### Table des matières

# Web-page of the Geneva University tropical group

PhD graduated: Kristin Shaw (December 2011), Lionel Lang (December 2014), Nikita Kalinin (December 2015), Mikhail Shkolnikov (June 2017), Johannes Josi (February 2018).

Current members: Thomas Blomme, Francesca Carocci, Slava Goncharov, Grigory Mikhalkin, Antoine Toussaint.

Alumni: Ivan Bazhov, Johan Bjorklund, Rémi Crétois, Weronika Czerniawska, Yi-Ning Hsiao, Jens Forsgard, Maxim Karev, Ilya Karzhemanov, Sergei Lanzat, Michele Nesci, Alina Pavlikova, Mikhail Pirogov, Johannes Rau, Arthur Renaudineau.

We organize several seminars:

Séminaire "Fables Géométriques".

pre-2017 (historical) Battelle Seminar and

Tropical working group Seminar.

Also, you can check how tropical curves (and hypersurfaces, in general) emerge from abelian sandpile models: tropicalsand

# Seminars and conferences

Diego MATESSI (Milano), Dec 4, 15h, Salle 06-13

“Tropical mirror symmetry and real Calabi-Yaus”

I will present some work in progress jont with Arthur Renaudineau. The goal is to understand the topology of real Calabi-Yaus by combining the Renaudineau-Shaw spectral sequence with mirror symmetry. We will consider mirror pairs of Calabi-Yau hypersurfaces X and X' in toric varieties associated to dual reflexive polytopes. The first step is to prove an isomorphism between tropical homology groups of X and X', reproducing the famous mirror symmetry exchange in hodge numbers. We then expect that the boundary maps in the Renaudineau-Shaw spectral sequence, computing the homology of the real Calabi-Yaus, can be interpreted, on the mirror side, using classical operations on homology.

Thomas Blomme, université de Genève, Thursday, Nov 9, 16h15, Room 1-15.

“Gromov-Witten invariants of bielliptic surfaces”

Bielliptic surfaces were classified by Bagnera & de Francis more than a century ago. They form a family spread into seven subfamilies of the Kodaira-Enriques surface classification which have nearly trivial canonical class in the sense that it is non-zero, but torsion. Thus, the virtual dimension of the moduli space of curves only depends on the genus, and contrarily to abelian and K3 surfaces, it yields non-zero invariants. In this talk we'll focus on some techniques to compute GW invariants of these surfaces along with some regularity properties.

Antoine Toussaint, université de Genève, Monday, Oct 23, at 15h, Salle 06-13

“Real Structures of Phase Tropical Surfaces”

Phase tropical surfaces can appear as a limit of a 1-parameter family of smooth complex algebraic surfaces. A phase tropical surface admits a stratified fibration over a smooth tropical surface. We study the real structures compatible with this fibration and give a description in terms of tropical cohomology. As an application, we deduce combinatorial criteria for the type of a real structure of a phase tropical surface. Time permitting, we will also discuss the connection with Renaudineau and Shaw's spectral sequence and Kalinin's spectral sequence.

Ozgur CEYHAN (University of Luxembourg), Monday, Oct 16, at 15h, Salle 06-13

“Complexities in backpropagation and tropicalization in neural networks”

The backpropagation algorithm and its variations are the primary training method of multi-layered neural networks. The backpropagation is a recursive gradient descent technique that works on large matrices. This talk explores backpropagation via tropical linear algebra and introduces multi-layered tropical neural networks as universal approximators. After giving a tropical reformulation of the backpropagation algorithm, we verify the algorithmic complexity is substantially lower than the usual backpropagation as the tropical arithmetic is free of the complexity of usual multiplication.

Gurvan Mével (Université de Nantes), Wednesday, Oct 18, at 14h15, Salle 06-13

“Universal polynomials for coefficients of tropical refined invariant in genus 0”

In enumerative geometry, some numbers of curves on surfaces are known to behave polynomially when the cogenus is fixed and the linear system varies, whereas it grows more than exponentially fast when the genus is fixed. In the first case, Göttsche's conjecture expresses the generating series of these numbers in terms of universal polynomials.

Tropical refined invariants are polynomials resulting of a weird way of counting curves, but linked with the previous enumerations. When the genus is fixed, Brugallé and Jaramillo-Puentes proved that some coefficients of these polynomials behave polynomially, bringing back a Göttsche's conjecture in a dual and refined setting. In this talk we will investigate the existence of universal polynomials for these coefficients.

# Geneva-Neuchâtel Symplectic Geometry Seminar

Schedule and more details: seminar page

We had our page http://www.unige.ch/math/folks/langl/battelle/

Also, there is **Séminaire de Géométrie Tropicale** in Paris:http://erwan.brugalle.perso.math.cnrs.fr/Seminaires/Geotrop/Geotrop.html

# About this page

You should be approved user to edit pages. You can write here something. (Create a small web page about you, write about you interests, explain tropical philosophy of our group, upload articles etc).

Page creating is simple, just add internal link on non-existing page from existing.

For example еуые: .

Google “docuwiki …” to know how to do something if it is not clear.

Send me your nickname and password - to have right to create and edit pages.

Latex support is here. $\int \mathbb C + \prod\limits_{x\to \infty} f^g$