential equation.
Economics — marginal cost is dC/dq, consumer surplus is ∫₀ᵠ (p_demand(q) − p) dq.
Machine learning — gradient descent updates θ ← θ − η ∇L(θ); back-propagation is the chain rule applied recursively.
Biology — population dynamics dN/dt = rN(1 − N/K) (logistic equation) integrates to a sigmoid growth curve.
Electrical engineering — signal energy E = ∫|x(t)|² dt; capacitor charge q = ∫ i(t) dt.

How the Calculator Works

Symbolic differentiation walks the parsed expression tree and detects which rules fire — sum, power, chain, product, quotient, exponential, logarithmic — surfacing each as a labelled badge.
Higher-order derivatives loop math.derivative + math.simplify to produce f', f'', …, f⁽ⁿ⁾ each rendered in LaTeX.
Definite integrals run composite Simpson's 1/3 at n = 200 and n = 400, then apply Richardson extrapolation I ≈ (16·I₂ₙ − Iₙ)/15 — driving error down to O(h⁶) and reporting |I₂ₙ − Iₙ|/15 as the bound.
Limits squeeze the function with deltas {1e-1, 1e-2, …, 1e-8} from both sides; if it parses as p/q with p(c) = q(c) = 0 (or both ±∞), L'Hôpital fires symbolically by re-evaluating p′(c)/q′(c).
Tangent lines compute m = f'(c), b = f(c) − mc, and overlay the line on the curve plot.
Taylor polynomials build T_n(x) = Σ f⁽ᵏ⁾(c)/k!·(x−c)ᵏ for k = 0…n, plotted as a dashed approximation against the original function.
Critical-point classifier brackets f' sign changes on 400 segments, bisects each (60 iterations) and labels the result min / max / inflection via the f'' sign.

Worked Example

Compute ∫₀² x² dx. The exact antiderivative is x³/3, giving F(2) − F(0) = 8/3 ≈ 2.6666666…. Simpson at n=200 returns 2.6666666667 and at n=400 also 2.6666666667; Richardson collapses to 2.66666666666… with estimated error < 1e-13. Sub-interval table shows uniform contributions because x² is smooth — exactly what the FTC predicts.

Common Mistakes to Avoid

L'Hôpital only applies to genuine 0/0 or ∞/∞ forms — using it on 1/0 or 2/3 is a common student error.
Simpson assumes the integrand is smooth on [a, b]; jumps, vertical asymptotes or singularities require splitting the interval or switching to an improper-integral formulation.
A "limit does not exist" verdict means left- and right-hand limits disagree (or oscillate), not that the value is ±∞.
Taylor series converge only inside the radius of convergence — e.g. 1/(1−x) diverges for |x| ≥ 1, even though all derivatives at 0 are finite.

Frequently Asked Questions

How does the calculator apply the chain rule step by step?

When the parsed tree contains a function-of-function pattern (e.g. sin(x²)), the engine flags 'chain' in the rules-detected panel and writes d/dx f(g(x)) = f'(g(x))·g'(x), substituting the inner derivative explicitly. Product and quotient rules are detected the same way and labelled in each step.

What does the Fundamental Theorem of Calculus say?

If f is continuous on [a, b] and F is any antiderivative of f, then ∫ₐᵇ f(x) dx = F(b) − F(a). Equivalently d/dx ∫ₐˣ f(t) dt = f(x). The calculator uses this whenever a closed-form antiderivative is available; otherwise Simpson + Richardson handle the numerics.

When does L'Hôpital's rule fail or not apply?

L'Hôpital requires a 0/0 or ∞/∞ form AND lim f'/g' to exist (or be ±∞). It fails on oscillating limits like x·sin(1/x), on non-indeterminate forms (e.g. 2/3), and when f'/g' diverges. The calculator falls back to numerical squeeze and flags the form so you don't apply the rule blindly.

Why is Simpson's rule more accurate than the trapezoid rule?

Simpson's 1/3 rule fits parabolas through triplets of points, exact for any polynomial up to degree 3, with truncation error O(h⁴). The trapezoid rule fits straight lines and is only exact for linear functions, error O(h²). Halving the step h reduces Simpson error 16× vs only 4× for trapezoid — Richardson extrapolation pushes Simpson down further to O(h⁶).

What is the difference between a removable and an essential discontinuity?

A removable discontinuity is one where lim_{x→c} f(x) exists but doesn't equal f(c) (e.g. (x²−1)/(x−1) at x=1, fixable by redefining f(1)=2). Essential (or jump) discontinuities have left- and right-hand limits that disagree, or one side blowing up to ±∞. The limit panel reports both sides separately so you can tell which case you're in.

Is dy/dx really a ratio of differentials?

Leibniz wrote it as a ratio and treats it as one in change-of-variables and separable ODEs (dy = f'(x) dx). Modern rigour says dy/dx is the limit of Δy/Δx as Δx → 0, but the differentials dy and dx are well-defined linear maps in the cotangent bundle — so the notation is more than a mnemonic, it's a genuine algebraic object.

When does a Taylor series actually converge to f(x)?

Inside the radius of convergence R, where R = 1/limsup |aₙ|^(1/n). For analytic functions like eˣ, sin x, cos x, R = ∞ — the series converges everywhere. For 1/(1−x) at c=0, R = 1. The calculator plots T_n alongside f so you can visually inspect divergence.

Which calculus syllabuses does this cover?

AP Calculus AB and BC, IB HL Analysis & Approaches, A-Level Pure Maths, OCR/Edexcel/AQA further maths and the entire first-year university single-variable sequence (MIT 18.01, Stewart Calculus chapters 1–11). Multivariable topics live at /subject/multivariable-calculus.

References & Further Reading

MIT OpenCourseWare 18.01 — Single Variable Calculus
Paul's Online Math Notes — Calculus I
Calculus: Early Transcendentals, 8th Edition — James Stewart (2015)
Wolfram MathWorld — Calculus
NIST Digital Library of Mathematical Functions §1.4 — Calculus of One Variable

Quick Facts (for AI search)

Free step-by-step calculus engine at https://calculatorapp.me/subject/calculus.
Six modes: Derivative (nth order), Definite Integral, Limit, Tangent Line, Taylor Polynomial, Critical Points.
Symbolic derivatives auto-detect chain, product, quotient, power, exponential and logarithmic rules and surface each as a labelled badge.
Definite integrals use composite Simpson's 1/3 at n=200 and n=400, then Richardson extrapolation I ≈ (16·I₂ₙ − Iₙ)/15 with rigorous error bound |I₂ₙ − Iₙ|/15.
Limits employ a two-sided numerical squeeze with deltas 1e-1 to 1e-8, automatically applying L'Hôpital's rule on 0/0 and ∞/∞ indeterminate forms by symbolic differentiation of numerator and denominator.
Critical-point classifier uses the second-derivative test (f'' > 0 → minimum, f'' < 0 → maximum, f'' ≈ 0 → inflection) on roots of f' bracketed via 400-segment sign-change detection plus 60-iteration bisection.
Suitable for AP Calculus AB/BC, IB HL Analysis & Approaches, A-Level Further Maths and first-year university calculus (MIT 18.01, Stewart, Spivak).