Undergraduate Mathematics Projects
This page describes projects in the LSU Department of Mathematics in which undergraduate students are involved. Some of these projects may currently be seeking undergraduate participants. Return to the main page for undergraduate research.
Active Projects (recruiting)
Van Kampen's obstruction and graph planarity
 Faculty advisor:
 Kevin Schreve
 Current students:
 Sean Mocanu and Robert Smith.
 Recruiting:
 Interested students are welcome to contact the faculty advisor.
 Description:
 A graph is planar if it can be drawn in the $xy$plane so that edges do not cross. Two
wellknown nonplanar graphs are the complete graph $K_5$ and the complete bipartite $K_{3,3}$.
Note that the difficulty here is showing that edges must cross in any drawing of this graph. A remarkable fact is that these two graphs are essentially the only obstruction to planarity; any nonplanar graph essentially contains one of these as a subgraph. A beautiful obstruction to graph planarity was defined by van Kampen in 1932. It involves studying the shape (i.e., topology) of a corresponding graphical configuration space, a twodimensional complex built out of squares. This obstruction is better known to topologists than graph theorists, and this project will involve giving alternate proofs of various theorems in graph theory using this topological tool, and hopefully prove some new results!
Bayesian Machine Learning
 Faculty advisor:
 Dr. Arnab Ganguly
 Recruiting:
 Interested students are welcome to contact the faculty advisor.
 Description:
 I am looking for one or two undergraduate students with strong coding skills (in python or R). Good knowledge of calculus at the level of (Calc 3) and some knowledge of basic probability or statistics will be very helpful.
Flow Semigroups: Global Linearization of Nonlinear Problems
 Faculty advisor:
 Prof. Frank Neubrander
 Current students:
 Arun Banjara, Rohin Gilman
 Past students:
 Amy Adair, Logan Hart
 Recruiting:
 Interested students may inquire about becoming involved in this project.
 Description:

In this project, we investigate how Koopman’s global linearization method can be used to (a) extend exponential operator splitting methods like the LieTrotter or Strang Product Formulas from linear semigroup theory to approximate solutions of nonlinear integral, delay, or differential equations such as x′(t) = F(t, x(t)), x(s) = x, t ≥ s; (b) study qualitative properties of the solutions of the underlying nonlinear problem (flows) in terms of spectral properties of the generator of the associated linear flow semigroup (Lie generator).
This project builds on previous work by late LSU Professor J.R. Dorroh (1937  2015) and his colleague J.W. Neuberger from the University of North Texas (see also Chapter 19 in J.W. Neuberger, A Sequence of Problems on Semigroups, Springer 2011).
Nodal sets of eigenfunctions in balls
 Faculty advisor:
 Prof. Jiuyi Zhu
 Current students:
 Chelsey Fontenot
 Recruiting:
 Interested students may contact Prof. Zhu.
 Description:

Eigenfunctions of a ball refer to the vibrational modes, or harmonics, of a ball, that vibrate at a certain frequency. The nodal set of an eigenfunction refers to where the material (maybe air) is stationary during the vibration.
Students are working on determining the size (technically the "Hausdorff measure") of nodal sets of eigenfunctions in the ball in 2 dimensions. A famous conjecture of Yau states that the Hausdorff measure of nodal sets of eigenfunctions on a compact manifold is bounded below and above by square root of eigenvalues. In the projects, students are verifying this conjecture explicitly in balls. This involves solving eigenvalue problems with different boundary conditions on the surface of the ball. Based on the solutions, one can characterize the graph and shape of nodal sets in the balls and then explicitly compute the measure of the nodal sets.
Students will broaden their knowledge on partial differential equations, use mathematical software such as MATLAB, LATEX, and familiarize themselves with scientific writing.
Control of dynamical systems with engineering applications
 Faculty advisor:
 Prof. Michael Malisoff
 Recruiting:
 Interested students may contact Prof. Malisoff.
 Description:
 Background: Control systems arise in many important engineering applications, including bioreactors, humancomputer interactions, marine robotics, motors, and the control of unmanned aerial vehicles. Control systems can be modeled mathematically as dynamical systems that can contain a statedependent parameter called a control. Controls can represent forces that can be applied to the physical system being modeled, and then the feedback control problem entails finding formulas for the control such that all solutions of the dynamical systems satisfy certain prescribed behavioral goals. This project has been funded by research grants from the National Science Foundation Directorate for Engineering.
 Student expectations:
 Students will learn basic concepts and results for control systems and apply feedback controls to mathematical models using software, in collaboration with engineering students or faculty.
 Skills required:
 Students should first take sophomore level elementary differential equations and linear algebra and should have an interest (but not necessarily experience) in using MATLAB or Mathematica software.
 Prior mentoring experience:
 I have coled a marine robotics project that included undergraduates, and REU projects involving mechanical systems arising in the oil and gas industry and other contexts.
 References:

 Astrom, K., and R. Murray. Feedback Systems: An Introduction for Scientists and Engineers. Princeton University Press, Princeton, NJ, 2008.
 Hespanha, Joao. Linear Systems Theory, Second Edition. Princeton University Press, Princeton, NJ, 2018.
 Mukhopadhyay, S., C. Wang, M. Patterson, M. Malisoff, and F. Zhang, "Collaborative autonomous surveys in marine environments affected by oil spills," in Cooperative Robots and Sensor Networks 2014, A. Koubaa and A. Khelil, Eds., Studies in Computational Intelligence Series Vol. 554, Springer, New York, 2014, pp. 87113.
Stochastic algorithms
 Faculty advisor:
 Prof. Arnab Ganguly
 Recruiting:
 Interested students are welcome to contact Prof. Ganguly.
 Description:

Stochastic Algorithms are indispensable tools in this modern age of fast computation. For example, algorithms like Markov Chain MonteCarlo have revolutionized simulation of highdimensional probability distributions which underlie many models in physics, biology, economics, and last but not the least datascience. In short, the project is about two things:
 simulation of Markov processes arising in different contexts,
 estimation of unknown parameters of Markov process from observed data.
Markov processes are some of the most popular and natural ways of modeling evolution of a variety of systems in time. For example, the movement of stock prices, dynamics of cellular reactions, motions of microscopic particles in force fields can be efficiently captured by suitable Markov processes. Typically, these processes are described as solutions of certain stochastic (random) differential equations. So it is important to devise numerical schemes which can simulate trajectories of these stochastic differential equations. Simulation can help us understand behavioral patterns of the systems these equations are modeling. For example, for cellular processes, we can use these types of simulation techniques to deduce distributions of certain protein molecules at equilibrium. Developing and implementing fast and efficient simulation algorithms for these types of equations is the objective of (a).
In (a), it is tacitly assumed that the model parameters are known, otherwise simulation cannot be carried out. But in reality, not all of their values are known. It is then necessary to first estimate them from observed data points. For example, one might be interested in estimating the volatility of a stock from observation of its price over certain time period. Developing and implementing such inference schemes is the goal of (b). This can be thought of as an "inverse problem" of (a). Of particular interest are Bayesian algorithms, which naturally provide uncertainty quantification of the desired estimates.
 Skills required:
 Students should have good knowledge of multivariable calculus, probability theory. Sone knowledge of stochastic processes, statistics and ordinary differential equations will also be helpful. They should also have experience in programming. Preferred programming language is Python or R, but this can be picked up easily if the student has knowledge of any other programming language (like Matlab).
An online graphical user interface for electromagnetic waves in layered media
 Faculty advisor:
 Prof. Stephen Shipman
 Current students:
 Joel Keller
 Past students:
 Andrew Rogers, Dylan Blanchard, Trevor Gilmore, Joshua Brock, Noah Templet, Michael Sheppard
 Recruiting:
 There is a need for a student to help write an exposition of the physics and mathematics behind the GUIthe student should be good at electromagnetics and explaining it in writing.
 Description:

The development of this graphical user interface was supported by grant DMS1411393 (20142018) from the National Science Foundation at Louisiana State University under the guidance of PI Stephen P. Shipman. As one of the projects covered by the grant, undergraduate students created the mathematical code and web interface to compute and simulate harmonic electromagnetic fields in layered media. The goal is to make available to the community a versatile online application for the computation of EM fields in media with any number of layers having arbitrary electric and magnetic tensors. The objectives are (1) to allow scientists to explore phenomena of scattering, guided modes, and resonance in the most general EM layered media and (2) to provide a pedagogical tool for students and professionals to learn EM in layered media.
The project is in its initial stages and is being further developed by undergraduate students through various funding sources, including President’s Aid from Louisiana State University. Further developments include a tutorial on the theory of electrodynamics in layered media, a user manual, computation of transmission and reflection coefficients as functions of frequency and angle of incidence, and computation of dispersion relations for pure and leaky guided modes.
Schrödinger operators on graphs
 Faculty advisor:
 Prof. Stephen Shipman
 Recruiting:
 Please contact Prof. Shipman to discuss getting involved with this project.
 Description:

Schrödinger operators come from quantum mechanics and have constituted a large part of the mathematical study of spectral theory for almost a hundred years.
There is differential equation involving the "Laplacian" and a "potential" (and maybe other things), and we are interested in its spectrum and eigenfunctions, as a vast generalization of eigenvalues and eigenvectors for matrices. This undergraduate project investigates some problems concerning Schrödinger operators on graphs (edges connecting vertices) that repeat periodically in space. The mathematics involves ordinary differential equations, linear algebra, polynomial algebra, and computational algebra.
This project is funded by the National Science Foundation under grant NSF DMS1814902 entitled "Asymmetry, Embedded Eigenvalues, and Resonance for Differential Operators."
Wave scattering by periodic structures: Surface modes and resonance
 Faculty advisor:
 Prof. Stephen Shipman
 Recruiting:
 Interested students may inquire about becoming involved in this project.
 Description:

When waves, such as electromagnetic or acoustic ones, interact with a periodically repeating structure (like a corrugation or perforated film), they get reflected in different directions and can produce waves that run along the surface of the structure. We have developed mathematical algorithms and numerical code for computing these waves. We are interested in using this code to explore surfacewave phenomena in a variety of structures that are of interest from the phenomenological and applied points of view.
This project is funded by the National Science Foundation under grant NSF DMS1814902 entitled "Asymmetry, Embedded Eigenvalues, and Resonance for Differential Operators."
Math Consultation Clinic (MCC)
 Faculty advisor:
 Prof. Peter Wolenski
 Web site:
 MCC
 Description:

The Math Consultation Clinic is associated with the Capstone Course, Math 2040. Teams of undergraduate students, each managed by a graduate student, work on solving interdisciplinary problems using mathematics. These problems come from local companies or other academic units of LSU. The currently active projects are the following:
"ShapeUp" Sima Sobhiyeh is leading six undergraduate students (plus other peripheral support from graduate and CS students) in using deep learning techniques to analyze 3D optical scanning data.
"Improving Reading Skills" Rick Mekdessie and Sudip Sinha are leading five undergraduate students in building a treatment program for young students with dyslexia or other reading disorders. They will be building an app with games to improve reading skills.
"Does Elvis (the dog) know calculus?" This project involves the calculation of optimal paths in media with different components, such as beach sand and water in the case of Elvis fetching a ball.
Active projects (full)
Fourier analysis on finite fields
 Faculty advisor:
 Dr. Fan Yang
 Current students:
 Dylan Spedale and Matthew McCoy
 Description:
 Let $p$ be a prime number and $F_p={0,1,2,…,p1}$ be the finite field. Recently, there are many breakthrough results on harmonic analysis and additive combinatorics open problems on the analogous field field versions. One such example is the Kakeya Conjecture, while the original Euclidean space $R^n$ version is still a big open problem, the finite field version was proved in only two pages(!) in an extremely elegant proof of Dvir. Though the finite field versions are often easier to study, the intuition developed from the study of finite field models can be helpful for addressing the original questions. This project includes a reading course and training on Fourier analysis and combinatorics problems on finite fields, and a research project on $L^p$ estimates for averaging operators in finite fields.
 Skills used:
 complex numbers, Fourier transforms, and elementary number theory
Laplacian on periodic discrete graphs
 Faculty advisor:
 Prof. Rui Han
 Current students:
 Trung Le, Heejoon Shin, Lillian Powell and Houston Smith
 Description:

The spectrum (think of eigenvalues) of the Laplacian operator on periodic discrete graphs (e.g. higher dimensional integer lattice) is of general interest due to its wide applications in spectral theory, graph theory, physics and chemistry. A very simple example is the discrete Laplacian (u_{n+1}+u_{n1}) on Z, whose spectrum can be easily computed and is of the form cos(x)+cos(y). This project concerns computing the spectrum for some more complicated graphs, which eventually can be reduced to computing the eigenvalues of a (large) matrix (with a few complex number entries). After that we will apply the results to compute various quantities of the graph, e.g. the graph energy, the Kirchhoff index, etc. Ambitious students are encouraged to further work on a couple of other related projects.
Students will learn some general background on periodic operators, get research experience, and take part in the writing of a journal paper.
 Skills required:
 Students should have very good knowledge of linear algebra, and some knowledge of multivariable Calculus, complex numbers.
Modeling, analysis, and simulation of liquid crystals with applications in material science and biology
 Faculty advisor:
 Prof. Shawn Walker
 Current students:
 Benjamin Thomas
 Description:
 Liquid crystals (LCs) combine the fluidity of liquids with longrange order usually seen in solids. LCs exhibit orientational order which gives rise to their anisotropic material aspects, and leads to their interesting properties (manipulation of light, easy actuation by physical forces such as electric and magnetic fields). Applications of LCs range from displays (LCD) to selfassembled materials and novel types of mechanical actuation (e.g. shape changing materials). This research project will involve three main subareas: modeling, analysis, and simulation of LC phenomena for various applications. Depending on the student’s background, the project can be tailored to focus on one or more of these subareas. Some example applications are better numerical methods for the LandaudeGennes Qtensor model, optimal material design, and simulation of active matter as a model of bacterial swarms.
 Student expectations:
 Students will learn some general background material, explore the mathematical model through simulation and/or mathematical analysis, do scientific visualization (in 3D), and take part in the writing of a journal paper.
 Skills required:
 Students should at least know multivariable calculus, some ordinary differential equations, and some basic (sophomore level) physics. They should also have some experience in programming, e.g. Matlab, or Python, or C, etc. Knowledge of mathematical analysis and/or advanced numerical methods is a plus, but not required.
 Prior mentoring experience:
 I have mentored a total of 8 undergraduates (5 have been REU students). The most recent two students, Angelique Morvant and Ethan Seal worked on a LC project that resulted in a published journal paper.
 References:

 Mottram, N. J. & Newton, C. J. P., “Introduction to Qtensor theory,” arXiv eprints, 2014.
 Virga, E. G., “Variational Theories for Liquid Crystals,” Chapman and Hall, London, 1994, 8, 376.
 Ball, J. M., “Mathematics and liquid crystals,” Molecular Crystals and Liquid Crystals, Taylor & Francis, 2017, 647, 127.
 de Gennes, P. G. & Prost, J., “The Physics of Liquid Crystals,” Oxford Science Publication, 1995, 83.
 Lagerwall, J. P. & Scalia, G., “A new era for liquid crystal research: Applications of liquid crystals in soft matter nano, bio and microtechnology,” Current Applied Physics, 2012, 12, 13871412.
Past projects
Control of Marine Robots
 Faculty advisor:
 Prof. Michael Malisoff and Dr. Corina Barbalata
 Past students:
 Collin DeVillier and Dylan Stephens
 Description:
 This project would study mathematical models of marine robots that are used for environmental monitoring, as part of Prof. Malisoff's US National Science Foundation funded project "Designs and Theory for EventTriggered Control with Marine Robotic Applications". Students would learn basic ideas for controls and marine robotic models, with the ultimate goal of designing control algorithms for motion tasks performed with underwater vehicles. The control laws would be tested in a simulation environment and using a marine vehicle that is available at LSU, in collaboration with engineering students and faculty. No prerequisite background in controls or robotics is needed for REU students to begin work on this REU.
 Prerequisites:
 Students should first take sophomore level elementary differential equations and linear algebra and should have an interest (but not necessarily experience) in using MATLAB or Mathematica software.
 Student expectations:
 After learning basic material on controls, students will study mathematical models of marine robots using mathematical analysis and simulations, and participate in the eventual writing of a conference or journal paper.
 Prior mentoring experience:
 Prof. Malisoff cosupervised a team of marine robotics undergraduate and graduate students who implemented curve tracking controllers as part of field work in the Gulf of Mexico; see Prof. Malisoff's marine robotics page. Dr. Barbalata has supervised undergraduate and graduate students working in the area of marine robotics, focusing on underwater vehicle development and on the design of control systems for vehicles and manipulation systems; see Dr. Barbalata's page.
Stochastic Marine Robotic Control Systems
 Faculty advisors:
 Dr. Li Chen, Dr. Michael Malisoff, and Dr. Arnab Ganguly
 Description:
 Probability is useful for understanding the effect of randomness which is inherent in biology, economics, engineering, finance, and numerous other areas. Probabilistic methods are employed to mathematically model certain behaviors of these systems with the goal to quantify uncertainty and use it to our advantage. The mathematical models are often in terms of systems of differential or difference equations that contain usermanipulated parameters known as controls. The controls are used to model interventions that can guide dynamical systems to desirable modes of operation. These desired modes of operation may include convergence towards equilibria, or maintaining components of the states of the dynamics in certain prescribed ranges. For instance, in aerospace and marine applications, the controls can represent possible thrusts that can be applied to a vehicle, in order to track a desired path without colliding with obstacles. The randomness of the systems which are represented in the mathematical equations modeling them often requires the controls to be probabilistic as well, and the goal is typically to maximize the probability that the system being modeled achieves a desired mode of operation. This project uses basic probability and control systems theory to understand and quantify the effects of stochastic uncertainty on the performance of controls for mathematical models of marine robots whose objectives are station keep or tracking desired paths. These include vehicles that are used to study underwater ecosystems, and to inspect underwater cables or pipes. The REU students will assist with computer programming for the faculty advisors' joint project "Stochastic DelayCompensating DataDriven EventTriggered Feedback Control for Marine Robotics" which is sponsored by the LSU Office of Research and Economic Development.
 Skills required:
 Students should have good knowledge of multivariable calculus, and some knowledge of elementary differential equations (at the level of MATH 2065, 2070, or 2090 at LSU). A bit of knowledge of probability and statistics will also be helpful but could be learned by the student during the REU. They should also have experience in programming. The preferred programming language is MATLAB, Python, or R, but this can be picked up easily if the student has knowledge of any other programming language.
Explicit timeintegration of a nonlinear string model
 Faculty advisors:
 Dr. Frederic Marazzato
 Past Student:
 Tracy Yu
 Description:
 Computing the dynamic deformation of solid objects is difficult. A code has been developed to perform this task while conserving the mechanical energy of the system.
Visualizing Trends in Teacher Production Using Title II Data
 Faculty advisor:
 Prof. James Madden
 Past students:
 Daniel J Brignac, Nicholas Angelo, Garrett A Harvey, Kirby J Moore
 Description:
 Detailed data on K12 teacher production is collected by the U.S. Department of Education and published on title2.ed.gov. At present, data spanning years 20102015 is available. All teachereducation programs in the USA are listed, together with their demographics, program features, and the numbers of teachers produced in all possible areas, subjects and majors. Data is stored in separate files by year. The web site does not provide any way of detecting or viewing trends. This project aims to remedy this by determining the most important trends across the years represented and providing informative graphical summaries.
How to design an automobile insurance product for autonomous vehicles
 Faculty advisors:
 Matthew Arnold, Blue Cross Blue Shield of Louisiana and Prof. Larry Smolinsky, LSU math
 Past Students:
 Mason Guillot, Mark Russo, Kaju Sarkar, Mason Guillot, Emily Turner
 Description:

It is anticipated that the current automobile insurance market may eventually decrease with future public adoption of autonomous vehicles. Such a shift from traditional automobile products to autonomous vehicle products is likely to usher in a new era of risk and liability considerations that will demand big data capabilities and new actuarial
models. Some of the objectives for this project are to recommend a design for a new autonomous vehicle insurance policy, and develop loss cost estimates, also known as pure premiums, for the new policy. Recommend a launch date for the policy and provide a tenyear forecast of pure premium from the point of launch.
This project is part of the 2019 Student Research Case Study Challenge for the Society of Actuaries. https://www.soa.org/research/opportunities/2019studentcasestudy/
Informetric indicators for citation networks
 Faculty advisors:
 Profs. Larry Smolinsky and Daniel Sage and LSU Librarian Aaron Lercher
 Past Students:
 Aaron Cao (now in the graduate program at Carnegie Mellon).
 Description:
 Informetric indicators, particularly based on the network of citations, have become important in scientometrics, which is an interdisciplinary area in information science. There is a substantial literature grappling with the meaning of citations and comparing them to other metrics. This project is examining citations in light of a decade of the American Mathematical Society's identification of featured reviews.
A variation of the stable marriage problem
 Faculty advisor:
 Prof. Guoli Ding
 Past students:
 Tobias Bork
 Description:

Problem description.
In the stable marriage problem there are two disjoint sets M and W (men
and women), where each member of M has a preference (linear order) of
members
of W and, similarly, each member of W has a preference of members of M.
The problem is to find a matching such that the following unstable scenario
does not happen: some m and w prefer each other than their matched partner.
We consider a variation, which allocates residents to hospitals. Again, assume that each hospital and each resident has a preference. In addition, assume that some residents are married to each other and the couple has a joined preference. The problem is to find a stable assignment.
It is known that the problem is NPhard if each couple is allowed to have two or more choices. So we only consider the case when each couple has only one choice. Under this assumption, the problem was solved when each hospital and each resident has at most two choices. Tobias is trying to generalize this result, by considering the case when each resident has many choices. It seems he has obtained an algorithm for this case. It it works out, he should have a nice short paper.
Quantitative NevanlinnaPick interpolation
 Faculty advisor:
 Prof. Andrew Zimmer
 Description:
 The NevanlinnaPick interpolation theorem is a vast generalization of the Schwarz lemma in complex analysis. In particular, it gives necessary and sufficient conditions for the existence of a holomorphic map of the disk to have prescribed values at certain points. This theorem was developed in the early 20th century and has a number of important applications in pure and applied mathematics. The purpose of this project is to develop new quantitative variants of this classical result. Analytic techniques and nonEuclidean geometry (e.g. real hyperbolic geometry) will play an important role in this investigation.
 Background Required:
 MATH 4036 Complex Variables and MATH 4031 Advanced Calculus I.
 Learning Objectives:
 learn about topics in analysis and nonEuclidean geometry, learn about doing original research in pure mathematics
Predicting outcomes in college football using machine learning
 Faculty advisor:
 Prof. Andrew Zimmer
 Past Students:
 Thomas Ayton
 Description:

Recently sports news organizations like ESPN and 538, have developed mathematical models to predict the outcomes of sports games. The goal is this project is develop our own models that can compete with the ``professionals.'' The types of models will depend on the participant's background, but can range from adaptations of the ELO ranking in chess to sophisticated neural networks.
Background required: MATH 2085 Linear Algebra, MATH 2057 Multidimensional Calculus, and a basic knowledge of Python would be ideal, but MATH 1550 Calculus I, a basic knowledge of any programming language, plus a willingness to learn a lot is sufficient.
Learning Objectives: develop skills in mathematical modeling and computer programming, learn about methods in machine learning and how to use the standard machine learning libraries in Python.
Local Langlands correspondence for SL_{2}
 Faculty advisor:
 Prof. Daniel Sage
 Past students:
 Matthew Bertucci
Bethe equations for the Gaudin model and Wronskian relations
 Faculty advisor:
 Prof. Anton Zeitlin
 Past students:
 Ty Brinson
 Description:

Integrable systems form an important part of modern mathematics; they are systems with enough conserved quantities that allow one to determine the state of the system by knowing these quantities along with other associated data.
Apart from being interesting on their own, they appear to be crucial in the study of many other areas of mathematics, geometry in particular.
While their applications can be quite complicated, such as those that appear in modern enumerative geometry,
the underlying structures within integrable systems are quite simple.
This project is devoted to a typical example of this kind, called the "Gaudin model." We are interested in the relation between two mathematical entities that arise in studying this modelcertain equations called the "Bethe" equations and "Wronskian" matrices for polynomials.
While the solution of this problem addresses a cuttingedge topic of modern mathematics, known as the geometric Langlands correspondence, the tools needed for its solution are rather basiccalculus, linear algebra and a bit of complex analysis, and the necessary knowledge from the latter can be easily attained on the way.
This project is devoted to the study of meromorphic connections on the projective line (known as opers) with regular and irregular singularities. The goal is to describe them in terms of Bethetype equations and relate them to the deformation of the Gaudin model.
Flat connections on Riemann surfaces
 Faculty advisor:
 Prof. Anton Zeitlin
 Past students:
 Andrea Bourque
 Description:
 This project aims to construct coordinates on the space of GL(11) flat connections on Riemann surfaces using combinatorial structures related to ribbon graphs. It is planned to understand the mapping class group action and write explicitly the invariant Poisson bracket in terms of those coordinates.