The Student Room Group

STEP I, II, III 1999 solutions

Scroll to see replies

I don't have the question in front of me, but I don't see why x1>x2    x1>x2>...>xnx_1 > x_2 \implies x_1 > x_2 > ... > x_n (that is, I would want more explanation than you've given).
**Ignore me**
(edited 12 years ago)
Do you think that is sufficient?
No.

Note that you've made an arbitrary decision that x_1 > x_2. This has nothing to do with the "symmetrical equations" and so there is no reason to expect this to imply similar relations for x_2 > x_3 etc.

And in fact, now I look up the question, your claim is not actually true. Look at (ii):

You have x1x2=x3x2x2x3x_1 - x_2 = \frac{x_3-x_2}{x_2x_3}. Since x_2x_3 > 0, the RHS is > 0 when the LHS is. So x1>x2    x3>x2x_1 > x_2 \implies x_3 > x_2.

(This would also easily be seen by doing a "sanity check": e.g. if x_3 = 10, then x_2 = 11/10 and x_1 = 21/11, so x_1 > x_2 < x_3).
**Ignore me**
(edited 12 years ago)
You haven't established a contradiction. Your statement "If (1) holds then x1 > x2 > ..." is not true.

Your choice that x1 > x2 is arbitrary. You say yourself, "either x1 > x2, or x1 < x2". Yes it's fine to look at each case and see if you can show a contradiction. But you CAN'T say "the equations are all the same, so if x1 > x2 then it must also be true that x2 > x3 ...". Because your assumption that x1 > x2 has NOTHING to do with the equations.

If the equations were instead:

x1 = 1 / x2, x2 = 1/x3 ..., x9 = 1 / x10, x10 = 1/x1 (i.e. same set of equations, but without the "1 + " bit and forcing n = 10) then you'd find x1 can be any non-zero number and if x1 is positive and not equal to 1 you have x1 > x2, x2 < x3, x3 > x4 ... with no contradiction.
ahhhh, I see.
I'd never really used a proof like this before so I thought it would be a good idea to ask first. I guess I'll have to be more careful next time I try something like it.
Thanks very much for your help.
Original post by jasperleeabc
The number of currants X in a portion x is actually Po(4x)\sim Po(4x)

So the integrals are actually (2x)[e4x(4x)44!]dx\displaystyle \int (2x)[e^{-4x}\frac{(4x)^{4}}{4!}] \, dx


Pretty sure my solution was ok, and that Poisson isn't really appropriate here (taking X=1 shows this most easily). However, I must apologise for my mis-spelling of currant...

Hope the STEP revision is going ok!
Nevermind!
(edited 12 years ago)
Reply 149
Original post by Zhen Lin
Aaah. I see. I will try that now.

In the meantime, these are my solutions for questions 1, 3, 4, 5, and 7 of paper III. Incidentally, one of my interview questions was to prove that there were only five regular polyhedra (last part of question 4), but by considering the sum of angles at each vertex instead... Much less long-winded, I think, even though it would be necessary to assume that all regular polyhedra are convex.

Also, I think this post needs updating... (There's also a thread for 2000 now.)


A typo I think in Q 4, the tetrahedron should have 6 edges but not 4.
Reply 150
STEP II 1999 question 4 part ii

Prove that
2t=1nt(nt)2=n(2nn)2\sum_{t=1}^n t\binom{n}{t}^2=n\binom{2n}{n}.

A simple-minded alternative solution.

Let C be the function that returns the value of the coefficient of x^n in its argument.

2t=1nt(nt)2=2C((1+x)nd(1+x)ndx)2\sum_{t=1}^n t\binom{n}{t}^2=2C\left( (1+x)^n \frac{d(1+x)^n}{dx}\right)

=2C(n(1+x)2n1)=2C(n(1+x)^{2n-1})

=2n(2n1n)=2n\binom{2n-1}{n}

=2n(2n1)!n!(n1)!=\frac{2n(2n-1)!}{n!(n-1)!}

=(2n)!n!(n1)!=\frac{(2n)!}{n!(n-1)!}

=n(2n)!n!n!=\frac{n(2n)!}{n!n!}

=n(2nn)=n\binom{2n}{n}
Reply 151
STEP II/3 is much easier if when considering the induction assume that Sn(x) is of the form exp(x3)exp(-x3)Pqa(x) where Pqa(x) is a polynomial of order q with cofficient of highest power a. The clue is the "By means of your induction argument argument" in the question.

You can then express Sn+1(x) as -3x2Pqa(x) + P'qa(x), which makes what's going on really obvious.
Reply 152
I did question 3, STEP II, yesterday, this is my solution. Be critical please. Im **** at using latex so i just scanned and uploaded it.

Going to upload it again, some sort of wizardy occured.. ok tell me if the attachments work...

Read the 2nd attachment first
Read the 1st attachment second
other two are in the right order, dont know why the first two swapped around
(edited 11 years ago)
Reply 153
@desijut

There is a really easy way to do the last part of II/3 and show that Sn(a)Sn+1(a)<=0 :

Sn+1(a)=-3a2Sn(a)

hence Sn(a)Sn+1(a)=-3(aSn(a))2

The part in the brackets is squared, so must be >=0, but it is multiplied by -3, so it must be <=0 .

Aside from that, when you have the function notation S(x), you can save a lot of writing by writing S'(x) instead of (dS(x))/(dx) !
Reply 154
Yeah cheers, the thing was it was under timed conditions so at the time going case by case was the easiest thing.

And, i rewrote that, in my initial working i used S'(x) but then i saw their notation on the last part, so when i quickly rewrote it (my initial working was really messy, and since it was a proof, i wanted it to be neat) i used the notation they used
Reply 155
Original post by squeezebox
STEP III - Question 8

i)

0x2πn=1 0 \leq x \leq 2\pi \Rightarrow n = 1 .

So the differential equation becomes:

d2ydx2+y=0 \frac{d^{2}y}{dx^{2}} + y = 0, whose general solution is: y(x)=Acos(x)+Bsin(x) y(x) = A\cos(x) + B\sin(x) .
Now, since when x=0, y=0 and dydx=1\frac{dy}{dx} = 1 , the particular solution is y(x)=sin(x)y(x) =\sin(x) (*).

ii) Since 2πx4π,n=2 2\pi \leq x \leq 4\pi, \Rightarrow n = 2 .

So this time the differential equation is:

d2ydx2+4y=0\frac{d^{2}y}{dx^{2}} + 4y = 0

general solution: y(x)=Ccos(2x)+Dsin(2x)y(x) = C\cos(2x) + D\sin(2x) (**).

Now, since y and dydx\frac{dy}{dx} are continuous at x=2πx=2\pi, (**) has the same value as (*) at x=2πx = 2\pi and also their derivatives take the same value. From this we can work out the values of C and D, (C=0 and D= 12\frac{1}{2}). Which gives us: y(x)=12sin(2x)y(x) = \frac{1}{2}\sin(2x).
Using the same method in the previous part, we can deduce that y(x)=1nsin(nx)y(x) = \frac{1}{n}\sin(nx).

iii)

0y2dx=02πy2dx+2π4πy2dx+4π6πy2dx+...........+2(r1)π2rπy2dx+.... \displaystyle\int^\infty_0 y^{2} \, dx = \displaystyle\int^{2\pi}_{0} y^{2} \, dx + \displaystyle\int^{4\pi}_{2\pi} y^{2} \, dx + \displaystyle\int^{6\pi}_{4\pi} y^{2} \, dx + ...........+ \displaystyle\int^{2r\pi}_{2(r-1)\pi} y^{2} \, dx + ....




= 02πsin2(x)dx+2π4π122sin2(x)dx+4π6π132sin2(x)dx+..... \displaystyle\int^{2\pi}_{0} \sin^{2}(x) \, dx + \displaystyle\int^{4\pi}_{2\pi} \frac{1}{2^{2}}\sin^{2}(x) \, dx + \displaystyle\int^{6\pi}_{4\pi} \frac{1}{3^{2}}\sin^{2}(x) \, dx + .....



= 02π(1cos(2x))dx+2π4π122(1cos(4x))dx+4π6π132(1cos(6x))dx+..... \displaystyle\int^{2\pi}_{0} (1-\cos(2x)) \, dx + \displaystyle\int^{4\pi}_{2\pi} \frac{1}{2^{2}}(1-\cos(4x)) \, dx + \displaystyle\int^{6\pi}_{4\pi} \frac{1}{3^{2}}(1-\cos(6x))\, dx + .....

(using cos(2x)=12sin2(x) \cos(2x) = 1 - 2\sin^{2}(x) )

= 12[2π0]+122[12(4π2π)]+132[12(6π4π)]+.... \frac{1}{2}[ 2\pi - 0] + \frac{1}{2^{2}}[\frac{1}{2}( 4\pi - 2\pi )] + \frac{1}{3^{2}}[\frac{1}{2}( 6\pi - 4\pi)] + ....


= π+π22+π32+.....=πn=11n2 \pi + \frac{\pi}{2^{2}} + \frac{\pi}{3^{2}} + ..... = \pi \displaystyle\sum_{n=1}^\infty \frac{1}{n^{2}}


For this question i did:

d2ydx2+n2y=0 \frac{d^{2}y}{dx^{2}} + n^2y = 0, whose general solution is: y(x)=Acos(nx)+Bsin(nx) y(x) = A\cos(nx) + B\sin(nx) .
Now, since when x=0, y=0 and dydx=1\frac{dy}{dx} = 1 , the particular solution is y(x)=1/nsin(nx)y(x) =1/n\sin(nx) (*).

Which meant i just subbed in n=1 and n=2 for (i) and (ii) and stated that due to it being continous the << changed to \leq

However, this made me confused for find y(x) for all xox\geq o

But then for 0y2dx \displaystyle\int^\infty_0 y^{2} \, dx i changed it to n=11n20sin2(nx)dx\displaystyle\sum_{n=1}^\infty \frac{1}{n^{2}} \displaystyle\int^\infty_0 sin^{2}(nx)dx (infinity and 0 should be 2n(pi) and 2(n-1)(pi))

and calculated 0sin2(nx)dx=π\displaystyle\int^\infty_0 sin^{2}(nx)dx = \pi
(edited 11 years ago)
Reply 156
Original post by squeezebox
STEP III - Question 6

Parametrization: x=acosn(t)x=a\cos^{n}(t) and y=asinn(t)y=a\sin^{n}(t)

When n=3, x=acos3(t)x=a\cos^{3}(t) and y=asin3(t)y=a\sin^{3}(t).

Using the chain rule: dydx=dydt.dxdt=3asin2(t)cos(t)3acos2(t)sin(t)=tan(t).\frac{dy}{dx} = \frac{dy}{dt}.\frac{dx}{dt} = \frac{3a\sin^{2}(t)\cos(t)}{-3a\cos^{2}(t)\sin(t)} = -\tan(t).

So dydx=0\frac{dy}{dx} = 0 when t=0,π t= 0, \pi and 2π2\pi and as tπ2 t \longrightarrow \frac{\pi}{2} and 3π2,dydx±.\frac{3\pi}{2}, \frac{dy}{dx} \longrightarrow \pm \infty.

And also x=a when y=0, and y=a when x=0.

*graph is attached*

see later in thread for the rest.


how many marks would the graph be? I got x= a^(2/3), y=0 accidentally but i got the turning points, etc
Reply 157
Brianeverit 1999IIQ8.png
Original post by brianeverit
1999 Paper 3 numbers 9,11 and 14


Step II, Q.8- Can anyone walk me through Brianeverit's solution, from the part attached below.

How to jump from cos(1/2)pi/n - cos (n+1/2)pi/n to -2sin((n+1)2n)pi x sin(-pi/2n).

And also, the next step- which seems to imply that -sin(-pi/2n)=cos(pi/2), which isn't clear to me at all...

Please quote/PM me if you can help walk me through. Thanks.

Brianeverit 1999IIQ8.png
He has simply used the formula (given in the formula booklet you should have or have met) that cosA-cosB=-2sin(A+B/2)sin(A-B/2)
Reply 159
Original post by snaptwig


Step II, Q.8- Can anyone walk me through Brianeverit's solution, from the part attached below.



I actually arrived at the required answer through absolutely no algebraic manipulation, hence why I am curious about Brianeverit's method.

My method is as in the attached (sorry I did through Paint, I don't know how to Latex). It seems to produce the required answer, but far less rigid than Brian's approach.

Brianeverit 1999IIQ8.png

Quick Reply

Latest

Trending

Trending