Mordell–Weil, Heights, and the Birch–Swinnerton-Dyer Conjecture

Job: develop the Mordell–Weil theorem / canonical height / descent / Selmer, the BSD conjecture (rank = \(L\)-order, refined BSD / Tamagawa / regulator / Sha), the Gross–Zagier / Kolyvagin partial BSD, and the Hasse–Weil \(L\)-function / modularity / \(p\)-adic BSD — with the BSD for \(y^2=x^3-x\) (rank 0) / the Heegner point (rank 1) as the worked example.

Core

The Mordell–Weil theorem and the canonical height

Theorem (Mordell–Weil). \(A(K)\cong A(K)_{\mathrm{tors}}\oplus\mathbb{Z}^r\) (the finitely-generated abelian group for an abelian variety \(A\) over a number field \(K\)).

The canonical height \(\hat{h}:A(K)\to\mathbb{R}_{\ge0}\) (the Néron–Tate quadratic form, \(\hat{h}(P)=\lim_{n\to\infty}h(nP)/n^2\), the kernel the torsion). The descent: the Selmer group \(\mathrm{Sel}^{(n)}(A/K)=\ker(H^1(K,A[n])\to H^1(K,A))\) controls \(A(K)/nA(K)\), the weak Mordell–Weil \(|A(K)/nA(K)|<\infty\), the descent / the rank computation.

Mordell–Weil / height / descent Description
Mordell–Weil \(A(K)\cong A(K)_{\mathrm{tors}}\oplus\mathbb{Z}^r\) (the finitely-generated)
Canonical height \(\hat{h}\) the Néron–Tate quadratic (\(\hat{h}(P)=\lim h(nP)/n^2\), the kernel the torsion)
Descent / Selmer \(\mathrm{Sel}^{(n)}=\ker(H^1(K,A[n])\to H^1(K,A))\) controls \(A(K)/nA(K)\)
Weak Mordell–Weil $

The Birch–Swinnerton-Dyer conjecture

Conjecture (BSD). For \(E/K\) an elliptic curve, \(\mathrm{rank}\,E(K)=\mathrm{ord}_{s=1}L(E,s)\) (the rank = the order of vanishing of the \(L\)-function at \(s=1\)).

Conjecture (refined BSD). The leading coefficient: $$ \lim_{s\to1}(s-1)^{-r}L(E,s)=\frac{|\mathrm{Sha}|\,\Omega_E\,R_E\,\prod c_p}{|E(K)_{\mathrm{tors}}|^2}, $$ the Tamagawa numbers \(c_p\), the regulator \(R_E\), the Tate–Shafarevich group \(|\mathrm{Sha}|\).

BSD / refined Description
BSD (rank) \(\mathrm{rank}\,E(K)=\mathrm{ord}_{s=1}L(E,s)\) (the rank = the \(L\)-order)
Refined BSD the leading coefficient $=
Tamagawa / regulator / Sha the \(c_p\) the Tamagawa, \(R_E\) the regulator, $
Analytic rank \(\mathrm{ord}_{s=1}L(E,s)\) the analytic rank (the BSD the algebraic = the analytic)

The Gross–Zagier / Kolyvagin and the partial BSD

Theorem (Gross–Zagier). The Heegner point / the \(L'(E,1)\ne0\) the rank 1: the Gross–Zagier the height of the Heegner point the \(L'\), the analytic rank 1 = the algebraic rank 1.

Theorem (Kolyvagin). The Euler system / the Heegner point the non-torsion \(\Rightarrow\) the rank 1 and \(|\mathrm{Sha}|<\infty\).

Gross–Zagier / Kolyvagin Description
Gross–Zagier the Heegner point / the \(L'(E,1)=c\cdot\hat{h}(P)\) (the height the \(L'\), the rank 1)
Kolyvagin the Euler system / the Heegner non-torsion \(\Rightarrow\) rank 1 and $
Partial BSD the Gross–Zagier / Kolyvagin the rank \(\le1\), the full BSD open

The \(L\)-function and the modularity

Hasse–Weil \(L\) / modularity / Iwasawa Description
Hasse–Weil \(L\)-function \(L(E/K,s)=\prod_v L_v(E/K,s)\) (the Euler product, the Frobenius at the good primes, the bad / the conductor)
Modularity the Wiles–Taylor–Wiles / the Breuil–Conrad–Diamond–Taylor: \(L(E,s)=L(f,s)\) for a modular form \(f\) of weight 2 (the modularity theorem)
Iwasawa / \(p\)-adic BSD the Mazur / the Greenberg, the \(p\)-adic \(L\)-function / the \(p\)-adic BSD

See it

BSD for the rank-0 elliptic curve \(y^2=x^3-x\) and the Heegner point for rank 1. The curve \(E:y^2=x^3-x\) (the congruent number curve, the conductor 32, the analytic rank 0, the \(L(E,1)\ne0\)): the BSD rank 0 (\(E(\mathbb{Q})=\{(0,0),(0,1),(0,-1),\mathcal{O}\}\) the rank 0, the \(L(E,1)\ne0\), the Kolyvagin / the rank 0). The Heegner point: for a rank-1 curve (e.g. \(E:y^2=x^3-157^2 x\) the congruent number 157, the rank 1), the Heegner point \(P\in E(\mathbb{Q})\) (the CM point on \(X_0(N)\) / the modular parametrization), the Gross–Zagier: \(L'(E,1)=c\cdot\hat{h}(P)\) (the \(L'\) the canonical height of the Heegner point), the rank 1 (the \(L'(E,1)\ne0\) the non-torsion \(P\), the Gross–Zagier / the Kolyvagin). The key point: the BSD the rank = the \(L\)-order, the rank 0 the \(L\ne0\) (the Kolyvagin), the rank 1 the \(L'\ne0\) (the Gross–Zagier Heegner) — the analytic / the algebraic the bridge.

Check

Show that the canonical height \(\hat{h}(P)=\lim_{n\to\infty}h(nP)/n^2\) is a quadratic form with kernel the torsion, and explain the descent.

The canonical height \(\hat{h}(P)=\lim_{n\to\infty}h(nP)/n^2\) (the Néron–Tate, the limit of the naive height \(h(nP)\) divided by \(n^2\)). The quadratic: \(\hat{h}(P+Q)+\hat{h}(P-Q)=2\hat{h}(P)+2\hat{h}(Q)\) (the parallelogram law, the Néron–Tate). The kernel: \(\hat{h}(P)=0\) iff \(P\) torsion (the height the positive definite modulo the torsion, the kernel the torsion). The descent: the Selmer group \(\mathrm{Sel}^{(n)}(A/K)=\ker(H^1(K,A[n])\to H^1(K,A))\) (the Kummer sequence \(0\to A[n]\to A\xrightarrow{n}A\to0\) the \(H^1\)), controls \(A(K)/nA(K)\) (the Selmer the finite, the \(A(K)/nA(K)\) the finite quotient). The weak Mordell–Weil: \(|A(K)/nA(K)|<\infty\) (the Selmer finite \(\Rightarrow\) the quotient finite), the descent / the rank computation (the rank \(r\) from the Selmer / the descent).

Explain the refined BSD: the leading coefficient \(=|\mathrm{Sha}|\,\Omega_E\,R_E\,\prod c_p/|E(K)_{\mathrm{tors}}|^2\), and identify the terms.

The refined BSD: the leading coefficient \(\lim_{s\to1}(s-1)^{-r}L(E,s)=\frac{|\mathrm{Sha}|\,\Omega_E\,R_E\,\prod c_p}{|E(K)_{\mathrm{tors}}|^2}\). The terms: \(|\mathrm{Sha}|\) the Tate–Shafarevich group (the locally trivial torsors, the Hasse failure, the conjecturally finite), \(\Omega_E\) the real / complex period (the Néron differential the integral), \(R_E\) the regulator (the canonical height pairing of the generators, \(\det(\hat{h}(P_i,P_j))\)), \(c_p\) the Tamagawa numbers (the local factors, the component groups at the bad primes), \(|E(K)_{\mathrm{tors}}|\) the torsion (the denominator). The refined BSD the deep: the leading coefficient the arithmetic / the geometric invariants (the Sha, the period, the regulator, the Tamagawa, the torsion), the full BSD the open (the partial the Gross–Zagier / Kolyvagin the rank \(\le1\)).

Pitfalls

  • The BSD is a conjecture (the rank = the \(L\)-order, the refined — the open, the partial the Gross–Zagier / Kolyvagin the rank \(\le1\)). Do not state the BSD as a theorem; state the proven cases (the rank \(\le1\), the Gross–Zagier / Kolyvagin).
  • The Sha the conjecturally finite (the BSD the \(|\mathrm{Sha}|\) the analytic, the finiteness the open). The \(|\mathrm{Sha}|\) in the refined BSD the conjectural (the finiteness the open, the analytic the BSD).
  • The modularity the Wiles (the elliptic / the rank-2, the general / the higher-dimensional the partial, the Fontaine–Mazur the converse). The modularity theorem (the Wiles / the BCDT) the elliptic / the rank-2, the higher-dimensional the partial.
  • The \(p\)-adic BSD / the Iwasawa (the Mazur / the Greenberg) the \(p\)-adic \(L\) / the \(p\)-adic BSD, the \(p\)-adic the analog, not the direct (the \(p\)-adic the different \(L\)-function).

Refs

  • Silverman, The Arithmetic of Elliptic Curves / Advanced Topics; Hindry–Silverman, Diophantine Geometry
  • Gross–Zagier, "Heegner points and derivatives of \(L\)-series" (Invent. 1986); Kolyvagin, "Euler systems" (1990)
  • Wiles, "Modular elliptic curves and Fermat's Last Theorem" (Annals 1995); Breuil–Conrad–Diamond–Taylor (2001)
  • Mazur, "On the arithmetic of elliptic curves" (the \(p\)-adic BSD); Greenberg, the Iwasawa / the \(p\)-adic BSD

Next

AG index · Foundations · Diophantine & Arakelov · Étale & Weil · Galois Reps & Fontaine–Mazur · Analytic NT / L-functions · Langlands