Samart’s conjecture $n_{4}(81)=40M_{7}$:
the exact CM evaluation and the two obstructions
A status report, companion to Mahler measures at interior CM points

Let

$$P(x,y,z):=\frac{x^{4}+y^{4}+z^{4}+1}{xyz},$$

and denote by $\mathrm{m}$ the logarithmic Mahler measure,

$$\mathrm{m}(Q)=\int_{\mathbb{T}^{3}}\log|Q(x,y,z)|\,\frac{\mathrm{d}x}{2\pi\mathrm{i}x}\frac{\mathrm{d}y}{2\pi\mathrm{i}y}\frac{\mathrm{d}z}{2\pi\mathrm{i}z},\qquad\mathbb{T}=\{|w|=1\}.$$

Since $\mathrm{m}(xyz)=0$, one has $\mathrm{m}(P+c)=\mathrm{m}(x^{4}+y^{4}+z^{4}+1+c\,xyz)$, and the substitution $x\mapsto\mathrm{i}x$ (Haar-invariant, sending $c\mapsto\mathrm{i}c$) shows that the value is unchanged when $c$ is multiplied by a fourth root of unity. Following Samart’s table notation [3] we therefore write, unambiguously,

$$n_{4}(s):=4\,\mathrm{m}\bigl(x^{4}+y^{4}+z^{4}+1+s^{1/4}xyz\bigr),\qquad s\in\mathbb{C},$$

(1)

for either choice of the fourth root. (In [2] the same function is denoted $f_{4}$.)

Conjecture (Samart, [3, Table 6]). $n_{4}(81)=40M_{7}$, where $M_{7}:=L^{\prime}(g_{7},0)$ and $g_{7}(\tau)=\eta(\tau)^{3}\eta(7\tau)^{3}$ is the unique normalized newform of $S_{3}(\Gamma_{0}(7),\chi_{-7})$ (LMFDB label 7.3.b.a [10]).

Among Samart’s open interior-point entries this one is structurally the cleanest: the attached CM point has discriminant $-7$ (class number $1$), and the conjectural value is a single $L$-value with no Dirichlet term, in contrast to the two-term formula of the $n_{2}$ conjugate pair proved in [1]. Fei [4] proved the identity at the level of the real part of the holomorphic (modified) Mahler measure $\operatorname{Re}\widetilde{\mathrm{m}}$; explicitly, in Fei’s table the row $c=81$ is placed between a pair of double lines marking the entries for which $\operatorname{Re}\widetilde{\mathrm{m}}$ is not guaranteed to equal the genuine Mahler measure [4, Note (3) and Caution 2.3]. The passage to the genuine Mahler measure at the interior parameter $s^{1/4}=3$ was left open, and Section 6 below determines what actually happens there: the two sides differ in the fifth significant digit.

The companion paper [1] proved Samart’s conjectures for the $n_{2}$-family at $s=1$ (Theorem A, differential-comparison continuation) and at $s=(47\pm 45\sqrt{-7})/2$ (Theorem B, the Fricke trick). The present conjecture was considered as a third target and dropped: two independent obstructions, documented quantitatively in Sections 4 and 5 below, kill both the continuation machine and the Fricke shortcut for the $n_{4}$-family. This note archives the state of the problem for future work:

1. (P1)

   Done, exact. The $L$-value side. Samart’s Eisenstein–Kronecker expression $\mathrm{EK}_{4}$ at the CM point $\tau_{2}=(7+\sqrt{-7})/4$ (where $s_{4}(\tau_{2})=81$) evaluates exactly to $40M_{7}$ (Theorem 3.4, full proof in Section 3; certified to $2.7\times 10^{-59}$ by the $45$ checks of verify_P1_n4_81.py).

2. (P2)

   Blocked. The Mahler side $n_{4}(81)\stackrel{{\scriptstyle?}}{{=}}\mathrm{EK}_{4}(\tau_{2})$. Obstruction 1: the critical image $P(\mathbb{T}^{3})$ is the two-dimensional astroid disc $\{|{\operatorname{Re}c}|^{2/3}+|{\operatorname{Im}c}|^{2/3}\leq 4^{2/3}\}$, and $c=3$ is an interior point, so the open mapping theorem forbids any path continuation (Section 4). Obstruction 2, the decisive one: Samart’s $U$-series is proved to equal $n_{4}(s_{4}(\tau))$ only for $\operatorname{Im}\tau\geq 1/\sqrt{2}$, and below that threshold it leaves the geometric sheet of the holomorphic Mahler measure everywhere — not only inside the critical image (Section 5).

In one sentence: for the $n_{2}$-family the bad locus was the one-dimensional slit $[0,64]$, which a certified path can go around; for the $n_{4}$-family the bad locus is a two-dimensional disc and the analytic formula itself is on the wrong sheet below $\operatorname{Im}\tau=1/\sqrt{2}$, so there is nothing left to continue.

We keep the notation of [1]. Throughout, $\tau=x+\mathrm{i}y\in\mathbb{H}$, $q=\mathrm{e}^{2\pi\mathrm{i}\tau}$, $\eta(\tau)=q^{1/24}\prod_{n\geq 1}(1-q^{n})$, and $\sum^{\prime}$ denotes omission of the zero term. The field $K=\mathbb{Q}(\sqrt{-7})$ has ring of integers $\mathcal{O}_{K}=\mathbb{Z}[\varpi]$, class number $h(-7)=1$ and units $\{\pm 1\}$, where

$$\varpi:=\frac{1+\sqrt{-7}}{2},\qquad\varpi+\bar{\varpi}=1,\qquad\varpi\bar{\varpi}=N(\varpi)=2,\qquad\varpi^{2}=\varpi-2,$$

$$\varpi^{2}+\bar{\varpi}^{2}=-3,\qquad\operatorname{Re}\varpi^{2}=\operatorname{Re}\bar{\varpi}^{2}=-\tfrac{3}{2},\qquad(\bar{\varpi})=\bar{\varpi}\mathcal{O}_{K}=2\mathbb{Z}+\bar{\varpi}\mathbb{Z}.$$

(2)

We write $L_{3}:=L(g_{7},3)$. The theta identity [1, Appendix B] gives

$${\sum_{\alpha\in\mathcal{O}_{K}}}^{\!\prime}\frac{\alpha^{2}}{|\alpha|^{2s}}=2L(g_{7},s),\qquad{\sum_{\alpha\in\mathcal{O}_{K}}}^{\!\prime}\frac{\bar{\alpha}^{2}}{|\alpha|^{6}}=2L_{3},\qquad{\sum_{\alpha\in\mathcal{O}_{K}}}^{\!\prime}|\alpha|^{-4}=2\zeta_{K}(2),$$

(3)

the middle equality by conjugation ($L_{3}\in\mathbb{R}$ since $a_{n}(g_{7})\in\mathbb{Z}$), and the functional equation of $g_{7}$ (root number $+1$) gives

$$M_{7}=L^{\prime}(g_{7},0)=\frac{7\sqrt{7}}{4\pi^{3}}\,L_{3}.$$

(4)

Both (3) and (4) are quoted theorems, certified numerically in [1] (verify_P1.py, lvalue_g7.py) and re-certified for this note in verify_P1_n4_81.py (check [L1]: the theta identity to $q^{60}$ as exact integers, (4) to $10^{-61}$).

For $z\in\mathbb{C}\setminus\{0\}$ let

$$F(z):=\frac{4(\operatorname{Re}z)^{2}}{|z|^{6}}-\frac{1}{|z|^{4}},$$

and for $\tau\in\mathbb{H}$ and $d\geq 1$ define the double series

$$T_{d}(\tau):={\sum_{(m,n)\in\mathbb{Z}^{2}}}^{\!\prime}F(dm\tau+n),$$

(5)

summed by rows in $m$; each row is absolutely convergent and the row sums decay like $|m|^{-2}$ (Poisson summation), and the $m=0$ row is $\sum_{n\neq 0}(4n^{2}\cdot n^{-6}-n^{-4})=6\zeta(4)=\pi^{4}/15$ [1, §2.4]. Define, for all $\tau\in\mathbb{H}$,

$$\mathrm{EK}_{4}(\tau):=\frac{10\,\operatorname{Im}\tau}{\pi^{3}}\bigl(-T_{1}(\tau)+4T_{2}(\tau)\bigr),$$

(6)

real-analytic on all of $\mathbb{H}$. Since $4(\operatorname{Re}\lambda)^{2}=(\lambda+\bar{\lambda})^{2}=\lambda^{2}+2|\lambda|^{2}+\bar{\lambda}^{2}$, the summand rewrites as $F(\lambda)=2\operatorname{Re}\bigl(\bar{\lambda}^{2}/|\lambda|^{6}\bigr)+|\lambda|^{-4}$, so with $\Lambda_{d}(\tau):=\mathbb{Z}+\mathbb{Z}\,d\tau$,

$$T_{d}(\tau)={\sum_{\lambda\in\Lambda_{d}(\tau)}}^{\!\prime}\Bigl[2\operatorname{Re}\frac{\bar{\lambda}^{2}}{|\lambda|^{6}}+\frac{1}{|\lambda|^{4}}\Bigr],$$

(7)

the form used for the CM evaluation.

Define also, for $j\geq 1$,

$$U_{j}(\tau):=\sum_{m\neq 0}\frac{1}{m}\sum_{n\in\mathbb{Z}}(jm\tau+n)^{-3}=2\pi^{3}\sum_{m\geq 1}\frac{\cos(j\pi m\tau)}{m\,\sin^{3}(j\pi m\tau)},$$

absolutely convergent on $\mathbb{H}$ exactly as in [1, §2.4].

The $U$-series form (8) is Samart’s exponential-sum formula for the $n_{4}$-family; it converges for every $\tau\in\mathbb{H}$.

Define

$$c(\tau):=\Bigl(\frac{\eta(2\tau)}{\eta(\tau)}\Bigr)^{6}\bigl(16A(\tau)^{4}+A(\tau)^{-4}\bigr),\qquad A(\tau):=\frac{\eta(\tau)\,\eta(4\tau)^{2}}{\eta(2\tau)^{3}},$$

(9)

holomorphic and nonvanishing-free on $\mathbb{H}$, and $s_{4}(\tau):=c(\tau)^{4}$ (this agrees with Samart’s $\Delta$-quotient definition of his signature-$4$ modular function, checked numerically to $40$ digits at sample points; diag_n4_astroid.py, Part 1). With Samart’s normalization [2, Lemma 2.2],

$$s_{4}(\mathrm{i})=648,\qquad s_{4}\bigl(\mathrm{i}/\sqrt{2}\bigr)=256,\qquad s_{4}(\mathrm{i}y)\geq 256\ \ (y>0),$$

so on the imaginary axis $c(\mathrm{i}y)$ is real with $c(\mathrm{i}y)\geq 4$.

We take this as a quoted theorem. Its proof region is essential to what follows: the identity is not asserted, and in fact fails, below $\operatorname{Im}\tau=1/\sqrt{2}$ (Section 5).

Both lattices are $\mathcal{O}_{K}$-ideals: unlike the $\tau_{w}$ evaluation of [1, §5] there is no imprimitive order and no Euler factor at the primes above $2$.

For a fractional $\mathcal{O}_{K}$-ideal $\mathfrak{a}$ define

$$B(\mathfrak{a}):={\sum_{\gamma\in\mathfrak{a}}}^{\!\prime}|\gamma|^{-4},\qquad G(\mathfrak{a}):={\sum_{\gamma\in\mathfrak{a}}}^{\!\prime}\frac{\bar{\gamma}^{2}}{|\gamma|^{6}}.$$

By (3),

$$B(\mathcal{O}_{K})=2\zeta_{K}(2),\qquad G(\mathcal{O}_{K})=2L_{3}.$$

(13)

For the principal ideal $(\bar{\varpi})$ write $\alpha=\bar{\varpi}\gamma$, $\gamma\in\mathcal{O}_{K}\setminus\{0\}$; then $|\alpha|^{-4}=|\bar{\varpi}|^{-4}|\gamma|^{-4}$ and $\bar{\alpha}^{2}|\alpha|^{-6}=(\varpi^{2}/|\bar{\varpi}|^{6})\,\bar{\gamma}^{2}|\gamma|^{-6}$, so with $N(\bar{\varpi})=2$ and $|\bar{\varpi}|^{6}=8$,

$$B\bigl((\bar{\varpi})\bigr)=\frac{2\zeta_{K}(2)}{N(\bar{\varpi})^{2}}=\frac{\zeta_{K}(2)}{2},\qquad G\bigl((\bar{\varpi})\bigr)=\frac{\varpi^{2}}{8}\cdot 2L_{3}=\frac{\varpi^{2}L_{3}}{4}.$$

(14)

Note $G((\bar{\varpi}))$ is genuinely complex ($(\bar{\varpi})$ is not conjugation-stable); only its real part enters the $T$-sums, and by (2),

$$\operatorname{Re}G\bigl((\bar{\varpi})\bigr)=\frac{\operatorname{Re}\varpi^{2}}{4}\,L_{3}=-\frac{3}{8}\,L_{3}.$$

(15)

(Direct Poisson row summation confirms (13)–(15), including the imaginary part $\operatorname{Im}G((\bar{\varpi}))=\sqrt{7}\,L_{3}/8$, to $\sim 10^{-60}$; verify_P1_n4_81.py, checks [S1]–[S4].)

Unlike the $\tau_{w}$ case of [1] (where the surviving $\zeta_{K}(2)$-term produced the Dirichlet value $d_{7}$), the coefficients $-8$ and $+8$ cancel exactly, leaving a single $L$-value — the structural prerequisite for the one-term formula $40M_{7}$.