Processing math: 15%

Saturday, May 30, 2015

Inequality

Mathematics Olympiad ~ Vinod Singh, Kolkata Problem #3 Find all real numbers x for which 3xx+1>12 Let f(x) =3xx+1. First note that f(x) is defined for 1x3 f(x)=123x12x+1=(123x+12x+1)<0f(x) is strictly decreasing Now f(1)=2>12 and f(3)=2<12 Since f(x) is continuous, at least one x (1,3) suct that f(x)=12 f(x)=123xx+1=1264x2128x+33=0x=1±318 but x=1+318 does not satisfy 3xx+1=12 Check yourself! So the only solution is x=1318 Since f(x) is strictly decreasing, the given inequality is true for x[1,1318)

Matrices & Determinants

Mathematics Olympiad ~ Vinod Sing, Kolkata Problem #1 If A and B are different matrices satisfying A3=B3 and A2B=B2A, find det(A2+B2) Since A and B are different matrices ABO, Now (A2+B2)(AB)=A3A2B+B2AB3 =O since A3=B3 and A2B=B2A This shows that (A2+B2) has a zero divisor, so it is not invertible hence det(A2+B2)=0

Indian Statistical Institute B.Math & B.Stat : Number Theory

Indian Statistical Institute : Number Theory Let k be any odd integer greater that 1. Then show that 1k+2k+3k++2006k is divisible by 2013021. We will prove the general case. Let S=1k+2k+3k++nk where n2 and nN. 2S=1k+2k+3k++nk+1k+2k+3k++nk=(1k+nk)+(2k+(n1)k)++(nk+1k) Using the result, nk+mk is always divisible by n+m if k is odd we see that 2S is divisible by (n+1) Again 2S=1k+2k+3k++nk+1k+2k+3k++nk =(1k+(n1)k)+(2k+(n2)k)++((n1)k+1k)+2nk Using the same result and noting that 2nk is divisible by n we see that 2S is divisible by n. Now both n an n+1 divides 2S and g.c.d(n.n+1)=1 we see that n(n+1) divides 2S this implies n(n+1)2 divides S. ( n(n+1) is always even) In the given problem n=2006, thus 1k+2k+3k++2006k is divisible by 2006(2006+1)2=2013021

Indian Statistical Institute B.Math & B.Stat : Combinatorics

Indian Statistical Institute : Combinatorics In how many ways one can choose three distinct numbers from the set {1,2,3,,19,20} such that their product is divisible by 4? We partition the set {1,2,3,,19,20} into three disjoint sets S1={4,8,12,16,20},S2={2,6,10,14,18},S3={1,3,5,7,9,11,13,15,17,19} Three selected (distinct) numbers will not be divisible by 4 iff all the three numbers are selected form S3 or two of them are selected from S3 and one of them from S1. Numbers of such numbers are \binom{10}{3}+\binom{5}{2} \times \binom{5}{1} = 345 So numbers of selection such that their product is divisible by 4 is \binom{20}{3}-345= 795

Indian Statistical Institute B.Math & B.Stat : Number Theory

Indian Statistical Institute : Number Theory Find the digit at the unit place of \big(1!-2!+3!-\dots \dots +25!\big )^{\big(1!-2!+3!-\dots \dots +25!\big )} First note that k! \equiv 0 (mod\ 10) for all k \geq 5 , k \in \mathbb{N} So, 5!-6!+7!-\dots \dots +25! \equiv 0 (mod\ 10) and 1!-2!+3!-4! = -19 \equiv 1 (mod\ 10) (Using the property of congruences). Using the above two congruences \big(1!-2!+3!-\dots \dots +25!\big ) \equiv 1 (mod\ 10) So, \big(1!-2!+3!-\dots \dots +25!\big )^{\big(1!-2!+3!-\dots \dots +25!\big )} \equiv 1^{\big(1!-2!+3!-\dots \dots +25!\big )} \equiv 1 (mod 10) giving 1 as the last digit. Let a \equiv a' (mod\ m) and b \equiv b' (mod\ m), then important properties of congruences include the following, where \implies means "implies": 1. Reflexivity: a\equiv a (mod- m). 2. Symmetry: a\equiv b (mod\ m) \implies b\equiv a (mod\ m). 3. Transitivity: a\equiv b (mod\ m) and b \equiv c (mod\ m)\implies a\equiv c (mod\ m). 4. a+b \equiv a'+b' (mod\ m) 5. a-b\equiv a'-b' (mod\ m). 6. ab\equiv a'b' (mod\ m). 7. a\equiv b (mod\ m)\implies ka \equiv kb (mod\ m). 8. a\equiv b (mod\ m)\implies a^n\equiv b^n (mod\ m). 9. ak\equiv bk (mod\ m)\implies a\equiv b \big(mod\ \frac{m}{(k,m)}\big), where (k,m) is the greatest common divisor. 11. If a \equiv b (mod\ m), then P(a) \equiv P(b) (mod\ m), for P(x) a polynomial with integer coefficients.

Friday, May 29, 2015

Indian Statistical Institute B.Math & B.Stat : Number Theory

Indian Statistical Institute B.Math & B.Stat Find the sum of all even positive divisors of 1000. 1000 = 2^3 \times 5^3 . Now any even divisor of 1000 must contain a factor of the form 2^j where j \in \{1,2,3\}. We note that 2 and 5 are the only prime factors of 1000 , so a even factor must be of the form 2^j\times 5^i where j \in \{1,2,3\} and i \in \{0,1,2,3\}. So the required sum is \sum_{j=0}^{3} \sum_{i=1}^{3} 2^i \times 5^j =\sum_{j=0}^{3} 14 \times 5^j = 14 \times \frac{5^4-1}{5-1} = 2184

Indian Statistical Institute B.Math & B.Stat : Integration

Indian Statistical Institute B.Math & B.Stat Let \alpha and \beta be two positive real numbers. For any integer n>0, define a_n = \int_{\beta}^{n} \frac{\alpha}{u(u^\alpha+2+u^{-\alpha})}du. Then find \lim_{n \to \infty} a_n . Multiplying u^{\alpha-1} to the numerator and denominator of the integrand, we have a_n = \int_{\beta}^{n} \frac{\alpha u^{\alpha-1}}{u\times u^{\alpha-1}(u^\alpha+2+u^{-\alpha})}du Substituting u^{\alpha}=t we get the transformed integral as a_n = \int_{\beta^\alpha}^{n^\alpha}\frac{dt}{(t+1)^2}dt = \frac{n^\alpha-\beta^\alpha}{(1+\beta^\alpha)(1+n^\alpha)} Therefore, \lim_{n \to \infty} a_n = \lim_{n \to \infty} \frac{n^\alpha-\beta^\alpha}{(1+\beta^\alpha)(1+n^\alpha)}= \lim_{n \to \infty} \frac{1-\big({\frac{\beta}{n}}\big)^\alpha}{(1+\beta^\alpha)(1+\frac{1}{n})}=\frac{1}{1+\beta^\alpha}

Indian Statistical Institute B.Math & B.Stat : Number Theory

Indian Statistical Institute B.Math & B.Stat Let S be the set of all integers k, 1 \leq k \leq n, such that g.c.d(k,n)=1. What is the arithmetic mean of the integers in S?. First note that |S| = \phi(n) . Now let k \in S then there exists intergers u,v such that ku+nv=1\dots (A). The integer n-k \in \{1,2,\dots,n-1\} because k can never be equals n, for g.c.d(n,n)=n and k \in S. We will now show that g.c.d(n-k,n)=1. Adding -nu to both sides of (A) we get -nu+ku+nv=-nu+1 \implies -u(n-k)+(v+u)n = 1 \implies g.c.d(n-k,n)=1 So, for k \in \{1,2,\dots,n-1\} if k \in S \implies n-k \in S Thus S can be written in the form, S=\{k_1,k_2,k,.....,k_r,n-k_r,......,n-k_2,n-k_1\} where |S| = \phi(n) Clearly the sum of the elements of S is (k_1+n-k_1)+(k_2+n-k_2)+\dots+(k_r+n-k_r) = \frac{n\phi(n)}{2} ( Pairing reduces the terms to half the original (\phi(n))). arithmetic mean =\frac{\frac{n\phi(n)}{2}}{\phi(n)} = \frac{n}{2}

Indian Statistical Institute B.Math & B.Stat : Inequality

Indian Statistical Institute B.Math & B.Stat If a,b,c \in (0,1) satisfy a+b+c=2, prove that \frac{abc}{(1-a)(1-b)(1-c)} \geq 8. Let p = 1-a, q = 1-b, r = 1-c . p+q+r= 3-(a+b+c)=1.Clearly p,q,r are positive. Substituting in the given inequality, it transforms to \frac{(1-p)(1-q)(1-r)}{pqr} \geq 8 \iff {(1-p)(1-q)(1-r)} \geq 8{pqr} \iff 1-(p+q+r)+qr+rp+pq-pqr \geq 8{pqr} \iff qr+rp+pq-pqr \geq 8{pqr} \iff qr+rp+pq \geq 9{pqr} \iff \frac{1}{p}+\frac{1}{q}+\frac{1}{r} \geq 9 Thus proving the above inequality reduces to proving \frac{1}{p}+\frac{1}{q}+\frac{1}{r} \geq 9 subjected to p+q+r=1 Since p,q,r are positive, apllying A.M \geq H.M we have \frac{p+q+r}{3} \geq \frac{3}{\frac{1}{p}+\frac{1}{q}+\frac{1}{r}} Noting p+q+r=1, we have \frac{1}{p}+\frac{1}{q}+\frac{1}{r} \geq 9

Thursday, May 28, 2015

Indian Statistical Institute B.Math & B.Stat : Combinatorics

Indian Statistical Institute B.Math & B.Stat A point P with coordinates (x,y) is said to be good if both x and y are positive integers. Find the number of good points on the curve xy=27027. First note that, 27027 = 3^3 \times 13 \times 11 \times 7 . Now for any good point on the curve xy=27027, x must be of the form 3^p \times 13^q \times 11^r \times 7^s and y must be of the form 3^{p'} \times 13^{q'} \times 11^{r'} \times 7^{s'}. Where p+p'=3,q+q'=1,r+r'=1 and s+s'=1 Now considers the coordinates (3^3 \times ?, 3^0 \times ?),(3^2 \times ?, 3^1 \times ?),(3^1 \times ?, 3^2 \times ?),(3^0 \times ?, 3^3 \times ?) where in the place of ? in the x- coordinate we can have any combinations of the products 13^q \times 11^r \times 7^s and in the place of ? in the y- coordinate we can have any combinations of the products 13^{q'} \times 11^{r'} \times 7^{s'} such that q+q'=1,r+r'=1 and s+s'=1. Note that for the equation q+q'=1 the possible non-negative solutions are (1,0), (0,1) . Similarly for the other two equations r+r'=1 and s+s'=1. Counting the combinations for ? is the x-coordinate for the coordinate (3^3 \times ?, 3^0 \times ?) we can have CASE I : All the three numbers 13,11,7 appears which can be done in \binom {3}{3} ways. ? in the y- coordinate the must contain 13^0,11^0,7^0. CASE II : Any two of the three numbers 13,11,7 appears which can be done in \binom {3}{2} ways. ? in the y- coordinate the must contain 13 or 11 or 7. CASE III : Any one of the three numbers 13,11,7 appears which can be done in \binom {3}{1} ways. ? in the y- coordinate the must contain 13,11 or 11,7 or 7,13. CASE IV : None of the three numbers 13,11,7 appears which can be done in \binom {3}{0} ways. ? in the y- coordinate the must contain 13,11,7. Giving a total of \binom {3}{3}+ \binom {3}{2}+ \binom {3}{1}+ \binom {3}{0}=8 possibilities. Similarly we have 8 possibilities for each of the coordinates of the form (3^2 \times ?, 3^1 \times ?),(3^1 \times ?, 3^2 \times ?),(3^0 \times ?, 3^3 \times ?), giving in total 4\times8=32 good coordinates Another problem of same flavor form Indian- Statistical- Institute What is the number of ordered triplets (a, b, c), where a, b, c are positive integers (not necessarily distinct), such that abc = 1000? First note that, 1000 = 2^3 \times 5^3 . Any ordered triplet with the given condition must be of the form (2^l\times5^m,2^p\times5^q,2^r\times5^s) where l+p+r=3 and m+q+s=3. Number of non-negative solutions of the above equations is \binom {3+3-1}{2} = 10 in both cases. Now consider one ordered triplet (2^1\times5^m,2^0\times5^q,2^2\times5^s) ( 10 such triplet are possible, varying powers of 2 subjected to l+p+r=3 ), for this triplet now we can vary m,q,s subjected to m+q+s=3 in 10 ways.Thus in total 10\times10=100 ordered triplets are possible. Another good problem with a kick of Derangement!! There are 8 balls numbered 1,2,\dots,8 and 8 boxes numbered 1,2,\dots,8. Find the number of ways one can put these balls in the boxes so that each box gets one ball and exactly 4 balls go in their corresponding numbered boxes. 4 balls can be selected in \binom{8}{4} ways and for each for such selection, say \{Ball_2,Ball_5,Ball_3,Ball_7\} the balls can go in their corresponding boxes in exactly one way (why?). Now the remaining 4 balls \{Ball_3,Ball_4,Ball_1,Ball_8\} has to be put in the boxes in such a way none of the balls goes to the box of same number. Since, exactly 4 balls go in their corresponding numbered boxes. Which is nothing but D_4, so total number of ways is \binom{8}{4} \times D_4 where D_4 = 4!\big( 1 - \frac{1}{1!}+\frac{1}{2!}-\frac{1}{3!}+\frac{1}{4!}\big) = 9 In general D_n is Derangement of n objects.

Wednesday, May 27, 2015

Indian Statistical Institute B.Math & B.Stat : Combinatorics

Indian Statistical Institute B.Math & B.Stat Using only the digits 2, 3 and 9, how many six digit numbers can be formed which are divisible by 6? First note that a number is divisible by 6, iff it is divisible by 2 and by 3. (2 and 3 being co-prime) The above argument clearly shows that the number 222222 is divisible by 6. Use divisibility test of 2 and 3. First note that a number will be divisible by 2, iff the digit at the unit place is 2. Now observe that, if the number which is divisible by 2 has to be divisible by 3, the sum of the digits must be divisible by 3. So the 5 places of the number (except the digit at the unit place, which is 2) is a combination of the digits 2, 3 and 9 such that the sum including the digit at the unit place is divided by 3. Now observe that among the 5 places to be filled, exactly two places can be the digit 2. (Convince yourself why?! thin if not...) And the remaining three places can be filled by either 3 or 9. Thus giving \binom {5} {2} \times 2\times2\times2 possibilities with desired condition. therefore, total number of numbers = \binom {5} {2} \times 2\times2\times2+1 = 80+1=81

Indian Statistical Institute B.Math & B.Stat : Integration

Indian Statistical Institute B.Math & B.Stat Evaluate: \int_\frac{1}{2014}^{2014} \frac{tan^{-1}x}{x} dx Let I=\int_\frac{1}{2014}^{2014} \frac{tan^{-1}x}{x} dx, Put x=\frac{1}{t} \implies I = -\int_{2014}^\frac{1}{2014} \frac{tan^{-1}\frac{1}{t}}{t} dt =\int_\frac{1}{2014}^{2014} \frac{tan^{-1}\frac{1}{t}}{t} dt = \int_\frac{1}{2014}^{2014} \frac{tan^{-1}\frac{1}{x}}{x} dx \mathbb{Therefore,} 2I=\int_\frac{1}{2014}^{2014} \frac{tan^{-1}x}{x} dx+\int_\frac{1}{2014}^{2014} \frac{tan^{-1}\frac{1}{x}}{x} dx =\int_\frac{1}{2014}^{2014} \frac{tan^{-1}x+tan^{-1}\frac{1}{x}}{x} dx =\int_\frac{1}{2014}^{2014} \frac{\frac{\pi}{2}}{x} dx =\frac{\pi}{2}\int_\frac{1}{2014}^{2014} \frac{dx}{x} =\frac{\pi}{2}[\log x]_{\frac{1}{2014}}^{2104} =\frac{\pi}{2}\big[\log 2014-\log \frac{1}{2014}\big] =\frac{\pi}{2}\log 2014^2 =\pi\log 2014

Monday, May 25, 2015

Indian Statistical Institute B.Math & B.Stat : Continuity

Let P: \mathbb{R} \rightarrow \mathbb{R} be a continuous function such that P(x)=x has no real solution. Prove that P(P(x))=x has no real solution. If possible let, P(P(x))=x has a real solution for x=x_0. Then P(P(x_0))=x_0\dots(1) Now let, P(x_0)=y_0 \implies P(y_0)=x_0 using (1) Note that x_0 \neq y_0, otherwise we will have a solution to the equation P(x)=x! A contradiction to the hypothesis. Without loss of generality assume that x_0 < y_0 Construct a function Q:[x_0,y_0]\rightarrow \mathbb{R} where Q(x)=P(x)-x,since P is given to be continuous on \mathbb{R}, Q is continuous on [x_0,y_0]. Observe that, Q(x_0)=P(x_0)-x_0=y_0-x_0 > 0 and Q(y_0)=P(y_0)-y_0=x_0-y_0 < 0 \implies Q(x_0)Q(y_0) < 0 \implies there exists a point c\in [x_0,y_0] such that Q(c)=0 using Intermediate-Value-Theorem \implies P(c)-c=0 \implies P(c)=c for a real value, which contradicts the hypothesis, thus the assumption P(P(x))=x has a real solution is not tenable.

Saturday, May 23, 2015

Combinatorics :Indian Statistical Institute B.Math & B.Stat

Indian Statistical Institute B.Math & B.Stat How many three-digits numbers of distinct digits can be formed by using the digits 1,2,3,4,5,9 such that the sum of digits is at least 12? First note that since the digits must be distinct there must be no repetition and further note that the maximum sum of such three digits is 9+5+4=18. We will find all such possible by considering their sum of the digits. Case I: Sum of the digits is 12. In this case the selection of digits can be \{1,2,9\},\{3,4,5\} So, a total of 3!+3!=12 three-digits numbers possible. Case II: Sum of the digits is 13. In this case the selection of digits can be \{1,3,9\} So, a total of 3!=6 three-digits numbers possible. Case III: Sum of the digits is 14. In this case the selection of digits can be \{1,4,9\},\{2,3,9\} So, a total of 3!+3!=12 three-digits numbers possible. case IV: Sum of the digits is 15. In this case the selection of digits can be \{1,5,9\},\{2,4,9\} So, a total of 3!+3!=12 three-digits numbers possible. Case V: Sum of the digits is 16. In this case the selection of digits can be \{2,5,9\},\{2,5,9\} So, a total of 3!+3!=12 three-digits numbers possible. Case VI: Sum of the digits is 17. In this case the selection of digits can be \{3,5,9\} So, a total of 3!=6 three-digits numbers possible. Case VII: Sum of the digits is 18. In this case the selection of digits can be \{4,5,9\} So, a total of 3!=6 three-digits numbers possible. Adding all the cases we have 12+6+12+12+12+6+6=66, three-digits numbers of distinct digits can be formed.

Thursday, May 21, 2015

Complex Numbers :Indian Statistical Institute B.Math & B.Stat

Indian Statistical Institute B.Math & B.Stat Let \omega be the complex cube root of unity. Find the cardinality of the set S where S = \{(1+\omega+\omega^2+\dots+\omega^n)^m | m,n = 1,2,3,......\} n must be of the form 3k, 3k+1 or 3k+2 where k \in N when n is multiple of 3,1+\omega+\omega^2+\dots+\omega^n = 0 whence (1+\omega+\omega^2+\dots+\omega^n)^m = 0 \forall n,m \in N when n is of the form 3k+1,1+\omega+\omega^2+\dots+\omega^n = 1 whence (1+\omega+\omega^2+\dots+\omega^n)^m = 1^m=1 \forall n,m \in N when n is of the form 3k+2,1+\omega+\omega^2+\dots+\omega^n = 1+\omega(**) whence (1+\omega+\omega^2+\dots+\omega^n)^m = (1+\omega)^m =(-\omega)^{2m} \forall n,m \in N In this case the possible values are possible values of (-\omega)^{2m} which are -1,1,\omega,-\omega,\omega^2,-\omega^2, note m varies over the set of Natural numbers Combining the three cases we see that S = \{0,-1,1,\omega,-\omega,\omega^2,-\omega^2\} , therefore |S|=7. (**) To understand these first observe that \omega^n=1,\omega,\omega^2 for any natural number n. So when n is of the for 3k+2 we can couple the three consecutive recurring terms 1,\omega,\omega^2 to get the sum as 0. After that we are left with two more terms 1 and \omega, since the series 1+\omega+\omega^2+\omega^3+\omega^4+\omega^5\dots+\omega^n has repeated terms after every 3 terms (similar to the first three terms). Similarly for the othere cases the same argument follows. This concludes the result .

Saturday, May 9, 2015

Common terms of two A.P Series : Indian Statistical Institute B.Math & B.Stat

Indian Statistical Institute B.Math & B.Stat Consider the two arithmetic progressions 3,7,11,\dots,407 and 2,9,16,\dots,709. Find the number of common terms of these two progressions. Let a_n and a_m be the last terms of the progressions respectively \Rightarrow 407 = 3+(n-1)4 and 709 = 2+(m-1)7 Solving we get, n,m = 102. To find the common terms, assume that the n^{th} term of the first progression is equal to the m^{th} term of the second progression. \Rightarrow 3+(n-1)4=2+(m-1)7 \Rightarrow 3+4n-4-2+7=7m \Rightarrow 4(n+1)=7m, where n,m \in \{1,2,3,\dots,102\} R.H.S is a multiple of 7, while L.H.S is 4(n+1). Since g.c.d(4,7)=1 L.H.S will be multiple of 7, iff n+1 is a multiple of 7. \Rightarrow n = 6,13,20,\dots Again since n is bounded by 102. The largest possible value of n is 97. So, n \in \{6,13,20,\dots\,97} Which has 14 terms. Thus the number of common terms of the progression is 14

Application of Rolle's Theorem

Let a_0,a_1,a_2 and a_3 be real numbers such that a_0+\frac{a_1}{2}+\frac{a_2}{3}+\frac{a_3}{4}=0. Then show that the polynomial f(x)=a_0+a_1x+a_2x^2+a_3x^3 has at least one root in the interval ( 0 , 1 ) . Consider the polynomial g(x) = a_0x+\frac{a_1}{2}x^2+\frac{a_2}{3}x^3+\frac{a_3}{4}x^4 Clearly g(0)=0 and g(1) = a_0+\frac{a_1}{2}+\frac{a_2}{3}+\frac{a_3}{4} = 0 { given in the problem } Since g(x) is a polynomial it is continuous in [0,1] and differentiable in (0,1) Thus by Rolle's Theorem, g'(x) = 0 for at least one x \in (0,1) \Rightarrow a_0+a_1x+a_2x^2+a_3x^3 = 0 for at least one x \in (0,1)

Tuesday, May 5, 2015

Problem from Indian Statistical Institute: B.Stat. (Hons.)

Let P(x) be a polynomial of degree 11 such that P(x) = \frac{1}{x+1}, for x = 0,1,2,\dots,11. Find the value of P(12) Solution: Let f(x) = (x+1)P(x)-1, clearly f(x) is a polynomial of degree 12. Now for x \in \{0,1,2,\dots,11\}, f(x) = (x+1)P(x)-1=\frac{x+1}{x+1}-1=1-1=0 This shows that f(x) vanishes at the points x = 0,1,2,\dots,11. f(x) being of degree 12, the above statement assures that x = 0,1,2,\dots,11. are the possible roots of f(x) Therefore f(x)=(x+1)P(x)-1=a_{0}(x-0)(x-1)(x-2)\dots(x-11) Letting x=-1 in the above equality, we have -1=a_{0}(-1)(-2)(-3)\dots(-12) \Rightarrow a_{0} = \frac{-1}{12!} Therefore f(x)=(x+1)P(x)-1=\frac{-1}{12!}(x-0)(x-1)(x-2)\dots(x-11) Now, letting x=12 we have 13P(12)-1=\frac{-1}{12!}(12)(11)(10)\dots(1)=\frac{-12!}{12!}=-1 \Rightarrow 13P(12)-1=-1 \Rightarrow P(12) = 0

Monday, May 4, 2015

Sunday, May 3, 2015

Remainder Theorem

The term containing the highest power of x in the polynomial f(x) is 2x^4. Two of the roots of the equation f(x)=0 are -1 and 2. Given that x^2-3x+1 is a quadratic factor of f(x), find the remainder when f(x) is divided by 2x-1. Since degree of f(x) is 4 and x^2-3x+1 is a factor of f(x), it can be written as product of two quadratics Therefore, f(x) = ( x^2-3x+1 ) (ax^2+bx+c ) . Again since, 2x^4 is the leading term a, must be equal to 2 So, f(x) = ( x^2-3x+1 ) (2x^2+bx+c ) . Given -1 and 2 are the roots of f(x) \Rightarrow f(-1)=0 & f(2)=0 Note that x^2-3x+1 does not vanishes at x = -1 , 2 \Rightarrow 2x^2+bx+c must vanishes at this two points. \Rightarrow 2-b+c = 0 , 8+2b+c=0 Solving the equations we get b=-2, c=-4 Thus f(x) = ( x^2-3x+1 ) (2x^2-2x-4 ) . Required remaninder is f(\frac{1}{2}) = \frac{9}{8}