Professional Documents
Culture Documents
Automated Reasoning 10th International Joint Conference IJCAR 2020 Paris France July 1 4 2020 Proceedings Part I Nicolas Peltier
Automated Reasoning 10th International Joint Conference IJCAR 2020 Paris France July 1 4 2020 Proceedings Part I Nicolas Peltier
https://textbookfull.com/product/automated-reasoning-10th-
international-joint-conference-ijcar-2020-paris-france-
july-1-4-2020-proceedings-part-ii-nicolas-peltier/
https://textbookfull.com/product/hci-
international-2020-posters-22nd-international-conference-
hcii-2020-copenhagen-denmark-july-19-24-2020-proceedings-part-i-
constantine-stephanidis/
https://textbookfull.com/product/computational-science-and-its-
applications-iccsa-2020-20th-international-conference-cagliari-
italy-july-1-4-2020-proceedings-part-vii-osvaldo-gervasi/
https://textbookfull.com/product/computational-science-and-its-
applications-iccsa-2020-20th-international-conference-cagliari-
italy-july-1-4-2020-proceedings-part-vi-osvaldo-gervasi/
Automated Reasoning 7th International Joint Conference
IJCAR 2014 Held as Part of the Vienna Summer of Logic
VSL 2014 Vienna Austria July 19 22 2014 Proceedings 1st
Edition Stéphane Demri
https://textbookfull.com/product/automated-reasoning-7th-
international-joint-conference-ijcar-2014-held-as-part-of-the-
vienna-summer-of-logic-vsl-2014-vienna-austria-
july-19-22-2014-proceedings-1st-edition-stephane-demri/
https://textbookfull.com/product/artificial-intelligence-in-
education-21st-international-conference-aied-2020-ifrane-morocco-
july-6-10-2020-proceedings-part-i-ig-ibert-bittencourt/
https://textbookfull.com/product/hci-
international-2020-posters-22nd-international-conference-
hcii-2020-copenhagen-denmark-july-19-24-2020-proceedings-part-ii-
constantine-stephanidis/
https://textbookfull.com/product/hci-
international-2020-posters-22nd-international-conference-
hcii-2020-copenhagen-denmark-july-19-24-2020-proceedings-part-
Nicolas Peltier
Viorica Sofronie-Stokkermans (Eds.)
LNAI 12166
Automated Reasoning
10th International Joint Conference, IJCAR 2020
Paris, France, July 1–4, 2020
Proceedings, Part I
123
Lecture Notes in Artificial Intelligence 12166
Series Editors
Randy Goebel
University of Alberta, Edmonton, Canada
Yuzuru Tanaka
Hokkaido University, Sapporo, Japan
Wolfgang Wahlster
DFKI and Saarland University, Saarbrücken, Germany
Founding Editor
Jörg Siekmann
DFKI and Saarland University, Saarbrücken, Germany
More information about this series at http://www.springer.com/series/1244
Nicolas Peltier Viorica Sofronie-Stokkermans (Eds.)
•
Automated Reasoning
10th International Joint Conference, IJCAR 2020
Paris, France, July 1–4, 2020
Proceedings, Part I
123
Editors
Nicolas Peltier Viorica Sofronie-Stokkermans
CNRS, LIG, Université Grenoble Alpes University Koblenz-Landau
Saint Martin d’Hères, France Koblenz, Germany
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
These volumes contain the papers presented at the 10th International Joint Conference
on Automated Reasoning (IJCAR 2020) initially planned to be held in Paris, but – due
to the COVID-19 pandemic – held by remote conferencing during July 1-4, 2020.
IJCAR is the premier international joint conference on all aspects of automated
reasoning, including foundations, implementations, and applications, comprising sev-
eral leading conferences and workshops. IJCAR 2020 united CADE, the Conference on
Automated Deduction, TABLEAUX, the International Conference on Automated
Reasoning with Analytic Tableaux and Related Methods, FroCoS, the International
Symposium on Frontiers of Combining Systems, and ITP, the International Conference
on Interactive Theorem Proving. Previous IJCAR conferences were held in Siena
(Italy) in 2001, Cork (Ireland) in 2004, Seattle (USA) in 2006, Sydney (Australia) in
2008, Edinburgh (UK) in 2010, Manchester (UK) in 2012, Vienna (Austria) in 2014,
Coimbra (Portugal) in 2016, and Oxford (UK) in 2018.
150 papers were submitted to IJCAR: 105 regular papers, 21 system description,
and 24 short papers, describing interesting work in progress. Each submission was
assigned to three Program Committee (PC) members; in a few cases, a fourth additional
review was requested. A rebuttal phase was added for the authors to respond to the
reviews before the final deliberation. The PC accepted 62 papers, resulting in an
acceptance rate of about 41%: 46 regular papers (43%), 11 system descriptions (52%),
and 5 short papers (20%).
In addition, the program included three invited talks, by Clark Barrett, Elaine
Pimentel, and Ruzica Piskac, plus two additional invited talks shared with the con-
ference on Formal Structures for Computation and Deduction (FSCD), by John
Harrison and René Thiemann (the abstract of the invited talk by René Thiemann is
available in the proceedings of FSCD 2020).
The Best Paper Award was shared this year by two papers: “Politeness for The
Theory of Algebraic Datatypes” by Ying Sheng, Yoni Zohar, Christophe Ringeissen,
Jane Lange, Pascal Fontaine, and Clark Barrett, and “The Resolution of Keller’s
Conjecture” by Joshua Brakensiek, Marijn Heule, John Mackey, and David Narvaez.
IJCAR acknowledges the generous sponsorship of the CNRS (French National
Centre for Scientific Research), Inria (French Institute for Research in Computer
Science and Automation), the Northern Paris Computer Science (LIPN: Laboratoire
d’Informatique de Paris Nord) at the University of Paris North (Université Sorbonne
Paris Nord), and of the Computer Science Laboratory of Ecole Polytechnique (LIX:
Laboratoire d’Informatique de l’École Polytechnique) in the École Polytechnique.
The EasyChair system was extremely useful for the reviewing and selection of
papers, the organization of the program, and the creation of this proceedings volume.
The PC chairs also want to thank Springer for their support of this publication.
vi Preface
We would like to thank the organizers of IJCAR, the members of the IJCAR PC,
and the additional reviewers, who provided high-quality reviews, as well as all authors,
speakers, and attendees.
The COVID-19 pandemic had a strong impact on the organization of IJCAR
and significantly weighted the burden on authors, reviewers, and organizers. We are
very grateful to all of them for their hard work under such difficult and unusual
circumstances.
Program Committee
Takahito Aoto Niigata University, Japan
Carlos Areces Universidad Nacional de Córdoba (FaMAF), Argentina
Jeremy Avigad Carnegie Mellon University, USA
Franz Baader TU Dresden, Germany
Peter Baumgartner CSIRO, Australia
Christoph Benzmüller Freie Universität Berlin, Germany
Yves Bertot Inria, France
Armin Biere Johannes Kepler University Linz, Austria
Nikolaj Bjørner Microsoft, USA
Jasmin Blanchette Vrije Universiteit Amsterdam, The Netherlands
Maria Paola Bonacina Università degli Studi di Verona, Italy
James Brotherston University College London, UK
Serenella Cerrito IBISC, Université d'Évry, Université Paris-Saclay,
France
Agata Ciabattoni Vienna University of Technology, Austria
Koen Claessen Chalmers University of Technology, Sweden
Leonardo de Moura Microsoft, USA
Stéphane Demri LSV, CNRS, ENS Paris-Saclay, France
Gilles Dowek Inria, ENS Paris-Saclay, France
Marcelo Finger University of São Paulo, Brazil
Pascal Fontaine Université de Liège, Belgium
Didier Galmiche Université de Lorraine, CNRS, LORIA, France
Silvio Ghilardi Università degli Studi di Milano, Italy
Martin Giese University of Oslo, Norway
Jürgen Giesl RWTH Aachen University, Germany
Valentin Goranko Stockholm University, Sweden
Rajeev Gore The Australian National University, Australia
Stefan Hetzl Vienna University of Technology, Austria
Marijn Heule Carnegie Mellon University, USA
Cezary Kaliszyk University of Innsbruck, Austria
Deepak Kapur University of New Mexico, USA
Laura Kovacs Vienna University of Technology, Austria
Andreas Lochbihler Digital Asset GmbH, Switzerland
viii Organization
Conference Chair
Kaustuv Chaudhuri Inria, Ecole Polytechnique Paris, France
Workshop Chairs
Giulio Manzonetto Université Paris-Nord, France
Andrew Reynolds University of Iowa, USA
Additional Reviewers
Clark Barrett
John Harrison
Abstract. I have focused a lot of the applied side of my work over the last 20
years on formal verification of arithmetic in some sense. Originally my main
focus was on verification of floating-point algorithms for division, square root,
transcendental functions, etc. More recently my interests have shifted to discrete
arithmetical primitives, large integer arithmetic, and elliptic curve operations. As
well as many contrasts and special problems, there are a number of common
themes running through all this work: the challenges of verification at the
unstructured machine-code level or indeed even getting adequate specifications
for machine instruction sets, the countervailing benefit of generally having clear
and incontrovertible specifications of the functions themselves, and the value of
special customized decision procedures in making verifications of this kind
practical. I will give an overview of some of the highlights of this work, as well
as talking in more detail about my current project.
Focusing, Axioms and Synthetic Inference
Rules (Extended Abstract)
Proving a sequent in sequent-based systems often involves many choices. For example,
at every node of a tree-derivation one could: apply an introduction rule over a non-
atomic formula; apply a structural rule; introduce a lemma; apply initial rules, etc.
Hence, there is a need for discipline in structuring such choices and taming proof-
search. One such discipline is focusing [1].
Focused proof systems combines two devices: polarization and focused rule
application. In classical/intuitionistic first order systems, polarized formulas are built
using atomic formulas and polarized versions of logical connectives and constants. The
positive and negative versions of connectives and constants have identical truth con-
ditions but have different inference rules inside the polarized proof systems. For
example, left introduction rules for positive connectives are invertible while left
introduction rules for negative connectives are not necessarily invertible. The polarity
of a non-atomic formula is determined by its top-level connective. Since every
polarized formula is classified as positive or negative, a polarity to atomic formulas
must also be provided. As it turns out, this assignment of polarity to atomic formulas
can be arbitrary [1].
When focusing on a formula, the focus is transferred to the active formulas in the
premises (focused rule applications). This process goes on in all branches of the
derivation, until: an initial rule/introduction rule on constants is applied (and the
derivation ends at that branch); either the polarity of the focused formula changes or the
side (left/right) of the focus flips (but not both). In this case, focus is released and the
formula is eagerly decomposed into its negative-left, positive-right and/or atomic
subformulas, that are stored in the context. Reading derivations from the root upwards,
this forces a sequent derivation to be organized into focused phases, each of them
corresponding to an application of a synthetic inference rule [2], where the focused
formula is rewritten into (some of) its subformulas.
There is a class of formulas corresponding to particularly interesting synthetic rules:
the bipolars. Bipolars are negative formulas in which polarity can change at most once
among its subformulas. This means that left-focusing on a bipolar A gives rise to
xvi E. Pimentel et al.
(possibly many) synthetic inference rules having simple shape, with leaves involving
only atomic subformulas of A. We call a synthetic inference rule corresponding to the
bipolar A a bipole for A.
In this talk, we will present a careful study of bipoles, giving a fresh view to an old
problem: how to incorporate inference rules encoding axioms into proof systems for
classical and intuitionistic logics.
We start by considering LKF and LJF [6,7] as the basic focused proof systems for
classical and intuitionistic logics, respectively. In such systems, leafs of focused phases
can be composed of either: (i) a conclusion-sequent of the application of introduction
rule on constants; (ii) a (focused) conclusion-sequent of the application of the initial
rule; (iii) an (unfocused) sequent after the storage of the remaining formulas. As an
example, consider the following first order formula, that relates the subset and mem-
bership predicates in set theory:
A ¼ 8yz:ð8x:x 2 y x 2 zÞ yz:
Assuming that the predicate is given negative polarity, in the focused phase given by
(left) focusing on A
Γ, x y x z, Δ
storel , storer
Γ x y x z Δ
∀ ,⊃
Γ x.x y x z Δ r r
releaser initl
Γ x.x y x z Δ Γ y z Δ
⊃l
Γ ( x.x y x z) y z Δ
∀l
Γ yz.( x.x y x z) y z Δ
the right leaf has shape (ii) while the left one is of the form (iii). The formula between
the + and ‘ is the focus of that sequent.
Observe that it must be the case that yz 2 D (since yz is atomic, negative and
under focus), while x 2 y; x 2 z end-up being stored into contexts. This is not by
chance: restricted to bipoles, leaves of the shape (ii) forces atoms to belong to the
context, while leaves of the shape (iii) adds atoms to the context. This implies that
principal and active formulas in bipoles for A (if any) are atomic formulas. That is:
bipoles can be seen, in a sense, as introduction rules for atoms. For example, the bipolar
above corresponds to the (unpolarized) synthetic rule
x y, Γ x z, Δ .
Γ y z, Δ
Similarly, when reasoning about container data structures that can hold
duplicate elements, multisets are the obvious choice of an abstraction. Multisets
are collections of objects where an element can occur several times. They can be
seen as “sets with counting”. Although multisets are interesting mathematical
objects that can be widely used in verification, there was no efficient reasoner for
multisets and sets with cardinality constraints until recently [6]. Moreover, for a
long time it was not known if the logic of multisets with cardinality constraints
is even decidable [7]. Nevertheless, researchers have recognized the importance
of this logic and they have been studied multisets in combination with other
theories.
Zarba [13] investigated decision procedures for quantifier-free multisets but
without the cardinality operator. He showed how to reduce a multiset formula
to a quantifier-free defining each multiset operation pointwise on the elements
of the set. Adding the cardinality operator makes such a reduction impossible.
Lugiez studied multiset constraints in the context of a more general result
on multitree automata [7] and proved the decidability of multiset formulas with
a weaker form of cardinality operator that counts only distinct elements in a
multiset.
In this paper we revive the first decision procedure for multisets with cardi-
nality constraints [9,10]. We represent multisets (bags) with their characteristic
functions. A multiset m is a function E → N, where E is the universe used for
populating multisets and N is the set of non-negative integers. The value m(e) is
the multiplicity (the number of occurrences) of an element e in a multiset m. We
assume that the domain E is fixed and finite but of an unknown size. We consider
the logic of multisets constraints with the cardinality operator (MAPA), given
in Fig. 1. An atom in MAPA is either a multiset comparison, or it is a stan-
dard quantifier-free linear integer arithmetic atom, or it is a quantified formula
(∀e.Fin ), or it is a collecting sum formula. We allow only universal quantification
over all elements of E. This way we can express, for example, that for a multiset
k it holds ∀e.k(e) = 0 ∨ k(e) = 1 – in other words, k is a set. A collecting sum
atom is used to group several formulas involving sums into a single atom. This
is needed for the next step of the decision procedure. The sums are used in the
definition of the cardinality operator:
|m| = m(e)
e∈E
Piskac and Kuncak [9] showed that every MAPA formula can be translated
to an equisatisfiable LIA formula. The translation is linear and described in [9].
This way reasoning about MAPA formulas reduces to reasoning about LIA
formulas.
Efficient Automated Reasoning About Sets and Multisets 5
top-level formulas:
F ::= A | F ∧ F | ¬F
A ::= M =M | M ⊆ M | ∀e.Fin | Aout
outer linear arithmetic formulas:
Fout ::= Aout | Fout ∧ Fout | ¬Fout
Aout ::= tout ≤ tout | tout =tout | (tout , . . . , tout )= (tin , . . . , tin )
Fin
tout ::= k | |M | | C | tout + tout | C · tout | ite(Fout , tout , tout )
inner linear arithmetic formulas:
Fin ::= Ain | Fin ∧ Fin | ¬Fin
Ain ::= tin ≤ tin | tin =tin
tin ::= m(e) | P | tin + tin | P · tin | ite(Fin , tin , tin )
multiset expressions:
M ::= m | ∅ | M ∩ M | M ∪ M | M M | M \ M | M \\ M | set(M )
terminals:
m - multiset variables; e - index variable (fixed)
k - integer variable; C - integer constant; P - non-negative integer constant
The result of the star operator applied to set S is a set if all linear additive
combinations of vectors from S. Its syntax is given in Fig. 2.
Fig. 2. Linear integer arithmetic (LIA) and an extension with the Star Operator.