MA342J: Introduction to Modular Forms Tutorial 4, March 15 & 21 Transforming an elliptic curve into a Weierstrass form Recall that an elliptic curve over the field k is a nonsingular projective plane cubic curve CF (F ∈ k[X, Y, Z], homogeneous, irreducible, of degree 3) with a point O ∈ CF (k). Theorem. (i) For any a, b, c ∈ k the curve Ea,b,c : Y 2 Z = X 3 + aX 2 Z + bXZ 2 + cZ 3 , O = (0 : 1 : 0) . is an elliptic curve over k whenever the polynomial x3 + ax2 + bx + c has no multiple roots in k. (ii) Let E = (CF , O) be an elliptic curve over the field k. If the characteristics of the field char(k) 6= 2, 31 then E is isomorphic over k to an elliptic curve of the above form Ea,b,c . (This is called a Weierstrass form of E, though it is not unique.) Here is the algorithm to transform a cubic F (X, Y, Z) = 0 with a point O = (X0 : Y0 : Z0 ) into a Weierstrass form (or to discover that this cubic is not an elliptic curve). The tangent line to CF at O must intersect CF in three points. Moreover, the multiplicity of O itself is at least 2, therefore the third point (let us call it P ) is defined over k. Hence there are two possibilities: P = O and P 6= O. We will consider them below. Step 0. Performing a linear change of coordinates in P2 (k) if necessary, we can assume that O = (0 : 1 : 0). Then the curve is given by F3 (X, Z) + F2 (X, Z)Y + F1 (X, Z)Y 2 = 0 (Fi is homogeneous of degree i). Moreover, since the point O is nonsingular F1 (X, Z) = c1 X + c2 Z is nonzero, and performing the change of coordinates (X 0 : Y 0 : Z 0 ) = (X : Y : c1 X + c2 Z) we can assume that F1 (X, Z) = Z. Case 1: P = O. Step 2. Check that P = O implies that F2 (X, Z) = a1 XZ + a2 Z 2 (no X 2 term), then change the coordinates (X 0 : Y 0 : Z 0 ) = (X : Y + a21 X + a2 2 Z) to get equation of the form b0 X 3 + b1 X 2 Z + b2 XZ 2 + b3 Z 3 + ZY 2 = 0 . Step 3. Rescale the coordinates to get the Weierstrass form. Case 2: P 6= O. 1 char(k) is the characteristics of the field k. It is either 0, e.g. char(Q) = char(R) = char(C) = 0, or a prime number, e.g. char(Fp ) = p. 1 Step 2. Observe that it is possible to make a linear change of coordinates so that O = (0 : 1 : 0) and P = (0 : 0 : 1). In the affine coordinates Y x= X Z , y = Z the point P has coordinates (0, 0) and the curve is given by f3 (x, y) + f2 (x, y) + f1 (x, y) = 0 (fi is homogeneous of degree i). Let t = xy . Dividing the last equation by y we get y 2 φ3 (t) + yφ2 (t) + φ1 (t) = 0 φi (t) := fi (t, 1) and therefore √ −φ2 (t) ± δ , y = 2φ3 (t) δ = φ2 (t)2 − 4φ1 (t)φ3 (t) . √ Approximately, we want x0 = t and y 0 = δ to be our new (affine) coordinates since they determine x and y. Since φi (t) = fi (t, 1) is a polynomial of degree i in t, we have φ2 (t)2 − 4φ1 (t)φ3 (t) = b0 t4 + b1 t3 + b2 t2 + b3 t + b4 and equation of the curve in these new ”coordinates” is given by (y 0 )2 = b0 (x0 )4 + b1 (x0 )3 + b2 (x0 )2 + b3 x0 + b4 . Moreover, the condition δ(t) = 0 means that the line of the slope t through the origin P is a tangent line to a curve to some point. One of these points is O and it is easy to check that the corresponding value is t = 0. It follows that actually b4 = 0. Therefore in the coordinates 1 1 y = = 0 x t x y0 2φ3 (t)y + φ2 (t) = = = ... (x0 )2 t2 x00 = y 00 the curve has the equation (y 00 )2 = b0 + b1 x00 + b2 (x00 )2 + b3 (x00 )3 . Step 3. Rescale the coordinates y 00 = b23 y 000 , x00 = b3 x000 to get the Weierstrass form. Step 4. Write the final transformation in projective coordinates and check that it is a bijection from CF to Ea,b . Observe that in the case P = O this transformation was linear (and therefore was obviously 1-to-1) while here it is given by homogeneous polynomials of higher degree. Question 1: Write the curve X3 + Y 3 = Z3 O = (1 : 0 : 1) 2 in a Weierstrass form. Question 2: Let t ∈ Q be a parameter. Consider the cubic curve (X + Y + Z)(XY + XZ + Y Z) = tXY Z . One can easily find 6 rational points on this curve, namely (1 : 0 : 0) , (0 : 1 : 0) , (0 : 0 : 1) , (1 : −1 : 0) , (1 : 0 : −1) , (0 : 1 : −1) . Up to simple permutations of coordinates, there are then two elliptic curves to be considered: with O = (0 : 0 : 1) and O = (0 : −1 : 1). Write the Weierstrass form for every case. For which values of t the curves are not elliptic. (Here the answer should be independent of the choise of O because the reason would be that the above cubic curve is either singular or reducible.) 3