On a definite integral ∫abf(x)dx where the integrand is awkward but the interval is symmetric under reflection, substitute x↦a+b−x to get a second form I=∫abf(a+b−x)dx, then add: 2I=∫ab(f(x)+f(a+b−x))dx. The trigger is any integrand that gets simpler when you swap the endpoints.
Instances: (i) g(x)+g(a+b−x)g(x), the two copies add to ∫ab1dx, so I=21(b−a); (ii) x↦2π−x to swap sin↔cos, or θ↦−θ on [−2π,2π] to flip the sign of an odd term in the denominator; (iii) x↦π−x on ∫0πxf(sinx)dx to pull the linear factor out: ∫0πxf(sinx)dx=2π∫0πf(sinx)dx; (iv) reflect about the centre to expose rotational symmetry g(x)+g(2a−x)=2b, so a symmetric integral equals (width)×(centre height).
Linked questions (8)
STEP 2 2006 — Q4
By making the substitution x=π−t, show that ∫0πxf(sinx)dx=21π∫0πf(sinx)dx, where f(sinx) is a given function of sinx.
Evaluate the following integrals:
∫0π3+sin2xxsinxdx;
∫02π3+sin2xxsinxdx;
∫0π3+sin2xxsin2xdx.
Voir la correctionMasquer la correction
The key idea is to exploit the symmetry of sinx about x=2π.
Establishing the identity.
Let I=∫0πxf(sinx)dx. Apply the substitution x=π−t, so dx=−dt. When x=0, t=π; when x=π, t=0. The integral becomesI=∫π0(π−t)f(sin(π−t))(−dt)=∫0π(π−t)f(sint)dt,using the identity sin(π−t)=sint. Since t is a dummy variable, ∫0πtf(sint)dt=I, so expanding the right-hand side givesI=π∫0πf(sint)dt−I.Rearranging: 2I=π∫0πf(sint)dt, which yields∫0πxf(sinx)dx=2π∫0πf(sinx)dx.Part (i).
Here f(sinx)=3+sin2xsinx, so the identity gives∫0π3+sin2xxsinxdx=2π∫0π3+sin2xsinxdx.Substitute c=cosx, dc=−sinxdx. The limits x=0,π become c=1,−1, and sin2x=1−c2, so the right-hand integral becomes2π∫−113+(1−c2)dc=2π∫−114−c2dc.Using partial fractions, 4−c21=41(2−c1+2+c1), so2π⋅41[−ln∣2−c∣+ln∣2+c∣]−11=8π[ln(2−c2+c)]−11=8π(ln3−ln31)=8π⋅2ln3=4πln3.Part (ii).
Split the integral at π:∫02π3+sin2xxsinxdx=∫0π3+sin2xxsinxdx+∫π2π3+sin2xxsinxdx.For the second piece, substitute y=x−π, so x=y+π and sinx=sin(y+π)=−siny:∫0π3+sin2y(y+π)(−siny)dy=−∫0π3+sin2yysinydy−π∫0π3+sin2ysinydy.The first integral is the result from part (i), 4πln3. For the second, the same c=cosy substitution gives π⋅4ln3 (using the same working as part (i) without the 2π factor). Therefore the total is4πln3−4πln3−2πln3=−2πln3.Part (iii).
Write ∣sin2x∣=2∣sinxcosx∣. On [0,π] one has sinx≥0, so3+sin2xx∣sin2x∣=3+sin2x2x∣cosx∣sinx.Now ∣cosx∣=1−sin2x, which is a function of sinx, so the identity applies with f(sinx)=3+sin2x2∣cosx∣sinx:∫0π3+sin2xx∣sin2x∣dx=π∫0π3+sin2x∣cosx∣sinxdx.Split at x=2π where cosx changes sign, and use c=cosx on each half:π(∫0π/23+sin2xcosxsinxdx+∫π/2π3+sin2x−cosxsinxdx)=2π∫0π/23+sin2xcosxsinxdx.With u=sin2x, du=2sinxcosxdx, the integral becomes2π∫012(3+u)du=π[ln(3+u)]01=π(ln4−ln3)=πln34.
STEP 3 2006 — Q2
Let I=∫−21π21π1−sinθsin2αcos2θdθ and J=∫−21π21π1+tan2θcos22αsec2θdθ where 0<α<41π.
Show that I=∫−21π21π1+sinθsin2αcos2θdθ and hence that 2I=∫−21π21π1+tan2θcos22α2dθ.
Find J.
By considering Isin22α+Jcos22α, or otherwise, show that I=21πsec2α.
Evaluate I in the case 41π<α<21π.
Voir la correctionMasquer la correction
Part (i).
To show the first result, substitute θ↦−θ in the original integral I. Since cos2(−θ)=cos2θ and sin(−θ)=−sinθ, the integrand becomes 1+sinθsin2αcos2θ, and the limits −2π to 2π are unchanged. HenceI=∫−π/2π/21+sinθsin2αcos2θ,dθ.Adding this to the original expression for I gives2I=∫−π/2π/2cos2θ(1−sinθsin2α1+1+sinθsin2α1)dθ=∫−π/2π/21−sin2θsin22α2cos2θ,dθ.Divide numerator and denominator by cos2θ:2I=∫−π/2π/2sec2θ−tan2θsin22α2,dθ.Since sec2θ=1+tan2θ, the denominator becomes 1+tan2θ(1−sin22α)=1+tan2θcos22α, giving2I=∫−π/2π/21+tan2θcos22α2,dθ.Part (ii).
To evaluate J=∫−π/2π/21+tan2θcos22αsec2θ,dθ, set u=cos2αtanθ. Then du=cos2αsec2θ,dθ, so sec2θ,dθ=cos2αdu. Since 0<α<4π, we have cos2α>0, so the substitution is orientation-preserving and as θ ranges from −2π to 2π, u ranges from −∞ to +∞. ThereforeJ=cos2α1∫−∞∞1+u2du=cos2α1⋅π=πsec2α.Part (iii).
From Part (i), 2I=∫−π/2π/21+tan2θcos22α2,dθ, so I=J⋅⋯cos22α+1... A cleaner route is to form the combination Isin22α+Jcos22α.
Note that J=∫−π/2π/21+tan2θcos22αsec2θ,dθ and, from Part (i), 2I=∫−π/2π/21+tan2θcos22α2,dθ. Multiply the second by 21sin22α and add cos22α times the J-integral:Isin22α+Jcos22α=∫−π/2π/21+tan2θcos22αsin22α+cos22αsec2θ,dθ.The numerator equals sin22α+cos22α(1+tan2θ)=1+tan2θcos22α, which cancels the denominator exactly. HenceIsin22α+Jcos22α=∫−π/2π/2dθ=π.Substituting J=πsec2α:Isin22α=π−πsec2αcos22α=π(1−cos2α).Using the identity 1−cos2α=2sin2α and sin22α=4sin2αcos2α:I=4sin2αcos2α2πsin2α=2cos2απ=21πsec2α.Part (iv).
For 4π<α<2π, the constant cos2α<0. In the substitution for J, the factor cos2α is now negative, so u=cos2αtanθ runs from +∞ to −∞ as θ increases, reversing orientation. Thus J=−πsec2α in this range.
Returning to Isin22α+Jcos22α=π (this combinatorial identity holds regardless of the sign of cos2α, since it was established by purely algebraic cancellation in the integrand):Isin22α=π−(−πsec2α)cos22α=π(1+cos2α).Using 1+cos2α=2cos2α:I=4sin2αcos2α2πcos2α=2sin2απ=21π,cosec2α.
STEP 2 2010 — Q4
Let I=∫0af(x)+f(a−x)f(x)dx. Use a substitution to show that I=∫0af(x)+f(a−x)f(a−x)dx and hence evaluate I in terms of a.
Use this result to evaluate the integrals ∫01ln(2+x−x2)ln(x+1)dx and ∫02πsin(x+4π)sinxdx.
Evaluate ∫212x(sinx+sinx1)sinxdx.
Voir la correctionMasquer la correction
Part (i).
The key observation is that the integrand has a hidden symmetry when you swap x with a−x. To exploit it, substitute x=a−y (so dx=−dy), and note that the limits flip: when x=0, y=a, and when x=a, y=0. ThereforeI=∫0af(x)+f(a−x)f(x)dx=∫a0f(a−y)+f(y)f(a−y)⋅(−dy)=∫0af(y)+f(a−y)f(a−y)dy.Since y is just a dummy variable, this is exactlyI=∫0af(x)+f(a−x)f(a−x)dx.Now add the two expressions for I:2I=∫0af(x)+f(a−x)f(x)+f(a−x)dx=∫0a1dx=a.Hence I=2a, regardless of what f is (as long as the denominator never vanishes).
First integral. We need to match ln(2+x−x2)ln(x+1) to the form f(x)+f(a−x)f(x). Factorise the denominator's argument:2+x−x2=(1+x)(2−x).So ln(2+x−x2)=ln(1+x)+ln(2−x). Set f(x)=ln(1+x) and a=1. Then f(a−x)=ln(2−x), and the integral becomes ∫01f(x)+f(1−x)f(x)dx=21.
Second integral. For ∫0π/2sin(x+4π)sinxdx, expand the denominator:sin(x+4π)=21sinx+21cosx.So the denominator is 21(sinx+cosx). Set f(x)=sinx and a=2π; then f(a−x)=sin(2π−x)=cosx. The integral is∫0π/2sinx+cosxsinxdx=2a=4π.Part (ii).
The structure is different: the limits are 21 and 2, and the integrand involves sinx and sinx1. The hint is to mimic the same averaging trick with the substitution u=x1.
Let u=x1, so dx=−u21du. When x=21, u=2; when x=2, u=21. Write J for the integral:J=∫1/22x(sinx+sinx1)sinxdx=∫21/2u1(sinu1+sinu)sinu1⋅(−u21)du.Simplifying (noting u1⋅u2=u):J=∫1/22u(sinu+sinu1)sinu1du.Adding this to the original:2J=∫1/22x(sinx+sinx1)sinx+sinx1dx=∫1/22x1dx=[lnx]1/22=ln2−ln21=2ln2.Hence J=ln2.
STEP 2 2015 — Q6
Show that sec2(41π−21x)=1+sinx2. Hence integrate 1+sinx1 with respect to x.
By means of the substitution y=π−x, show that ∫0πxf(sinx)dx=2π∫0πf(sinx)dx, where f is any function for which these integrals exist.
Hence evaluate ∫0π1+sinxxdx.
Evaluate ∫0π(1+sinx)22x3−3πx2dx.
Voir la correctionMasquer la correction
Part (i). We use the compound-angle identity for cosine. Write 41π−21x=θ; thencos2θ=1−2sin2θ,so cos(2π−x)=1−2sin2(4π−2x). Since cos(2π−x)=sinx, this givessinx=1−2sin2(4π−2x).Divide through by cos2(4π−2x) and use sec2=1+tan2:sec2(4π−2x)=cos2(4π−2x)1.Equivalently, 1+sinx=2cos2(4π−2x), sosec2(4π−2x)=1+sinx2.✓Hence 1+sinx1=21sec2(4π−2x). Since dxdtan(4π−2x)=−21sec2(4π−2x),∫1+sinxdx=−tan(4π−2x)+C.Part (ii). Set y=π−x, so x=π−y and dx=−dy. When x=0, y=π; when x=π, y=0. Also sin(π−y)=siny. Therefore∫0πxf(sinx)dx=∫π0(π−y)f(siny)(−dy)=∫0π(π−y)f(siny)dy.Call this common value I. Adding the two expressions:2I=∫0πxf(sinx)dx+∫0π(π−x)f(sinx)dx=π∫0πf(sinx)dx,yielding I=2π∫0πf(sinx)dx. ✓
Apply this with f(sinx)=1+sinx1:∫0π1+sinxxdx=2π∫0π1+sinxdx=2π[−tan(4π−2x)]0π.At x=π: tan(−4π)=−1. At x=0: tan(4π)=1. So∫0π1+sinxxdx=2π(1−(−1)⋅(−1))=2π⋅2=π.Wait — more carefully: [−tan(4π−2x)]0π=(−tan(−4π))−(−tan(4π))=1−(−1)=2. Hence the integral equals 2π⋅2=π.∫0π1+sinxxdx=π.Part (iii). Write 2x3−3πx2=x2(2x−3π). Try the substitution y=π−x again. Under x↦π−y, we have (1+sinx)2→(1+siny)2 and dx=−dy, soJ=∫0π(1+sinx)22x3−3πx2dx=∫0π(1+siny)22(π−y)3−3π(π−y)2dy.Adding the two expressions for J:2J=∫0π(1+sinx)2[2x3−3πx2]+[2(π−x)3−3π(π−x)2]dx.Expand the numerator: 2x3−3πx2+2(π−x)3−3π(π−x)2. Direct expansion gives −π3, a constant. Hence2J=−π3∫0π(1+sinx)2dx=−π3∫0π41sec4(4π−2x)dx.Substitute u=4π−2x, so dx=−2du; limits u:4π→−4π:∫0π(1+sinx)2dx=41⋅2∫−π/4π/4sec4udu=21∫−π/4π/4(1+tan2u)sec2udu.This equals 21[tanu+3tan3u]−π/4π/4=21⋅2(1+31)=34.
Show that ∫0af(x)dx=∫0af(a−x)dx,(∗) where f is any function for which the integrals exist.
Use (∗) to evaluate ∫021πcosx+sinxsinxdx.
Evaluate ∫041πcosx+sinxsinxdx.
Evaluate ∫041πln(1+tanx)dx.
Evaluate ∫041πcosx(cosx+sinx)xdx.
Voir la correctionMasquer la correction
Preliminary result. Let u=a−x, so du=−dx. When x=0, u=a; when x=a, u=0. Thus∫0af(x)dx=∫a0f(a−u)(−du)=∫0af(a−u)du=∫0af(a−x)dx,where the last step renames the dummy variable. □
Part (i). Let I=∫0π/2cosx+sinxsinxdx. Applying (∗) with a=2π, and using sin(2π−x)=cosx, cos(2π−x)=sinx:I=∫0π/2sinx+cosxcosxdx.Adding the two expressions for I:2I=∫0π/2cosx+sinxsinx+cosxdx=∫0π/21dx=2π.Hence I=4π.
Part (ii). Let J=∫0π/4cosx+sinxsinxdx. Write cosx+sinx=2sin(x+4π), and expand:cosx+sinxsinx=21−21cot(x+4π).Integrating term by term:J=[2x−21lnsin(x+4π)]0π/4.At x=4π: 8π−21ln1=8π. At x=0: 0−21ln21=4ln2. HenceJ=8π−4ln2.Part (iii). Let K=∫0π/4ln(1+tanx)dx. Apply (∗) with a=4π. Since tan(4π−x)=1+tanx1−tanx:K=∫0π/4ln(1+1+tanx1−tanx)dx=∫0π/4ln(1+tanx2)dx.Adding the two expressions for K:2K=∫0π/4[ln(1+tanx)+ln2−ln(1+tanx)]dx=4πln2.Hence K=8πln2.
Part (iv). Let L=∫0π/4cosx(cosx+sinx)xdx. Apply (∗) with a=4π. Note thatcos(4π−x)+sin(4π−x)=2cosx,cos(4π−x)=2cosx+sinx,so the symmetrised integrand simplifies to2cosx+sinx⋅2cosx4π−x=cosx(cosx+sinx)4π−x.Adding the two expressions for L:2L=4π∫0π/4cosx(cosx+sinx)dx.To evaluate this remaining integral, substitute u=tanx (so dx=1+u2du and the limits become 0 to 1):∫0π/4cosx(cosx+sinx)dx=∫0π/41+tanxsecxdx=∫01(1+u)(1+u2)du.Partial fractions: (1+u)(1+u2)1=1+u1/2+1+u21/2−u/2. Integrating:∫01(1+u)(1+u2)du=[2ln(1+u)+2arctanu−4ln(1+u2)]01=2ln2+8π−4ln2=4ln2+8π.Therefore2L=4π(4ln2+8π)=16πln2+32π2,givingL=32πln2+64π2.
STEP 2 2018 — Q3
Let f(x)=1+tanx1 for 0⩽x<21π.
Show that f′(x)=−1+sin2x1 and hence find the range of f′(x).
Sketch the curve y=f(x).
The function g(x) is continuous for −1⩽x⩽1.
Show that the curve y=g(x) has rotational symmetry of order 2 about the point (a,b) on the curve if and only if g(x)+g(2a−x)=2b.
Given that the curve y=g(x) passes through the origin and has rotational symmetry of order 2 about the origin, write down the value of ∫−11g(x)dx.
Show that the curve y=1+tankx1, where k is a positive constant and 0<x<21π, \\[3mm] has rotational symmetry of order~2 about a certain point (which you should specify) and evaluate ∫61π31π1+tankx1dx.
Voir la correctionMasquer la correction
Part (i). We differentiate f(x)=1+tanx1 using the quotient rule. The derivative of tanx is sec2x, sof′(x)=−(1+tanx)2sec2x.To reach the target form, multiply top and bottom by cos2x:f′(x)=−(cosx+sinx)21.Expanding the denominator and applying the double-angle identity sin2x=2sinxcosx gives (cosx+sinx)2=1+sin2x, sof′(x)=−1+sin2x1.For 0⩽x<21π, the argument 2x ranges over [0,π), so sin2x∈[0,1] and hence 1+sin2x∈[1,2]. Therefore f′(x)∈[−1,−21].
For the sketch: f is strictly decreasing from f(0)=1 down to 0 as x→2π−. The derivative is most negative (steepest) where sin2x=1, i.e. at x=4π, giving the point (4π,21). Since f′′ changes sign there, this is a point of inflexion. The curve is concave up near x=0 and concave down near x=2π.
Part (ii). A 180∘ rotation about (a,b) sends (x,y) to (2a−x,,2b−y). The curve y=g(x) is invariant under this rotation if and only if, whenever (x,g(x)) lies on the curve, so does (2a−x,,2b−g(x)). That second point lies on the curve precisely when g(2a−x)=2b−g(x), i.e.g(x)+g(2a−x)=2b.This must hold for all x in the domain, establishing the equivalence.
Now suppose g passes through the origin and has rotational symmetry of order 2 about the origin, so (a,b)=(0,0). The condition becomes g(x)+g(−x)=0, meaning g is an odd function. For any odd continuous function on a symmetric interval,∫−11g(x),dx=0.Part (iii). Set h(x)=1+tankx1 on (0,2π). We test the symmetry condition about the candidate point (4π,21). We need h(x)+h(2π−x)=1.
Using tan(2π−x)=cotx=tan−1x:h!(2π−x)=1+cotkx1=tankx+1tankx.Thereforeh(x)+h!(2π−x)=1+tankx1+1+tankxtankx=1.By part (ii) with a=4π, b=21, the curve has rotational symmetry of order 2 about (4π,21).
For the integral, the interval [6π,3π] is centred on 4π and has width 6π. Since the curve has rotational symmetry about the midpoint of this interval at height 21, the area under the curve equals the area of the rectangle of width 6π and height 21:∫π/6π/31+tankx1,dx=21⋅6π=12π.
STEP 2 2020 — Q8
In this question, f(x) is a quartic polynomial where the coefficient of x4 is equal to 1, and which has four real roots, 0, a, b and c, where 0<a<b<c.
F(x) is defined by F(x)=∫0xf(t)dt.
The area enclosed by the curve y=f(x) and the x-axis between 0 and a is equal to that between b and c, and half that between a and b.
(i) Sketch the curve y=F(x), showing the x-coordinates of its turning points. Explain why F(x) must have the form F(x)=51x2(x−c)2(x−h), where 0<h<c. Find, in factorised form, an expression for F(x)+F(c−x) in terms of c, h and x.
(ii) If 0≤x≤c, explain why F(b)+F(x)≥0 and why F(b)+F(x)>0 if x=a. Hence show that c−b=a or c>2h. By considering also F(a)+F(x), show that c=a+b and that c=2h.
(iii) Find an expression for f(x) in terms of c and x only. Show that the points of inflection on y=f(x) lie on the x-axis.
Voir la correctionMasquer la correction
Part (i).
Begin by thinking about the shape of F(x)=∫0xf(t)dt. Since f is a quartic with positive leading coefficient, F is a quintic. As x increases from 0 through a, the integrand f(t) is negative (the curve dips below the axis), so F decreases to a local minimum at x=a. From a to b, f(t) is positive, so F rises to a local maximum at x=b. From b to c, f(t) is negative again, and the area condition (equal areas between 0–a and b–c) forces F(c)=F(0)=0. The key structural facts are:
- F(0)=0 and F(c)=0, each with F touching zero from above/below (turning points on the x-axis), so x=0 and x=c are double roots of F. - The leading term of f(x)=x4+⋯ gives F(x)=51x5+⋯, fixing the leading coefficient. - F has exactly one more root, somewhere in (a,b) where F crosses zero; call it x=h, with 0<h<c.
Hence F(x)=51x2(x−c)2(x−h).Now compute F(x)+F(c−x). Substituting c−x into the formula:F(c−x)=51(c−x)2[(c−x)−c]2[(c−x)−h]=51(c−x)2x2(c−x−h).Adding:F(x)+F(c−x)=51x2(x−c)2(x−h)+51x2(x−c)2(c−x−h)=51x2(x−c)2[(x−h)+(c−x−h)]=51x2(x−c)2(c−2h).Part (ii).
Denote the area between 0 and a by A>0. The area conditions give F(a)=−A, F(b)=A, F(c)=0.
On [0,c], the minimum of F is F(a)=−A, so for any x∈[0,c]:F(b)+F(x)≥F(b)+F(a)=A−A=0.Equality holds precisely when F(x)=F(a), i.e. when x=a. Hence F(b)+F(x)>0 for x=a.
Set x=b in the identity from part (i):F(b)+F(c−b)=51b2(b−c)2(c−2h).Since b=0 and b=c, the factor b2(b−c)2>0. If c−b=a, then F(b)+F(c−b)>0, forcing c−2h>0, i.e. c>2h. So either c−b=a or c>2h.
Now apply the same reasoning to F(a)+F(x). On [0,c], the maximum of F is F(b)=A, so:F(a)+F(x)≤F(a)+F(b)=−A+A=0,with equality iff x=b.
Set x=a in the identity:F(a)+F(c−a)=51a2(a−c)2(c−2h).If c−a=b, then F(a)+F(c−a)<0, so c−2h<0, i.e. c<2h.
Having both c>2h (from the first argument) and c<2h simultaneously is impossible. Therefore both conditions force equality: c−b=a (giving c=a+b) and c=2h.
Part (iii).
Substitute h=c/2 into F(x):F(x)=51x2(x−c)2(x−2c)=101x2(x−c)2(2x−c).Differentiate using the product rule on u=x2(x−c)2 and v=2x−c:f(x)=F′(x)=101[2x(x−c)2(2x−c)+2x2(x−c)(2x−c)+2x2(x−c)2].Factor out 2x(x−c):f(x)=51x(x−c)[(x−c)(2x−c)+x(2x−c)+x(x−c)].Expand the bracket:(x−c)(2x−c)+x(2x−c)+x(x−c)=2x2−3cx+c2+2x2−cx+x2−cx=5x2−5cx+c2.Hence f(x)=51x(x−c)(5x2−5cx+c2).The roots of f are 0, c, and the two roots of 5x2−5cx+c2=0, which are a and b — consistent with the setup.
Points of inflection. Differentiating f(x)=101(2x−c)(x2−cx)2... it is more efficient to expand f(x) first. From the factored form, expanding:f(x)=51x(x−c)(5x2−5cx+c2)=51(5x4−10cx3+6c2x2−c3x).So f(x)=x4−2cx3+56c2x2−5c3x.
Then:f′(x)=4x3−6cx2+512c2x−5c3,f′′(x)=12x2−12cx+512c2=512(5x2−5cx+c2).The roots of f′′(x)=0 are the roots of 5x2−5cx+c2=0, which are exactly x=a and x=b. Since f(a)=f(b)=0, the points of inflection are (a,0) and (b,0), which lie on the x-axis.
STEP 3 2022 — Q5
(i) Show that∫−aa1+ex1dx=afor all a≥0.(ii) Explain why, if g is a continuous function and∫0ag(x)dx=0for all a≥0,then g(x)=0 for all x≥0.
Let f be a continuous function with f(x)≥0 for all x. Show that∫−aa1+f(x)1dx=afor all a≥0if and only if1+f(x)1+1+f(−x)1−1=0for all x≥0,and hence if and only if f(x)f(−x)=1 for all x.
(iii) Let f be a continuous function such that, for all x, f(x)≥0 and f(x)f(−x)=1. Show that, if h is a continuous function with h(x)=h(−x) for all x, then∫−aa1+f(x)h(x)dx=∫0ah(x)dx.(iv) Hence find the exact value of∫−21π21πcoshxe−xcosxdx.
Voir la correctionMasquer la correction
Part (i). The trick is to exploit the symmetry of the integrand. LetI=∫−aa1+ex1dx.Substitute t=−x (so dt=−dx, and the limits swap from −a to a):I=∫a−a1+e−t1(−1)dt=∫−aa1+e−t1dt=∫−aaex+1exdx.Adding the two expressions for I:2I=∫−aa1+ex1dx+∫−aaex+1exdx=∫−aa1+ex1+exdx=∫−aa1dx=2a.Hence I=a.
Part (ii). Define G(t)=∫0tg(x)dx. By hypothesis G(a)=0 for all a≥0. Differentiating gives G′(t)=g(t), so g(t)=0 for all t≥0.
For the main result, split the integral at 0 and substitute t=−x on [−a,0]:∫−aa1+f(x)1dx=∫0a1+f(−t)1dt+∫0a1+f(x)1dx=∫0a(1+f(x)1+1+f(−x)1)dx.For this to equal a=∫0a1dx, we need∫0a(1+f(x)1+1+f(−x)1−1)dx=0for all a≥0.By the first part of (ii) (with g(x)=1+f(x)1+1+f(−x)1−1, which is continuous and non-negative... actually the argument applies directly since all steps are reversible), this holds if and only if1+f(x)1+1+f(−x)1−1=0for all x≥0.Multiplying through by (1+f(x))(1+f(−x)):1+f(−x)+1+f(x)=(1+f(x))(1+f(−x))=1+f(x)+f(−x)+f(x)f(−x),which simplifies to f(x)f(−x)=1.
Part (iii). Split and substitute t=−x on the negative half as before:∫−aa1+f(x)h(x)dx=∫0a1+f(−x)h(−x)dt+∫0a1+f(x)h(x)dx.Since h(−x)=h(x):=∫0ah(x)(1+f(−x)1+1+f(x)1)dx.From part (ii), the condition f(x)f(−x)=1 is equivalent to 1+f(x)1+1+f(−x)1=1, so the bracket equals 1 for all x≥0. Therefore:∫−aa1+f(x)h(x)dx=∫0ah(x)dx.Part (iv). Rewrite the integrand by notingcoshxe−x=21(ex+e−x)e−x=e2x+12,so the integral becomes∫−π/2π/2coshxe−xcosxdx=∫−π/2π/21+e2x2cosxdx.Take f(x)=e2x and h(x)=2cosx. Then f(x)f(−x)=e2x⋅e−2x=1 and h(−x)=2cos(−x)=2cosx=h(x), so part (iii) applies with a=2π:∫−π/2π/21+e2x2cosxdx=∫0π/22cosxdx=[2sinx]0π/2=2.