Calc · whole subject on one page

Short notes: Calculus

Every topic in revision form: the key ideas, formulas, the question types GATE asks, and the traps. Read it the night before.

1. 🎯 Limits & Continuity

High yield ●●●6 PYQs →

The most repeated calculus question in GATE: evaluate a 0/0 limit and give a number. Continuity of a piecewise function and the intermediate value theorem come right after.

  • What a limit means. limx→a f(x) = L means f(x) gets as close to L as you like by taking x close enough to a (but x ≠ a). The value f(a) itself is irrelevant. A two-sided limit exists only if the left-hand and right-hand limits both exist and are equal. If either side blows up, oscillates or is undefined, the limit does not exist. Estimate by tabulating f at a ± 0.1, 0.01, 0.001.
    → The limit only cares about nearby values. Both sides must agree.
  • Evaluating limits: factor, rationalise, substitute. First substitute x = a. If you get a number, you are done. If you get 0/0: factor and cancel (x⁴ − 81 = (x − 3)(x + 3)(x² + 9)), or rationalise a square root (multiply by the conjugate), or use a standard limit. Then substitute again. Example (GATE 2019): limx→3 (x⁴ − 81)/(2x² − 5x − 3) = (x + 3)(x² + 9)/(2x + 1) at 3 = 6·18/7 = 108/7.
    → Substitute, then factor or rationalise, then substitute again.
  • Indeterminate forms and L'Hôpital's rule. The seven indeterminate forms are 0/0, ∞/∞, 0·∞, ∞ − ∞, 0⁰, 1^∞, ∞⁰. L'Hôpital: for 0/0 or ∞/∞, lim f/g = lim f′/g′ if the latter exists (differentiate top and bottom separately; repeat if still indeterminate). Turn 0·∞ into a quotient, and take logarithms for 0⁰, 1^∞, ∞⁰. Example (GATE 2021): limx→−3 (√(2x + 22) − 4)/(x + 3) = 1/√(2x + 22) = 1/4 = 0.25.
    → Only for 0/0 or ∞/∞. Differentiate top and bottom separately.
  • Standard limits and limits at infinity. sin x / x → 1, (1 − cos x)/x² → ½, tan x / x → 1, (eˣ − 1)/x → 1, ln(1 + x)/x → 1, (aˣ − 1)/x → ln a, (1 + 1/x)ˣ → e as x → ∞ (and (1 + x)^{1/x} → e as x → 0), sin(x − a)/(x − a) → 1 as x → a (GATE 2016: answer 1). At infinity compare growth: ln x ≪ xᵖ ≪ eˣ ≪ x!. For a rational function divide by the highest power of x.
    → Memorise sin x/x, (eˣ−1)/x, ln(1+x)/x, (1+1/x)ˣ.
  • Continuity and types of discontinuity. f is continuous at a if (1) f(a) exists, (2) lim f exists, (3) they are equal. A polynomial, sin, cos, eˣ are continuous everywhere; a rational function is continuous except where the denominator is 0. Discontinuities: removable (limit exists but f(a) is missing or wrong), jump (both one-sided limits exist but differ), infinite (blows up), oscillating (sin(1/x)). For a piecewise function compare left value, right value and f(a) (GATE 2013).
    → Continuous = value exists, limit exists, and they are equal.
  • Intermediate value theorem. If f is continuous on [a, b] and k lies between f(a) and f(b), then f(c) = k for some c in [a, b]. In particular, if f(a) and f(b) have opposite signs, f has a root in (a, b); bisection uses this. Typical use: apply IVT to a new continuous function built from f, such as g(y) = f(y) − f(y + 1) (GATE 2014: g(0) = −2, g(1) = 2, so f(y) = f(y + 1) somewhere in (0, 1)). A cubic always has a real root (its ends go to ±∞).
    → Continuous + opposite signs at the ends ⇒ a root in between.

Formulas

Two-sided limit
lim f(x) exists ⇔ lim from the left = lim from the right (both finite)
Continuity at a
lim f(x) as x → a = f(a) (value exists and matches)
Standard limits
sin x / x → 1; (1 − cos x)/x² → 1/2; (eˣ − 1)/x → 1; ln(1 + x)/x → 1; tan x / x → 1
Definition of e
lim (1 + 1/x)ˣ = e as x → ∞; lim (1 + x)^{1/x} = e as x → 0
L'Hôpital (0/0 or ∞/∞)
lim f/g = lim f′/g′ (only if that limit exists)
Growth order at ∞
ln x ≪ xᵖ ≪ aˣ (a > 1) ≪ x!
Squeeze theorem
g ≤ f ≤ h and lim g = lim h = L ⇒ lim f = L

What GATE asks

  • ★★★Evaluate a limit (NAT or MCQ). A 0/0 form after substitution. Factor and cancel, rationalise, or L'Hôpital. Check the answer with a table of values.
  • ★★★Is a piecewise function continuous at a point?. Compute the left limit, right limit and the value. All three must agree.
  • ★★★Existence-type statements using IVT. Define a helper function (a difference) and check signs at the ends.

Traps

Do not substitute blindly: 0/0 means "not decided yet", not "0" and not "undefined".
L'Hôpital differentiates the numerator and denominator separately; it is not the quotient rule.
A limit can exist even where f is undefined (a hole). Do not confuse the limit with f(a).
With √x or ln x the function exists on one side only, so only a one-sided limit is meaningful (GATE 2022 used x → 0⁺).
When x → ∞ in a ratio of polynomials, the answer depends only on the leading terms.

2. 📐 Differentiation & Mean Value Theorems

High yield ●●●6 PYQs →

Differentiability of piecewise and max/abs functions, monotonicity and the mean value theorems are tested almost every other year, usually as a one-mark conceptual question.

  • The derivative as slope, and the rules. f′(a) = limh→0 (f(a + h) − f(a))/h is the slope of the tangent. Rules: (uv)′ = u′v + uv′, (u/v)′ = (u′v − uv′)/v², chain rule (f(g(x)))′ = f′(g)·g′. Table: (xⁿ)′ = nxⁿ⁻¹, (eˣ)′ = eˣ, (ln x)′ = 1/x, (sin x)′ = cos x, (cos x)′ = −sin x, (tan x)′ = sec²x. Higher derivatives: f″ is the derivative of f′. Example (GATE 2014): f = x sin x ⇒ f″ + f = 2 cos x.
    → Derivative = slope of the tangent = limit of secant slopes.
  • Differentiable vs continuous. Differentiable ⇒ continuous, but not the other way round. A function fails to be differentiable at a corner (|x| at 0: slopes −1 and +1), a cusp or vertical tangent (x^{1/3} at 0), or a jump. For max{f, g} or min{f, g} look at the points where f = g and compare the two slopes: GATE 2024, f = max{x, x³} crosses at x = −1, 0, 1, with different slopes at each, so it is not differentiable exactly at {−1, 0, 1}.
    → Corners and cusps break differentiability; hunt them at crossings.
  • Piecewise functions: matching value and slope. For a piecewise function to be differentiable at the join x = c you need two equations: the pieces have the same value there (continuity) and the same derivative. GATE 2025: f = ax + b for x < 1, x³ + x² + 1 for x ≥ 1. Value: a + b = 3. Slope: a = 3x² + 2x at 1 = 5. So a = 5, b = −2. If only continuity is asked, one equation is enough.
    → Differentiable at a join: value AND slope must match.
  • Rolle's theorem and Lagrange's mean value theorem. Rolle: f continuous on [a, b], differentiable on (a, b) and f(a) = f(b) ⇒ f′(c) = 0 for some c in (a, b). Lagrange (MVT): same hypotheses without f(a) = f(b) ⇒ f′(c) = (f(b) − f(a))/(b − a). Both hypotheses matter: |x| on [−1, 1] and 1/x on [−1, 1] break them. GATE 2014: f(θ) is a determinant that is 0 at both π/6 and π/3, so Rolle gives f′(θ) = 0 inside, and f is not identically 0, so f′ ≠ 0 elsewhere: both statements true.
    → Chord slope = tangent slope at some c (needs continuity and differentiability).
  • Increasing, decreasing and polynomial facts. f′ > 0 on an interval ⇒ f increasing there; f′ < 0 ⇒ decreasing. "Increasing everywhere on [0, 1]" needs f′ ≥ 0 on the whole interval: x² − sin x has f′ = 2x − cos x = −1 at 0, so it is not. For polynomials: deg f′ = deg f − 1; the derivative of an even function is odd and of an odd function is even; f(x) + f(−x) keeps only the even-power terms (GATE 2016: if that has degree 10 then g(x) − g(−x), where g = f′, has degree 9). A cubic has a local max and a local min only if f′ has two distinct real roots.
    → f′ ≥ 0 everywhere on the interval ⇒ increasing; a single negative spot breaks it.

Formulas

Definition
f′(a) = lim (f(a + h) − f(a)) / h as h → 0
Product / quotient
(uv)′ = u′v + uv′; (u/v)′ = (u′v − uv′)/v²
Chain rule
d/dx f(g(x)) = f′(g(x))·g′(x)
Common derivatives
(xⁿ)′ = nxⁿ⁻¹; (eˣ)′ = eˣ; (ln x)′ = 1/x; (sin x)′ = cos x; (cos x)′ = −sin x; (tan x)′ = sec² x
Rolle
f(a) = f(b) ⇒ ∃ c ∈ (a, b): f′(c) = 0
Lagrange MVT
∃ c ∈ (a, b): f′(c) = (f(b) − f(a)) / (b − a)
Smooth join of two pieces at c
value: left(c) = right(c); slope: left′(c) = right′(c)

What GATE asks

  • ★★★Where is f NOT differentiable? (max, min, abs). Find where the pieces meet, then compare the left and right slopes. List every such point.
  • ★★★Find a constant that makes a piecewise function differentiable. Two equations: equal values and equal derivatives at the join. Solve for the two unknowns.
  • ★★★Derivative of a given expression (identity). Differentiate twice and match terms (for example f = x sin x).
  • ★★★Statements using Rolle / MVT or monotonicity. Check the hypotheses; use the sign of f′ for increasing or decreasing.

Traps

Continuous does not imply differentiable (|x|). The reverse is always true.
For a piecewise differentiability question, matching only the values is half the job; the slopes must match too.
MVT and Rolle guarantee that a c exists; they do not say it is unique.
An increasing function can still have f′ = 0 at isolated points (x³ at 0).
Differentiating a product term by term (uv)′ = u′v′ is a classic mistake.

3. ⛰️ Maxima, Minima & Taylor Series

Medium ●●3 PYQs →

Finding local extrema of a polynomial or a quadratic arch, and Taylor series approximations, are easy marks if you can run the f′ = 0 and f″ tests quickly.

  • Critical points and the first-derivative test. A local extremum of a differentiable f can only occur where f′(x) = 0 (a critical point; also check where f′ does not exist). If f′ changes from + to − the point is a local maximum; from − to + a local minimum; if the sign does not change it is neither (x³ at 0). GATE 2023: f = x³ + 15x² − 33x − 36, f′ = 3(x + 11)(x − 1), so critical points are −11 and 1.
    → Set f′ = 0, then check the sign change. A zero slope alone is not an extremum.
  • Second-derivative test, inflection points and endpoints. At a critical point: f″ > 0 ⇒ local min, f″ < 0 ⇒ local max, f″ = 0 ⇒ inconclusive (use the first-derivative test). An inflection point is where f″ changes sign (concavity changes). On a closed interval the global max and min are the best among critical points and endpoints. GATE 2023: f″(−11) = −36 < 0 (max) and f″(1) = 36 > 0 (min), so both a local max and a local min exist. GATE 2012: on [π/4, 7π/4] the key counted minima at π/4 (endpoint) and 3π/2.
    → f″ > 0 min, f″ < 0 max. Do not forget the endpoints.
  • Optimisation problems. Write the quantity to optimise as a function of one variable, set its derivative to 0, and check the type. For a quadratic ax² + bx + c the extremum is at x = −b/2a, a minimum if a > 0 and a maximum if a < 0, with value c − b²/4a. GATE 2012: arch y = 2x − 0.1x² has its maximum at x = 10, so the greatest height is 20 − 10 = 10 m. Always verify that the optimum lies inside the allowed range.
    → Quadratic vertex: x = −b/2a. a < 0 means maximum.
  • Taylor and Maclaurin series. Taylor series of f about a: f(x) = Σ f⁽ⁿ⁾(a)(x − a)ⁿ / n!. Maclaurin means a = 0. A polynomial of degree n made from the first terms approximates f well near a; the error shrinks as n grows (inside the radius of convergence). The linear approximation f(x) ≈ f(a) + f′(a)(x − a) is the tangent line. Lagrange remainder: Rn = f⁽ⁿ⁺¹⁾(ξ)(x − a)ⁿ⁺¹/(n + 1)!.
    → Taylor: Σ f⁽ⁿ⁾(a)(x − a)ⁿ/n!. Maclaurin is the case a = 0.
  • Standard series and where they work. eˣ = Σ xⁿ/n! (all x), sin x = x − x³/3! + x⁵/5! − … (all x), cos x = 1 − x²/2! + x⁴/4! − … (all x), ln(1 + x) = x − x²/2 + x³/3 − … (−1 < x ≤ 1), 1/(1 − x) = 1 + x + x² + … (|x| < 1), (1 + x)ᵐ = 1 + mx + m(m − 1)x²/2! + …, arctan x = x − x³/3 + x⁵/5 − …. Use them to evaluate limits: (1 − cos x)/x² → 1/2 because cos x = 1 − x²/2 + ….
    → Memorise the standard series and where each one is valid.
  • Functions of two variables: partial derivatives and saddle points. Partial derivatives fx, fy differentiate with respect to one variable, treating the other as a constant. A stationary point has fx = fy = 0. With D = fxxfyy − (fxy)²: D > 0 and fxx > 0 ⇒ local min, D > 0 and fxx < 0 ⇒ local max, D < 0 ⇒ saddle, D = 0 ⇒ test fails. Mixed partials are equal when they are continuous: fxy = fyx.
    → D = fxx·fyy − fxy². D < 0 means saddle.

Formulas

Critical points
f′(x) = 0 (or f′ undefined)
Second-derivative test
f″ > 0 → local min; f″ < 0 → local max; f″ = 0 → inconclusive
Quadratic ax² + bx + c
extremum at x = −b / 2a, value c − b² / 4a (min if a > 0, max if a < 0)
Taylor series about a
f(x) = Σ f⁽ⁿ⁾(a) (x − a)ⁿ / n!
Maclaurin series
eˣ = Σ xⁿ/n!; sin x = x − x³/3! + …; cos x = 1 − x²/2! + …; 1/(1 − x) = Σ xⁿ (|x| < 1)
Series of ln
ln(1 + x) = x − x²/2 + x³/3 − … for −1 < x ≤ 1
Two-variable test
D = fxx·fyy − fxy²; D > 0 & fxx > 0: min; D > 0 & fxx < 0: max; D < 0: saddle

What GATE asks

  • ★★★Does f have a local max / local min? (polynomial). Solve f′ = 0 and use the sign of f″ at each root. A cubic with two real critical points has one of each.
  • ★★★Maximum height / profit of a quadratic. Use the vertex x = −b/2a, then plug in. No calculus required.
  • ★★★Number and location of local minima on an interval. Remember to consider the endpoints of the interval.

Traps

f′(c) = 0 does not guarantee an extremum (x³ at 0).
A global max/min on a closed interval may sit at an endpoint, where f′ is not zero.
The Taylor series of ln(1 + x) and 1/(1 − x) only converge for |x| < 1 (ln also at x = 1); outside that, more terms make the answer worse.
D = 0 is inconclusive: do not call it a saddle.
Do not mix up "f(x) = max value" with "x where the max occurs".

4. ∫ Integration

High yield ●●●7 PYQs →

Definite integrals with a symmetry trick (odd/even, a + b − x) and triple integrals that vanish are quick marks; the other half are routine substitutions and parts.

  • The integral as area and the fundamental theorem. The definite integral ∫ab f dx is the signed area under f (area below the axis counts negative). FTC part 1: d/dx ∫ax f(t) dt = f(x). FTC part 2: ∫ab f dx = F(b) − F(a) where F′ = f. Also d/dx ∫ag(x) f(t) dt = f(g(x))·g′(x). Example (GATE 2025): ∫1x t ln t dt = x²ln x/2 − x²/4 + 1/4, which equals 1/4 when ln x = 1/2, i.e. x = √e.
    → F′ = f, and ∫ from a to b = F(b) − F(a).
  • Standard integrals, substitution and parts. ∫xⁿ dx = xⁿ⁺¹/(n + 1), ∫1/x = ln|x|, ∫eˣ = eˣ, ∫sin = −cos, ∫cos = sin, ∫sec² = tan. Substitution: if you see f(g(x))·g′(x), let u = g(x). Example (GATE 2018): ∫0π/4 x cos(x²) dx, u = x², = ½ sin(x²) from 0 to π/4 = ½ sin(π²/16) ≈ 0.289. By parts: ∫u dv = uv − ∫v du, choosing u by L-I-A-T-E (Log, Inverse trig, Algebraic, Trig, Exponential).
    → f(g(x))·g′(x) means substitute. Product of unlike functions means by parts.
  • Properties of definite integrals (symmetry tricks). Odd function over [−a, a]: integral = 0. Even function: = 2∫0a. King's property: ∫ab f(x) dx = ∫ab f(a + b − x) dx. If f(x) + f(a + b − x) is a constant k, then the integral is k(b − a)/2. GATE 2024: f(x) = 1 − f(2 − x) means f(x) + f(2 − x) = 1, so ∫02 f dx = 1·2/2 = 1. Also ∫ab = ∫ac + ∫cb, and swapping the limits flips the sign.
    → Odd: 0. Even: double. f(x) + f(a+b−x) = k gives k(b−a)/2.
  • Numerical integration: Riemann, trapezoid, Simpson. Split [a, b] into n strips of width h = (b − a)/n. Trapezoidal rule: h[(f₀ + fₙ)/2 + f₁ + … + fₙ₋₁]; it over-estimates a convex function. Simpson's 1/3 rule (n even): (h/3)[f₀ + 4f₁ + 2f₂ + 4f₃ + … + fₙ]; exact for polynomials up to degree 3. GATE 2013: points of f(x) = x² at spacing 0.3 on [0, 3] give 0.3 × [(0 + 9)/2 + 25.65] = 9.045 (exact value 9).
    → Trapezoid: h[(f₀+fₙ)/2 + middle]. Simpson needs even n.
  • Improper integrals. An improper integral has an infinite limit or an integrand that blows up inside the range; it is defined as a limit. p-test (type I): ∫1∞ dx/xᵖ converges iff p > 1 (value 1/(p − 1)). Type II: ∫01 dx/xᵖ converges iff p < 1 (value 1/(1 − p)). ∫0∞ e⁻ˣ dx = 1, ∫0∞ xⁿe⁻ˣ dx = n! (Gamma function). Use comparison with a p-integral for convergence questions.
    → At ∞: converges iff p > 1. Near 0: converges iff p < 1.
  • Multiple integrals. Evaluate an iterated integral from the inside out, treating outer variables as constants. Over a box with constant limits the order does not matter (Fubini) and a product integrand factors. Symmetry kills terms: a term that is odd in a variable, over limits symmetric about 0 in that variable, integrates to 0. GATE 2023: ∭ (4x²y − z³) dz dy dx over [−3, 3] × [−2, 2] × [−1, 1]: 4x²y is odd in y and z³ is odd in z, so the answer is 0.
    → Do the inside variable first. Odd term over symmetric limits is 0.

Formulas

FTC
∫ₐᵇ f dx = F(b) − F(a); d/dx ∫ₐˣ f(t)dt = f(x)
Substitution
∫ f(g(x)) g′(x) dx = ∫ f(u) du with u = g(x)
By parts
∫ u dv = uv − ∫ v du (order by L-I-A-T-E)
Odd / even
odd: ∫₋ₐᵃ f = 0; even: ∫₋ₐᵃ f = 2 ∫₀ᵃ f
King's property
∫ₐᵇ f(x) dx = ∫ₐᵇ f(a + b − x) dx; if f(x) + f(a + b − x) = k then the integral = k(b − a)/2
Trapezoidal / Simpson
T = h[(f₀ + fₙ)/2 + f₁ + … + fₙ₋₁]; S = (h/3)[f₀ + 4f₁ + 2f₂ + … + 4fₙ₋₁ + fₙ]
p-integrals
∫₁^∞ dx/xᵖ converges iff p > 1; ∫₀¹ dx/xᵖ converges iff p < 1
Gamma
∫₀^∞ xⁿ e⁻ˣ dx = n! (n a non-negative integer)

What GATE asks

  • ★★★Evaluate a definite integral (NAT). Look for a symmetry (odd/even, a + b − x) before computing. Otherwise substitute.
  • ★★★Solve ∫ = value for the upper limit. Integrate by parts, set the antiderivative equal to the value, solve for x.
  • ★★★Trapezoidal or Simpson estimate from a table. Apply the formula directly with h from the table spacing.
  • ★★★Probability density normalisation. The total integral of a density must be 1; solve for the constant or the limit.

Traps

The area below the axis counts negative in a definite integral. For total area use |f|, split at the roots (GATE 2014: ∫₀^{2π} |x sin x| dx = 4π).
Do not forget to change the limits when you substitute (or go back to x).
p-test conditions are opposite for ∞ limits (p > 1) and for blow-ups at 0 (p < 1).
Symmetry needs symmetric limits about 0 and the right parity of the whole integrand.
Simpson needs an even number of strips.

5. ➕ Sequences & Series

Low ●

A small but listed part of the syllabus: convergence of sequences and series and the radius of a power series. It rarely appears directly, but it underlies Taylor series and asymptotics.

  • Sequences and their limits. A sequence aₙ converges to L if aₙ gets and stays arbitrarily close to L. Every convergent sequence is bounded. A monotonic and bounded sequence always converges. Examples: n/(n + 1) → 1, (1 + 1/n)ⁿ → e, (−1)ⁿ diverges (oscillates), n² → ∞. Squeeze: if bₙ ≤ aₙ ≤ cₙ and bₙ, cₙ → L, then aₙ → L (so sin n / n → 0).
    → Settles to one number: converges. Bounded and monotonic: converges.
  • Series and the geometric series. The series Σ aₙ converges if its partial sums Sₙ = a₁ + … + aₙ approach a finite limit. Geometric: Σ_{n≥0} rⁿ = 1/(1 − r) if |r| < 1 and diverges otherwise; the first term matters: a + ar + ar² + … = a/(1 − r). A telescoping series such as Σ 1/(n(n + 1)) collapses to 1. Examples: Σ 1/n! = e, Σ n/2ⁿ = 2.
    → Geometric series: a/(1 − r) when |r| < 1.
  • Convergence tests. nth-term test: if aₙ does not tend to 0 the series diverges (but aₙ → 0 does not prove convergence: the harmonic series Σ 1/n diverges). p-series: Σ 1/nᵖ converges iff p > 1. Comparison: compare with a known series (0 ≤ aₙ ≤ bₙ). Ratio test: L = lim |aₙ₊₁/aₙ|; converges if L < 1, diverges if L > 1, no decision if L = 1. Alternating series test: terms decrease to 0 ⇒ converges (Σ (−1)ⁿ/n converges but not absolutely).
    → aₙ → 0 is necessary, not sufficient.
  • Power series and radius of convergence. A power series Σ cₙ(x − a)ⁿ converges for |x − a| < R and diverges for |x − a| > R, where R = lim |cₙ/cₙ₊₁| (ratio) or 1/lim sup |cₙ|^{1/n}. The endpoints x = a ± R must be tested separately. Examples: Σ xⁿ (R = 1), Σ xⁿ/n! (R = ∞), Σ n! xⁿ (R = 0), Σ xⁿ/n converges on [−1, 1) only. Taylor series of eˣ, sin x, cos x have R = ∞; ln(1 + x) and 1/(1 − x) have R = 1.
    → R = lim |cₙ/cₙ₊₁|. Always test the endpoints separately.

Formulas

Geometric series
Σ_{n=0}^∞ a rⁿ = a / (1 − r) for |r| < 1
p-series
Σ 1/nᵖ converges iff p > 1 (harmonic series p = 1 diverges)
Ratio test
L = lim |aₙ₊₁ / aₙ|: L < 1 converges; L > 1 diverges; L = 1 undecided
nth-term test
aₙ ↛ 0 ⇒ Σ aₙ diverges
Radius of convergence
R = lim |cₙ / cₙ₊₁|; check the endpoints a ± R by hand
Famous sums
Σ 1/n! = e; Σ 1/n² = π²/6; Σ (−1)ⁿ⁺¹/n = ln 2; Σ 1/(n(n + 1)) = 1

What GATE asks

  • ★★★Convergence of a series or sequence (rare). Apply the ratio test or compare with a p-series; remember that aₙ → 0 is necessary but not sufficient.

Traps

aₙ → 0 does not mean Σ aₙ converges. The harmonic series is the standard counter-example.
The ratio test says nothing when L = 1 (p-series).
A power series needs its endpoints checked separately.
For a geometric series start-index matters: Σ_{n=1}^∞ rⁿ = r/(1 − r).