\section{Discriminant \texorpdfstring{\(-8\)}{-8}}\label{sec:minus8} Put \[ Q_8(x,y)=x^2+2y^2,\qquad P_8=\{Q_8(x,y)>0:\gcd(x,y)=1\}, \] and let \(C_8\) contain \(1\) and the positive integers whose prime factors are \(1\) or \(3\pmod8\). \begin{theorem}\label{thm:minus8-exact} There are no exceptional admissible targets: \[ \begin{aligned} C_8+C_8&=\{N\ge2:N\ {\rm even},\ 8\nmid N\},\\ C_8+2C_8&=\{N\ge3:N\ {\rm odd}\},\\ P_8+P_8&=\{N\ge2:16\nmid N\}. \end{aligned} \] For every \(N\ge2^{78}\) with \(16\nmid N\), \[ \#\{(a,b)\in P_8^2:a+b=N\} \ge N^{2/3}/4800 . \] \end{theorem} Write \(Q,P,C\) for these objects in this section. The exact dyadic shell formulas will supply both a large-core theorem and an analytic prime boundary. Sharper selected-shell bounds then reduce the remaining work to \(107\) rational interval tests and complete finite core coverage. \subsection{Primitive values and dyadic reduction} The ring \(R=\Z[\sqrt{-2}]\) is norm Euclidean: rounding both coordinates leaves error norm at most \(1/4+2/4=3/4<1\). Thus it is principal. An inert prime, congruent to \(5,7\pmod8\), dividing a norm divides both coordinates. Primitivity permits the ramified prime ideal of norm two to exponent at most one. For split primes, orient each full prime power to one of the two conjugate principal prime ideals. These observations prove \begin{equation} P=C\mathbin{\dot\cup}2C. \label{eq:m8-value} \end{equation} Every \(C\)-value is \(1\) or \(3\pmod8\). Consequently \begin{equation} P+P=(C+C)\cup2(C+C)\cup(C+2C). \label{eq:m8-union} \end{equation} The first two terms have residues respectively \(2,4,6\pmod8\) and \(4,8,12\pmod {16}\), whereas the last has odd targets at least three. This proves every local exclusion in the theorem; the two core identities imply the full identity. Multiplication by \(\varpi=\sqrt{-2}\) acts by \[ (x,y)\longmapsto(-2y,x). \] It bijects primitive odd-norm vectors with primitive even-norm vectors; the latter have even first coordinate and odd second coordinate, and the inverse is \((X,Y)\mapsto(Y,-X/2)\). At odd primes the determinant is a unit. Thus every division or multiplication by two below is justified on the indicated parity class; no general closure of primitive vectors under doubling is being assumed. \subsection{Exact dyadic shell formulas} Let \(r(N)\) count the full signed ordered \(Q\oplus Q\) shell, let \(s(N)\) count \(Q\oplus4Q\), and let \(G_2(N)\) count shell points for which neither binary coordinate pair is divisible by two. Only dyadic primitivity is imposed in \(G_2\). With \(\sigma(t)=0\) for nonpositive or nonintegral \(t\), define \[ f_{32}(\tau)=\eta(4\tau)^2\eta(8\tau)^2 =\sum_{N\ge1}a_Nq^N =q\prod_{j\ge1}(1-q^{4j})^2(1-q^{8j})^2 . \] Then \begin{equation} \begin{aligned} r(N)&=4\sigma(N)-4\sigma(N/2)+8\sigma(N/4)-32\sigma(N/8),\\ s(N)&=\sigma(N)-\sigma(N/2)+8\sigma(N/16)-32\sigma(N/32)+a_N,\\ G_2(N)&=r(N)-2s(N)+r(N/4). \end{aligned}\label{eq:m8-theta} \end{equation} We include the modular verification, so these formulas do not depend on extrapolating a finite list of shell counts. The even Gram matrices of \(Q\oplus Q\) and \(Q\oplus4Q\) have levels \(8,32\) and square determinants \(64,1024\). Their theta series therefore belong respectively to \(M_2(\Gamma_0(8))\) and \(M_2(\Gamma_0(32))\), with trivial character. Put \(E_2(\tau)=1-24\sum_{n\ge1}\sigma(n)q^n\). The proposed Eisenstein series are \[ \begin{aligned} E_r&=-\tfrac1{24} \{4E_2(\tau)-4E_2(2\tau)+8E_2(4\tau)-32E_2(8\tau)\},\\ E_s&=-\tfrac1{24} \{E_2(\tau)-E_2(2\tau)+8E_2(16\tau)-32E_2(32\tau)\}. \end{aligned} \] To verify holomorphy at all cusps, write \[ D_d(\tau)=dE_2(d\tau)-E_2(\tau) =\frac1{2\pi i}\frac d{d\tau} \log\frac{\Delta(d\tau)}{\Delta(\tau)} . \] The quotient is a modular function, nonzero in the upper half-plane and meromorphic at cusps. If at a cusp of width \(h\) it is \(q_c^\nu U(q_c)\), \(U(0)\ne0\), its logarithmic derivative has expansion \[ \frac1h\left(\nu+ q_c\frac{U'(q_c)}{U(q_c)}\right). \] There is no cusp pole. Hence \(D_d\) is holomorphic of weight two, and \[ E_r=D_2/12-D_4/12+D_8/6,\qquad E_s=D_2/48-D_{16}/48+D_{32}/24 \] have the required levels. Both have constant term one. Martin--Ono \cite[Theorem 1, level-32 row]{MartinOno} identify \(f_{32}\) as a normalized weight-two newform. The Sturm bounds are \(2\) at level \(8\) and \(8\) at level \(32\) \cite[Theorem 1]{Sturm}. The necessary coefficient lists are \[ \begin{array}{c|rrrrrrrrr} N&0&1&2&3&4&5&6&7&8\\ \hline {}[q^N]\Theta_{Q\oplus Q}&1&4&8&&&&&&\\ {}[q^N]E_r&1&4&8&&&&&&\\ {}[q^N]\Theta_{Q\oplus4Q}&1&2&2&4&4&4&8&8&8\\ {}[q^N](E_s+f_{32})&1&2&2&4&4&4&8&8&8 \end{array} \] They follow by direct lattice enumeration, divisor sums, and truncated multiplication of the displayed eta product. All coefficients are integers. Apply Sturm's congruence theorem modulo every rational prime: each coefficient of the difference is divisible by every prime, so is zero. This proves the first two formulas in \eqref{eq:m8-theta}; the last is inclusion-exclusion. The CM formula in \cite[Theorems 2 and 3]{MartinOno} gives \(|a_p|\le2\sqrt p\) at every odd prime, with \(a_p=0\) for \(p\equiv3\pmod4\). Its unramified Hecke roots have modulus \(\sqrt p\); the Euler recurrence and multiplicativity imply \(|a_N|\le\tau(N)\sqrt N\) for odd \(N\). The eta product is supported on \(N\equiv1\pmod4\), so its even coefficients vanish. Write \(N=2^em\), \(m\) odd. For \(0\le e\le3\), \[ r(N)=4c_e\sigma(m),\qquad (c_0,c_1,c_2,c_3)=(1,2,6,6), \] and \begin{equation} (G_2(m),G_2(2m),G_2(4m),G_2(8m)) =(2\sigma(m)-2a_m,\ 4\sigma(m),\ 20\sigma(m),\ 16\sigma(m)). \label{eq:m8-dyadic} \end{equation} For \(e\ge4\), \(G_2(N)=0\). \subsection{The nonmaximal dyadic order and Hecke bound} Let \(H\) be the Hurwitz order in Hamilton's quaternion algebra, with generators \(I,J,K\). Put \(\iota=I+J\), \(j=K\), and \[ O=\Z[\iota]\oplus\Z[\iota]j =\{a+bI+cJ+dK:a,b,c,d\in\Z,\ b\equiv c\pmod2\}. \] Here \(\iota^2=-2\), \(j^2=-1\), \(j\iota=-\iota j\), and the norm is \(Q\oplus Q\). This order has index four in \(H\); it is not maximal at two. We verify its locally principal right-ideal class number directly. The Hurwitz order is norm Euclidean. After rounding a real quaternion to an integer vector, let its coordinate errors be \(|r_i|\le1/2\), and put \(s=\sum r_i^2\). The squared distance to an appropriate half-integer vector is \(s-\sum|r_i|+1\). Since \(s\le\frac12\sum|r_i|\), one of the two distances is at most \(1/2\). Thus all Hurwitz right ideals are principal. Away from two, \(O_\ell=H_\ell\). Normalize the additive volume of \(H_2\) to one. Its half-integer coset consists of odd-norm elements, as does half its integer coset; hence \(\operatorname{vol}(H_2^\times)=3/4\). The additive volume of \(O_2\) is \(1/4\), and half its elements have odd norm, so \(\operatorname{vol}(O_2^\times)=1/8\). The unit index is six. The \(24\) Hurwitz units have stabilizer \[ O^\times=\{\pm1,\pm K\} \] of size four on these cosets, so act transitively. Principalization in \(H\), followed by this unit transitivity at two, principalizes every locally principal \(O\)-ideal. The dyadic unit norm map is onto \(\Z_2^\times\). Indeed \(1\), \(1+\iota\), \(1+\iota+\iota j\), \(1+\iota+2j\) have norms \(1,3,5,7\). Their inverses belong to \(O_2\). Every \(2\)-adic unit is one of these four norms times a unit square, supplied by a rational \(2\)-adic scalar. For odd inert \(p\), reduction gives \(O/pO=M_2(\mathbb F_p)\). The projective unit image \(\Gamma_p\) has order two, and the fine quotient is \(Y_p=\Gamma_p\backslash\mathrm{PGL}_2(\mathbb F_p)\), with uniform measure. Class number one identifies this quotient with the adelic congruence quotient: its stabilizer is precisely the projective unit image. To check this, local scalar corrections have even norm valuations and combine into one rational scalar, leaving an \(O\)-unit. For \(p\nmid n\), define concretely \begin{equation} (T_nh)(\Gamma_pg) =\frac14\sum_{\operatorname{nrd}\alpha=n} h(\Gamma_p\rho_p(\alpha)g). \label{eq:m8-Hecke} \end{equation} For odd \(m\), this is the all-integral-ideal operator, of degree \(\sigma(m)\). Inverse principal generators give the displayed orientation by quaternion conjugation. The global norm characters on \(Y_p\) are exactly \(1\) and the determinant square character \(\chi_p\). The norm surjectivity just proved prevents an additional dyadic character. At \(p\), central scalars and the reduction kernel leave only determinant square classes. Lemma~\ref{lem:hecke-transfer}, including its strong-approximation exclusion of finite-dimensional split-place factors, therefore yields \begin{equation} \|T_m\|_{\{1,\chi_p\}^{\perp}} \le\tau(m)\sqrt m \quad(m\ {\rm odd},\ p\nmid m). \label{eq:m8-Ramanujan} \end{equation} All primes dividing \(m\) have maximal split local level. No unramified theorem at two has been used. For \(0\le e\le3\), let \(A_e=T_{2^e}\) be given by \eqref{eq:m8-Hecke}. It is self-adjoint with nonnegative entries, row sum \(c_e\), and norm at most \(c_e\). It has \(\chi_p\)-eigenvalue \((2/p)^ec_e\). Coprime local factorization gives \begin{equation} T_{2^em}=A_eT_m=T_mA_e. \label{eq:m8-factor} \end{equation} For example, for \(\alpha\in O\) of norm \(2^em\), the right superideal with localization \(O_2\) at two and \(\alpha O_\ell\) at odd primes is integral, locally principal, and of norm \(m\). Write it as \(\beta O\), \(\operatorname{nrd}\beta=m\); then \(\alpha=\beta\gamma\), \(\operatorname{nrd}\gamma=2^e\). Exactly four choices differ by an \(O\)-unit. Reversing the coprime local factors gives commutation. Thus the complementary norm of \(T_N\) is at most \(c_e\tau(m)\sqrt m\). \subsection{The uniform sieve and its infinite ranges} Put \[ \begin{aligned} \mathcal I&=\{p\ {\rm prime}:p\equiv5,7\pmod8\},\\ E(m)&=\frac{\tau(m)\sqrt m}{\sigma(m)},\\ M(X)&=\max_{1\le t\le X} \prod_{\substack{\ell\mid t\\\ell\equiv1,3(8)}} (v_\ell(t)+1). \end{aligned} \] Let \(b_p(N)\) count full-shell points with at least one binary coordinate pair divisible by \(p\). For \(p\nmid N\), that condition is the nonsplit-torus normalizer \(H_p\), of order \(2(p+1)\). Its fraction in \(\mathrm{PGL}_2(\mathbb F_p)\) is \(\mu_p=2/(p(p-1))\). Its indicator \(u\) has mean \(\mu_p\), zero coefficient on \(\chi_p\), and \(\|u-\mu_p\|^2=\mu_p(1-\mu_p)\), since the torus norm is onto. The operator commutes with the right action and is constant on the support of \(u\). Thus \(\langle u,T_Nu\rangle=\mu_pb_p(N)/4\). Equations \eqref{eq:m8-Ramanujan}--\eqref{eq:m8-factor} give \begin{equation} \frac{b_p(N)}{r(N)} \le\mu_p+(1-\mu_p)E(m) \le\mu_p+E(m). \label{eq:m8-small} \end{equation} If \(p\mid N\), anisotropy forces both blocks divisible by \(p\). The spoiled count is zero at valuation one; otherwise its ratio is \(\sigma(m/p^2)/\sigma(m)
P}}p^{-2}
&\le1/(4P)+2/P^2. \end{aligned}\label{eq:m8-progression}
\end{equation}
The last two statements follow from the progressions \(5,7\pmod8\).
For the first, sum inert primes below \(64\) exactly and bound
the rest by
\[
\frac52\left(\frac1{69^2}+\frac1{8\cdot69}
+\frac1{71^2}+\frac1{8\cdot71}\right).
\]
The resulting bound is
\(B<3/16\); its exact rational value is in the supplement.
By \eqref{eq:m8-dyadic}, \(G_2(N)/r(N)\ge1/2-E(m)/2\).
A sufficient condition for a positive surviving shell is
\begin{equation}
B+E(m)(P/4+3/2)
+L_eM(N)\{1/(4P)+2/P^2\}<1/2. \label{eq:m8-coarse}
\end{equation}
Surviving norms are positive, have no inert factor, and satisfy
the dyadic condition in \eqref{eq:m8-value}; their values lie in \(P\).
The complete finite Euler products give
\begin{equation}
\begin{aligned}
M(X)&\le6X^{1/6},& M(X)&\le2X^{1/4},\\
\tau(m)&\le7m^{1/6}\quad(m\ {\rm inert\!-\!supported}),&
\tau(m)&\le42m^{1/6}\quad(m\ {\rm odd}). \end{aligned}\label{eq:m8-Euler}
\end{equation}
For the sixth-power bounds, maximize
\((a+1)^6/p^a\) at each odd prime below \(64\), over \(0\le a\le6\).
The factors decrease afterwards, and factors at primes at least
\(64\) are at most one by \(a+1\le2^a\).
The split and inert products are respectively below \(6^6,7^6\);
their product gives \(42^6\).
For the fourth-power split bound only primes \(3,11\) contribute,
and their product is \(4096/297<2^4\).
These calculations establish \eqref{eq:m8-Euler} for all \(X,m\).
Since \(N\le8m\), \(M(N)\le9m^{1/6}\).
For arbitrary odd \(m\), use \(P=m^{1/4}\) in
\eqref{eq:m8-coarse}. The terms after \(B\) are at most
\[
\frac{45}{2}m^{-1/12}+159m^{-1/3}.
\]
If \(m\ge2^{75}\), use \(m^{1/12}>224/3\) to obtain
\begin{equation}
\frac3{16}+\frac{135}{448}
+\frac{159}{2^{25}}<\frac12.
\label{eq:m8-globalnumeric}
\end{equation}
This proves the stated global onset \(N\ge2^{78}\).
The margin is
\(2620327/234881024>1/100\).
In every eligible stratum \(r(N)\ge3N\), and each ordered value
pair has at most \((2M(N))^2\le144N^{1/3}\) witnesses.
The count is therefore at least \(N^{2/3}/4800\).
Two additional uniform families follow from the same inequality.
For inert-supported \(m\), use \(P=2m^{1/4}\) and
\(\tau(m)\le7m^{1/6}\).
The error after \(B\) is at most
\[
\frac{19}{2}m^{-1/12}
+\frac{69}{2}m^{-1/3}.
\]
For \(m\ge2^{60}\), its total with \(B\) is bounded by
\begin{equation}
\frac3{16}+\frac{19}{64}
+\frac{69}{2^{21}}<\frac12.
\label{eq:m8-coretail}
\end{equation}
Thus all \(2^em\), \(0\le e\le3\), in this family lie in \(P+P\).
At \(m\) the two values must be in different parity classes;
at \(2m\) both are in \(C\); at \(8m\) both are in \(2C\),
and division by two gives \(4m\in C+C\).
This proves the three core conclusions above the onset.
For an odd prime \(n\), use \(E(n)\le2n^{-1/2}\),
\(M(n)\le2n^{1/4}\), \(L_0=4\), and \(P=2n^{3/8}\).
The error after \(B\) is at most \(2n^{-1/8}+7n^{-1/2}\).
For \(n\ge2^{22}\), it is dominated by
\(3/10+7/2048\), and
\begin{equation}
3/16+3/10+7/2048<1/2. \label{eq:m8-prime}
\end{equation}
Every such prime lies in \(P+P\); since it is odd,
it lies in \(C+2C\). The finite mixed-core coverage below will
supply all smaller odd primes, so no separate prime-family
search is required.
\subsection{Sharper selected shells}
The improved lattice bounds are
\begin{equation}
\begin{aligned}
S_Q(X)&:=\#\{(x,y):0P}}p^{-2}.
\label{eq:m8-selectedtest}
\end{equation}
This follows by subtracting the union of all bad events from the
selected subset; no Hecke stability of that subset is needed.
After sieving the rows give respectively \(C+2C,C+C,C+C\).
\subsection{The complete interval and finite certificates}
For inert-supported odd \(m\), a sharper bound is
\begin{equation}
E(m)<\frac{18}{5}m^{-1/3}. \label{eq:m8-Esharp}
\end{equation}
Raise the normalized prime-power factors to the sixth power:
they are
\((a+1)^6p^{5a}/\sigma(p^a)^6\).
Primes at least \(64\) contribute at most one.
For each smaller inert prime, exponents through three suffice:
the subsequent ratio is strictly less than
\((5/4)^6/p\le(5/4)^6/5<1\).
The exact product of the finite maxima is below \((18/5)^6\).
Thus \eqref{eq:m8-Esharp} is an unrestricted bound.
For the bounded \(M\)-envelope, enumerate nonincreasing exponents
on initial segments of the smallest split primes.
Swapping exponents \(a2^{63}\); these bound both its depth
and exponent range.
Set \(R_0=65536\), \(D_0=2^{64}\). For \(P