This is the corrected exercise set for the partial fractions chapter of MATH 141, Calculus 2, the integral calculus course taken in first year at McGill University. It follows section 7.4 of Stewart. Every integral is built to be done by hand, as on the exam: the denominators factor over small integers, the constants come out as simple fractions, and every definite integral ends on an exact value such as ln2+85π. The solutions name each choice, the division, the factorization, the form, the method used for the constants, because that is where the marks are.
The thread running through the whole set: the SHAPE of the answer is decided before a single constant is computed, and in a fixed order. First the degree check, with a long division whenever the numerator is not of smaller degree. Then the denominator factored completely, each quadratic tested by its discriminant. Then one term per power of each factor, with a linear numerator over each irreducible quadratic. Each kind of term then integrates to one kind of function: a logarithm for a simple linear factor, a power for a repeated one, a logarithm plus an arctangent for an irreducible quadratic.
The traps named explicitly in the solutions: decomposing an improper fraction and losing the quotient that the cover-up cannot see, forgetting the factor p1 in ∫px+qdx, dropping the absolute values on an interval where the factor is negative, leaving out the lower power of a repeated factor, integrating (x−a)21 into a logarithm, taking a splitting quadratic for an irreducible one, writing a logarithm over a quadratic whose derivative is not on top, and forgetting dx=udu after u=ex.
Exercise 1: The degree check comes first: long division before anything else
Partial fractions only apply to a PROPER rational function, one whose numerator has a strictly smaller degree than its denominator. When degN≥degD, the first line of the answer is a long division, which writes DN=Q+DR with a polynomial quotient Q and a remainder R of degree less than degD. Only DR is then decomposed.
The figure shows y=x+1x2+3x with the dashed line y=x+2 and the vertical line x=−1.
a) Divide x2+3x by x+1 and write x+1x2+3x as a polynomial plus a proper fraction. Explain what the dashed line of the figure is, and on which side of it the curve lies when x>−1. Then find ∫x+1x2+3xdx.
b) Find ∫x2−4x3dx. After the division, is a decomposition still needed?
c) For ∫x2+x−2x2+1dx, a classmate writes (x−1)(x+2)x2+1=x−1A+x+2B and finds A=32 and B=−35 by the cover-up method. Test his identity at x=0, explain what went wrong, and find the integral correctly.
d) Evaluate ∫01x+12x2+xdx exactly.
Show the solution
Answers
a)x+1x2+3x=x+2−x+12; the curve lies below y=x+2; integral 2x2+2x−2ln∣x+1∣+C
b)x2−4x3=x+x2−44x; 2x2+2ln∣x2−4∣+C, no decomposition needed
c)At x=0: −21=−23. x2+x−2x2+1=1+x−12/3−x+25/3; integral x+32ln∣x−1∣−35ln∣x+2∣+C
d)ln2
a) The degrees are 2 and 1, so the fraction is improper and the division is compulsory. x2+3x divided by x+1: first term x, and x(x+1)=x2+x leaves 2x; next term 2, and 2(x+1)=2x+2 leaves −2. So x2+3x=(x+1)(x+2)−2 and x+1x2+3x=x+2−x+12. The quotient x+2 is the dashed line: as x→±∞ the proper part −x+12 tends to 0, so the curve hugs y=x+2, its slant asymptote. For x>−1 the gap −x+12 is negative, so the curve lies BELOW the line, as the figure shows. Integrating term by term: ∫x+1x2+3xdx=2x2+2x−2ln∣x+1∣+C. Check by differentiating: x+2−x+12, which is the integrand.
b) Degrees 3 and 2: divide. x3=x(x2−4)+4x, so x2−4x3=x+x2−44x. The remainder is now proper, and before reaching for A and B it is worth reading its SHAPE: the numerator 4x is twice the derivative of x2−4. The substitution u=x2−4, du=2xdx gives ∫x2−44xdx=2ln∣x2−4∣+C. So ∫x2−4x3dx=2x2+2ln∣x2−4∣+C. Partial fractions give the same thing, x2−44x=x−22+x+22, whose integral 2ln∣x−2∣+2ln∣x+2∣ is 2ln∣x2−4∣: the decomposition is correct but unnecessary. The division was not optional; the decomposition was.
c) At x=0 the left side is −21=−21, while the right side is −12/3+2−5/3=−32−65=−23. The identity is false. The numerator and the denominator both have degree 2, so the fraction tends to 1 at infinity, while any sum x−1A+x+2B tends to 0: no choice of A and B can work. The cover-up did not warn him, because it reads the behaviour near each pole and is blind to the quotient: the constants it gives are right, but a whole term is missing. Correct route: x2+1=1⋅(x2+x−2)+(−x+3), so x2+x−2x2+1=1+(x−1)(x+2)−x+3. Then −x+3=A(x+2)+B(x−1); x=1 gives 2=3A, A=32; x=−2 gives 5=−3B, B=−35. Hence ∫x2+x−2x2+1dx=x+32ln∣x−1∣−35ln∣x+2∣+C. His error costs the term x, and on a definite integral it costs the whole length of the interval.
d) Degrees 2 and 1: divide. 2x2+x=(x+1)(2x−1)+1, since (x+1)(2x−1)=2x2+x−1. So x+12x2+x=2x−1+x+11, and the pole x=−1 is outside [0,1], so the integrand is continuous there. ∫01(2x−1+x+11)dx=[x2−x+ln∣x+1∣]01=(1−1+ln2)−(0−0+ln1)=ln2. Sanity check: the integrand rises from 0 at x=0 to 23 at x=1, so an area of about 0.69 is believable.
Exercise 2: Distinct linear factors: the cover-up, the inner coefficient and the absolute values
When the denominator is a product of distinct linear factors, each factor (px+q) contributes ONE term px+qA. The constants come fastest from the cover-up method: multiply the identity by one factor and evaluate at its root, where every other term vanishes.
The figure shows y=x2−11 and the region between the curve and the x-axis for −21≤x≤21.
a) Find ∫x3−x6x2−x−1dx. Justify the cover-up step once in full, then check your three constants with one quick test.
b) Find ∫2x2+5x−3dx and write the answer as a single logarithm.
c) Evaluate ∫−1/21/2x2−1dx exactly. Explain from the figure why the answer must be negative, and what happens to your antiderivative on this interval if the absolute values are dropped.
d) The integrand of c) is even. Use this to recompute the answer from ∫01/2 only.
Show the solution
Answers
a)ln∣x∣+2ln∣x−1∣+3ln∣x+1∣+K
b)71ln∣2x−1∣−71ln∣x+3∣+C=71lnx+32x−1+C
c)−ln3; without absolute values the logarithm of the negative number x+1x−1 is undefined.
d)2⋅21[lnx+1x−1]01/2=ln31=−ln3
a) Degree 2 over degree 3: proper, no division. Factor: x3−x=x(x−1)(x+1), three distinct linear factors, so the form is x(x−1)(x+1)6x2−x−1=xA+x−1B+x+1C. Multiply both sides by x(x−1)(x+1): 6x2−x−1=A(x−1)(x+1)+Bx(x+1)+Cx(x−1), an identity of polynomials, true for EVERY x. At x=0 the B and C terms vanish: −1=−A, so A=1. At x=1: 6−1−1=4=2B, so B=2. At x=−1: 6+1−1=6=2C, so C=3. Quick test: the coefficient of x2 on the right is A+B+C=6, which matches the left. Therefore ∫x3−x6x2−x−1dx=ln∣x∣+2ln∣x−1∣+3ln∣x+1∣+K, where K is the constant of integration (the letter C is taken). The step that earns the method mark is the polynomial identity: the values x=0,1,−1 are forbidden in the FRACTION but allowed in the identity, which is why the cover-up is legitimate.
b) 2x2+5x−3 has discriminant 25+24=49=72, so it factors: (2x−1)(x+3). Form: (2x−1)(x+3)1=2x−1A+x+3B, so 1=A(x+3)+B(2x−1). At x=21: 1=27A, A=72. At x=−3: 1=−7B, B=−71. Now the trap of the chapter: ∫2x−12/7dx is NOT 72ln∣2x−1∣. With u=2x−1, du=2dx, it is 72⋅21ln∣2x−1∣=71ln∣2x−1∣. So ∫2x2+5x−3dx=71ln∣2x−1∣−71ln∣x+3∣+C=71lnx+32x−1+C. Differentiating the answer confirms it: 71(2x−12−x+31)=71⋅(2x−1)(x+3)2x+6−2x+1=(2x−1)(x+3)1.
c) x2−11=(x−1)(x+1)1=x−11/2−x+11/2 (cover-up at x=1 gives 21, at x=−1 gives −21). An antiderivative is F(x)=21ln∣x−1∣−21ln∣x+1∣=21lnx+1x−1. The poles ±1 lie outside [−21,21], so the integrand is continuous there and the Fundamental Theorem applies. F(21)=21ln3/2−1/2=21ln31 and F(−21)=21ln1/2−3/2=21ln3. So the integral is 21ln31−21ln3=−ln3. The figure agrees: on this interval x2−1<0, the curve is below the axis (it passes through (0,−1)), and a region below the axis contributes a NEGATIVE signed area; −ln3≈−1.10 against a width of 1 and depths between 1 and 34 is consistent. Without absolute values, x+1x−1 is negative on the whole interval and ln of it is undefined: the absolute values are not decoration, they are what makes the antiderivative exist on this interval.
d) (−x)2−11=x2−11: the integrand is even, so ∫−1/21/2=2∫01/2. With the same F: 2[F(21)−F(0)]=2[21ln31−21ln1]=ln31=−ln3. Same value, one evaluation fewer, and a useful cross-check of c).
Exercise 3: Repeated linear factors: one term per power, and a power in the answer
A factor (x−a)m of the denominator contributes m terms, x−aA1+(x−a)2A2+⋯+(x−a)mAm, one for EACH power up to m, whatever the numerator looks like. The terms of power 2 and higher do not integrate to logarithms: ∫(x−a)2dx=−x−a1+K.
Throughout, K denotes the constant of integration, since the letters A, B, C are taken by the decomposition.
a) A student writes x2(x+2)3x2−4=xA+x+2C. Prove that no constants A and C can make this true.
b) Write the correct form, find its constants, and find ∫x3+2x23x2−4dx.
c) Evaluate ∫12x3+2x23x2−4dx exactly, as a logarithm minus an integer.
d) Find ∫x(x−1)2x2+1dx. One constant turns out to be 0: was the term useless in the form? A classmate integrates the last term as 2ln∣(x−1)2∣; show by differentiating that this is wrong.
Show the solution
Answers
a)The identity would force 3x2−4=x[A(x+2)+Cx], false at x=0 (−4=0).
b)x1−x22+x+22; integral ln∣x∣+x2+2ln∣x+2∣+K
c)5ln2−2ln3−1=ln932−1
d)x1+x−10+(x−1)22; integral ln∣x∣−x−12+K. The form must keep the term; 2ln∣(x−1)2∣ differentiates to x−14.
a) Put the right side over the common denominator: xA+x+2C=x(x+2)A(x+2)+Cx=x2(x+2)x[A(x+2)+Cx]. The identity would therefore require 3x2−4=x[A(x+2)+Cx] for every x. At x=0 the right side is 0 and the left side is −4: impossible, whatever A and C are. The deeper reason: every term of the proposed form has at most ONE factor x in its denominator, so its sum cannot blow up like x21 near 0, while x2(x+2)3x2−4 behaves like 2x2−4=−x22 there. The factor x2 is a REPEATED linear factor, and it opens two boxes.
b) Form: x2(x+2)3x2−4=xA+x2B+x+2C, three unknowns for a denominator of degree 3. Clearing denominators: 3x2−4=Ax(x+2)+B(x+2)+Cx2. At x=0: −4=2B, so B=−2. At x=−2: 12−4=8=4C, so C=2. There is no third root to plug in, so compare the coefficients of x2: 3=A+C, so A=1. Check at x=1: left −1; right A⋅3+B⋅3+C=3−6+2=−1. Now integrate each term by its own rule: ∫xdx=ln∣x∣, ∫−x22dx=x2, ∫x+22dx=2ln∣x+2∣. So ∫x3+2x23x2−4dx=ln∣x∣+x2+2ln∣x+2∣+K.
c) The poles 0 and −2 are outside [1,2], so with F(x)=lnx+x2+2ln(x+2) on this interval: F(2)=ln2+1+2ln4=5ln2+1 and F(1)=0+2+2ln3. Hence ∫12x3+2x23x2−4dx=5ln2+1−2−2ln3=5ln2−2ln3−1=ln932−1. Order of magnitude with ln2≈0.69 and ln3≈1.10: 3.47−2.20−1≈0.27. The integrand goes from −31 at x=1 to 21 at x=2, so a small positive net area is what we should expect.
d) Proper (degree 2 over 3). Form: x(x−1)2x2+1=xA+x−1B+(x−1)2C, so x2+1=A(x−1)2+Bx(x−1)+Cx. At x=0: 1=A. At x=1: 2=C. Coefficient of x2: 1=A+B, so B=0. The box x−1B comes out empty for THIS numerator, but the form had to contain it: it is the numerator that decides the values, the denominator that decides the form, and a form without the B term would have been marked wrong even though B=0. So ∫x(x−1)2x2+1dx=∫(x1+(x−1)22)dx=ln∣x∣−x−12+K. The classmate's 2ln∣(x−1)2∣=4ln∣x−1∣ has derivative x−14, not (x−1)22: a squared factor integrates by the POWER rule, ∫(x−1)−2dx=−(x−1)−1.
Exercise 4: Irreducible quadratics: complete the square, split the numerator, log plus arctangent
A quadratic factor with NEGATIVE discriminant does not split over the reals. It contributes a term with a LINEAR numerator, ax2+bx+cBx+C, and that term integrates in two pieces: the part of the numerator proportional to the derivative of the quadratic gives a logarithm, the rest gives an arctangent through ∫u2+a2du=a1arctanau+K.
The figure shows the region under y=x2+2x+52x+7 for −1≤x≤1.
a) Find ∫x3+9x3x2+2x+9dx.
b) Find ∫x2+2x+52x+7dx by completing the square and splitting the numerator. Why is there no absolute value inside the logarithm?
c) Evaluate the shaded area ∫−11x2+2x+52x+7dx exactly, and check it against the figure.
d) Find ∫x2+2x−32x+7dx. Same numerator, a denominator that looks almost the same: what decides the shape of the answer?
c)ln2+85π, about 2.65, between the bounds 2.25 and 2.85 read on the figure
d)−41ln∣x+3∣+49ln∣x−1∣+K: the discriminant decides.
a) Proper. x3+9x=x(x2+9), and x2+9 has discriminant −36<0: irreducible. Form: x(x2+9)3x2+2x+9=xA+x2+9Bx+C, three unknowns for degree 3. Clearing: 3x2+2x+9=A(x2+9)+(Bx+C)x. At x=0: 9=9A, A=1. Coefficient of x2: 3=A+B, B=2. Coefficient of x: 2=C. So the integrand is x1+x2+92x+x2+92. The middle term is uu′ with u=x2+9, and the last one uses a=3: ∫x2+92dx=32arctan3x, not 2arctan3x and not 92arctanx. Result: ln∣x∣+ln(x2+9)+32arctan3x+K.
b) The discriminant of x2+2x+5 is 4−20=−16<0, so no partial fractions are possible: the quadratic IS the term. Complete the square: x2+2x+5=(x+1)2+4. Split the numerator around the derivative 2x+2 of the denominator: 2x+7=(2x+2)+5. Then ∫x2+2x+52x+2dx=ln(x2+2x+5), and with u=x+1, ∫(x+1)2+45dx=25arctan2x+1. So ∫x2+2x+52x+7dx=ln(x2+2x+5)+25arctan2x+1+K. No absolute value is needed because (x+1)2+4≥4>0 for every x: writing one is not wrong, but explaining why it is unnecessary shows the square was understood.
c) With F(x)=ln(x2+2x+5)+25arctan2x+1: F(1)=ln8+25arctan1=ln8+85π and F(−1)=ln4+25arctan0=ln4. So the area is ln8−ln4+85π=ln2+85π. The bounds were chosen so that x+1 runs from 0 to 2 and the arctangent lands on arctan1=4π. Check against the figure with the comparison property: on [−1,1] the curve stays between its value 89 at x=1 and a peak a little under 1.43, so the area lies between 2×1.125=2.25 and 2×1.43≈2.86. With ln2≈0.69 and π≈3.14, ln2+85π≈0.69+1.96=2.65: inside the bracket.
d) Here the discriminant is 4+12=16>0: x2+2x−3=(x+3)(x−1) splits, and there is NO arctangent. Form x+3A+x−1B with 2x+7=A(x−1)+B(x+3). At x=−3: 1=−4A, A=−41. At x=1: 9=4B, B=49. So ∫x2+2x−32x+7dx=−41ln∣x+3∣+49ln∣x−1∣+K. Two denominators that differ only in their constant term give answers of a different NATURE: the discriminant, computed before anything else, decides between logarithms and an arctangent. Completing the square on x2+2x−3=(x+1)2−4 would lead to a DIFFERENCE of squares, which is the signal that the quadratic factors.
Exercise 5: Rationalizing substitutions: u equals root x, u equals e to the x
Some integrands are not rational functions of x but become rational after a substitution. With u=x one writes x=u2 and dx=2udu; with u=ex one writes du=exdx, that is dx=udu. The whole of dx must be converted, not only the integrand, and after the substitution the chapter starts again from the degree check.
Throughout, K denotes the constant of integration.
a) Evaluate ∫14x+3x+2dx exactly, changing the bounds with the variable.
b) Find ∫x−4xdx for x>0, x=4.
c) Find ∫1+exdx. A classmate sets u=ex and writes ∫1+udu=ln(1+ex)+K. Find his error, and confirm your answer by a second route: multiply the numerator and the denominator by e−x.
d) Evaluate ∫0ln2e2x+3ex+2exdx exactly.
Show the solution
Answers
a)10ln2−6ln3=ln7291024
b)2x+2lnx+2x−2+K
c)x−ln(1+ex)+K, equal to −ln(1+e−x)+K; the classmate forgot dx=udu.
d)ln89
a) Let u=x, so x=u2, dx=2udu, and the bounds become u=1 and u=2. The integral becomes ∫12u2+3u+22udu, a proper rational function of u. Factor: u2+3u+2=(u+1)(u+2). Form: (u+1)(u+2)2u=u+1A+u+2B, 2u=A(u+2)+B(u+1). At u=−1: −2=A. At u=−2: −4=−B, B=4. So the integral is [−2ln(u+1)+4ln(u+2)]12=(−2ln3+4ln4)−(−2ln2+4ln3)=10ln2−6ln3=ln36210=ln7291024. It is positive since 1024>729, and small: the integrand lies between 121 (at x=4) and 61 (at x=1) on an interval of length 3, so the value is between 0.25 and 0.5, and ln7291024≈10(0.69)−6(1.10)=0.3. The trap is the factor 2u from dx: without it the integrand in u would be (u+1)(u+2)1 and every constant would be wrong.
b) u=x, x=u2, dx=2udu: ∫u2−4u⋅2udu=∫u2−42u2du. The substitution produced an IMPROPER fraction (degree 2 over degree 2), so the degree check comes back first: 2u2=2(u2−4)+8, hence u2−42u2=2+(u−2)(u+2)8. Cover-up: at u=2, 48=2; at u=−2, −48=−2. So the integral is ∫(2+u−22−u+22)du=2u+2ln∣u−2∣−2ln∣u+2∣+K. Back to x, and since x+2>0: ∫x−4xdx=2x+2lnx+2x−2+K. The absolute value on x−2 stays: it is negative for 0<x<4.
c) With u=ex, du=exdx=udx, so dx=udu and ∫1+exdx=∫u(1+u)du. The classmate replaced dx by du, dropping the factor u1; differentiating his answer shows it: dxdln(1+ex)=1+exex, not 1+ex1. Correctly: u(1+u)1=u1−1+u1 (cover-up at u=0 and u=−1), so the integral is lnu−ln(1+u)=x−ln(1+ex)+K, with no absolute values since u=ex>0. Second route: 1+ex1=e−x+1e−x, whose numerator is minus the derivative of the denominator, so ∫1+exdx=−ln(1+e−x)+K. The two answers agree: −ln(1+e−x)=−lnexex+1=x−ln(1+ex). Two correct antiderivatives may look different; they can only differ by a constant, here by 0.
d) u=ex, du=exdx: the numerator exdx is exactly du this time, and e2x=u2. The bounds become e0=1 and eln2=2. So the integral is ∫12u2+3u+2du=∫12(u+11−u+21)du=[lnu+2u+1]12=ln43−ln32=ln89. Compare with c): the same substitution, but here exdx was already present, so no u1 appears. Reading WHICH factor du absorbs is the whole gesture.
Part B: problems and reasoning (/50)
Exercise 6: Write the form before any constant: five denominators, five traps
On a MATH 141 exam, the question write the form of the partial fraction decomposition, do not solve for the constants is common and quick, provided nothing is taken on trust. The denominator must be factored COMPLETELY over the reals, each quadratic tested by its discriminant, and the degree check done first.
The five functions: f1(x)=x2(x2+4)x3+1, f2(x)=x3+85, f3(x)=(x2−6x+9)(x2+x+1)x, f4(x)=(x2+1)22x3+1 and f5(x)=x4−1x4. Throughout, K denotes the constant of integration.
a) Write the form of the decomposition of f1 and of f2.
b) Write the form of the decomposition of f3 and of f4.
c) Explain why the form of f5 cannot be written directly, then write it.
d) For each of the five, compare the number of unknown constants with the degree of the denominator. What do you notice, and why must it be so?
d)Always equal: 4, 3, 4, 4, 4. The identity of numerators is a square linear system.
e)x+41lnx+1x−1−21arctanx+K
a) f1: degree 3 over degree 4, proper. The factor x2 is the linear factor x REPEATED twice, so it opens two terms, and x2+4 (discriminant −16) is irreducible, so it takes a linear numerator: f1=xA+x2B+x2+4Cx+D. f2: x3+8 is NOT irreducible just because it has no obvious factor; x=−2 is a root, and the sum of cubes gives x3+8=(x+2)(x2−2x+4). The quadratic has discriminant 4−16=−12<0, so it stays: f2=x+2A+x2−2x+4Bx+C. Writing x3+8Ax2+Bx+C is not a decomposition at all, it is the original fraction with letters.
b) f3: the first quadratic has discriminant 36−36=0: it is the perfect square (x−3)2, a REPEATED linear factor, not an irreducible quadratic. The second, x2+x+1, has discriminant 1−4=−3<0. So f3=x−3A+(x−3)2B+x2+x+1Cx+D. Treating x2−6x+9 as irreducible and writing x2−6x+9Cx+D is the trap: the answer would then contain an arctangent of a quadratic that has a real root, which cannot be right. f4: degree 3 over degree 4, proper; x2+1 is irreducible and REPEATED, so it opens two terms, each with a linear numerator: f4=x2+1Ax+B+(x2+1)2Cx+D. Here in fact 2x3+1=2x(x2+1)+(1−2x), so A=2, B=0, C=−2, D=1; integrating the term (x2+1)21 would then call for the trigonometric substitution of the previous chapter, which is why exams usually stop at the form.
c) f5 has degree 4 over degree 4: it is improper, and a sum of proper fractions tends to 0 at infinity while f5 tends to 1. Divide first: x4=1⋅(x4−1)+1, so f5=1+x4−11. Then factor COMPLETELY: x4−1=(x2−1)(x2+1)=(x−1)(x+1)(x2+1). Stopping at (x2−1)(x2+1) and writing x2−1Ax+B would leave a quadratic that still splits. Form: f5=1+x−1A+x+1B+x2+1Cx+D.
d) Unknowns against degree of the denominator: f1, 4 and 4; f2, 3 and 3; f3, 4 and 4; f4, 4 and 4; f5, 4 and 4 (after the division). They always agree. After clearing denominators, the proper numerator of degree less than n=degD has n coefficients, and the combination on the right is also a polynomial of degree less than n whose n coefficients depend linearly on the n unknowns: a square system, which has exactly one solution. This is a free check on any form: if the count is off, a term is missing (usually the lower power of a repeated factor) or one is invented (usually a constant numerator over a quadratic that should have been linear).
e) From c), x4−11=x−1A+x+1B+x2+1Cx+D, so 1=A(x+1)(x2+1)+B(x−1)(x2+1)+(Cx+D)(x2−1). At x=1: 1=4A, A=41. At x=−1: 1=−4B, B=−41. At x=0: 1=A−B−D=21−D, so D=−21. Coefficient of x3: 0=A+B+C, so C=0. Check at x=2: left 151, right 41−121−101=6015−5−6=151. Integrating: ∫x4−1x4dx=x+41ln∣x−1∣−41ln∣x+1∣−21arctanx+K=x+41lnx+1x−1−21arctanx+K.
Exercise 7: The denominator arrives unfactored: grouping, the rational root test, a hidden square
On an exam the denominator is rarely handed over factored, and until it is, nobody knows how many terms the answer has. Two tools do almost all the work without a calculator: grouping, and the rational root test (a rational root of a polynomial with integer coefficients is a divisor of the constant term over a divisor of the leading coefficient), followed by a division.
Throughout, K denotes the constant of integration.
a) Factor x3+x2+x+1 by grouping, then find ∫x3+x2+x+1x2+3dx.
b) Evaluate ∫01x3+x2+x+1x2+3dx exactly.
c) Factor x3−3x+2 with the rational root test, then find ∫x3−3x+29dx.
d) Check the answer of c) by differentiating it, and explain what a student who stopped at the first root found would have written as the form.
Show the solution
Answers
a)(x+1)(x2+1); 2ln∣x+1∣−21ln(x2+1)+arctanx+K
b)23ln2+4π
c)(x−1)2(x+2); −ln∣x−1∣−x−13+ln∣x+2∣+K
d)The derivative is (x−1)2(x+2)9; stopping early gives a form without the (x−1)2B term, which cannot work.
a) Grouping: x3+x2+x+1=x2(x+1)+(x+1)=(x+1)(x2+1), and x2+1 is irreducible. Degree 2 over 3, proper. Form: (x+1)(x2+1)x2+3=x+1A+x2+1Bx+C, so x2+3=A(x2+1)+(Bx+C)(x+1). At x=−1: 4=2A, A=2. Then (Bx+C)(x+1)=x2+3−2x2−2=1−x2=(1−x)(1+x), so Bx+C=1−x: B=−1, C=1. The quadratic term splits as x2+1−x+x2+11, integrating to −21ln(x2+1)+arctanx. So ∫x3+x2+x+1x2+3dx=2ln∣x+1∣−21ln(x2+1)+arctanx+K. Note the factor 21: the numerator −x is −21 of the derivative 2x.
b) The only real pole is x=−1, outside [0,1]. With F(x)=2ln(x+1)−21ln(x2+1)+arctanx: F(1)=2ln2−21ln2+4π and F(0)=0−0+0=0. The integral is 23ln2+4π. Order of magnitude: 1.5(0.69)+0.785≈1.82, while the integrand goes from 3 at x=0 to 1 at x=1 and decreases in between: an area a little under 2 on an interval of length 1 is consistent.
c) Candidates for a rational root of x3−3x+2: ±1, ±2. 1−3+2=0, so x=1 is a root, and dividing by x−1 gives x2+x−2. That quadratic factors AGAIN, as (x−1)(x+2): the root 1 is DOUBLE, and x3−3x+2=(x−1)2(x+2). Form: (x−1)2(x+2)9=x−1A+(x−1)2B+x+2C, so 9=A(x−1)(x+2)+B(x+2)+C(x−1)2. At x=1: 9=3B, B=3. At x=−2: 9=9C, C=1. Coefficient of x2: 0=A+C, so A=−1. Integrating, with the power rule on the squared term: ∫x3−3x+29dx=−ln∣x−1∣−x−13+ln∣x+2∣+K=lnx−1x+2−x−13+K.
d) dxd[−ln∣x−1∣−x−13+ln∣x+2∣]=−x−11+(x−1)23+x+21. Over the common denominator (x−1)2(x+2), the numerator is −(x−1)(x+2)+3(x+2)+(x−1)2=−x2−x+2+3x+6+x2−2x+1=9. Confirmed. A student who found the root 1, wrote (x−1)(x2+x−2) and did not factor the quadratic would have posed x−1A+x2+x−2Bx+C, treating a splitting quadratic as irreducible; one who split it without noticing the repeated root would have written x−1A+x−1B+x+2C, two copies of the same term, and the system for A+B and C has no solution. The factorization is only finished when every quadratic has been tested and every repeated root counted.
Exercise 8: Five statements to correct
Each statement below was written on a MATH 141 paper, and each is false. Say what is wrong, give the correct statement or the correct antiderivative, and settle it with a computation: differentiating the proposed answer is usually the fastest proof.
a) ∫x2−2x−3dx needs an arctangent, since the denominator is a quadratic.
b) ∫2x+13dx=3ln∣2x+1∣+C.
c) A factor (x2+4)2 in the denominator contributes the single term (x2+4)2Ax+B.
d) ∫x2+4x+8dx=ln(x2+4x+8)+C.
e) For ∫x2−12xdx, one student finds ln∣x2−1∣+C by substitution and another ln∣x−1∣+ln∣x+1∣+C by partial fractions. The answers look different, so one of them is wrong.
Show the solution
Answers
a)False: discriminant 16>0; 41lnx+1x−3+C
b)False: 23ln∣2x+1∣+C
c)False: two terms, x2+4Ax+B+(x2+4)2Cx+D; e.g. (x2+4)2x2=x2+41−(x2+4)24
d)False: 21arctan2x+2+C
e)False: both are right, ln∣x−1∣+ln∣x+1∣=ln∣x2−1∣.
a) FALSE. A quadratic gives an arctangent only when it is IRREDUCIBLE, and the discriminant decides: here 4+12=16>0, so x2−2x−3=(x−3)(x+1). Cover-up: (x−3)(x+1)1=x−31/4−x+11/4. So ∫x2−2x−3dx=41lnx+1x−3+C. Completing the square would have warned too: (x−1)2−4 is a DIFFERENCE of squares, not a sum. Correct statement: compute the discriminant first; positive, it factors into logarithms; zero, it is a perfect square and gives a power; negative, it gives an arctangent.
b) FALSE. Differentiate the proposed answer: dxd3ln∣2x+1∣=2x+13⋅2=2x+16, twice the integrand. With u=2x+1, du=2dx: ∫2x+13dx=23ln∣2x+1∣+C. Correct statement: ∫px+qdx=p1ln∣px+q∣+C. This slip costs a factor on every term with a non-monic linear factor, which after a cover-up on (2x−1) or (3x+2) is most of them.
c) FALSE. A repeated factor, linear or quadratic, opens one term per POWER: (x2+4)2 contributes x2+4Ax+B+(x2+4)2Cx+D, four unknowns for degree 4. Counterexample to the single term: (x2+4)2x2 would need x2=Ax+B for every x, impossible since the left side has degree 2. The correct decomposition is (x2+4)2x2=(x2+4)2(x2+4)−4=x2+41−(x2+4)24, which uses both boxes. Counting unknowns against the degree of the denominator catches this at once: two unknowns for degree 4 is too few.
d) FALSE. The logarithm needs the derivative of the denominator on top: dxdln(x2+4x+8)=x2+4x+82x+4, not x2+4x+81. The discriminant is 16−32=−16<0, so complete the square: x2+4x+8=(x+2)2+4, and with u=x+2, a=2: ∫(x+2)2+4dx=21arctan2x+2+C. Correct statement: over an irreducible quadratic, a CONSTANT numerator gives an arctangent; only a numerator proportional to the derivative of the quadratic gives a logarithm.
e) FALSE. For x=±1, ln∣x−1∣+ln∣x+1∣=ln(∣x−1∣∣x+1∣)=ln∣x2−1∣: the two answers are the SAME function. More generally, two correct antiderivatives on an interval can only differ by a constant, so different-looking answers must be compared by simplifying or by differentiating both, never judged on sight. Here both derivatives are x−11+x+11=x2−12x. The substitution was the faster route: the numerator is exactly the derivative of the denominator, which is worth noticing before any decomposition.
Exercise 9: A second-order reaction: how long until the product reaches a given concentration
In the reaction A+B→C, with initial concentrations a=3 mol/L of A and b=1 mol/L of B, the chemistry of the rate law shows that the time needed for the concentration of the product C to reach x mol/L is t(x)=k1∫0x(3−s)(1−s)ds, for 0≤x<1, where k is the rate constant in L/(mol min). This formula is taken as given: the task is the integral.
The figure shows the integrand for 0≤s<1 and the region under it from 0 to 21.
a) Decompose the integrand and show that t(x)=2k1ln3(1−x)3−x. Watch the signs of ∫1−sds.
b) Take k=21 L/(mol min). Give the exact times, in minutes, for the product to reach 21 mol/L and 43 mol/L. Show, without a calculator, that the second takes more than twice as long as the first.
c) Still with k=21, solve for x as a function of t. Check your formula at t=ln3 and show that x(t)<1 for every t≥0.
d) A second run starts with EQUAL concentrations, a=b=2 mol/L, so that t(x)=k1∫0x(2−s)2ds for 0≤x<2. With k=21, find t(x) and the time to reach x=1 mol/L. Why is there no logarithm this time?
e) The shaded area equals kt(21). Using only the values of the integrand at s=0 and s=21, bracket that area, and check that your exact value from a) fits.
b)t(21)=ln35 min, t(43)=ln3 min, and ln3>2ln35 since 27>25
c)x(t)=3et−13(et−1); x(ln3)=43; x<1 since 3et−3<3et−1
d)t(x)=2−xx; t(1)=1 min; a repeated factor integrates to a power
e)Between 61 and 52; exact 21ln35<21ln2≈0.35
a) The integrand is proper with two distinct linear factors: (3−s)(1−s)1=1−sA+3−sB, so 1=A(3−s)+B(1−s). At s=1: 1=2A, A=21. At s=3: 1=−2B, B=−21. Check at s=0: 21−61=31, the value of the integrand. Now the sign trap: with u=1−s, du=−ds, so ∫1−sds=−ln∣1−s∣, and likewise ∫3−sds=−ln∣3−s∣. For 0≤s≤x<1 both factors are positive. Hence ∫0x(3−s)(1−s)ds=[−21ln(1−s)+21ln(3−s)]0x=−21ln(1−x)+21ln(3−x)−21ln3=21ln3(1−x)3−x, and dividing by k gives t(x)=2k1ln3(1−x)3−x. Units: the integral is in L/mol, divided by L/(mol min), which leaves minutes. Check: t(0)=2k1ln1=0.
b) With k=21, 2k1=1 and t(x)=ln3(1−x)3−x. At x=21: 3/25/2=35, so t=ln35 min. At x=43: 3/49/4=3, so t=ln3 min. Comparing: 2ln35=ln925 and 925<3 because 25<27; since ln is increasing, ln3>2ln35. Reaching 43 mol/L takes more than twice as long as reaching 21: the reaction slows down as B runs out, which the figure shows as a curve that climbs faster and faster.
c) Set t=ln3(1−x)3−x, so et=3(1−x)3−x and 3et(1−x)=3−x. Collect the x terms: 3et−3=x(3et−1), so x(t)=3et−13(et−1). At t=ln3, et=3 and x=9−13⋅2=43, matching b). For t≥0 the denominator 3et−1≥2 is positive and the numerator 3et−3 is smaller than it by 2, so x(t)<1: the product never reaches 1 mol/L, the amount of B available.
d) Now the denominator is a REPEATED linear factor, and a single term (2−s)21 is already its own decomposition. With u=2−s, ∫(2−s)2ds=2−s1 (the two minus signs cancel: dsd2−s1=(2−s)21). So t(x)=2[2−s1]0x=2−x2−1=2−xx minutes, and t(1)=11=1 min. There is no logarithm because a squared factor integrates by the POWER rule; writing −ln(2−s)2 here, by habit from a), would give a wrong time: its derivative is 2−s2, not (2−s)21. The kind of factor decides the kind of function, in chemistry as in the exercises before.
e) On [0,21] the integrand increases (both factors of the denominator decrease), from 31 at s=0 to (5/2)(1/2)1=54 at s=21. By the comparison property, the area lies between 21⋅31=61 and 21⋅54=52. The exact value is kt(21)=21ln35, positive, and since 35<2 it is less than 21ln2≈0.35: inside the bracket. A value outside it, for instance a negative one from a lost sign in a), would be caught at once.
Exercise 10: A final exam question: division, factoring, decomposition, arctangent, exact value
This is the shape of a long final exam question on the chapter: every step of the method is needed once, in order, and the final value is exact. Evaluate I=∫12x3−2x2+2xx4+3x+2dx.
The figure shows the integrand, the dashed line y=x+2, and the region whose area is I. Throughout, K denotes the constant of integration.
a) Carry out the long division of x4+3x+2 by x3−2x2+2x, writing the missing powers of the numerator explicitly. Relate the quotient to the figure.
b) Factor the denominator completely over the reals, justifying your conclusion, and write the form of the decomposition of the remainder fraction.
c) Find the constants.
d) Find an antiderivative of the integrand, completing the square where needed.
e) Evaluate I exactly, and check the order of magnitude against the figure using ln2≈0.69 and π≈3.14.
Show the solution
Answers
a)Quotient x+2, remainder 2x2−x+2; the quotient is the slant asymptote.
b)x(x2−2x+2), discriminant −4; form xA+x2−2x+2Bx+C
c)A=1, B=1, C=1
d)2x2+2x+ln∣x∣+21ln(x2−2x+2)+2arctan(x−1)+K
e)I=27+23ln2+2π≈6.1
a) Degree 4 over degree 3: improper, divide. Write the numerator with its missing powers, x4+0x3+0x2+3x+2, so that each column is lined up. First term of the quotient: x, and x(x3−2x2+2x)=x4−2x3+2x2; subtracting leaves 2x3−2x2+3x+2. Next term: 2, and 2(x3−2x2+2x)=2x3−4x2+4x; subtracting leaves 2x2−x+2, of degree 2<3, so the division stops. Hence x3−2x2+2xx4+3x+2=x+2+x3−2x2+2x2x2−x+2. Forgetting the zero placeholders is the usual way to subtract −2x3 from the wrong column. The quotient x+2 is the dashed line of the figure: far from the origin the remainder fraction tends to 0 and the curve approaches its slant asymptote.
b) x3−2x2+2x=x(x2−2x+2), and x2−2x+2 has discriminant 4−8=−4<0, so it is irreducible: equivalently x2−2x+2=(x−1)2+1≥1. The factorization is complete. Form of the remainder fraction, three unknowns for degree 3: x(x2−2x+2)2x2−x+2=xA+x2−2x+2Bx+C.
c) Clearing denominators: 2x2−x+2=A(x2−2x+2)+(Bx+C)x. At x=0: 2=2A, so A=1. Then (Bx+C)x=2x2−x+2−(x2−2x+2)=x2+x=x(x+1), so B=1 and C=1. Check at x=1: left 2−1+2=3; right A⋅1+(B+C)⋅1=1+2=3.
d) The integrand is x+2+x1+x2−2x+2x+1. For the last term, split the numerator around the derivative 2x−2 of the quadratic: x+1=21(2x−2)+2. Then ∫x2−2x+221(2x−2)dx=21ln(x2−2x+2), and with the completed square and u=x−1: ∫(x−1)2+12dx=2arctan(x−1). An antiderivative is F(x)=2x2+2x+ln∣x∣+21ln(x2−2x+2)+2arctan(x−1)+K. Differentiating the last two terms gives back x2−2x+2x−1+x2−2x+22=x2−2x+2x+1, as required.
e) The only real pole is x=0, outside [1,2]. F(2)=2+4+ln2+21ln2+2arctan1=6+23ln2+2π and F(1)=21+2+0+21ln1+2arctan0=25. So I=27+23ln2+2π. The bounds were chosen so that x−1 runs from 0 to 1, where the arctangent is known exactly. Order of magnitude: 3.5+1.5(0.69)+1.57≈6.1. On the figure, the integrand equals 6 at both x=1 and x=2 and stays close to 6 in between (about 6.2 at x=1.5), so the region is nearly a rectangle of width 1 and height 6: an area slightly above 6, as found.