The Student Room Group

STEP maths I, II, III 1991 solutions

Scroll to see replies

generalebriety
Um, because I copied your answer without thinking when you corrected me. Because I hate the Maclaurin series. :wink:

There's no problem. I just do stupid things like ln(1+x) = 1 when x=0. Or I differentiate wrongly (especially with complicated functions). Or my arithmetic goes haywire. And I can't remember any of them off the top of my head. :s-smilie:
For what it's worth, I think the only I'd try to memorize are ln(1+x), e^x and possibly tan^-1 x. The ln(1+x) (and related ln(1+x)-ln(1-x) series) seem to come up quite a lot.
arctan is unnecessary IMO, it is quite easy to derive by integrating 11+x2\frac{1}{1+x^2} which is a G.P.

David, no need to be sorry, I was just a little stunned at what I eventually had done wrong:tongue:
Reply 22
I have an alternative solution for post 7 part b (please add freely if you think it is useful).

Notice that 1+x+x2+x3=(1+x)(1+x2)1+x+x^2+x^3=(1+x)(1+x^2)
First part:

sin2θ+sin2ϕ+sin2ψ+2sinθsinϕsinψ\sin^2\theta+sin^2\phi+sin^2\psi+2\sin\theta\sin \phi \sin\psi\\
=sinθ[cos(ϕ+ψ)+2sinϕsinψ]+sin2ϕ+sin2ψ=\sin\theta[\cos(\phi+\psi)+2\sin\phi\sin\psi]+\sin^2\phi+\sin^2\psi\\
=sinθ[cosϕcosψ+sinϕsinψ]+sin2ϕ+sin2ψ=\sin\theta[\cos\phi\cos\psi+\sin\phi\sin\psi]+\sin^2\phi+\sin^2\psi\\
=cos(ϕ+ψ)cos(ϕψ)+sin2ϕ+sin2ψ=\cos(\phi+\psi)\cos(\phi-\psi)+\sin^2\phi+\sin^2\psi\\
=cos2ϕcos2ψsin2ϕsin2ψ+sin2ϕ+sin2ψ=\cos^2\phi\cos^2\psi-\sin^2\phi\sin^2\psi+\sin^2\phi+\sin^2\psi\\
=(1sin2ϕ)(1sin2ψ)sin2ϕsin2ψ+sin2ϕ+sin2ψ=(1-\sin^2\phi)(1-\sin^2\psi)-\sin^2\phi\sin^2\psi+\sin^2\phi+\sin^2\psi\\
=1=1

Second part:

Letting θ=ϕ=π/5\theta=\phi=\pi/5 and ψ=π/10\psi=\pi/10, using the first part we have:

2sin2π5+sin2π10+2sin2π5sinπ101=02\sin^2\frac{\pi}{5}+\sin^2\frac{\pi}{10}+2\sin^2 \frac{\pi}{5}\sin\frac{\pi}{10}-1=0\\
8sin2π10cos2π10+sin2π10+8sin3π10cos2π101=0\Rightarrow8\sin^2\frac{\pi}{10}\cos^2\frac{\pi}{10}+\sin^2\frac{\pi}{10}+8\sin^3\frac{\pi}{10}\cos^2\frac{\pi}{10}-1=0\\
8sin2π108sin4π10+sin2π10+8sin3π108sin5π101=0\Rightarrow8\sin^2\frac{\pi}{10}-8\sin^4\frac{\pi}{10}+\sin^2\frac{\pi}{10}+8\sin^3\frac{\pi}{10}-8\sin^5\frac{\pi}{10}-1=0\\
(8sin3π10+sin2π101)+sin2π10(1sin2π10sin3π10)=0\Rightarrow(8\sin^3\frac{\pi}{10}+\sin^2\frac{\pi}{10}-1)+\sin^2\frac{\pi}{10}(1-\sin^2\frac{\pi}{10}-\sin^3\frac{\pi}{10})=0
(sin3π10+sin2π101)cos2π10=0\Rightarrow(\sin^3\frac{\pi}{10}+\sin^2\frac{\pi}{10}-1)\cos^2\frac{\pi}{10}=0\\
sin3π10+sin2π101=0\Rightarrow\sin^3\frac{\pi}{10}+\sin^2\frac{\pi}{10}-1=0

justinsh
Sorry for misprints, it seems LaTex does funny things on this forum. If someone could tell me how to fix the errors, I would be glad to do so.

spaces sometimes does wonders... (didn't quote you just so you can look at the code)
justinsh
I have an alternative solution for post 7 part b (please add freely if you think it is useful).

Notice that 1+x+x2+x3=(1+x)(1+x2)1+x+x^2+x^3=(1+x)(1+x^2)

Nice thought. I'll gladly add it if you do the rest of the working because I'm certainly not doing any more Maclaurin today. :wink:
Reply 25
I may have done mistakes but here it goes:

ln(1+x+x2+x3)\ln(1+x+x^2+x^3)\\
=ln(1+x)+ln(1+x2)=\ln(1+x)+\ln(1+x^2)\\
=(x12x2+13x3+...+(1)n+1nxn)+(x212x4+13x6+...+(1)n+1nx2n)=(x-\frac{1}{2}x^2+\frac{1}{3}x^3+...+\frac{(-1)^{n+1}}{n}x^n)+(x^2-\frac{1}{2}x^4+\frac{1}{3}x^6+...+\frac{(-1)^{n+1}}{n}x^{2n})\\
=x+12x2+13x334x4+...+[(1)n+1n12n]x2n+x2n+12n+1=x+\frac{1}{2}x^2+\frac{1}{3}x^3-\frac{3}{4}x^4+...+\left[\frac{(-1)^{n+1}}{n}-\frac{1}{2n}\right]x^{2n}+\frac{x^{2n+1}}{2n+1}
Reply 26
STEP II question 3

First part:

Multiplying be xyz everywhere yelds:

x2yz+xz=xy2z+xy=xyz2+yzx^2yz+xz=xy^2z+xy=xyz^2+yz (*)

Using (*) three times:

x2yz+xz=xy2z+xyx^2yz+xz=xy^2z+xy
xyz(xy)=x(yz)\Rightarrow xyz(x-y)=x(y-z) (1)

xy2z+xy=xyz2+yzxy^2z+xy=xyz^2+yz
xyz(yz)=y(zx)\Rightarrow xyz(y-z)=y(z-x) (2)

xyz2+yz=x2yz+xzxyz^2+yz=x^2yz+xz
xyz(zx)=z(xy)\Rightarrow xyz(z-x)=z(x-y) (3)

Multiplying (1), (2) and (3) yelds:

(xyz)3(xy)(yz)(zx)=xyz(yz)(zx)(xy)(xyz)^3(x-y)(y-z)(z-x)=xyz(y-z)(z-x)(x-y)
x2y2z2=1\Rightarrow x^2y^2z^2=1

Second part:

(x+1y)3(x+\frac{1}{y})^3\\
=(x+1y)(y+1z)(z+1x)=(x+\frac{1}{y})(y+\frac{1}{z})(z+\frac{1}{x})\\
=xyz+1xyz+(x+1y)+(y+1z)+(z+1x)=xyz+\frac{1}{xyz}+(x+\frac{1}{y})+(y+\frac{1}{z})+(z+\frac{1}{x})\\
=xyz+1xyz+3(x+1y)=xyz+\frac{1}{xyz}+3(x+\frac{1}{y})
=±2+3(x+1y)=\pm2+3(x+\frac{1}{y}) since xyz=1xyz=-1 (from first part)

First case: (notation X=x+1yX=x+\frac{1}{y})

X3=2+3XX^3=2+3X
(X+1)2(X2)=0\Rightarrow(X+1)^2(X-2)=0
X{1;2}\Rightarrow X\in\{-1;2\}

Second case:

X3=2+3XX^3=-2+3X
(X1)2(X+2)=0\Rightarrow(X-1)^2(X+2)=0
X{1;2}\Rightarrow X\in\{1;-2\}

I don't know how to get rid of the degenerate solutions... (help!)

EDIT: sorry khaixiang, nice solution though!
Reply 27
Ahhh.. I wanted to do II/3! Since we're using completely different methods, I suppose this is still worth posting. And there're many more ways to do this question (I've seen at least 3 more different ways)

x+1y=y+1zy+1z=z+1xz+1y=z+1x\\x+\frac{1}{y}=y+\frac{1}{z}\\ y+\frac{1}{z}=z+\frac{1}{x}\\ z+\frac{1}{y}=z+\frac{1}{x}

Rewrite the above equations:
xy=1z1yyz=1x1zxz=1x1y\\x-y=\frac{1}{z}-\frac{1}{y}\\ y-z=\frac{1}{x}-\frac{1}{z}\\ x-z=\frac{1}{x}-\frac{1}{y}

xy=yzyzyz=zxxzxz=yzxy\\x-y=\frac{y-z}{yz}\\ y-z=\frac{z-x}{xz}\\ x-z=\frac{y-z}{xy}

Multiply them together:

(xy)(yz)(xz)=(xy)(yz)(xz)x2y2z2xyz=±1\\(x-y)(y-z)(x-z)=\frac{(x-y)(y-z)(x-z)}{x^2y^2z^2}\\ xyz=\pm 1

Now let:
x+1y=y+1z=m\\x+\frac{1}{y}=y+\frac{1}{z}=m

(1)x+1y=mxy+1=my\\x+\frac{1}{y}=m\\ xy+1=my

(2)y+1z=m If xyz=1 Then y+xy=m\\y+\frac{1}{z}=m\\ \text{ If } xyz=1\\ \text{ Then }y+xy=m

(3)y+1z=m If xyz=1 Then yxy=m\\y+\frac{1}{z}=m\\ \text{ If } xyz=-1\\ \text{ Then }y-xy=m

(1)-(2) and (1)+(3) gives m=±1m=\pm1
Reply 28
STEP II question 4 (under construction)

First part:

2y2=3y+22y^2=3y+2
y{12,2}\Leftrightarrow y\in\{\frac{-1}{2},2\}
Because y is composed of two cosines smaller than 1, y=12y=\frac{-1}{2}.

Second part:

[br]k=116coskθ=k=116eikθ=(k=116eikθ)=(1ei17θ1eiθ)=(111eiθ)=0[br]\displaystyle\sum_{k=1}^{16}\cos{k\theta}=\sum_{k=1}^{16}\Re e^{ik\theta}=\Re\left(\sum_{k=1}^{16}e^{ik\theta} \right) = \Re\left( \frac{1-e^{i17\theta}}{1-e^{i\theta}} \right)=\Re \left(\frac{1-1}{1-e^{i\theta}}\right)=0

Third part:

Considering the fact that the inverse of a complex number of modulus 1 is its conjugate:

S=2(cosθ+cos2θ+cos4θ+cos8θ)S=2(\cos\theta+\cos2\theta+\cos4\theta+\cos8\theta)
=eiθ+ei2θ+ei4θ+ei8θ+ei9θ+ei13θ+ei15θ+ei16θ=e^{i\theta}+e^{i2\theta}+e^{i4\theta}+e^{ i 8 \theta }+e^{i9\theta}+e^{i13\theta}+e^{i15\theta}+e^{i16 \theta }

We then compute S2S^2 realising that the square of the terms equals S and simplying the rest using the second part. We eventually find that S2S+4=0S^2-S+4=0.

A drawing then let us determine the sign of the correct expression.
Reply 29
Nvm.
XD
Reply 30
III/5

All derivatives with respect to t.

r'' + 4r = 5sin3t

CF -> m^2 + 4 = 0
m = +-2i
r = Acos2t + Bsin2t

PI -> r = psin3t + qcos3t
r'' = -9psin3t - 9qcos3t
-5psin3t - 5qcos3t = 5sin3t
p = -1, q = 0

r = Acos2t + Bsin2t - sin3t
r' = -2Asin2t + 2Bcos2t - 3cos3t

t = pi/2, r = 1
1 = -A + 1
A = 0

t = pi/2, r' = -2
-2 = -2B
B = 1

r = sin2t - sin3t

r(pi/5) = sin2pi/5 - sin(3pi/5) = sin(2pi/5) - sin(pi - 3pi/5) = 0
r(3pi/5) = sin(6pi/5) - sin(9pi/5) = -sin(pi/5) - sin(-pi/5) = 0
Since the function is clearly continuous, and the curve starts and finishes at the origin, it is a closed loop.

Area = (1/2) INT (pi/5 to 3pi/5) r^2 dt
= (1/2) INT sin^2(2t) + sin^2(3t) - 2sin2tsin3t dt
= (1/2) INT (1/2) - (1/2)cos4t + (1/2) - (1/2)cos6t + cos5t - cost dt
= (1/2)[t - (1/8)sin4t - (1/12)sin6t + (1/5)sin5t - sint](pi/5 to 3pi/5)
= (1/2)[3pi/5 - (1/8)sin(12pi/5) - (1/12)sin(18pi/5) + (1/5)sin(15pi/5) - sin(3pi/5)] - (1/2)[pi/5 - (1/8)sin(4pi/5) - (1/12)sin(6pi/5) + (1/5)sin(5pi/5) - sin(pi/5)]
= (1/2)[3pi/5 - (1/8)sin(2pi/5) + (1/12)sin(3pi/5) + 0 - sin(3pi/5)] - (1/2)[pi/5 - (1/8)sin(pi/5) + (1/12)sin(pi/5) + 0 - sin(pi/5)]
= pi/5 - (1/16)sin(2pi/5) - (11/24)sin(3pi/5) + (25/48)sin(pi/5)
= pi/5 - (3/48)sin(2pi/5) - (22/48)sin(2pi/5) + (25/48)sin(pi/5)

= pi/5 + (25/48)[sin(pi/5) - sin(2pi/5)]

Hope there's another straightforward differential equation question on this year's STEP III :smile:
Reply 31
II, 2:
I'm not entirely sure how to finish this one off, so someone else do that.
We're given
3y2+3x210xy+16y16x+15=03y^2 +3x^2 -10xy +16y - 16x +15=0
Differentiate and divide by 2 - I'll write m for dy/dx, since I hate fractions:
3ym+3x5[y+xm]+8m8=03ym + 3x - 5[y+xm] + 8m -8 = 0
[3y5x+8]m=83x+5y [3y-5x+8]m = 8-3x+5y
m=83x+5y3y5x+8 m = \frac{8-3x+5y}{3y-5x+8}
Now use yy1=m(xx1)y-y_1 = m(x-x_1), the infamous straight line formula, with x1=s, y1=t:

y=mxms+t y = mx-ms+t
Consider the ms-t bit, putting it all over the denominator of m:
ms+t=8s+3s25st+3t25st+8t3t5s+8-ms+t = \frac{-8s+3s^2-5st +3t^2 -5st + 8t}{3t -5s+8}
From our original equation, 3s2+3t210st=16s16t15 3s^2 + 3t^2 -10st = 16s - 16t -15 as (s,t) is on the curve. So -ms+t becomes
16s16t158s+8t3t5s+8=8s8t153t5s+8 \frac{16s-16t-15-8s+8t}{3t-5s+8} = \frac{8s-8t-15}{3t-5s+8}
Now you have to wonder why you're missing a minus sign, and realise that they've factored it out for some reason, so it's the same thing really.
Next bit:
As x, y (or s,t, whatever) get large, we can ignore the linear terms (and the constant of course) in the original equation:
The curve approximates another curve - 3y210xy+3x2=03y^2-10xy +3x^2=0
y2103xy+x2=0y^2 - \frac{10}{3} xy + x^2 =0
(y53x)2+x2259x2=0(y-\frac{5}{3} x)^2 +x^2 - \frac{25}{9} x^2 =0
y=53x±43x y = \frac{5}{3} x ± \frac{4}{3} x
Those = signs should probably be --> to represent some kind of tending to, but...hm. Anyway.

The above relation gives y = 3x and y=x/3. Putting those in (with s and t) gives the required result.

For the next bit, use dy/dx = tan{angle between x axis}.
And dx/dy = tan{angle between y axis}.

For the axes of the hyperbola, use y=mx+c, and you know that both axes cross through the intersection of the two asymptotes (which is to be found).

I'm not sure how to determine the gradient of the axes (they bisect the asymptotes, but still...)
Anyone want to tell me?

I'm guessing that the gradients are 1 and -1, since tan(m) is probably something to do with averaging the angles, ie, ½(arctan(3) + arctan(1/3)) = 45 (works for any arctan x and 1/x).
Reply 32
Oh crap. You post it then, I'll do away with mine :P
Sorry bout that.
Reply 33
Am I going crazy or are matrices no longer on the STEP syllabus?
http://www.ocr.org.uk/Data/publications/specifications_syllabuses_and_tutors_handbooks/Specificat98487.pdf


Also, does anyone have the '05 papers? (II and III '06 are available if you go back a couple hundred pages to last June in this forum).
Reply 34
Wow, I can't find Matrices there either.
(Thanks about the 06 papers, by the way...did you want to start threads about those newer ones some time, too?)
Decota
GE you might want to check this line.

Oh, ****.

:rolleyes: I'll fix it, just not now. Cheers. Thought it was too simple.
Reply 36
There are full sets of solutions online for 1999 onwards, so I'm not sure there's much point. I think we should continue going back in time for now.

EDIT: It would seem that vectors beyond C4 are also not on the syllabus :biggrin:
Speleo: correct, matrices aren't on it any more.
Rabite: it was DFranklin who suggested that. :smile:

Can someone check I've got that bloody question right this time please? :redface:
Reply 38
Okay. Just asking because I thought I saw someone asking if we should discuss the new ones too, but perhaps it's just my imagination.

Anyway, II, Q5, since it doesn't seem to be here yet:

y=tanx should be a familiar graph. (attached anyway)
In the given range, tanx <1, so increasing the power makes it 'smaller'. Note that the area under tan^k x approaches zero as k gets arbitrarily large.

For the integral, establish some sort of reduction formula. Let's call I_n the thing they want us to integrate:
By the identity tan²+1=sec²





So:

The terms in the 1/2n sum alternate their signs, so we can call it (m=1 to m=n) - but the last term (1/2n) is always positive, so we account for that with a (-1)^n, such that for odd n, it cancels out the minus sign on the last term of the sum.
As for the sign on , its sign is negative for odd n, and positive for even n (check the case with n=1 to make sure), so we can replace the ± with another (-1)^n. This gives



which is a derivative over its function:


As n approaches infinity, the integral tends to zero (see graph) and so the negative of the sum approaches ½ln2.
This gives the required result (factor the -1 into the summation to get the m-1 in the power).

For the next bit, redefine I_n:



as before (just replace all the 2n+1s with 2ns).


This time, I_0 is simply pi/4.
The required result follows from the same logic as before.

Erm, a question - what's a good way of justifying the choice of sign on the different parts? Here I just looked at n=1 and generalised that to odd n, which isn't terribly convincing.
Reply 39
II, Q7:
Again, not entirely sure if it's totally right. I'm one of those people who only ever get two thirds of anything right ><;;

Differentiating g(x)+g(y)=g(z) with respect to x, treating y, and g(y) as constants:
g`(x)=g`(z)*{dz/dx}
z=xyx+y+1 z = \frac{xy}{x+y+1} gives

dzdx=(x+y+1)yxy(x+y+1)2 \frac{dz}{dx} = \frac{(x+y+1)y-xy}{(x+y+1)^2}
=y(y+1)(x+y+1)2 = \frac{y(y+1)}{(x+y+1)^2}

By brute force:
z(z+1)=(xyx+y+1)(xy+x+y+1x+y+1)=(xyx+y+1)((y+1)(x+1)x+y+1)z(z+1) = (\frac{xy}{x+y+1})(\frac{xy+x+y+1}{x+y+1}) =(\frac{xy}{x+y+1})(\frac{(y+1)(x+1)}{x+y+1})
=xy(x+1)(y+1)(x+y+1)2=\frac{xy(x+1)(y+1)}{(x+y+1)^2}
ie z(z+1)x(x+1)=y(y+1)(x+y+1)2 \frac{z(z+1)}{x(x+1)} = \frac{y(y+1)}{(x+y+1)^2}

So we arrive at the required result.

Setting x=1 and z=u (or y=2u/(1-u) such that z=u if you prefer), we get
g(1)=u(1+u)2g(u)g'(1) = \frac{u(1+u)}{2} g'(u)

g`(z) is a sort of "d{g(z)}/dz" thing, so we can separate and integrate:
2g(1)duu(u+1)=dg2g'(1) \int \frac{du}{u(u+1)} = \int dg
2g(1)1u1u1du=g 2g'(1) \int \frac{1}{u} -\frac{1}{u-1} du = g
2g(1)[lnulnu1]+C=g 2g'(1)[\ln{u} - ln{u-1}] +C = g

g=2g(1)lnuu1+Cg = 2g'(1) \ln\frac{u}{u-1} +C
C is B. And B=0 since 2B=B from *.
A cannot be determined from *, since whatever values of x, y you take, A will always divide out.


Proceeding as before, for constant y:
f`(x) = f`(z) {dz/dx}
z=xy, so dz/dx = y
f`(x) = f`(z)y = f`(z) (z/x)

Now let x=a (some constant)and so z=u.
f`(a) = uf`(u)/a = u(df/du)/a
af(a)1udu=df af'(a)\int \frac{1}{u} du = \int df

af(a)lnu+c=f af'(a) \ln{u}+c = f
c=0 as before.

I'm not sure about all this treating things as constants...when they said "C", did they mean anything in particular? As above, the constant infront of the ln can't be determined, so I suppose not...

Quick Reply

Latest

Trending

Trending