Scribd
Upload
55 views2 pages

Number Theory: Elliptic Curves, Problem Sheet 3

This document is a problem sheet for a number theory course on elliptic curves. It provides 7 questions to give students background and context for statements made in class. Question 1 asks students to prove there are solutions to 3x^3 + 4y^3 = 5 in the p-adic fields Qp for any prime p, using Hensel's lemma. Question 4 begins a proof that the equation 3x^3 + 4y^3 = 5 has no rational solutions, using some Galois theory. Question 7 asks students to check that the discriminant of the cubic equation x^3 + ax + b is 4a^3 + 27b^2.

Uploaded by

sticker592
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
55 views2 pages

Number Theory: Elliptic Curves, Problem Sheet 3

This document is a problem sheet for a number theory course on elliptic curves. It provides 7 questions to give students background and context for statements made in class. Question 1 asks students to prove there are solutions to 3x^3 + 4y^3 = 5 in the p-adic fields Qp for any prime p, using Hensel's lemma. Question 4 begins a proof that the equation 3x^3 + 4y^3 = 5 has no rational solutions, using some Galois theory. Question 7 asks students to check that the discriminant of the cubic equation x^3 + ax + b is 4a^3 + 27b^2.

Uploaded by

sticker592
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

KMB, 18/11/10

Number Theory: Elliptic Curves, Problem Sheet 3


The questions on this sheet are not logically essential for your understanding of the course, and some do not even test your ability to apply theorems from the coursethey are just there to give you some background to some of the statements I have made recently in the course. I will perhaps make use of Q3 and Q7 in the course, but this doesnt mean you have to do the questions, it just means that you have to have read the statements. Q1 is quite good, it wont teach you too much about cubics, but it will make you good at applying Hensels lemma, which is important. 1) Here is a sketch of a proof there are solutions to 3x3 + 4y 3 = 5 in Qp for any prime p. I claimed this without proof in the course. If you do this question then it will teach you a bit about the power of Hensels lemma. Note that there are obviously real solutions to 3x3 + 4y 3 = 5 and hence this question shows that there are solutions to 3x3 + 4y 3 = 5 in any completion of Q. a) Prove that there exists y Q3 such that 4y 3 = 5. Deduce that there is a solution to 3 3x + 4y 3 = 5 in Q3 . b) Prove that there is x Q5 such that 3x3 = 1. Deduce that there is a solution to 3x3 +4y 3 = 5 in Q5 . Clearly strategies like the above may well deal with any xed p that you can think of. But how to deal with all p at once? There are pure thought methods, coming from deeper results about the existence of mod p solutions to cubic equations, but here is an elementary approach that works for the cubic in question. Let p be any prime that is not 3 or 5. c) Prove that either 3, 5, 15 or 45 is a cube mod p. Hint: consider (Z/pZ) quotiented out by the subgroup of cubes. If this doesnt have order 1 then it has order 3. d) (i) If 3 is a cube mod p then check that there is a Qp -solution to 3x3 + 4y 3 = 5 with y = 1. (ii) If 5 is a cube mod p then check that there is a solution with x = y . (iii) If 45 is a cube mod p then check that there is a solution with y = 0. (iv) If 15 is a cube mod p then check that there is a solution with y = 5/7. Conclude that there is a solution to 3x3 + 4y 3 = 5 in Qp for any prime p. In fact this proof is elementary but rather articial and if you knew the standard but tricky fact that smooth cubic curves over nite elds always had points then the result (and many more like it) follows easily from Hensels lemma 2) Here as promised is the change of coordinates which takes an irreducible cubic with a given smooth point O into the form y 2 + a1 xy + a3 y (x3 + a2 x2 + a4 x + a6 ) = 0. We work over a eld k . The exercise is to check the details. Firstly draw the tangent line to the cubic at O. The tangent line will meet the cubic at three points, at least two of which will be O. CASE 1: the third point of intersection is also O. In this case a linear change in coordinates (i.e., X1 = aX + bY + cZ , Y1 = eX + f Y + gZ , Z1 = hX + iY + jZ ) will suce: change variables of the homogeneous equation by a linear transformation so that O goes to [0 : 1 : 0] and the tangent line goes to the line Z1 = 0 in projective 2-space, that is, the line at innity. Check that this does it. CASE 2: the third point of intersection is P , a point not equal to O. Make a linear change of variables so that P is at (0, 0) and that the tangent to O is the y -axis in the x, y -plane. Lets work with the inhomogeneous cubic f (x, y ) = 0. The fact that (0, 0) is on the cubic implies that the constant term of f is zero, so we can write f (x, y ) = f1 (x, y ) + f2 (x, y ) + f3 (x, y ) where fi is homogeneous of degree i.

We know that the intersection of f = 0 with the line x = 0 (that is, the y -axis) is (0, 0) with multiplicity 1 and some other root (namely O) with multiplicity 2. Hence, setting x = 0, we deduce that the quadratic f1 (0, 1) + yf2 (0, 1) + y 2 f3 (0, 1) has a double root and hence f2 (0, 1)2 = 4f1 (0, 1)f3 (0, 1). Now draw lines through the origin. The line y = tx (think of t as a constant) meets the curve at three points whose x-coordinates are the three roots of xf1 (1, t) + x2 f2 (1, t) + x3 f3 (1, t) so this line hits a k -point other than (0, 0) if and only if the quadratic x2 f3 (1, t) + xf2 (1, t) + f1 (1, t) has roots in k , which happens i f2 (1, t)2 4f1 (1, t)f3 (1, t) is a square in k . Hence if we write s2 = f2 (1, t)2 4f1 (1, t)f3 (1, t) then the formula for the roots of a quadratic equation give us, for every solution to this equation, a root of x2 f3 (1, t) + xf2 (1, t) + f1 (1, t), namely x = (f2 (1, t) + s)/2f3 (1, t). The punchline however is that G(t) := f2 (1, t)2 4f1 (1, t)f3 (1, t), which looks like an equation of degree 4 in t, is actually of degree 3 because f2 (0, 1)2 = 4f1 (0, 1)f3 (0, 1). Hence the cubic has now become an equation of the form s2 = G(t) with G a cubic, which is what we were after. 3) An example of the easy case of the algorithm presented in Q2. Let d be a non-zero constant, and lets consider the cubic F (X, Y, Z ) = X 3 + Y 3 + dZ 3 , over a eld of characteristic not equal to 2 or 3. (i) Prove that F = 0 has no singular points. (ii) Prove that the point [1 : 1 : 0] is a point of inexion. (iii) Write down a linear change of coordinates (i.e., X1 = aX + bY + cZ , Y1 = eX + f Y + gZ , Z1 = hX + iY + jZ ) that takes this point to [0 : 1 : 0] and which takes the tangent line at this point to the line Z1 = 0. If you did it right, and then tidy up and complete the square and cube if necessary, you should 3 3 be left with an equation of the form Y12 Z1 = X1 432d2 Z1 , whose corresponding dehomogenisation 2 3 2 is y = x 432d . These are exactly the kinds of curves we will deal with later. 4) Here is the beginning of a proof that 3x3 + 4y 3 = 5 has no rational solutions, but it assumes a little Galois theory, or some common sense. Say there were a solution. Then by clearing denominators we get integers u, v , w, not all zero (and hence all non-zero), such that 3u3 + 4v 3 + 5w3 = 0 (change the sign of w if necessary). Now set = e2i/3 , a non-trivial cube root of 1, and work in the eld Q() = Q( 3). Set = 3u3 + 4v 3 + 52 w3 and let be the complex conjugate of (note that the conjugate of is 2 ). Check that + = 9u3 (recall + 2 = 1), 3 3 that + 2 = 15w3 and that 2 + = 12v 3 . Deduce that 3 + = 60 , where = 3uvw. 3 3 Now P := (/, / ) is a point on x + y = 60 dened over Q( 3), and so is its complex conjugate P . Draw the line through P and P ; this meets the cubic x3 + y 3 = 60 at P and P and a third point Q; applying complex conjugation to everything we deduce that Q = Q and hence that Q has rational coordinates. Furthermore, the point Q is not the point at innity, as the general line through innity is of the form x + y = c with c constant, and + cannot be rational. We conclude that if there is a rational solution to 3x3 +4y 3 = 5 then there is a rational solution to x3 + y 3 = 60. By Q3, we deduce that there is a rational solution to s2 = t3 60.432. Later on in the course well see how to prove that the only solution to this equation over the rationals is the point at innity, and nally our proof of the statement 3x3 + 4y 3 = 5 has p-adic points for all p, and real points, but no rational points will be complete. 5) It turns out that there are other equations that can be put into the form y 2 = f (x) with f a cubic. For example, perhaps surprisingly, the equation y 2 = g (x) with g (x) of degree 4, can be put into this form, as long as a smooth point is known. Read the proof on p35 of Cassels book. 6) Similarly, the intersection of two quadric surfaces, that is, the simultaneous solutions to two homogeneous equations of degree 2 in 4 variables, can be put into this form, if a smooth point is known. Read the proof on p36 of Cassels book. 7) Check that disc(x3 + ax + b) = 4a3 +27b2 . One can do this for example by calling the roots , and and then noting + + = 0 etc, and then explicitly checking that ( )2 ( )2 ( )2 agrees with 4a3 + 27b2 up to a minus sign (I think that the general rule is that discriminants are only dened up to sign).

You might also like