NOTES ON P3

ALGEBRA
 When you remove the modulus sign by squaring be fair and square each and every term
e.g. |2x+3|<5|6-x | that 5 must be squared.
 Please use difference if 2 squares with brackets to factorize rather than expanding and then
factorize.
[(2x+3)-5(6-x)][(2x+3)+5(6-x)]<0.
 For you to get full marks please draw the number line and draw the graph – NO SHORTCUT!
 For |2x+3|>3x DO NOT USE THE GRAPH. Use the CHECKING method after getting the roots.
 Use graphical approach for |2x+3|< 1+ sqrt (x) not the squaring method.

 Remainder theorem stated: P(x) = (x-a)Q(x) + R while factor theorem is P(x) = (x+a)Q(x) + 0.
If 2 is a root of an equation then x-2 is a FACTOR of that polynomial.
 When doing a long division ensure (i) divisor and the dividend are in DESCENDING powers of x ,
(ii) make provision for ALL THE POWERS of x in the dividend.

 Partial fractions : please check DEGREES first. When you run out of the values of x to make some
of the terms zero, equate COEFFICIENTS of the like terms INDEPENDENTLY i.e. the x squared on
their own , the x on their own and the constant terms on their own.

 Given (2-x)^(-3) remember the 2 must be raised to power( - 3) and and when you expand use
(-x/2) raised to the appropriate power. Do not forget the minus!

LOGARITHMIC AND EXPONENTIAL FUNCTIONS

 Log a + log b = log ab. DO NOT OPEN BRACKETS IF YOU HAVE log(a + b).
Log a – log b = log a/b. Loga/logb is not equal to log a – log b
Log a^n = nlog a. AVOID lna^2b = 2lnab
Log 1/b = - log b or log 2/3 = - log 3/2.

 ln e = 1; e^(lnx) = x; ln(2x-4) =3 goes to 2x-4 = e^3.

 The graph of y = e^x has y=0 as an asymptote and crosses y axis at (0;1)
The graph of y = ke^x has y=0 as an asymptote and crosses the y axis at (0;k)
The graph of y= e^(-x) is the reflection of y=e^x in the y axis.
The graph of y= - e^x is the reflection of y=e ^x in the x axis.

 The graph of y= lnx has x=0 as an asymptote and crosses the x axis at (1;0) it is also the reflection
of y = e^x in the line y=x.
 The graph of y=klnx has x =0 as an asymptote and crosses the x axis at (1;0)
 The graph of y = lnkx has x=0 as an asymptote and crosses the x axis at (1/k; 0), equate kx to 1.
 To get the asymptote for the graph y = ln(3x-1) equate 3x-1 to zero, x=1/3 and to get where it
cuts the x axis, equate 3x-1 to 1, x=2/3.

 To solve 3^x = 2 take ln on both sides. To solve e^x + e^-x =3 NEVER ln one by one. Use a
substitution. Given y = ab^(-x) use lny = lna – xlnb and RELATE it to Y = MX + C where lny = Y; -
lnb = M( note the minus); x = X and lna = C.
TRIGONOMETRY
 Recall secant X = sec X = 1/cos X ; cosecant x = cosec X = 1/sin X ; cotangent X = cot X = 1/tan X
 To solve sec X = 5 go 1/cos X = 5 cos X = 1/5 then go to quadrants.
 Graphs: for sec X start with cos X graph dotted and then draw the sec X from that. Same goes for
cosec X from sin X and cot X from tan X.

 Trig identities you are given on the list of formulae – know them; recognize them; move from
one form to another.

 Understand and recognize compound angles sin(A+B) etc. if you are given sin X + 2cos X and
asked to express that in the form Rcos(X – a) please write it as 2 cos X + sin X first for you to get
the right a.

 Understand and recognize DOUBLE angles. You double or halve the angles only.
Sinx = 2sin(X/2)cos(X/2) . Remember sin X cos X = (1/2)sin2x.

DIFFERENTIATION
 Exponential functions – ‘well behaved’ never destroyed d/dx(e^f(X)) = f’(X)e^f(X).
 Ln functions please apply theory of logs first if possible e.g. asked to differentiate lnx^3(1/X+3)
go 3lnx-ln(X+3) then differentiate term by term; asked to differentiate lncotx go lncosx- lnsinx
then differentiate term by term.
 Please, please, please see PRODUCTS X sin 2x is a product; xy is a product etc and use the
product formula to differentiate which simply says differentiate one and multiple by the other
and vice versa.
 For ( cosx)/X use the quotient rule.
 For implicit functions please see products and remember do not start with dy/dx and when you
differentiate y wrt X you get dy/dx and when you differentiate dy/dx wrt X you get d^2y/dx^2
 For trig functions recall d/dx(sin X ) = cos X and d/dx(cos X) = - sin X . Do not mix up
differentiation with integration.

INTEGRATION
 Understand , recognize and remember the 6 techniques. Each time you integrate you divide by
something.
 TECHNIQUE 1 is for algebraic functions kx^n or (ax+b)^n the highest power of X inside the
brackets must be one.
 TECHNIQUE 2 is for the exponential functions ke^(ax).
 TECHNIQUE 3 is for functions which go to ln . These are of the form f’(X)/f(X) i.e. denominator
in brackets must have power 1 and the numerator can be made to be the EXACT derivative of
the denominator . tanx and cot X fall into this category and not under trig.
 TECHNIQUE 4 is for trig functions ksin(ax+b) ; kcos(ax+b); ksec^2(ax+b).
For sin^2 and cos^2 use the double angle formulae. Note that 1/sec^2x is cos^2 X needs double
angle. 1/secx is cosx .
 TECHNIQUE 5 is the use of a substitution.
Remember to write dx in the original expression and substitute for it first.
Do not mix variables .
 Remember to change the limits as well. If these seem to be upside down do not change them
around.
 TECHNIQUE 6 is for products as xsinx; xe^x; xlnx; lnx etc. put u = X unless there is a ln. If it is
there, then put u = lnx the other term will be dv/dx not v.
If this process is to be repeated please use brackets so that signs are not messed up.

 TECHNIQUE 7 – The Trapezium Rule – h = (b-a)/n where n is the number of intervals.

If the question says using 5 ordinates(y values) n = 4 . Read questions carefully so that you know
I am I given intervals or ordinates. Your upper and lower limits must work out correctly when
you add h to the lower limit up to the upper limit.
I recommend use of a table.
Concave down leads to an underestimate and concave up leads to an overestimate.

ITERATION
 Convergent – series/sequence of numbers getting to some fixed value.
 Divergent – series/sequence of numbers getting bigger and bigger.
 Given a convergent formula you can go back to the original equation by dropping the subscripts
and simplifying.
 If a formula contains sin , cos, tan make sure your calculator is in radian mode.
 To hunt for a sign change please introduce f(X) first.

VECTORS
IN ALL CASES PLEASE MAKE A SKETCH
 Be familiar with the various forms of the equation of the line – using column vectors, using the
standard unit vectors i ; j; k and a parameter t, parametric equations of the line; Cartesian
equations of a line. To find the equation of any line you need the direction and position vector
of one point on the line.
 Parallel lines – just look at the direction vectors only without the parameters. One must be a
scalar multiple of the other.
 Perpendicular lines – dot product of direction vectors must be 0.
 Intersecting lines – the three equations must be satisfied. Use two and check with the third.
 To show lines are skew – show that they are not parallel and that they do not intersect.
 Perpendicular distance of a point P from a line – let N be foot of perpendicular: (i) find position
vector of N on the line in terms of the parameter; (ii) find PN which is N – P; (iii) use PN dot
direction of line = 0 to find the value of the parameter etc
 To find the equation of a plane you need the normal and position vector of one point on the
plane and use r dot a= a dot b which can also be written as (r – b) dot a = 0.
 To show that a point lies on the plane substitute r by position vector of the point and get an
equality.
 To show that a line lies in a plane replace r by the equation of the line(right hand side), the
parameter disappears and get an equality.
 Angle between planes is the angle between normals.
 Angle between a line and a plane use a dot b = |a||b|sin theta
 To find where a line meets the plane replace r by right hand side of the equation of the line and
solve for the parameter and go back to equation of the line and put the parameter.
 Distance of a point from a plane use the formula.
  Distance of a plane from the origin use r dot a/|a|= d/|a| right hand side gives the distance.
 Distance between parallel planes is the difference between their distances from the origin.

DIFFERENTIAL EQUATIONS
 Be able to write a rate in terms of symbols d something by dt. Pay attention to units they will
help you.
 Rates which indicate a decrease must be accompanied by a minus.
 If you have volume in and volume out dv/dt is the net rate.
 Note the initial or boundary conditions well. At the beginning or initially means t = 0
 Be able to write a rate as a product of two rates noting the rate you already have.
 Separate the variables and integrate and add + C always.
 To see what a given quantity approaches write the right hand side with the other variable in the
denominator, that denominator will go to infinity and that term goes to zero.

COMPLEX NUMBERS
 Be familiar with the three forms RECTANGULAR; POLAR; EXPONENTIAL.
 When you draw an Argand diagram please label it.
 When you go polar use the Argand diagram for you to get the right argument.
 Recall De Moivre’s theorem [r(cosx + i sinx)]^n = (r^n)(cosnx + i sin nx). Very useful when finding
square roots.
 For loci use the brackets correctly |z-(2+i)|= k ; centre is at 2 + i; |z – 2 + i|= k ; centre is at 2 – i
|z+2+i|= k ; centre is at – 2 – i.
 Be familiar with the three loci circle; perpendicular bisect and half line. Note to that
arg( – Z) = pi/4 is a half line from the origin to in the third quadrant directly opposite to the half
line for arg (z) = pi/4.
 If given an equation in complex number z e.g. Z^3 – 4z^2 + z – 1 = 0 and the equation has no i in
it then roots appear with their complex conjugates. If 2-i is a root, then 2 + i is a root also.
 Straight on from above: if 2 – i is a root the z – (2-i) which is z – 2 + i is a factor and in this case
z – 2 – i is another factor AND (z-2+i)(z-2-i) is another factor i.e. z^2 – 4z + 1 (no i) which can be
divided into the original equation to get the third factor.

NOTES ON P7
THE POISSON DISTRIBUTION
 This is for a discrete random variable. Values of X start from 0.
 Pay special attention to lambda. It is always per something.
 Before any calculations ensure you are using the right lambda by using proportion.
 If calculating a set of values for X , factor out e to minus lambda it simplifies working and you
save time.
 You can add the lambdas of two independent Poisson . The key word will be ‘total’.
 Be familiar with the 4 conditions for the Poisson. If you are given say call come in at the rate
of 5 in an hour and they ask you to state the conditions which these calls should satisfy so
that Poisson distribution can be used, do not say they must come at a uniform rate. That is
repeating what they gave you already. Give the other three.
 Poisson can approximate B(n;p) provided n>50 and np<5.
 Poisson distribution is the only one where mean = variance always.
  Normal distribution can be used to estimate Poisson distribution provided lambda is greater
than 15. Please recall Check Chido applies here.

LINEAR COMBINATION OF RANDOM VARIABLES

 You know you NEVER SUBTRACT variances.
 Variance of a constant is zero.
 How to tell whether SUM or MULTIPLE ask yourself the question how many items are involved.
If more than 1 then it is a SUM, if 1 then it is a MULTIPLE.
 SUM you do not square the factor, MULTIPLE you square the factor.

CONTINUOUS RANDOM VARIABLES

 Know the basic facts about PDFs- never negative and area = 1.
 E(X) is integral of xf(X)
 Var(X) is integral of x^2f(X) - mean squared.
 To find P(a<X<b) integrate f(X) from a to b.
 Be familiar with location of median ; quartiles, deciles it is all by integrating and equating to the
right fraction.
 Know how to interpret GIVEN THAT that is CONDITIONAL probability = P(a<X<b)/P(a<X<c).
 Be able to sketch PDFs. Remember to show the zero elsewhere.
 If your PDF has a line of symmetry the line of symmetry gives E(X) and the median as well.

SAMPLING AND ESTIMATION

 A sample is part of a population. A random sample is one where every item in the population
has the same chance of being picked. Always ask yourself is every being given the same
chance?
 It is convenient to use samples provided they are random, it saves time, money, for some it
means destroying the whole population if you do not take a sample like e.g. How long bulbs will
last!
 Population parameters are mew; sigma or sigma squared; pi(for proportion).
 Sample parameters are X bar; s or s squared; var(X); p.
 Mew can be estimated by X bar because sample mean is an unbiased estimator for mew since
E(X bar) = mew.
 S squared is a biased estimator for sigma squared because E(s squared is not equal to sigma
squared. Use (n/n-1) s squared or (1/n-1)[€X squared- (€X)squared /n] to get un unbiased
estimate for sigma squared.
 Sampling distribution of the mean: X bar ~N(mew; sigma squared/n).
 CLT says provided the sample size n is LARGE, then X bar ~N(mew; sigma squared/n) where
mew and sigma squared come from the distribution of X. YOU DO NOT USE CLT IF X COMES
FROM A NORMAL DISTRIBUTION NO MATTER THE SIZE OF THE SAMPLE. For all other
distributions you can use CLT provided n is large.
 NO CC FOR X BAR PLEASE!
 To get a random sample you can use SIMPLE RANDOM sampling where names can be on pieces
of paper or cards which are thoroughly shuffled and picking is done after each shuffling; you can
use RANDOM NUMBER TABLES after the population is enumerated appropriately.
 CONFIDENCE INTERVAL – it is an interval with a degree of certainty or confidence attached to it.
e.g. 97% CI mean you are 97% sure that the population parameter lies in that interval. IT IS NOT
THE PROBABILITY THAT THE PARAMETER LIES IN THE INTERVAL.
  BE FAMILIAR WITH THE FORMULAE FOR CI. IF GIVEN THE INTERVAL THE SAMPLE MEAN OR
PROPORTION IS THE MIDPOINT.
 The width of a CI is given by TWO times z(half alpha) sigma/sqrt of n. Same goes for a
proportion.
 For the proportion the CI is approximate because we are estimating it using the normal
distribution when p comes from a binomial.
 Questions which say so many intervals were calculated find the probability that so many contain
mew or population proportion you use the binomial with p = to CI percentage. Understand that
the question is actually saying YOU ARE SURE will contain the parameter. That phrase is implied ,
it is not explicitly given.
 To calculate the level of confidence when given some data: get z(half alpha) ; draw a normal
diagram and put it. Shade the tail and find its probability; MULTIPLY THIS BY TWO and
SUBTRACT FROM 100%

HYPOTHESIS TESTING AND ERRORS

 Follow the steps: STATE DISTRIBUTION OF X; STATE Ho and H1[ do not use sample statistics
unless it is the unbiased estimation usually for sigma squared; you will need the sample mean
to calculate t] DRAW A DIAGRAM TO SHOW THE CR. ; COMPUTE t THE TEST STATISTIC FOR the
mean; COMPARE WITH SIG LEVEL AND CONCLUDE BY SAYING ACCEPT HO OR REJECT HO i.e.
then ANSWER QUESTION ASKED.
 FOR THE DISCRETE VARIABLES CHECK WHETHER THE GIVEN OBSERVED VALUE IS IN THE CR OR
NOT AND THEN CONCLUDE.
 IF YOU HAVE USED THE NORMAL DISTRIBUTION FOR POISSON OR BINOMIAL, PLEASE
REMEMBER AND DO THE CORRECT CC.
 ERRORS : TYPE l you REJECT HO AND TYPE II TOU ACCEPT HO.
 If in your hypothesis testing you have accepted HO then you are likely to have made TYPE I error
and the other way round.
 CALCULATIONS: ALWAYS DRAW THE RELEVANT DIAGRAM. MARK THE REJECTION AND
ACCEPTANCE REGIONS AND THEN CALCULATE WHAT IS REQUIRED USING THE REJECTION CUT
OFF VALUE. FOR TYPE II YOU MUST HAVE A NEW PARAMETER.