Abstract:We apply the structure theory of finite dimensional algebras in order to deduce dimension formulas for spaces of period numbers, i.e., complex numbers defined by integrals of algebraic nature. We get a complete and conceptually clear answer in the case of $1$-periods, generalising classical results like Baker's theorem on the logarithms of algebraic numbers and partial results in Huber--W{ü}stholz \cite{huber-wuestholz}.
The application to the case of Mixed Tate Motives (i.e., Multiple Zeta Values) recovers the dimension estimates of Deligne--Goncharov \cite{deligne-goncharov}.
What problem does this paper attempt to address?
The core problem that this paper attempts to solve is about the dimension formula of the period space, especially deriving these formulas through motives and species theories. Specifically, the author attempts to provide a complete and conceptually clear upper bound estimate for the period space of mixed motives. This includes:
1. **Upper bound estimate of the dimension of the period space**: For a given mixed motive \(M\), the author aims to derive a general upper - bound formula for the \(\mathbb{Q}\)-dimension of its period space \(P\langle M\rangle\). This formula is especially applicable to 1 - motives, that is, the case where \(\omega\) is a 1 - form in the period integral.
2. **Interpretation of the "error term"**: The author introduces a new perspective, interpreting the mysterious "error term" that appeared in previous studies as the dimension of the Yoneda 2 - extension space. This provides a deeper conceptual framework for understanding the dimension of the period space.
3. **Recovering known results and generalizing new results**: In certain specific cases, such as Mixed Tate Motives, this formula can recover the results of Deligne - Goncharov and be generalized to more types of motives.
### Formula derivation and key conclusions
The main formulas in the paper are as follows:
Let \(M\) be a mixed motive, and denote \(\langle M\rangle\) as the full Abelian subcategory generated by \(M\), which is closed in the Abelian category of all motives. Let \(M_1,\ldots,M_n\) be the representatives of the isomorphism classes of simple objects in \(\langle M\rangle\), \(D_i = \text{End}(M_i)\) be a finite - dimensional division ring, \(d_i=\dim_{\mathbb{Q}}D_i\), \(m_i = \dim D_iH_{\text{sing}}(M_i)\), and \(S_M\) be the species of \(\langle M\rangle\).
Then we have the following theorem:
\[
\text{dim}_{\mathbb{Q}}P\langle M\rangle\leq\sum_{i = 1}^n m_i^2d_i+\sum_{i,j = 1}^n\sum_{\gamma:i\rightsquigarrow j}m_im_j\text{dim}_{\mathbb{Q}}E(\gamma)-\sum_{i,j = 1}^n m_im_j\sum_{k = 2}^{\infty}\text{Ext}^k_{\langle M\rangle}(M_i,M_j)
\]
where:
\[
\text{dim}_{\mathbb{Q}}E(\gamma)=\prod_{\epsilon:v\rightarrow w\text{ on }\gamma}\text{dim}_{\mathbb{Q}}\text{Ext}^1_{\langle M\rangle}(M_w,M_v)\bigg/\prod_{v\in\circ\gamma}d_v
\]
Here \(\circ\gamma\) represents all vertices on the path \(\gamma\) that are different from the endpoints.
If the period conjecture holds for \(M\), then the equality holds in the following two cases:
- \(\langle M\rangle\) is hereditary, that is, all \(\text{Ext}^k_{\langle M\rangle}\) vanish for \(k\geq2\);
- The maximum length of paths in the quiver of the species \(S_M\) is 2, especially \(\text{Ext}^k_{