PGCE Secondary Mathematics Pre-Interview Task: 1 Dividing by Fractions
PGCE Secondary Mathematics Pre-Interview Task: 1 Dividing by Fractions
PGCE Secondary Mathematics Pre-Interview Task: 1 Dividing by Fractions
Jonathan Powell
1 Dividing by Fractions
The rule for dividing two fractions is stated as turn the second fraction upside down then multiply. That is p r p s = (1) q s q r
1.1 Explanation 1
The eld of rationals Q includes all numbers that can be written in the form a b where a Z and b Z \ 0. Division of a by b is dened as multiplication of a by b1 the multiplicative inverse of b. That is a b = a b1 (2)
r 1 1 = 1. It is easy let a = p q and b = s then we want to nd b , i.e. the solution to b b s to see that a solution to this is b1 = r as,
(3)
That is that dividing by a unit fraction 1 s is equivalent to multiplying the numerator by s p 1 p ps = s= . (4) q s q q Combining the two ideas will hopefully not present too many diculties, and can be reinforced by similar graphical examples.
Figure 2: Division by
1 3
is equivalent to multiplication by 3.
(5)
and so e.g. =
b+ b2 4ac 2a
and =
b b2 4ac . 2a
Whether or not these solutions exist depends on whether we are working in a number system that admits the operations of taking the square root, and arbitrary division. Let us consider the following cases: a, b, c C All the operations are valid on complex numbers, so we can always nd complex roots. a, b, c R R C, so we can always nd complex roots. However we can only nd real roots when b2 4ac 0, as then the square root is a valid operation in R. a, b, c Q We can always convert p(x) = 0 to an equivalent problem with coecients in Z by multiplying by the lowest common denominator of a, b and c, so we shall not consider this case further. a, b, c Z Z R C, so we can reuse the results above for complex and real roots. Rational roots occur when b2 4ac is a proper square as otherwise b2 4ac is irrational. Integer roots occur if and only if, a|b, a|c and b2 4ac is a proper square. Theorem 2.1. Given a quadratic polynomial, p(x) = ax2 + bx + c, where a, b, c Z. The roots of p(z ) are integers, if and only if, a|b and a|c and b2 4ac is a perfect square. Proof. Let, ax2 + bx + c = a(x )(x ) = a(x2 ( + )x + ), 3 (6)
integers, that is a divides b and c. Further b2 4ac must be a rational number, if we are coprime then,
b c a and a are 2 4ac , this implies that because = b+ 2b a 2 assume that b 4ac = p q , where p and q
b2 4ac =
p2 q2
(7)
but b2 4ac is an integer, which implies that q = 1 and we have that b2 4ac is a perfect square. To show the converse we assume a|b and a|c and that b2 4ac = k 2 is a perfect square. b+k b+k We know from the quadratic formula that = 2 a and = 2a are rational, and also that + and are integers. By Lemma 2.2 and are integers. Lemma 2.2. If p, q Q and p + q Z and pq Z then p, q Z. Proof. Let p + q = n and pq = m, where n, m Z. We have p = n q . Let q = a b where, a2 a2 a a gcd(a, b) = 1. Then (n b ) b = m = na b = mb = b Z. This implies a b Z, since a and b are coprime. Hence p, q Z.
Figure 3: C and D are the points of intersection of the two circles of identical radius and center A and B . E is the intersection of AB and CD. Lemma 3.1. CD is the perpendicular bisector of AB . Proof. AC, AD, BC and BD are all line segments of equal length by construction. Therefore ABC, ABD, ACD and BCD are all isoceles. ACB and ABD are congruent, as they share three side lengths, therefore BAC = ABC = BAD = ABD A similar argument also shows ACD = BCD = ADC = BDC . Therefore ACE, BCE, AED and BDE are all congruent as they share two angles and an included side. This implies AEC = BEC = AED = BED and furthermore these are all right angles as the sum of angles around a point sum to 360 . Therefore we have that AE = BE and AB CD.
D B E C F B
D F E C
Figure 4: Given an acute angle ABC , construct a circle centered at B , dening its intersection with AB as D and with BC as E . Construct two more circles of radius BD at D and E . Dene the intersection of these two circles as F . Lemma 3.2. BF bisects ABC . Proof. By a similar argument to the proof of Lemma 3.1, it can be shown that BDF and BEF are congruent. Therefore the corresponding angles ABF and CBF are equal. Hence BF bisects ABC .
Figure 5: Begin with a line segment AB , construct two circles of radius AB centered at A and B . Dene one of their points of intersection as C . By construction AB, BC, AC are all of equal length. Therefore ABC is equilateral.
4 Hypothesis Testing
In order to test a hypothesis about the relation between two sets of data it is rst important to have a well posed question. The hypothesis the girls in [a] class are taller than the boys is ambiguous as stated. Is the claim that all the girls are taller than the tallest boy? Or rather a claim that the average height of the girls is greater than the average height of the boys. If we assume that the claim is of the second type, then there is still the matter of how to dene the average. Colloquially speaking an average is a typical member of a set, and often in a mathematical sense refers to the arithmetic mean. However there exist other equally sensible averages, such as the mode, or the median, which have their own useful properties in certain situations. The median for instance is not as sensitive to outliers in the data set as the mean. If we assume that Mary is actually making the hypothesis that the mean height of the girls is greater than the mean height of the boys then that is a clear statement about a nite set. We can settle the question immediately by calculating the mean height of all the boys and girls, and simply comparing these two means. However Marys approach is to measure the mean of the rst 5 boys and 5 girls listed on the class register. This problem is then one of determining what can be inferred from knowledge about these limited samples. Assuming that the class register was not sorted pathologically (e.g. girls in descending height order and boys in ascending height order), it would be a reasonable assumption that register order would not be correlated with height. Marys choice of sampling could however be improved by some form of random selection to eliminate such possible correlations. The problem is that there is no way to make denitive statements about the mean of the entire population based on a subset sampling. The result must be couched in the language of probability. A standard way to achieve this is to formulate a null hypothesis, in this case that would be the hypothesis that the two means do not dier, and then attempt to disprove this hypothesis. A variety of statistical tools exist to perform such hypothesis testing, and the choice of such a test is informed by any knowledge of the underlying data. In this case, an appropiate test may be the independent two-sample Students t-test. The test statistic t is dened by 1 + X 2 X t= (8) s2 s2 1 2 N1 + N2 i , s2 , and Ni respectively. where the i-th samples mean, variance and size are denoted X i If t is less than a threshold, say 0.05, then this implies that it is quite unlikely that the two samples have come from distributions with equal means, and the null hypothesis can be rejected. If t > 0.05 then the interpretation is that we do not have enough evidence to reject the null hypothesis. It is important to understand that this provides evidence in favour of the hypothesis, and does not prove or disprove it. In this particular case there is virtually no added cost to just working out the true population means and comparing them. However in situations where it is costly to perform additional experiments it is important to have these statistical methods to correctly interpret the data. In the given case where Mary nds the mean height of the girls to be 162cm, and the mean height of the boys to be 7
158cm, she has insucent evidence to judge whether this supports her hypothesis or not. With the addition of the variance the calculation of the t value becomes possible, and this may enable her to be more condent of her hypothesis or attempt to formulate an alternative.