アブストラクト事後公開 — 2018年度年会(於:東京大学)
数学基礎論および歴史分科会
特別講演 Settheoretic reflection principles 渕野 昌 (神戸大システム情報) For a class $\mathcal{C}$ of structures (with a fixed notion $\mathcal{N}$ of substructure) and a property $\mathcal{P}$, the reflection cardinal of $(\mathcal{C},\mathcal{P})$ is the minimal cardinal $\kappa$ such that, for any $M\in\mathcal{C}$ of cardinality $>\kappa$, if $M$ does not satisfies the property $P$, then there are stationarily many substructures $N$ of $M$ of cardinality $<\kappa$. If $kappa$ is the reflection cardinal of $(\mathcal{C},\mathcal{N})$, we shall write $\kappa=Ref(\mathcal{C}, \mathcal{P})$. By choosing $\mathcal{C}$, $\mathcal{N}$ and $\mathcal{P}$, we can represent many settheoretic reflection statements. If, for example $\mathcal{P}$ is simply a contradiction, and $\mathcal{N}$ is the elementary submodel relation for some logic $\mathcal{L}$, then $\kappa=Ref(\mathcal{C}, \mathcal{P})$ is the strong form of Downward LöwenheimSkolem Theorem down to $<\kappa$ for $\mathcal{L}$. Of these reflection statements, the cases $\alpha_2=Ref(\mathcal{C}, \mathcal{P})$ and $2^{\aleph_0}=Ref(\mathcal{C}, \mathcal{P})$ seems to be of special interest. The former may be interpreted as a pronouncement that the first uncountable cardinal $\aleph_1$ captures the situation $\neg\mathcal{P}$ good enough while the latter as the pronouncement that the continuum is large enough in connection with the property $\mathcal{P}$. For a class $\mathcal{C}$ of structures (with a fixed notion $\mathcal{N}$ of substructure) and a property $\mathcal{P}$, the reflection cardinal of $(\mathcal{C},\mathcal{P})$ is the minimal cardinal $\kappa$ such that, for any $M\in\mathcal{C}$ of cardinality $>\kappa$, if $M$ does not satisfies the property $P$, then there are stationarily many substructures $N$ of $M$ of cardinality $<\kappa$. If $\kappa$ is the reflection cardinal of $(\mathcal{C},\mathcal{N})$, we shall write $\kappa=R{e}fl(\mathcal{C}, \mathcal{P})$. By choosing $\mathcal{C}$, $\mathcal{N}$ and $\mathcal{P}$, we can represent many settheoretic reflection statements. If, for example $\mathcal{P}$ is simply a contradiction, and $\mathcal{N}$ is the elementary submodel relation for some logic $\mathcal{L}$, then $\kappa=R{e}fl(\mathcal{C}, \mathcal{P})$ is the strong form of Downward LöwenheimSkolem Theorem down to $<\kappa$ for $\mathcal{L}$. Of these reflection statements, the cases $\aleph_2=R{e}fl(\mathcal{C}, \mathcal{P})$ and $2^{\aleph_0}=R{e}fl(\mathcal{C}, \mathcal{P})$ seems to be of special interest. The former may be interpreted as a pronouncement that the first uncountable cardinal $\aleph_1$ captures the situation $\neg\mathcal{P}$ good enough while the latter as the pronouncement that the continuum is large enough in connection with the property $\mathcal{P}$. The stronger assertions among $\aleph_2=R{e}fl(\mathcal{C}, \mathcal{P})$ imply the Continuum Hypothesis while assertions of the form $2^{\aleph_0}=R{e}fl(\mathcal{C}, \mathcal{P})$ tend to imply that the continuum is extremely large. Most of the natural assertions of the form $\aleph_2=R{e}fl(\mathcal{C}, \mathcal{P})$ or $2^{\aleph_0}=R{e}fl(\mathcal{C}, \mathcal{P})$ involves some kind of countability in the property $\mathcal{P}$. This is the case with the reflection assertion $\aleph_2=R{e}fl(\mathcal{C}, \mathcal{P})$ where $\mathcal{C}$ is the class of all graphs with induced subgraphs as the notion of substructure and $\mathcal{P}$ is the property “of countable coloring number”. It is shown that this assertion is equivalent to the Fodortype Reflection Principle (FRP). We can also consider the reflection number for the property obtained from these properties of countable character by replacing the countability by of cardinality $\kappa$. Recently many interesting results about reflection statements in this vein are obtained. In this talk we will give a survey on these reflection statements. msjmeeting2018mar01i001 

特別講演 計算機が大学入試数学問題を解く 松﨑拓也 (名大工) In a joint work with many people, we have developed a computer system that solves preuniversity level math problems written in natural language. The system is comprised of two parts. One is a language processing pipeline, which translates a math problem into a logical formula. The other is a computer algebra system that derives an answer from the translated problem. In the talk, I will mainly talk about the former part. The main obstacle in the translation from a natural language into a logical language is the flexibility of the natural language, which enables us to convey complex meaning in a concise expression but makes the sentences highly ambiguous for a machine. I will explain how we combat with it using both logical and statistical means. msjmeeting2018mar01i002 

1. 
分数関数をローラン展開と衰垜展開で表現し対比する 田中昭太郎 The fraction $1/(z^2+1)$ has two singularities $z=i, i$. Expand in $z=2i$. (1) Devide the domain into $0 \le z2i < 1$, $1 < z2i < 3$, $3 < z2i$. (2) Decomposite it into partials: $(i/2) \{1/(z+1)i/(zi)\}$. Ex. $1/(z+1)=1/\{(z2i)+3i\}$. Let us omit Laurent’s expansion. Wada’s theorem: $1/(\square  \triangle)^{p+1}=\Sigma (k=1 \rightarrow \infty) d_p(k) \square^{pk}\triangle^{k1}, 0<\triangle/\square<1$, where integers $p \ge 0$; $k \ge 1$ and $d_p(k) \equiv (k+p1)!/p!(k1)!$, named Suida expansion. If $1 \div \{(z2i)(3i)\}^{0+1}=\Sigma (k=1 \rightarrow \infty) d_0(k)(z2i)^{0k} (3i)^{k1}$, $0<3i/(z2i)<1$. Then $\Sigma(k=1\rightarrow \infty)$ $(3i)^{k1}(z2i)^{k}$, $3<z2i$. If $1/\{3i((z2i))\}^{0+1}=\Sigma (k=1 \rightarrow \infty)d_0(k)(3i)^{0k}((z2i))^{k1}$, $0<(z2i)/3i<1$. Then $\Sigma (k=1\rightarrow \infty)d_0(k)(1)^{k1}(1/3i)^k(z2i)^{k1}$, $0<z2i<3$. 

2. 
Mathematical principles treated in mechanics by Poisson 増田 茂 (京大数理研) Maupertuis treats principle of minimum action and Clairaut discuss effort and principles of equilibrium on earth. Poisson issues Study of Mechanics in 1833, which consists the second book of three books, entitled A Study of Mathematical Physics. He discusses the mathemetical principles from many side of mathematics. Our present to this session shows the points of mathematical scopes in mechanics by Poisson. 

3. 
Proof of rise of capillary surface by Poisson 増田 茂 (京大数理研) Providing capillary action in the equilibrium, Poisson assures that the rise of the surface of water is due to the abrupt variation of density in the neighborhood of the wall and of the surface. Poisson discusses this problem in 1831, in the rivalry to the paper/book of Laplace 1806–7 and Gauss 1831. We show Poisson’s discussion. 

4. 
20世紀初頭のハミルトン・ヤコビ理論と変換論 中根美知代 In the early twentieth century, mathematicians began to eagerly discuss transformations of variables that keep the canonical form of the differential equations, the socalled ^^ ^^ contact transformation” or ^^ ^^ canonical transformation”. This paper examines origins of these two transformations. We also discuss a process how these two transformations were bound and how they were introduced to the Hamilton–Jacobi theory. 

5. 
『数』の斗食算題について 張替俊夫 (大阪産大全学教育機構) Two ‘Toshoku (doushi)’ problems contained in the book “Shu”, and one of them had been left undeciphered. In this talk, we decipher the problem from the other deciphered problem, and discuss that the method used in Toshoku problems is ‘shaoguangshu.’ 

6. 
NMFによる和算図形問題類似評価 脇 克志 (山形大理)・土橋拓馬 (明大先端数理)・阿原一志 (明大総合数理) We try to evaluate the similarity of geometric problems in WASAN using NMF (Nonnegative Matrix Factorization). In our study, we calculate characteristic vectors of geometric problems in “SANPOUTENSEISHINAN” written by “AIDA YASUAKI” by NMF. And we try to evaluate the similarity of geometric problems by these characteristic vectors. 

7. 
大成算経の日用術について 森本光生 (四日市大関孝和数学研／上智大名誉教授) Volumes 8 and 9 of th Taisei Sankei (Great Accomplished Mathematical Treatise, 1710 or 11) are entitled Daily Mathematics and contain 220 problems of mathematics stemmed from daily lives. Because of their elementary character, they have been neglected by historians of Japanese mathematics; for example, in the Meijizen Hihon Sūgakushi (History of Mathematics before the Meiji Restoration), Fujiwara Matsusaburo paid almost no attention to them. Takebe Katahiro, one of three authors of the Taisei Sankei published the Sangaku Keimō Genkai Taisei (Great Colloquial Commentary on the Suanxue Qimeng) in 1690. The Suanxue Qimeng (Introduction to Mathematics, 1299), written by Zhu Shijie of the Yuan dynasty, was a collection of mathematical problems. We shall discuss the relation of the Daily Mathematics with the Suanxue Qimeng. 

8. 
会田安明の数学思想(その2) 小川 束 (四日市大環境情報) Aida Yasuaki (1747–1817) wrote the 6 volume Sanpou Kokon Tsuuran in 1797, which incisively criticized 19 already published books of mathematics. I have discussed the first volume of Aida’s work, which contained comments on 7 books from the Sanpou Kongen Ki (1666) to the Katsuyou Sanpou (1712). In this instance, I examined the second volume, which contained comments on 8 books from the Kagaku Sanpou (1715) to the Meigen Sanpou (1764). Though some comments were the same as those found in volume 1, some were new. I will comment on the history of difficult problems, and on the relationship between solutions using equations and solutions using the abacus. 

9. 
画像認識に基づく和算図形問題への自動タグ付け 土橋拓馬 (明大先端数理)・脇 克志 (山形大理)・阿原一志 (明大総合数理) In recent years, construction of image databases from digitized historical documents of Japanese mathematics (wasan) has been progressing. Since previous studies, the authors have proposed ideas for automatic tagging in these image databases of wasan based on geometric elements (triangles, squares, circles and so on) and those relationship (tangency of circles, number of elements and so on). In this study, the authors implement the program that tags images of geometric problems in wasan automatically and verify the effectiveness of our proposal for geometric problems included in actual documents of wasan. As a result, it is found that automatic tagging succeeds for geometric problems of more than 80 percent in “Sanpo tenseiho shinan” by Yasuaki Aida. 

10. 
Nondepthfirst search of an ANDOR tree 鈴木登志雄 (首都大東京理工) Liu and Tanaka (2007) asserted that among independent distributions on a uniform binary ANDOR tree, the minimum cost (achieved by an algorithm) is maximized only by an independent and identical distribution. In this decade, the assertion has been justified under a hypothesis that only depthfirst algorithms are taken into consideration. The uniform binary tree case was shown by S. and Niida (2015). Balanced multibranching tree case was shown (with a certain hypothesis) by Peng et al. (2017). We extend the results of S.–Niida and Peng et al. to the case where nondepthfirst algorithms are taken into consideration. 

11. 
重み付きANDOR木における固有分布の一意性について 田中一之 (東北大理)・沖坂祥平 (東北大理) In 2007, Liu and Tanaka characterized the eigendistributions that achieve the distributional complexity for ANDOR trees, and among others, they proved the uniqueness of eigendistribution for a uniform binary tree. Later, Suzuki and Nakamura showed that the uniqueness fails if only directional algorithms are considered. In this talk, we introduce the weighted trees, namely, trees with weighted cost depending on the value of a leaf. Using such models, we prove that for balanced multibranching trees, the uniqueness of eigendistribution holds w.r.t. all deterministic algorithms, but fails w.r.t. only directional algorithms. 

12. 
偽ソロベイ還元性に関する幾つかの結果 水澤勇気 (首都大東京理工) Solovay reducibility is a wellknown and inmoprtant notion in theory of randomness. We defined pseudo Solovay reducibility to generalize Solovay reducibility. We have some results of pseudo Solovay reducibility. 

13. 
指標関数, 強制法と証明の変換 横山啓太 (北陸先端大情報) The indicator argument is a modeltheoretic framework to obtain independence and conservation results in the study of firstorder arithmetic. In the talk, we will consider new formulation of indicator arguments with the idea of generic cuts and forcing. With this method, we will analyze the conservation results for bounding principle and Ramsey’s theorem for pairs. 

14. 
量子力学では確率1の事象は必ず起こる 只木孝太郎 (中部大工) The notion of probability plays a crucial role in quantum mechanics. It appears as the Born rule. In this talk we reveal that every event with probability one occurs certainly in quantum measurement. 

15. 
Undecidability of the existence of $C^*$embedded but not $C$embedded subsets in a product of natural numbers 矢島幸信 (神奈川大工)・平田康史 (神奈川大工) It is proved that every $C^*$embedded subset in $\Bbb N^{\omega_1}$ is $C$embedded in $\Bbb N^{\omega_1}$ under a certain settheoretic assumption, where $\Bbb N^{\omega_1}$ denotes the product of $\omega_1$ copies of natural numbers $\Bbb N$. As a consequence, it is independent of ZFC that there is a (closed) $C^*$embedded subset in $\Bbb N^{\omega_1}$. 

16. 
Products of Lindelöf spaces 薄葉季路 (早大理工) For a topological space $X$, the Lindelöf degree of $X$ is the minimal cardinal $\kappa$ such that every open cover of $X$ has a subcover of size $\le \kappa$. If $S$ is the Sorgenfrey line, then its product $S \times S$ has the Lindelöf degree $2^\omega$. On the other hand, it is unknown whether there are Lindelöf spaces $X$ and $Y$ with $L(X \times Y)>2^\omega$. In this talk, we prove that, in the Cohen forcing extension, the Lindelöf degree of the product of two regular Lindelöf spaces can be arbitrary large up to the least $\omega_1$strongly compact cardinal. We also show that if there is no such Lindelöf spaces, then $\omega_2$ is weakly compact in $L$. 

17. 
Aronszajn 木が持つふたつの組合せ的性質の違い 依岡輝幸 (静岡大理) The forcing notion which adds an uncountable antichain through an Aronszajn tree has two similar combinatorial properties, which are called the rectangle refining property and the property $\mathsf{R}_{1,\aleph_1}$. By a viewpoint of specialization of an Aronszajn tree, we can conclude that two properties are different. 

18. 
強制法理論及び連続体上の組合せ論 D. A. Mejía (静岡大理) We present recent developments in forcing theory that have been motivated by problems on combinatorics of the real line. Concretely, the construction of treedimensional arrangements of forcing generic extensions (joint work with Fischer, Friedman and Montoya), and the incorporation of ultrafilter limits in twodimensional arrangements of generic extensions (joint with Brendle and Cardona). These techniques work to construct models where the cardinals in Cicho\’n’s diagram (classical diagram of cardinal numbers associated with combinatorial properties of the real line) can be divided into 7 different values, which is the maximum number known modulo ZFC alone. 

19. 
Embeddability of uncountable LO into models generated by uncountable indiscernible sequences 酒井拓史 (神戸大システム情報) We discuss what kinds of uncountable linearly ordered sets are embeddable into models generated by uncountable indiscernible sequences. 

21. 
集合の概念の定義とその存在定理 伊東由文 (徳島大名誉教授) In this paper, we give the definition of sets and prove its existence theorem by using the axiomatic method. The system of axioms I used here is the system of axioms ZFC and the axiom of ordinary numbers and the axiom of transfinite induction. 

22. 
A predicate extension of the logic of provability 田中義人 (九州産大経済) In this talk, we introduce a Kripke complete predicate extension of the logic of provability, that is, the propositional modal logic defined by the Löb formula. The proof system for the logic of the talk is a modal extension of Gentzenstyle sequent calculus for predicate logic. It has a standard derivation rule for necessitation, but does not include the Löb formula as an axiom schema. Instead, it has a noncompact inference rule. We show the logic is complete with respect to the class of Kripke frames of bounded length. 

23. 
存在特性を持ち選言特性を持たない中間述語論理をたくさん作る 鈴木信行 (静岡大理) The disjunction and existence properties in intermediate predicate logics were revealed to be independent in our previous paper. In that paper, we constructed a continuum of intermediate predicate logics having existence property but lacking disjunction property. In this talk, we report the existence of a continuum of intermediate predicate logics having disjunction property but lacking existence property. 

24. 
The Church–Rosser Theorem and quantitative analysis of witnesses 藤田憲悦 (群馬大工) We analyze quantitative properties of witnesses of the Church–Rosser Theorem for betaequality in terms of Takahashi translation and by using the notion of parallel reduction. We show that the proof method developed here can be applied to other reduction systems such as lambdacalculus with betaetareduction, Girard’s system F, and Gödel’s system T as well. 

25. 
縮減フレームで特徴づけられる適切様相論理 関 隆宏 (新潟大経営戦略本部) In Routley–Meyer semantics, relevant logics or relevant modal logics are characterized by unreduced frames in general. In this talk, we consider modal extensions of Slaney’s reduced modeling theorem for relevant logics without WI. 

26. 
理論の分解と証明可能性論理 倉橋太志 (木更津工高専) We proved that if $L$ is one of the modal logics $\mathsf{GL}_\alpha$, $\mathsf{D}_\beta$, $\mathsf{S}_\beta$ and $\mathsf{GL}_\beta^$ where $\alpha \subseteq \omega$ is $\Sigma_1$ and $\beta \subseteq \omega$ is cofinite, then for any $\Sigma_1$definable consistent extension $U$ of Peano Arithmetic $\mathsf{PA}$, there exists a $\Sigma_1$ definition $\tau(v)$ of some extension of $I\Sigma_1$ such that the provability logic $\mathsf{PL}_\tau(U)$ of $\tau(v)$ relative to $U$ is exactly $L$. We proved this theorem by using Jeroslow’s method of decomposing theories. 

27. 
A model theoretic Rieffel’s theorem of quantum 2tori 板井昌典 (東海大理) We associate quantum 2tori $T_\theta$ with the structure over ${\mathbb C}_\theta = ({\mathbb C}, +, \cdot, y = x^\theta),$ where $\theta \in {\mathbb R} \setminus {\mathbb Q}$, and introduce the notion of geometric isomorphisms between such quantum 2tori. We show that this notion is closely connected with the fundamental notion of Morita equivalence of noncommutative geometry. Namely, we prove a model theoretic version of Rieffel’s theorem of quantum 2tori. 

28. 
ジェネリック構造上の自己同型群の幅について 岡部峻典 (神戸大システム情報) The automorphism group on the random graph is simple. Moreover, for all non identical element $g$ of the automorphism group, every element can be denoted by a product of at most three conjugates of $g$ or $g^{1}$, by Truss in 2003. We call the number of the product the width of the automorphism group. It’s known that the similar fact holds in generic structures analogues of the random graph. That is, if the generic structure has FAP (Free Amalgamation Property) and its automorphism group is transitive, then the group is simple and the width is less than or equal to 32. In this talk, we proof we can improve the width to 12. 

29. 
Hrushovskiのab initio融合クラスについて 桔梗宏孝 (神戸大システム情報)・岡部峻典 (神戸大システム情報) Hrushovski defined an amalgamation class $K_f$ by defining a concave increasing function $f$ referring to an irrational number $\alpha$ with $1/2 < \alpha < 2/3$. His construction works for any real number $\alpha$ with $0 < \alpha < 1$. We obtained the following: (1) $f$ is concave and strictly increasing. (2) $K_f$ is an amalgamation class for any $\alpha$. (3) $f$ is unbounded if $\alpha$ is rational. (4) If $\alpha = m/d$ then $f(2x) \leq f(x) + 1/d$. Therefore, the generic model of $K_f$ has a model complete theory. (5) There is $\alpha$ (e.g. $\alpha = 1/\sqrt{2}$) such that $f$ is bounded. (6) If $f$ is bounded then the generic model of $K_f$ has no model complete theory. (7) We can give an $\alpha$ in a form of continued fractions where $f$ is unbounded. 