Sato-Tate conjecture

Intuition Beginner

Pick a non-CM elliptic curve over the rationals — for example $y^2=x^3+x+1$ — and a prime $p$. Count the rational points on the reduction of this curve modulo $p$. The count $N_p$ is a positive integer, and the Hasse bound (Hasse 1933) constrains it to lie within $2\sqrt{p}$ of $p+1$: the deviation $a_p=p+1-N_p$ satisfies $|a_p|\le 2\sqrt{p}$. The deviation is a small whole number rolling inside a narrow window whose width grows like the square root of the prime.

The Sato-Tate conjecture asks the next question: as $p$ runs over all primes (of good reduction), how is the deviation distributed inside that window? The right way to look is the normalised quantity $a_p/(2\sqrt{p})\in [-1,1]$, or equivalently the angle ${\theta }_p\in [0,\pi ]$ defined by $a_p=2\sqrt{p}\cos {\theta }_p$. The conjecture predicts a specific bell-shaped curve: the angles are not uniformly distributed in $[0,\pi ]$, nor concentrated near the centre. They follow the Sato-Tate distribution with density $(2/\pi ){\sin }^2\theta$, the same density that records how a random rotation of three-dimensional space distributes its rotation angle.

The conjecture was formulated by Mikio Sato in 1963 (computer experiment) and John Tate in 1965 (theoretical framing), and was proved for every non-CM elliptic curve over $\mathbb{Q}$ by Taylor and his collaborators between 2006 and 2011. The proof goes through the modularity theorem and a deep web of symmetric-power $L$-functions, but the empirical statement is concrete: plot a histogram of ${\theta }_p$ over the first thousand primes and you see the semicircle.

Visual Beginner

A histogram of the angles ${\theta }_p\in [0,\pi ]$ for the first $1000$ primes of good reduction of a non-CM elliptic curve, overlaid with the smooth Sato-Tate density $(2/\pi ){\sin }^2\theta$. The histogram has small bins on the left and right edges, growing toward the middle, peaking at $\theta =\pi /2$, and falling back to small bins at the far edge — the characteristic semicircle shape.

The picture says: the angles concentrate near $\pi /2$ (corresponding to $a_p\approx 0$, the centre of the Hasse window) and thin out near $0$ and $\pi$ (corresponding to $a_p\approx \pm 2\sqrt{p}$, the extremes). The bell shape is the Sato-Tate semicircle, and the rate at which the histogram converges to it as the number of primes grows is the quantitative content of the conjecture.

Worked example Beginner

Run a computational check of Sato-Tate for the elliptic curve $E:y^2=x^3+x+1$ over the first few primes. This curve has discriminant $\Delta =-16(4\cdot 1^3+27\cdot 1^2)=-496=-16\cdot 31$, so the primes of bad reduction are $2$ and $31$; every other prime is a prime of good reduction.

Step 1. Count points at $p=3$. Substitute every $x\in {\mathbb{F}}_3=\{0,1,2\}$ into $x^3+x+1$ and read the quadratic-residue status mod $3$: at $x=0$, $y^2=1$ gives $y\in \{1,2\}$, $2$ solutions; at $x=1$, $y^2=3\equiv 0$ gives $y=0$, $1$ solution; at $x=2$, $y^2=11\equiv 2$ is a non-residue mod $3$, $0$ solutions. Affine count: $3$. Add the point at infinity: $N_3=4$. So $a_3=3+1-4=0$, and ${\theta }_3=\arccos (0/(2\sqrt{3}))=\pi /2$.

Step 2. Count points at $p=5$. Substitute every $x\in {\mathbb{F}}_5$: at $x=0$, $y^2=1$, $2$ solutions; at $x=1$, $y^2=3$, a non-residue, $0$ solutions; at $x=2$, $y^2=11\equiv 1$, $2$ solutions; at $x=3$, $y^2=31\equiv 1$, $2$ solutions; at $x=4$, $y^2=69\equiv 4$, $2$ solutions. Affine count: $8$. Add infinity: $N_5=9$. So $a_5=5+1-9=-3$, and ${\theta }_5=\arccos (-3/(2\sqrt{5}))=\arccos (-0.6708)\approx 2.30$ radians.

Step 3. Count points at $p=7$. Substituting every $x\in {\mathbb{F}}_7$ and counting quadratic residues gives $N_7=9$, so $a_7=7+1-9=-1$, and ${\theta }_7=\arccos (-1/(2\sqrt{7}))\approx 1.76$ radians. Continuing through the first ten primes of good reduction $\{3,5,7,11,13,17,19,23,29,37\}$ yields the angle sample $\{\pi /2,2.30,1.76,1.95,1.07,1.91,1.36,1.62,1.31,2.06\}$ (to two decimal places). All ten angles fall within $[1.0,2.4]$, clustered around $\pi /2\approx 1.57$.

Step 4. Compare to the Sato-Tate prediction. The Sato-Tate density is $(2/\pi ){\sin }^2\theta$. For angles in $[1.0,2.4]$ the density takes values in roughly $[0.4,0.64]$, peaking at $0.64$ near $\pi /2$. A sample of ten angles is far too small to see the shape — but extending the count to the first $1000$ primes of good reduction (an exercise in any modern computer-algebra system) gives a histogram that tracks the smooth curve $(2/\pi ){\sin }^2\theta$ to within sampling fluctuation. The histogram bins near $\theta =0$ and $\theta =\pi$ are sparsely populated; the bin near $\theta =\pi /2$ is densely populated; the overall shape is the predicted semicircle.

What this tells us: the angles ${\theta }_p$ are not arbitrary integers within the Hasse window. They follow a specific bell-shaped statistical law that emerges out of the modular structure on the elliptic curve. The law was a computer-experimental observation in 1963 and a theorem in 2011.

Check your understanding Beginner

Formal definition Intermediate+

Fix a non-CM elliptic curve $E$ over $\mathbb{Q}$ with conductor $N_E$. At every prime $p\nmid N_E$ (a prime of good reduction) the Hasse coefficient $$ a_p := p + 1 - #E(\mathbb{F}p) \in \mathbb{Z} $$ satisfies the Hasse bound $|a_p|\le 2\sqrt{p}$ (Hasse 1933 Nachr. Ges. Wiss. Göttingen 1933). Equivalently, the characteristic polynomial of the Frobenius endomorphism on the $\ell$-adic Tate module $T\ell E$(foranyprime$\ell \neq p$)factorsas$X^2 - a_p X + p = (X - \alpha_p)(X - \beta_p)$over$\mathbb{C}$,withconjugateroots$\alpha_p, \beta_p \in \mathbb{C}$satisfying$\alpha_p \bar\alpha_p = p$.So$|\alpha_p| = |\beta_p| = \sqrt{p}$and$\alpha_p = \sqrt{p}, e^{i\theta_p}$,$\beta_p = \sqrt{p}, e^{-i\theta_p}$foraunique\ast \ast Frobeniusangle\ast \ast$\theta_p \in [0, \pi]$. The normalisation is $$ a_p = \alpha_p + \beta_p = 2\sqrt{p}\cos\theta_p, \qquad \theta_p \in [0, \pi]. $$

Definition (Sato-Tate measure). The Sato-Tate measure on the interval $[0,\pi ]$ is the Borel probability measure $$ \mu_{ST} = \frac{2}{\pi}\sin^2\theta , d\theta. $$ The density $(2/\pi ){\sin }^2\theta$ vanishes at $\theta \in \{0,\pi \}$ and attains its maximum $2/\pi$ at $\theta =\pi /2$. The mass under the curve is ${\int }_0^{\pi }(2/\pi ){\sin }^2\theta d\theta =1$.

Sato-Tate conjecture (Sato unpublished 1963, Tate 1965). Let $E$ be a non-CM elliptic curve over $\mathbb{Q}$. The sequence of Frobenius angles $\{{\theta }_p:p\text{ prime},p\nmid N_E\}$ is equidistributed in $[0,\pi ]$ with respect to ${\mu }_{ST}$. Equivalently, for every continuous function $f:[0,\pi ]\to \mathbb{R}$, $$ \lim_{X \to \infty} \frac{1}{\pi(X)} \sum_{\substack{p \leq X \ p \nmid N_E}} f(\theta_p) = \int_0^\pi f(\theta) \frac{2}{\pi}\sin^2\theta , d\theta, $$ where $\pi (X)=\#\{p\le X:p\text{ prime}\}$.

Equivalent unitarized formulation. The matrix $u_p:=\mathrm{diag}(e^{i{\theta }_p},e^{-i{\theta }_p})\in \mathrm{SU}(2)$ is the unitarized Frobenius conjugacy class in the maximal compact subgroup $\mathrm{SU}(2)\subset {\mathrm{GL}}_2(\mathbb{C})$, defined up to conjugation. The Sato-Tate conjecture is equivalent to the assertion that the conjugacy classes $\{u_p\}$ equidistribute in the space of conjugacy classes of $\mathrm{SU}(2)$ with respect to the pushforward of the Haar probability measure on $\mathrm{SU}(2)$. Under the Weyl integration formula $$ \int_{\mathrm{SU}(2)} f , dg = \frac{1}{\pi} \int_0^\pi f\bigl(\mathrm{diag}(e^{i\theta}, e^{-i\theta})\bigr) \cdot 2\sin^2\theta , d\theta, $$ the pushforward to $[0,\pi ]$ is exactly ${\mu }_{ST}$, so the two formulations coincide.

Equivalent formulation as $a_p/(2\sqrt{p})$ equidistribution. Let $b_p:=a_p/(2\sqrt{p})=\cos {\theta }_p\in [-1,1]$. The unitarized Hasse coefficient $b_p$ equidistributes on $[-1,1]$ with respect to the Wigner semicircle measure $(2/\pi )\sqrt{1-x^2}dx$, the pushforward of ${\mu }_{ST}$ under $\theta \mapsto \cos \theta$. This is the form in which Sato's 1963 computer experiment exhibited the conjecture.

Counterexamples to common slips

• "Sato-Tate applies to every elliptic curve over $\mathbb{Q}$." False in the form stated: the conjecture requires $E$ to be non-CM. For an elliptic curve $E$ with complex multiplication by an order in an imaginary quadratic field $K$, the Galois representation ${\rho }_{E,\ell }$ is abelian, the image lies in a maximal torus of ${\mathrm{GL}}_2({\mathbb{Q}}_{\ell })$ rather than being open in $\mathrm{SU}(2)$, and the angles split into two distinct distributions: at primes that split in $K$, ${\theta }_p$ equidistributes with respect to the uniform measure $d\theta /\pi$ on $[0,\pi ]$; at primes that are inert in $K$, ${\theta }_p=\pi /2$ deterministically (so $a_p=0$). The Sato-Tate measure is the wrong answer in the CM case; the substitute is the Hecke equidistribution for the associated Grössencharacter (Hecke 1918-20).

• "The Sato-Tate measure is the uniform measure on $[0,\pi ]$." False — that would predict equal probability for every angle, with density $1/\pi$. The correct density $(2/\pi ){\sin }^2\theta$ vanishes at the endpoints and peaks at $\pi /2$. The shape reflects the structure of $\mathrm{SU}(2)$: there are more conjugacy classes with intermediate eigenvalues than with eigenvalues near $\pm 1$, weighted by the Vandermonde-style factor $|e^{i\theta }-e^{-i\theta }{|}^2=4{\sin }^2\theta$ from the Weyl integration formula.

• "The Hasse coefficient $a_p$ itself equidistributes." False — the unnormalised $a_p$ takes values in the growing window $[-2\sqrt{p},2\sqrt{p}]$, which expands as $p\to \infty$. The Sato-Tate conjecture is about the normalised quantity $a_p/(2\sqrt{p})\in [-1,1]$ or the angle ${\theta }_p\in [0,\pi ]$. The unnormalised distribution has no limit measure because the support escapes to infinity.

• "Sato-Tate is true for $L$-functions whose Euler factor at $p$ has degree $2$ but is unrelated to elliptic curves." The conjecture as stated is specific to non-CM elliptic curves over $\mathbb{Q}$ (or over a totally real field, BLGHT 2011). The general motivic generalisation — Serre's framework with the Sato-Tate group $ST(M)\subseteq \mathrm{USp}$ or $\mathrm{SO}$ attached to a motive $M$ — predicts equidistribution of the unitarized Frobenius in $ST(M)$ relative to its Haar measure; for non-CM elliptic curves $ST(E)=\mathrm{SU}(2)$ and the prediction collapses to the Sato-Tate semicircle, but for abelian surfaces the Sato-Tate group can be one of $52$ distinct subgroups of $\mathrm{USp}(4)$ (Fité-Kedlaya-Rotger-Sutherland 2012).

Key theorem with proof Intermediate+

The signature theorem of this unit is the Sato-Tate conjecture itself, proved by Taylor and his collaborators 2006-2011. The full proof is the deepest result in modern arithmetic geometry and cannot be reconstructed inside a single curriculum unit. What can be proved at Intermediate level is the Weyl-criterion-style reduction from equidistribution to non-vanishing of all symmetric-power $L$-functions, which is the analytic skeleton of the argument.

Theorem (Sato-Tate reduction to non-vanishing of symmetric-power $L$-functions; Serre 1968 §A.2; Taylor 2008). Let $E$ be a non-CM elliptic curve over $\mathbb{Q}$ with conductor $N_E$. For every integer $n\ge 1$, define the $n$-th symmetric-power $L$-function $$ L(s, \mathrm{Sym}^n E) := \prod_{p \nmid N_E} \prod_{j = 0}^n \bigl(1 - \alpha_p^{n - j} \beta_p^j \cdot p^{-s}\bigr)^{-1} \cdot \prod_{p \mid N_E} L_p(s, \mathrm{Sym}^n E), $$ where ${\alpha }_p,{\beta }_p$ are the Frobenius eigenvalues at $p$ and $L_p(s,{\mathrm{Sym}}^{n}E)$ is the appropriate local factor at primes of bad reduction. Suppose that for every $n\ge 1$, $L(s,{\mathrm{Sym}}^{n}E)$ admits an analytic continuation to the closed half-plane $\mathrm{Re}(s)\ge 1$ that is holomorphic and non-vanishing on the line $\mathrm{Re}(s)=1$ (with at most a simple pole permitted at $s=1$ that occurs only when $n$ is even). Then the Frobenius angles ${\theta }_p$ are equidistributed in $[0,\pi ]$ with respect to the Sato-Tate measure.

Proof. The argument splits into three steps: Weyl's criterion in terms of characters of $\mathrm{SU}(2)$; identification of the trace of ${\mathrm{Sym}}^n$ at the unitarized Frobenius with a Chebyshev polynomial in $\cos {\theta }_p$; and a Tauberian transition from non-vanishing on the critical line to the asymptotic.

Step 1: Weyl's criterion. A sequence $\{u_p\}$ in $\mathrm{SU}(2)$ equidistributes with respect to Haar measure if and only if for every irreducible representation $\rho :\mathrm{SU}(2)\to {\mathrm{GL}}_N(\mathbb{C})$ other than the identity representation, $$ \lim_{X \to \infty} \frac{1}{\pi(X)} \sum_{p \leq X} \mathrm{tr},\rho(u_p) = 0. $$ The irreducible representations of $\mathrm{SU}(2)$ are exactly the symmetric powers ${\mathrm{Sym}}^n$ of the standard $2$-dimensional representation, $n\ge 0$, with ${\mathrm{Sym}}^0$ the identity representation and $\dim {\mathrm{Sym}}^n=n+1$. The character of ${\mathrm{Sym}}^n$ at the diagonal matrix $u=\mathrm{diag}(e^{i\theta },e^{-i\theta })$ is the Chebyshev polynomial of the second kind: $$ \mathrm{tr},\mathrm{Sym}^n(u) = U_n(\cos\theta) := \frac{\sin((n + 1)\theta)}{\sin\theta} = \sum_{j = 0}^n e^{i(n - 2j)\theta}. $$ Weyl's criterion then says equidistribution is equivalent to $$ \lim_{X \to \infty} \frac{1}{\pi(X)} \sum_{p \leq X, p \nmid N_E} U_n(\cos\theta_p) = 0 \quad \text{for every } n \geq 1. $$

Step 2: Chebyshev sums and partial sums. The eigenvalues of ${\mathrm{Sym}}^n(u_p)$ are ${\alpha }_p^{n-j}{\beta }_p^j/p^{n/2}$ for $j=0,\dots ,n$ (the unitarized eigenvalues of the $n$-th symmetric power, divided through by $p^{n/2}$ to normalise). So $$ \mathrm{tr},\mathrm{Sym}^n(u_p) = p^{-n/2} \sum_{j = 0}^n \alpha_p^{n - j}\beta_p^j. $$ The sum on the right is exactly $-\log L_p(s,{\mathrm{Sym}}^{n}E)/\log p$ evaluated at $s=n/2$, after expanding the local Euler factor at $s=n/2+1$ as a logarithmic derivative and using the relation $-\log (1-x)={\sum }_{k\ge 1}{x}^k/k$. More precisely, the partial sum $$ S_n(X) := \sum_{p \leq X, p \nmid N_E} p^{-n/2} \sum_{j = 0}^n \alpha_p^{n - j}\beta_p^j $$ is the partial sum of the Dirichlet coefficients of $-L^{\prime }(s,{\mathrm{Sym}}^{n}E)/L(s,{\mathrm{Sym}}^{n}E)$ at $s=1$, up to lower-order corrections from prime powers $p^k$ with $k\ge 2$ that vanish in the limit by the absolute convergence of the corresponding Euler factor.

Step 3: Wiener-Ikehara Tauberian transition. The Wiener-Ikehara Tauberian theorem (Ikehara 1931 J. Math. Phys. Mass. Inst. Tech. 10) states: if $L(s)={\sum }_n{a}_n{n}^{-s}$ is a Dirichlet series with non-negative coefficients, absolutely convergent on $\mathrm{Re}(s)>1$, with an analytic continuation to $\mathrm{Re}(s)\ge 1$ that is holomorphic on $\mathrm{Re}(s)=1$ except for a simple pole at $s=1$ with residue $r$, then ${\sum }_{n\le X}{a}_n\sim rX$ as $X\to \infty$.

Applied to the logarithmic derivative of $L(s,{\mathrm{Sym}}^{n}E)$ at $s=1$, the theorem gives an asymptotic for the prime sum $S_n(X)$: under the hypothesis that $L(s,{\mathrm{Sym}}^{n}E)$ is holomorphic and non-vanishing on $\mathrm{Re}(s)=1$ for every $n\ge 1$ (with the exception that $L(s,{\mathrm{Sym}}^{n}E)$ may have a simple pole at $s=1$ when $n$ is even, corresponding to the identity subrepresentation in the decomposition of ${\mathrm{Sym}}^n\otimes {\mathrm{Sym}}^n$), the partial sum $S_n(X)$ grows like ${\delta }_{n,0}\pi (X)+o(\pi (X))$ where the Kronecker symbol ${\delta }_{n,0}$ is $1$ for $n=0$ and $0$ otherwise.

For $n\ge 1$, the conclusion is $S_n(X)/\pi (X)\to 0$, which (after converting ${\sum }_j{\alpha }_p^{n-j}{\beta }_p^j/p^{n/2}$ to $U_n(\cos {\theta }_p)$ via the trigonometric identity ${\sum }_{j=0}^n{e}^{i(n-2j){\theta }_p}=U_n(\cos {\theta }_p)$) is exactly the Weyl criterion. Equidistribution follows.

The hypothesis of non-vanishing and analytic continuation of every $L(s,{\mathrm{Sym}}^{n}E)$ on the critical line is the deep input: it is established by the potential-automorphy theorem of Taylor 2008 + Clozel-Harris-Taylor 2008, refined by Harris-Shepherd-Barron-Taylor 2010 and Barnet-Lamb-Geraghty-Harris-Taylor 2011. $\square$

Bridge. The reduction builds toward the analytic-number-theoretic content of the Langlands programme, the chapter-closing theme that appears again in 21.10.01 (Langlands programme). The foundational reason the Sato-Tate semicircle has the specific density $(2/\pi ){\sin }^2\theta$ is exactly the Weyl integration formula on $\mathrm{SU}(2)$ — the measure is rigidly determined once the Galois image is constrained to lie in $\mathrm{SU}(2)$ rather than in a smaller subgroup. This is exactly the same Plancherel-formula structure that appears again in 21.04.02 (Hecke operators), where the spectral decomposition of the Hecke algebra on cusp forms also reflects the Weyl integration formula at the level of ${\mathrm{GL}}_2$. The central insight is that Sato-Tate identifies the empirical statistics of point counts on an elliptic curve with the representation-theoretic statistics of $\mathrm{SU}(2)$, mediated by the modular Galois representation of 21.05.01. The bridge is from the algebraic-geometric object $E$ to the equidistribution measure ${\mu }_{ST}$, via the modularity of 21.06.01 and the potential automorphy of every symmetric power. Putting these together, the Sato-Tate conjecture generalises beyond elliptic curves: for a general motive $M$ over $\mathbb{Q}$, the unitarized Frobenius traces equidistribute with respect to Haar measure on the Sato-Tate group $ST(M)$, a compact subgroup of $\mathrm{USp}$ or $\mathrm{SO}$, and is dual to the Tate conjecture under the algebraic-cycles / Galois-invariants correspondence. The bridge is exactly the recognition that point-count statistics on an arithmetic variety are predicted by the representation theory of a compact Lie group attached to the variety's motive.

Exercises Intermediate+

Advanced results Master

Theorem (Sato-Tate for non-CM elliptic curves over a totally real field; Taylor 2008 + BLGHT 2011). Let $F$ be a totally real number field and $E$ a non-CM elliptic curve over $F$. The Frobenius angles ${\theta }_v$ at primes $v$ of good reduction are equidistributed in $[0,\pi ]$ with respect to the Sato-Tate measure ${\mu }_{ST}=(2/\pi ){\sin }^2\theta d\theta$, where $a_v=\mathrm{N}(v)+1-\#E({\mathbb{F}}_v)=2\sqrt{\mathrm{N}(v)}\cos {\theta }_v$ at each prime $v$ of good reduction.

The chain of proofs runs Taylor 2008 (multiplicative-reduction case) → Clozel-Harris-Taylor 2008 (automorphy-lifting in regular weight on unitary groups) → Harris-Shepherd-Barron-Taylor 2010 (Calabi-Yau geometry for the Dwork pencil) → Barnet-Lamb-Geraghty-Harris-Taylor 2011 (removal of the multiplicative-reduction hypothesis). The proof strategy is the symmetric-power tower: ${\mathrm{Sym}}^n{\rho }_{E,\ell }$ is shown to be potentially automorphic for every $n$, which forces $L(s,{\mathrm{Sym}}^{n}E)$ to be entire (or to have only the predicted pole) and non-vanishing on $\mathrm{Re}(s)=1$. Newton-Thorne 2021 strengthens this to unconditional ${\mathrm{Sym}}^n$ functoriality for cuspidal cohomological forms on ${\mathrm{GL}}_2(\mathbb{Q})$, refining the Sato-Tate proof over $\mathbb{Q}$ but not changing the totally-real-field generality of BLGHT.

Theorem (Sato-Tate for CM elliptic curves; Hecke 1918, Deuring 1953). Let $E$ be an elliptic curve over $\mathbb{Q}$ with complex multiplication by an order in an imaginary quadratic field $K$. The Frobenius angles ${\theta }_p$ at primes $p$ of good reduction equidistribute as follows: at primes $p$ split in $K$ (density $1/2$), ${\theta }_p$ equidistributes uniformly with respect to $d\theta /\pi$ on $[0,\pi ]$; at primes $p$ inert in $K$ (density $1/2$), ${\theta }_p=\pi /2$ deterministically.

The CM Sato-Tate distribution is the equidistribution attached to the Sato-Tate group $ST(E)=N(T)$ — the normaliser of a maximal torus of $\mathrm{SU}(2)$ — relative to its Haar measure. The split primes are precisely those at which the unitarized Frobenius lies in the identity component of $N(T)$ (a copy of $T$), and the inert primes are those at which the unitarized Frobenius lies in the other coset (the anti-diagonal block, with trace $0$). The proof is classical and predates the non-CM case by a century: Hecke 1918-20 Math. Z. 1, 6 established the equidistribution of Hecke Grössencharacters on the idele class group, and Deuring 1953-57 Abh. Math. Sem. Hamb. 14, 16 attached such a Grössencharacter to every CM elliptic curve.

Theorem (vertical Sato-Tate / Birch distribution; Birch 1968). Fix a prime $p$. For an elliptic curve $E$ chosen uniformly at random from the finite set of elliptic curves over ${\mathbb{F}}_p$ (up to ${\mathbb{F}}_p$-isomorphism, weighted by the inverse of $|{\mathrm{Aut}}_{{\mathbb{F}}_p}(E)|$), the unitarized Hasse coefficient $a_p(E)/(2\sqrt{p})$ equidistributes with respect to a measure ${\mu }_{Birch,p}$ on $[-1,1]$ that converges weakly to the Wigner semicircle $(2/\pi )\sqrt{1-x^2}dx$ as $p\to \infty$.

The vertical statement is dual to the (horizontal) Sato-Tate conjecture: instead of fixing $E$ and varying $p$, one fixes $p$ and varies $E$. The proof uses the Eichler-Selberg trace formula and the Plancherel measure for ${\mathrm{GL}}_2({\mathbb{F}}_p)$; the explicit measure ${\mu }_{Birch,p}$ for finite $p$ deviates from the semicircle by terms of order $1/p$, controlled by the dimension formula for cusp forms of small weight. The vertical Sato-Tate result is unconditional — Birch's proof in 1968 predates the proof of horizontal Sato-Tate by more than four decades — and is one of the few cases where a Sato-Tate-type statement is fully proved at a fixed prime.

Theorem (Sato-Tate group classification for abelian surfaces; Fité-Kedlaya-Rotger-Sutherland 2012). Let $A$ be a polarised abelian surface over a number field $F$. The Sato-Tate group $ST(A)\subseteq \mathrm{USp}(4)$ is determined by the $\mathbb{Q}$-algebraic Galois action on the endomorphism algebra ${\mathrm{End}}_{\bar{F}}(A)\otimes \mathbb{Q}$, and there are exactly $52$ possibilities up to conjugacy. The generalised Sato-Tate conjecture predicts that the unitarized Frobenius conjugacy classes $u_v\in \mathrm{USp}(4)$ at primes $v$ of good reduction equidistribute with respect to the Haar measure on $ST(A)$.

For a non-CM elliptic curve, the dimension is $g=1$, the Sato-Tate group is $\mathrm{USp}(2)=\mathrm{SU}(2)$, and the prediction collapses to the standard Sato-Tate semicircle. For an abelian surface that is not isogenous to a product or a CM-type variety, the Sato-Tate group is generically $\mathrm{USp}(4)$, with conjectural pushforward measure on $[-2,2]$ given by the Weyl integration formula on $\mathrm{USp}(4)$ — a more complex semicircle-like density that vanishes at $\pm 2$ and is supported on $[-2,2]$. Of the $52$ possibilities, $34$ occur for Jacobians of genus-$2$ curves over $\mathbb{Q}$. The classification was extended by Fité-Kedlaya-Sutherland 2019 to genus-$3$ Jacobians (approximately $410$ candidate Sato-Tate groups).

Theorem (effective Sato-Tate under GRH; Murty-Murty 1997, Bucur-Kedlaya 2016). Assume the generalised Riemann hypothesis for all symmetric-power $L$-functions $L(s,{\mathrm{Sym}}^{n}E)$. For a non-CM elliptic curve $E/\mathbb{Q}$ of conductor $N_E$ and any interval $I=[a,b]\subseteq [0,\pi ]$, $$ #{p \leq X : p \nmid N_E, \theta_p \in I} = \mu_{ST}(I) \cdot \pi(X) + O\bigl(X^{3/4}(\log(X N_E))^{1/2}\bigr), $$ with implicit constant absolute.

The error term $O(X^{3/4})$ is the GRH-conditional rate of convergence. Unconditionally, the Taylor-et-al. proof gives an asymptotic but no quantitative error term; the explicit power-saving rate is conjectural. The conjecturally best rate is $O(X^{1/2}\log X)$, the elliptic-curve analogue of the prime-counting error under GRH. Bucur-Kedlaya 2016 Math. Comp. 85 verified the conditional rate computationally for the first $10^{11}$ primes for selected curves.

Theorem (Sato-Tate and the Lang-Trotter conjecture). For a non-CM elliptic curve $E/\mathbb{Q}$ and a fixed integer $r$, define ${\pi }_E(X,r):=\#\{p\le X:p\nmid N_E,a_p=r\}$. Lang-Trotter 1976 Lect. Notes Math. 504 conjecture that, for $r\ne 0$, $$ \pi_E(X, r) \sim C_{E, r} \frac{\sqrt{X}}{\log X} $$ as $X\to \infty$, with $C_{E,r}\ge 0$ an explicit constant.

The Lang-Trotter conjecture is the fine version of Sato-Tate: it predicts the count of primes at which $a_p$ equals a specific integer, not just the density of angles in an interval. The Sato-Tate measure assigns measure zero to any single point, so it predicts only ${\pi }_E(X,r)=o(\pi (X))$; Lang-Trotter predicts the precise order $\sqrt{X}/\log X$, smaller than $\pi (X)=X/\log X$ by a factor of $\sqrt{X}$. The conjecture is open; the best unconditional bound is ${\pi }_E(X,r)=O(X^{4/5})$ (Elkies 1991 Inventiones Math. 102, conditional on Sato-Tate), with $O(X^{3/4})$ under GRH.

Synthesis. The Sato-Tate conjecture is the foundational reason every non-CM elliptic curve over $\mathbb{Q}$ has the same statistical fingerprint at the level of Frobenius traces, identifying the empirical statistics of point counts on the algebraic-geometric object $E$ with the representation-theoretic statistics of the compact Lie group $\mathrm{SU}(2)$. The central insight is that the unitarized Frobenius conjugacy class $u_p\in \mathrm{SU}(2)$ equidistributes according to Haar measure on $\mathrm{SU}(2)$, and the Sato-Tate semicircle is exactly the pushforward of Haar measure under the trace map — the Weyl integration formula made arithmetic. This is exactly the same compact-group equidistribution that appears again in 21.10.01 (Langlands programme), where every motivic $L$-function is predicted to be the $L$-function of an automorphic representation, with the unitarized Frobenius equidistributing according to the Haar measure on the Sato-Tate group of the motive. The bridge is the recognition that point-count statistics on an arithmetic variety are governed by the representation theory of a compact Lie group attached to the motivic Galois group, and that the rigid measure-theoretic structure of Haar measure on a compact group is what makes the equidistribution sharp. Putting these together, the Sato-Tate framework generalises beyond elliptic curves: for abelian surfaces it identifies the unitarized Frobenius with a point in one of $52$ subgroups of $\mathrm{USp}(4)$ (Fité-Kedlaya-Rotger-Sutherland 2012), for higher-genus motives it identifies it with a point in a Sato-Tate subgroup of $\mathrm{USp}$ or $\mathrm{SO}$, and for general motives it is dual to the Tate conjecture under the algebraic-cycles correspondence.

The duality with the modularity theorem is sharp. Modularity 21.06.01 identifies the $\ell$-adic Galois representation ${\rho }_{E,\ell }$ with the Galois representation attached to a weight-$2$ cusp newform $f_E$. Sato-Tate is the analytic consequence of modularity sharpened to all symmetric powers: ${\mathrm{Sym}}^n{\rho }_{E,\ell }={\mathrm{Sym}}^n{\rho }_{f_E,\ell }$ corresponds to a (potentially) automorphic representation on ${\mathrm{GL}}_{n+1}$, so its $L$-function has the standard analytic properties (entire or with simple pole at $s=1$, non-vanishing on $\mathrm{Re}(s)=1$, functional equation, Euler product), and the Tauberian transition converts the analytic properties to equidistribution. The bridge from modularity to Sato-Tate is exactly the bridge from the ${\mathrm{GL}}_2$ automorphy of $f_E$ to the ${\mathrm{GL}}_{n+1}$ automorphy of ${\mathrm{Sym}}^n{f}_E$, which is the simplest substantive case of Langlands functoriality — and which Newton-Thorne 2021 Publ. Math. IHÉS 134 proved unconditionally for cuspidal cohomological forms on ${\mathrm{GL}}_2(\mathbb{Q})$. The recursion stabilises at every level of the symmetric-power tower, and the entire tower produces the single semicircle measure ${\mu }_{ST}$ as its analytic shadow.

Full proof set Master

Proposition 1 (Hasse bound; Hasse 1933). For every elliptic curve $E$ over ${\mathbb{F}}_p$, $|a_p|\le 2\sqrt{p}$.

Proof. The Frobenius endomorphism $\varphi :E\to E$ over ${\mathbb{F}}_p$ satisfies the characteristic equation ${\varphi }^2-a_p\varphi +p=0$ in the endomorphism ring ${\mathrm{End}}_{{\mathbb{F}}_p}(E)$, with $a_p=p+1-\#E({\mathbb{F}}_p)$. The action of $\varphi$ on the $\ell$-adic Tate module $T_{\ell }E$ (for any $\ell \ne p$) has trace $a_p$ and determinant $p$, so the eigenvalues ${\alpha }_p,{\beta }_p$ satisfy ${\alpha }_p{\beta }_p=p$ and ${\alpha }_p+{\beta }_p=a_p$. The roots ${\alpha }_p,{\beta }_p$ of the characteristic polynomial $X^2-a_{p}X+p$ are complex conjugates with $|{\alpha }_p|=|{\beta }_p|=\sqrt{p}$ — this is the Riemann hypothesis for the local zeta function of $E$ over ${\mathbb{F}}_p$, proved by Hasse 1933 via the descent argument that produces the inequality $\#E({\mathbb{F}}_p)\le p+1+2\sqrt{p}$ on the elliptic-curve trace. Consequently $|a_p|=|{\alpha }_p+{\beta }_p|\le |{\alpha }_p|+|{\beta }_p|=2\sqrt{p}$. $\square$

Proposition 2 (Weyl integration formula on $\mathrm{SU}(2)$). Let $\mathrm{SU}(2)$ have its bi-invariant Haar probability measure. For every continuous class function $f:\mathrm{SU}(2)\to \mathbb{C}$, $$ \int_{\mathrm{SU}(2)} f(g) , dg = \frac{1}{\pi}\int_0^\pi f(\mathrm{diag}(e^{i\theta}, e^{-i\theta})) \cdot 2\sin^2\theta , d\theta. $$ Consequently the pushforward of Haar measure to the space $\mathrm{SU}(2)/\sim$ of conjugacy classes (parametrised by $\theta \in [0,\pi ]$) is the Sato-Tate measure ${\mu }_{ST}=(2/\pi ){\sin }^2\theta d\theta$.

Proof. Every element of $\mathrm{SU}(2)$ is conjugate to a diagonal matrix $\mathrm{diag}(e^{i\theta },e^{-i\theta })$ for a unique $\theta \in [0,\pi ]$, so a continuous class function $f$ is determined by its restriction to the maximal torus $T=\{\mathrm{diag}(e^{i\theta },e^{-i\theta })\}$. By the Weyl integration formula on a compact connected Lie group with maximal torus $T$ and Weyl group $W$, $$ \int_G f , dg = \frac{1}{|W|}\int_T f(t) \cdot |D(t)|^2 , dt, $$ where $D(t)={\prod }_{\alpha \in R^{+}}(1-e^{-\alpha (t)})$ is the Weyl denominator over positive roots. For $\mathrm{SU}(2)$: rank $1$, single positive root $\alpha$ with $\alpha (\mathrm{diag}(e^{i\theta },e^{-i\theta }))=2i\theta$, Weyl group of order $|W|=2$. So $|D(t){|}^2=|1-e^{-2i\theta }{|}^2=4{\sin }^2\theta$. The Haar measure on $T$ is $d\theta /(2\pi )$ over $\theta \in [-\pi ,\pi ]$, equivalently $d\theta /\pi$ over $\theta \in [0,\pi ]$ after symmetrising. Combining, $$ \int_G f , dg = \frac{1}{2}\int_0^\pi f(\mathrm{diag}(e^{i\theta}, e^{-i\theta})) \cdot 4\sin^2\theta \cdot \frac{d\theta}{\pi} = \frac{2}{\pi}\int_0^\pi f(\cdots)\sin^2\theta , d\theta. \quad \square $$

Proposition 3 (Chebyshev character of ${\mathrm{Sym}}^n$). Let ${\rho }_n={\mathrm{Sym}}^n:\mathrm{SU}(2)\to {\mathrm{GL}}_{n+1}(\mathbb{C})$ be the $n$-th symmetric power of the standard representation. Then $\mathrm{tr}{\rho }_n(\mathrm{diag}(e^{i\theta },e^{-i\theta }))=U_n(\cos \theta )=\sin ((n+1)\theta )/\sin \theta$, where $U_n$ is the Chebyshev polynomial of the second kind.

Proof. The eigenvalues of ${\mathrm{Sym}}^n(u)$ at $u=\mathrm{diag}(e^{i\theta },e^{-i\theta })$ are the $n+1$ monomial weights $e^{i\theta \cdot (n-j)}\cdot e^{-i\theta \cdot j}=e^{i(n-2j)\theta }$ for $j=0,1,\dots ,n$. The trace is $$ \sum_{j = 0}^n e^{i(n - 2j)\theta} = e^{in\theta} + e^{i(n - 2)\theta} + \cdots + e^{-in\theta} = \frac{e^{i(n + 1)\theta} - e^{-i(n + 1)\theta}}{e^{i\theta} - e^{-i\theta}} = \frac{\sin((n + 1)\theta)}{\sin\theta} = U_n(\cos\theta). \quad \square $$

Proposition 4 (Weyl's criterion on $\mathrm{SU}(2)$). A sequence $\{u_p\}$ in $\mathrm{SU}(2)$ equidistributes (modulo conjugation) with respect to Haar measure if and only if for every $n\ge 1$, $$ \lim_{X \to \infty} \frac{1}{\pi(X)} \sum_{p \leq X} U_n(\cos\theta_p) = 0. $$

Proof. Equidistribution of $\{u_p\}$ in the space of conjugacy classes means $\pi (X{)}^{-1}{\sum }_{p\le X}f(u_p)\to {\int }_{\mathrm{SU}(2)}fdg$ for every continuous class function $f$. By Peter-Weyl, the matrix coefficients of irreducible representations span a dense subspace of $L^2(\mathrm{SU}(2))$ in class-function form, the characters ${\chi }_{{\rho }_n}$ for $n\ge 0$. So it suffices to test against $f={\chi }_{{\rho }_n}$ for $n\ge 0$. For $n=0$ (the identity representation), both sides equal $1$. For $n\ge 1$, the right-hand side is ${\int }_{\mathrm{SU}(2)}{\chi }_{{\rho }_n}dg=0$ by orthogonality of characters. The left-hand side is $\pi (X{)}^{-1}{\sum }_{p\le X}{U}_n(\cos {\theta }_p)$ by Proposition 3. Equidistribution is exactly the vanishing of these sums in the limit. $\square$

Proposition 5 (Tauberian transition via Wiener-Ikehara). Suppose $L(s,{\mathrm{Sym}}^{n}E)$ admits an analytic continuation to $\mathrm{Re}(s)\ge 1$ that is holomorphic on $\mathrm{Re}(s)=1$ and (for $n\ge 1$) has no pole at $s=1$. Then $$ \sum_{p \leq X, p \nmid N_E} U_n(\cos\theta_p) = o(\pi(X)) $$ as $X\to \infty$.

Proof. Consider the Dirichlet-series logarithmic derivative $$ \Lambda_n(s) := -\frac{L'(s, \mathrm{Sym}^n E)}{L(s, \mathrm{Sym}^n E)} = \sum_p \sum_{k \geq 1} \frac{(\log p), a_n(p^k)}{p^{ks}}, \qquad a_n(p) = \sum_{j = 0}^n \alpha_p^{n - j}\beta_p^j \cdot p^{-n/2}. $$ The coefficient at the prime $p$ (the $k=1$ term) is $a_n(p)\cdot (\log p)/p^s$. Under the normalisation $|{\alpha }_p|=|{\beta }_p|=\sqrt{p}$ embedded in the unitarized form, $a_n(p)=U_n(\cos {\theta }_p)$ by Proposition 3 (with the eigenvalues ${\alpha }_p/\sqrt{p}=e^{i{\theta }_p}$). The prime-power terms $k\ge 2$ contribute a Dirichlet series absolutely convergent on $\mathrm{Re}(s)>1/2$, hence negligible in the Tauberian regime $\mathrm{Re}(s)\to 1^{+}$.

By the hypothesis, $L(s,{\mathrm{Sym}}^{n}E)$ is holomorphic and non-vanishing on the closed half-plane $\mathrm{Re}(s)\ge 1$ (with no pole at $s=1$ for $n\ge 1$), so the logarithmic derivative ${\Lambda }_n(s)$ is holomorphic on $\mathrm{Re}(s)\ge 1$ (no pole at $s=1$, no pole on the critical line from a zero of $L$). The Wiener-Ikehara Tauberian theorem (Ikehara 1931 J. Math. Phys. MIT 10; Newman 1980 Amer. Math. Monthly 87 for the modern elementary proof) then gives the asymptotic $$ \sum_{p \leq X, p \nmid N_E} U_n(\cos\theta_p) \cdot \log p = o(X) $$ as $X\to \infty$ (the dominant-pole residue at $s=1$ is zero). Partial summation against $1/\log p$ converts to ${\sum }_p{U}_n(\cos {\theta }_p)=o(\pi (X))$ by the prime number theorem. $\square$

Theorem 6 (Sato-Tate from non-vanishing of all symmetric powers; conditional version). Suppose, for a non-CM elliptic curve $E/\mathbb{Q}$, that for every $n\ge 1$ the $n$-th symmetric-power $L$-function $L(s,{\mathrm{Sym}}^{n}E)$ admits an analytic continuation to $\mathrm{Re}(s)\ge 1$ that is holomorphic and non-vanishing on $\mathrm{Re}(s)=1$ except for at most a simple pole at $s=1$ when $n$ is even (and never for $n$ odd). Then the Frobenius angles ${\theta }_p$ equidistribute on $[0,\pi ]$ with respect to ${\mu }_{ST}$.

Proof. By Proposition 5, the hypothesis forces ${\sum }_p{U}_n(\cos {\theta }_p)=o(\pi (X))$ for every $n\ge 1$. By Proposition 4 (Weyl's criterion), the unitarized Frobenius classes $u_p$ equidistribute in $\mathrm{SU}(2)/\sim$ with respect to the pushforward of Haar measure. By Proposition 2 (Weyl integration formula), the pushforward to the conjugacy-class space $[0,\pi ]$ is the Sato-Tate measure ${\mu }_{ST}$. Combining, the angles ${\theta }_p$ equidistribute with respect to ${\mu }_{ST}$ on $[0,\pi ]$. $\square$

Theorem 7 (Sato-Tate, unconditional; Taylor 2008, BLGHT 2011). For every non-CM elliptic curve $E/\mathbb{Q}$, the Frobenius angles ${\theta }_p$ equidistribute on $[0,\pi ]$ with respect to ${\mu }_{ST}$.

Proof (sketch). By Theorem 6 it suffices to verify the analytic hypothesis on every symmetric power. The proof is the potential automorphy argument of Taylor and his collaborators:

(a) Modularity of $E$ (Wiles 1995 + BCDT 2001): ${\rho }_{E,\ell }$ corresponds to a weight-$2$ cusp newform $f_E$ on ${\Gamma }_0(N_E)$ with rational Hecke eigenvalues, equivalently $E$ is isogenous to the modular abelian variety $A_{f_E}$. See 21.06.01.

(b) Potential automorphy of symmetric powers (Taylor 2008, BLGHT 2011): for each $n\ge 1$, there exists a totally real solvable Galois extension $F_n/\mathbb{Q}$ such that ${\mathrm{Sym}}^n{\rho }_{E,\ell }{|}_{G_{F_n}}$ corresponds to a cuspidal automorphic representation ${\Pi }_n$ of ${\mathrm{GL}}_{n+1}({\mathbb{A}}_{F_n})$ with appropriate weight and conductor. The proof goes through modularity-lifting on unitary groups: the residual mod-$\ell$ representation of ${\mathrm{Sym}}^n{\rho }_{E,\ell }$ is shown to be residually automorphic over an auxiliary field $F$ via Clozel-Harris-Taylor's $R=T$ theorem on $\mathrm{GU}(n+1)$, and the Calabi-Yau Dwork-family construction of Harris-Shepherd-Barron-Taylor 2010 supplies the residual modularity at a key auxiliary prime. BLGHT 2011 removes the multiplicative-reduction hypothesis required by Taylor 2008 via a $p$-adic deformation argument at additive primes.

(c) Brauer-induction descent: the $L$-function of ${\Pi }_n$ over $F_n$, combined with the standard properties of automorphic $L$-functions on ${\mathrm{GL}}_{n+1}$ over a number field (Jacquet-Shalika 1976 Amer. J. Math. 98: meromorphic continuation, functional equation, non-vanishing on $\mathrm{Re}(s)=1$), and Brauer-induced back down to $\mathbb{Q}$, yields the required analytic properties of $L(s,{\mathrm{Sym}}^{n}E)$ over $\mathbb{Q}$: meromorphic on $\mathbb{C}$, holomorphic on $\mathrm{Re}(s)=1$ (with possible simple pole at $s=1$ when $n$ is even, from the identity subrepresentation of ${\mathrm{Sym}}^{2k}$ paired with itself), and non-vanishing on $\mathrm{Re}(s)=1$ (Jacquet-Shalika).

(d) Apply Theorem 6 to conclude. $\square$

The Newton-Thorne 2021 strengthening removes the "potential" in (b): for cuspidal cohomological forms on ${\mathrm{GL}}_2(\mathbb{Q})$ (which includes every weight-$2$ newform $f_E$ for a non-CM elliptic curve), ${\mathrm{Sym}}^n{f}_E$ is automorphic on ${\mathrm{GL}}_{n+1}(\mathbb{Q})$ unconditionally, without passing to a solvable extension. This shortens the proof of Sato-Tate over $\mathbb{Q}$ but does not change the totally-real-field generality of BLGHT.

Proposition 8 (Wiener-Ikehara Tauberian theorem; Ikehara 1931). Let $L(s)={\sum }_{n\ge 1}{a}_n{n}^{-s}$ be a Dirichlet series with $a_n\ge 0$, absolutely convergent on $\mathrm{Re}(s)>1$. Suppose $L(s)-r/(s-1)$ has an analytic continuation to an open neighbourhood of $\mathrm{Re}(s)\ge 1$ that is continuous on the closed half-plane. Then ${\sum }_{n\le X}{a}_n\sim rX$ as $X\to \infty$.

Proof (sketch following Newman 1980). Set $A(X)={\sum }_{n\le X}{a}_n$. The Mellin transform identity $$ L(s) = s \int_1^\infty A(X) X^{-s - 1} dX $$ (valid for $\mathrm{Re}(s)>1$) shifts to the line $\mathrm{Re}(s)=1$ by contour deformation, the deformation being permitted by the continuity hypothesis on the closed half-plane. The dominant contribution at $s=1$ comes from the simple-pole residue $r$, giving $A(X)\sim rX$ by the inverse Mellin transform applied to the residue. Newman's elementary proof avoids the full machinery of Fourier-analytic Tauberian theorems by using contour integration on a finite rectangle and explicit error-estimate bookkeeping; the result is the same. $\square$

The Wiener-Ikehara theorem is the analytic engine of the Sato-Tate proof. It converts the analytic statement "non-vanishing on $\mathrm{Re}(s)=1$" into the arithmetic statement "${\sum }_{n\le X}{a}_n\sim rX$", and the arithmetic statement applied to symmetric-power $L$-functions is exactly what Weyl's criterion demands.

Connections Master

• Modularity theorem and BSD conjecture 21.06.01. Modularity is the input on which the Sato-Tate proof rests: every non-CM elliptic curve over $\mathbb{Q}$ corresponds to a weight-$2$ cusp newform $f_E$, and the symmetric-power $L$-functions $L(s,{\mathrm{Sym}}^{n}E)$ that Taylor et al. prove (potentially) automorphic are the symmetric powers of the modular Galois representation ${\rho }_{f_E,\ell }$. Without modularity (Wiles 1995 + BCDT 2001) Sato-Tate has no proof; with modularity, the further step is automorphy of every symmetric power. Sato-Tate is the equidistribution refinement that complements the rank refinement of BSD: BSD predicts the order of vanishing of $L(E,s)$ at $s=1$, Sato-Tate predicts the distribution of the local Euler factors that build $L(E,s)$ in the first place.

• $\ell$-adic Galois representations 21.05.01. The Frobenius trace $a_p=\mathrm{tr}{\rho }_{E,\ell }({\mathrm{Frob}}_p)$ at a prime of good reduction is the data on which Sato-Tate is stated. The Sato-Tate group $ST(E)\subseteq \mathrm{SU}(2)$ is the Zariski closure of the image of ${\rho }_{E,\ell }$ intersected with the maximal compact subgroup, and the Sato-Tate measure is the pushforward of Haar measure on $ST(E)$ under the trace map. The non-CM hypothesis is exactly the hypothesis that $ST(E)=\mathrm{SU}(2)$ rather than a smaller subgroup; Serre's open-image theorem (Serre 1972 Invent. Math. 15) confirms the Zariski-density side of this picture.

• Hecke operators and Hecke algebra 21.04.02. The symmetric-power $L$-function $L(s,{\mathrm{Sym}}^{n}E)$ is, on the modular-form side, the $L$-function of ${\mathrm{Sym}}^n{f}_E$, defined via the Hecke eigenvalues $a_p(f_E)$ at primes of good reduction. The Hecke algebra $\mathbb{T}$ acts on cusp forms, and the action of ${\mathrm{Sym}}^n$ at each prime is encoded in the Chebyshev polynomial $U_n(a_p/(2\sqrt{p}))$ of Hecke eigenvalues. The Plancherel formula for cusp forms, and the spectral decomposition of the Hecke algebra, are the modular-form-side counterpart of the Weyl integration formula on $\mathrm{SU}(2)$.

• Elliptic curves 04.04.03. The algebraic-geometric foundation. The Hasse coefficient $a_p$ is defined via the point count $\#E({\mathbb{F}}_p)$, computed on the reduction of the Weierstrass model at $p$. The Hasse bound, the Frobenius endomorphism, and the local zeta function are developed in 04.04.03; Sato-Tate is the statistical refinement that records how the deviations $a_p$ from $p+1$ distribute as $p$ varies. The bridge from 04.04.03 to the present unit is the recognition that point-count data, viewed in the Hasse window, follows a compact-group equidistribution.

• Langlands programme 21.10.01. Forward-connected (to be produced). Sato-Tate is the simplest analytic instance of Langlands functoriality: the symmetric-power transfer ${\mathrm{Sym}}^n:{\mathrm{GL}}_2\to {\mathrm{GL}}_{n+1}$ predicted by Langlands carries every weight-$2$ cusp newform on ${\mathrm{GL}}_2$ to a cuspidal automorphic form on ${\mathrm{GL}}_{n+1}$. Newton-Thorne 2021 proved this functoriality unconditionally for cuspidal cohomological forms over $\mathbb{Q}$. The analytic consequence is the non-vanishing and analytic continuation of every $L(s,{\mathrm{Sym}}^{n}E)$, which is exactly the input to the Sato-Tate theorem.

Historical & philosophical context Master

Mikio Sato conducted an unpublished computer experiment in 1963 ^{[source pending]} at the University of Tokyo, computing the Frobenius traces $a_p$ for several elliptic curves over $\mathbb{Q}$ at the first thousand primes and plotting the histogram of $a_p/(2\sqrt{p})$. The empirical distribution he observed matched the Wigner semicircle $(2/\pi )\sqrt{1-x^2}$, and Sato communicated the observation to Tate, who provided the theoretical framing in terms of analytic continuation of symmetric-power $L$-functions. Tate's article "Algebraic cycles and poles of zeta functions" in Arithmetical Algebraic Geometry (Schilling, ed., Harper & Row 1965) ^{[source pending]} is the first written record of the conjecture, with attribution to Sato and a connection drawn to the conjectural analytic properties of the $L$-functions $L(s,{\mathrm{Sym}}^{n}E)$.

Jean-Pierre Serre 1968 Abelian $\ell$-adic Representations and Elliptic Curves (Benjamin) ^{[source pending]} reformulated Sato-Tate in the modern Galois-theoretic / Lie-group framework, identifying the equidistribution measure as the pushforward of Haar measure on $\mathrm{SU}(2)$ under the trace map. Serre's monograph established the language of Sato-Tate groups as compact subgroups of $\mathrm{USp}$ controlled by the Galois action on Tate modules. Birch 1968 Mathematika 15 ^{[source pending]} independently established the vertical analogue of the conjecture — equidistribution of $a_p(E)/(2\sqrt{p})$ as $E$ varies over elliptic curves over a fixed ${\mathbb{F}}_p$, with $p\to \infty$ — via the Plancherel formula for ${\mathrm{GL}}_2({\mathbb{F}}_p)$. The vertical statement is unconditional from 1968 onward, four decades before the horizontal statement.

The horizontal Sato-Tate conjecture remained open for forty years. Partial results accumulated: Serre 1989 Astérisque 168-169 proved Sato-Tate for elliptic curves with a substantive Hecke-algebra action of dimension $\ge 2$ over $\mathbb{Q}$ (a narrow class); Kim-Sarnak 2002 Duke Math. J. 112 proved Sato-Tate for ${\mathrm{Sym}}^3$ and ${\mathrm{Sym}}^4$ via functoriality results of Kim 2003 J. Amer. Math. Soc. 16, Kim-Shahidi 2002 Duke Math. J. 112, and Shahidi's Langlands-Shahidi method. The full proof for non-CM elliptic curves over $\mathbb{Q}$ with at least one prime of multiplicative reduction was given by Taylor 2008 Publ. Math. IHÉS 108 ^{[source pending]} in tandem with Clozel-Harris-Taylor 2008 Publ. Math. IHÉS 108 ^{[source pending]} (the modularity-lifting companion on unitary groups). The Calabi-Yau Dwork-family input was supplied by Harris-Shepherd-Barron-Taylor 2010 Ann. Math. 171 ^{[source pending]}. The most general statement — Sato-Tate for every non-CM elliptic curve over a totally real number field — was completed by Barnet-Lamb-Geraghty-Harris-Taylor 2011 Publ. Math. IHÉS 114 ^{[source pending]}, removing the multiplicative-reduction hypothesis via a wild $p$-adic deformation argument at additive primes.

The Fité-Kedlaya-Rotger-Sutherland 2012 Compos. Math. 148 classification ^{[source pending]} of the Sato-Tate groups for abelian surfaces enumerates the $52$ possibilities and verifies the equidistribution prediction computationally for representative cases. Newton-Thorne 2021 Publ. Math. IHÉS 134 ^{[source pending]} proved unconditional ${\mathrm{Sym}}^n$ functoriality for cuspidal cohomological forms on ${\mathrm{GL}}_2(\mathbb{Q})$, refining the potential-automorphy chain of Taylor 2008-BLGHT 2011 and giving a cleaner proof of Sato-Tate over $\mathbb{Q}$.

Manin-Panchishkin Introduction to Modern Number Theory (Springer EMS 49, 2nd ed. 2005) Ch. 6 §6 ^{[source pending]} codifies the Sato-Tate framework for the modern graduate student, with the status of the conjecture recorded as of 2005 — at that date, partial results through ${\mathrm{Sym}}^9$ were known via Kim-Sarnak 2002 and Kim 2004, and the full conjecture was conjectural. The Manin-Panchishkin treatment frames Sato-Tate as the natural sharpening of the Hasse bound and as the analytic shadow of the Langlands functoriality conjecture for symmetric powers.

Bibliography Master

@incollection{Tate1965SatoTate,
  author    = {Tate, John},
  title     = {Algebraic cycles and poles of zeta functions},
  booktitle = {Arithmetical Algebraic Geometry},
  editor    = {Schilling, O. F. G.},
  publisher = {Harper and Row},
  year      = {1965},
  pages     = {93--110}
}

@book{Serre1968AbelianLAdic,
  author    = {Serre, Jean-Pierre},
  title     = {Abelian $\ell$-adic Representations and Elliptic Curves},
  publisher = {W. A. Benjamin},
  address   = {New York},
  year      = {1968}
}

@article{Taylor2008,
  author  = {Taylor, Richard},
  title   = {Automorphy for some $\ell$-adic lifts of automorphic mod $\ell$ {G}alois representations. {II}},
  journal = {Publications Math{\'e}matiques de l'IH{\'E}S},
  volume  = {108},
  year    = {2008},
  pages   = {183--239}
}

@article{ClozelHarrisTaylor2008,
  author  = {Clozel, Laurent and Harris, Michael and Taylor, Richard},
  title   = {Automorphy for some $\ell$-adic lifts of automorphic mod $\ell$ {G}alois representations},
  journal = {Publications Math{\'e}matiques de l'IH{\'E}S},
  volume  = {108},
  year    = {2008},
  pages   = {1--181}
}

@article{HarrisShepherdBarronTaylor2010,
  author  = {Harris, Michael and Shepherd-Barron, Nicholas and Taylor, Richard},
  title   = {A family of {C}alabi-{Y}au varieties and potential automorphy},
  journal = {Annals of Mathematics (2)},
  volume  = {171},
  year    = {2010},
  pages   = {779--813}
}

@article{BLGHT2011,
  author  = {Barnet-Lamb, Thomas and Geraghty, David and Harris, Michael and Taylor, Richard},
  title   = {A family of {C}alabi-{Y}au varieties and potential automorphy {II}},
  journal = {Publications Math{\'e}matiques de l'IH{\'E}S},
  volume  = {114},
  year    = {2011},
  pages   = {105--180}
}

@article{Birch1968,
  author  = {Birch, Bryan J.},
  title   = {How the number of points of an elliptic curve over a fixed prime field varies},
  journal = {Mathematika},
  volume  = {15},
  year    = {1968},
  pages   = {153--176}
}

@article{Deligne1980WeilII,
  author  = {Deligne, Pierre},
  title   = {La conjecture de {W}eil. {II}},
  journal = {Publications Math{\'e}matiques de l'IH{\'E}S},
  volume  = {52},
  year    = {1980},
  pages   = {137--252}
}

@article{Hasse1933,
  author  = {Hasse, Helmut},
  title   = {Beweis des {A}nalogons der {R}iemannschen {V}ermutung f{\"u}r die {A}rtinschen und {F}. {K}. {S}chmidtschen {K}ongruenzzetafunktionen in gewissen elliptischen {F}{\"a}llen},
  journal = {Nachrichten der Gesellschaft der Wissenschaften zu G{\"o}ttingen},
  year    = {1933},
  pages   = {253--262}
}

@article{Ikehara1931,
  author  = {Ikehara, Shikao},
  title   = {An extension of {L}andau's theorem in the analytic theory of numbers},
  journal = {Journal of Mathematics and Physics of the Massachusetts Institute of Technology},
  volume  = {10},
  year    = {1931},
  pages   = {1--12}
}

@article{Newman1980,
  author  = {Newman, Donald J.},
  title   = {Simple analytic proof of the prime number theorem},
  journal = {American Mathematical Monthly},
  volume  = {87},
  year    = {1980},
  pages   = {693--696}
}

@article{FiteKedlayaRotgerSutherland2012,
  author  = {Fit{\'e}, Francesc and Kedlaya, Kiran S. and Rotger, Victor and Sutherland, Andrew V.},
  title   = {Sato-{T}ate distributions and {G}alois endomorphism modules in genus 2},
  journal = {Compositio Mathematica},
  volume  = {148},
  year    = {2012},
  pages   = {1390--1442}
}

@article{NewtonThorne2021,
  author  = {Newton, James and Thorne, Jack A.},
  title   = {Symmetric power functoriality for holomorphic modular forms},
  journal = {Publications Math{\'e}matiques de l'IH{\'E}S},
  volume  = {134},
  year    = {2021},
  pages   = {1--116}
}

@article{KimSarnak2002,
  author  = {Kim, Henry H. and Sarnak, Peter},
  title   = {Refined estimates towards the {R}amanujan and {S}elberg conjectures},
  journal = {Duke Mathematical Journal},
  volume  = {112},
  year    = {2002},
  pages   = {177--197}
}

@article{LangTrotter1976,
  author    = {Lang, Serge and Trotter, Hale},
  title     = {Frobenius distributions in {$\mathrm{GL}_2$}-extensions},
  series    = {Lecture Notes in Mathematics},
  volume    = {504},
  publisher = {Springer},
  year      = {1976}
}

@book{Silverman2009ArithEC,
  author    = {Silverman, Joseph H.},
  title     = {The Arithmetic of Elliptic Curves},
  series    = {Graduate Texts in Mathematics},
  volume    = {106},
  edition   = {2nd},
  publisher = {Springer},
  year      = {2009}
}

@book{ManinPanchishkin2005MNT,
  author    = {Manin, Yuri I. and Panchishkin, Alexei A.},
  title     = {Introduction to Modern Number Theory},
  series    = {Encyclopaedia of Mathematical Sciences},
  volume    = {49},
  edition   = {2nd},
  publisher = {Springer},
  year      = {2005}
}

@article{Hecke1918,
  author  = {Hecke, Erich},
  title   = {{\"U}ber die {Z}etafunktion beliebiger algebraischer {Z}ahlk{\"o}rper},
  journal = {Mathematische Zeitschrift},
  volume  = {1},
  year    = {1918},
  pages   = {83--126}
}

@article{Deuring1953,
  author  = {Deuring, Max},
  title   = {Die {Z}etafunktion einer algebraischen {K}urve vom {G}eschlechte {E}ins},
  journal = {Nachrichten der Akademie der Wissenschaften in G{\"o}ttingen},
  year    = {1953},
  pages   = {85--94}
}