Single-variable calculus and applications
Introduced| English | Meaning |
|---|---|
| continuity/kɒntɪˈnjuːɪti/ | The property of a function having its value equal to its limit at a point. |
| chain rule/tʃeɪn ruːl/ | The rule for differentiating a composite function by multiplying the appropriate derivatives. |
| Riemann sum/ˈriːmən sʌm/ | A sum of function values multiplied by subinterval widths, used to approximate a definite integral. |
A corner with no derivative
- A model of distance from a fixed point can have the form $f(x)=|x|$. It has no gap at zero, but its left and right slopes are different. There is no unique tangent with a finite slope there.
- GRE calculus asks which conditions justify a method: a limit, continuity 连续性, a derivative or an integral. A familiar formula is useful only after checking those conditions.
A limit is not the point value
- A finite two-sided limit describes values approaching a point from both sides. Continuity is the property of a function having its value equal to its limit at a point. For an interior point, the two one-sided limits must exist and agree, and the function must have that value there.
- For $x\ne1$, $(x^2-1)/(x-1)=x+1$, so the limit at $1$ is $2$. The original expression is undefined there. If a separate definition assigns $f(1)=5$, the limit is still $2$; assigning $f(1)=2$ instead repairs the discontinuity. The form $0/0$ signals further work, not a numerical answer.
For $x\ne1$, $f(x)=(x^2-1)/(x-1)$, and $f(1)=5$. Find $\lim_{x\to1}f(x)$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
Away from 1 the expression equals x+1. Its limit is 2, independently of the assigned value 5.
A derivative needs a finite limit
- The derivative is $f'(a)=\lim_{h\to0}\dfrac{f(a+h)-f(a)}{h}$ when this limit exists as a finite real number. Differentiability implies continuity. For $|x|$ at zero, the quotient is $-1$ on the left and $1$ on the right, so no derivative exists.
- The cube-root function is also continuous at zero, but its quotient is $1/|h|^{2/3}$, which grows without bound: a vertical tangent rather than a corner. For $x\sin(1/x)$ with value zero at zero, the quotient oscillates. These are different ways for differentiability to fail.
What happens at x = 0?
Separate continuity from differentiability at a point.
At $x = 0$, the function $f(x) = |x|$ is
Difficulty 1/5 · 100 coins + a small bonus on first completion today
The one-sided derivatives are −1 and 1.
Differentiate a composition
- The chain rule 链式法则 is a rule for differentiating a composition by multiplying the outer derivative by the inner derivative. If $g$ is differentiable at $x$ and $f$ at $g(x)$, then $(f\circ g)'(x)=f'(g(x))g'(x)$. For $\ln(x^2+1)$ this gives $2x/(x^2+1)$.
- For $y=\sin(x^2)$, first find $y'=2x\cos(x^2)$. Differentiating this product gives two terms: $y''=2\cos(x^2)-4x^2\sin(x^2)$. At zero, $y''(0)=2$; omitting the derivative of the factor $2x$ loses that term.
Find the derivative of $\ln(x^2 + 1)$ at $x = 1$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
$\frac{2x}{x^2+1}$ at $x = 1$ is $1$.
If $y=\sin(x^2)$, find $y\prime\prime(0)$.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
$y\prime=2x\cos(x^2)$, so $y\prime\prime=2\cos(x^2)-4x^2\sin(x^2)$. Substituting zero gives 2.
What theorems guarantee
- The extreme value theorem guarantees a maximum and minimum for a continuous function on a closed bounded interval. Without the endpoints or continuity, the guarantee may fail: $f(x)=x$ on $(0,1)$ attains neither its supremum nor its infimum.
- The mean value theorem needs continuity on $[a,b]$ and differentiability on $(a,b)$, with $a<b$. It gives a point $c$ inside the interval where $f'(c)=\dfrac{f(b)-f(a)}{b-a}$. For $|x|$ on $[-1,1]$, the endpoint slope is zero but no derivative equals zero; the differentiability condition fails.
Why does the mean value theorem not guarantee a point with zero derivative for $f(x)=|x|$ on $[-1,1]$?
Difficulty 1/5 · 100 coins + a small bonus on first completion today
The function is continuous and has equal endpoint values, but the required derivative does not exist at 0. The theorem therefore cannot be applied.
Candidates are not conclusions
- To find global extrema of a continuous function on $[a,b]$, compare endpoint values with interior critical points where the derivative is zero or does not exist. A stationary point is a candidate, not automatic evidence of a maximum or minimum.
- For $f(x)=x^3$ on $[-1,2]$, $f'(x)=3x^2$ vanishes at zero. Compare $f(-1)=-1$, $f(0)=0$ and $f(2)=8$: the minimum is $-1$ and maximum is $8$. Zero is stationary but is not a local extremum.
A student finds $f\prime(0)=0$ for $f(x)=x^3$ on $[-1,2]$ and calls zero the maximum. Correct the solution.
Continuity on the closed bounded interval guarantees global extrema. Check endpoints and critical points: $f(-1)=-1$, $f(0)=0$, $f(2)=8$. The minimum is -1 at -1 and maximum 8 at 2. The function increases through zero, so zero is not a local extremum.
Accumulation and the fundamental theorem
- A definite integral represents signed accumulation: positive and negative contributions can cancel. If $f$ is continuous on an interval containing $a$ and $x$, then $A(x)=\int_a^x f(t)\,dt$ satisfies $A'(x)=f(x)$ at interior points. The integration variable $t$ is a placeholder, not the upper-bound variable $x$.
- If $G'=f$ and $f$ is continuous on $[a,b]$, then $\int_a^b f(t)\,dt=G(b)-G(a)$. With a differentiable upper bound $h(x)$ and continuous integrand along the relevant interval, the chain rule gives $\dfrac{d}{dx}\int_a^{h(x)}f(t)\,dt=f(h(x))h'(x)$.
For $F(x)=\int_0^{x^2}e^t\,dt$, the moving upper bound is $x^2$. Thus $F'(x)=2xe^{x^2}$ and $F'(1)=2e\approx5.44$. The factor $2x$ comes from the inner derivative.
If $F(x) = \int_0^{x^2} e^t\,dt$, find $F'(1)$ to two decimal places.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
$F'(x) = 2x e^{x^2}$, so $F'(1) = 2e$.
Both bounds can move
- When $u$ and $v$ are differentiable and $f$ is continuous on the relevant intervals, write the integral as the difference of two accumulation functions. For an integrand depending only on $t$:
- For $B(x)=\int_x^{x^2}e^t\,dt$, the derivative is $B'(x)=2xe^{x^2}-e^x$. At $x=1$, the bounds coincide but $B'(1)=e$: the two bounds move at different rates. If the integrand also depends on $x$, this bound-only formula may be incomplete.
For $B(x)=\int_x^{x^2}e^t\,dt$, find $B\prime(1)$ to two decimal places.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
$B\prime(x)=2xe^{x^2}-e^x$, so $B\prime(1)=e\approx2.72$. Equal bounds at a point do not make the derivative zero.
From rectangles to an integral
- A Riemann sum 黎曼和 is a weighted sum of sampled function values, using subinterval widths as weights. On $[a,b]$, a regular right-endpoint sum uses $\Delta x=(b-a)/n$ and $x_k=a+k\Delta x$. The integrand supplies the rectangle height, and $\Delta x$ supplies its width.
- For a Riemann-integrable function, tagged sums approach the integral as the largest subinterval width, the mesh, tends to zero. More subintervals alone are insufficient if a large interval remains unsplit. Every continuous function on a closed bounded interval is Riemann integrable; some discontinuous functions are too.
Evaluate $\lim_{n \to \infty} \frac{1}{n}\sum_{k=1}^{n} \left(\frac{k}{n}\right)^2$.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
It is $\int_0^1 t^2\,dt = 1/3$.
On $[0,1]$, a partition repeatedly subdivides only $[0,1/2]$, leaving $[1/2,1]$ intact. Explain why growing numbers of rectangles do not by themselves justify convergence to the integral.
The mesh stays at least $1/2$, since the right-hand interval never shrinks. Sampling on that interval can leave a persistent approximation error. The usual tagged-sum guarantee requires the largest width to tend to zero for a Riemann-integrable function; the count of intervals alone is insufficient.
Read the interval and the width
- In $S_n=\sum_{k=1}^n\dfrac{n}{n^2+k^2}$, divide numerator and denominator by $n^2$ to get $\dfrac1n\sum_{k=1}^n\dfrac{1}{1+(k/n)^2}$. This uses width $1/n$ and right endpoints $k/n$ on $[0,1]$, so the limit is $\int_0^1\dfrac{dt}{1+t^2}=\pi/4$.
- The width is not always $1/n$. On $[2,4]$, it is $2/n$, and the right endpoints are $2+2k/n$. Thus $\dfrac2n\sum_{k=1}^n(2+2k/n)$ tends to $\int_2^4t\,dt=6$. Read the interval, sample points and width together.
Find $\lim_{n\to\infty}\dfrac2n\sum_{k=1}^n(2+2k/n)$.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Use width $2/n$ and right endpoints $2+2k/n$ on $[2,4]$. The limit is $\int_2^4t\,dt=6$.
A series tests a fixed sequence
- A series $\sum_{k=1}^\infty a_k$ is defined by the limit of partial sums from one fixed sequence. If it converges, then $a_k\to0$. The converse fails: $\sum 1/k$ diverges even though its terms tend to zero. Grouping successive powers-of-two blocks shows each block contributes at least $1/2$.
- By contrast, $\sum 1/k^2$ converges by the $p$-series test, and the geometric series $\sum_{k=0}^\infty r^k$ converges for $|r|<1$. State which test applies; a collection of sums whose summands change with $n$ is not automatically the partial sums of a fixed series.
What does this information establish?
Distinguish Riemann sums, fixed series and insufficient convergence evidence.
Although $1/k \to 0$, the series $\sum_{k=1}^{\infty} 1/k$ diverges.
Difficulty 1/5 · 100 coins + a small bonus on first completion today
True. Terms tending to 0 is necessary but not sufficient; the harmonic series diverges (compare with $\int_1^\infty dx/x$ or group terms).
Both $1/k$ and $1/k^2$ tend to zero. Explain why this does not prove their series have the same convergence behaviour.
Convergence of a series requires convergence of its partial sums. Terms tending to zero are only a necessary condition. The harmonic series diverges, whereas the p-series with p=2 converges. A valid further argument is required after the term limit is zero.
Choose a method, then justify it
- Check values and limits before claiming continuity; check finite difference quotients before claiming differentiability. Use the hypotheses of the mean value and extreme value theorems, and compare all extremum candidates.
- For accumulation, track both bound derivatives and their signs. For a Riemann sum, track sample points, widths and mesh; for a series, test partial-sum convergence rather than only whether terms tend to zero.
Let $A(x)=\int_0^x|t|\,dt$. Does the corner of $|x|$ at zero prevent $A\prime(0)$ from existing? Explain, and state whether $A\prime\prime(0)$ exists.
The integrand is continuous at zero, so the fundamental theorem gives $A\prime(x)=|x|$, including $A\prime(0)=0$. Differentiability of the integrand is not required for this first derivative. However, $A\prime\prime(0)$ would be the derivative of $|x|$ at zero, which does not exist.