Arbitragis Trading has partnered with Ecole Centrale Paris to teach Computational Finance

1. Introduction

Arbitragis, a pioneer in quantitative finance, is a proprietary trading firm specialized in algorithmic trading.

It struck a partnership with Nvidia and has helped provide Ecole Centrale Paris with a computing power
amounting to 4 Teraflops, i.e. 70% of the computing power of Météo France, the French weather forecast agency.

The Applied Mathematics Department and the laboratory of Mathematics Applied to
Systems (MAS) of Ecole Centrale Paris are willing to set up research projects and to teach
courses on the use of Graphics Processing Units (GPUs) for scientific computing with Arbitragis.
 
This brief document describes these projects, the background for teaching and research at
Ecole Centrale Paris.

2. Non-exhaustive list of considered courses and projects

CUDA-related courses

● Computational Finance with Graphics Processing Units

Lecturers : Ioane Muni Toke, ECP ; Tuan Nguyen, Arbitragis
Date : January-March 2009
Graduate Level (M.Sc. Applied Mathematics for Finance)
Duration : 21 hours (lectures + practical work)
Prerequisites : a one semester C-programming course ;  a one-semester course on
Stochastic Calculus for Finance
This course is designed for future quantitative researchers in financial institution. The most
recent algorithms for American Monte Carlo pricing will be introduced in lectures, and will
be subjects of practical work.
CUDA-related research projects

● Pricing financial derivatives using GPUs (Graduate level)

Graduate level, M.Sc. research seminar
Date : October 2008 – March 2009
Part-time internship for 2 graduate students in the company Arbitragis.
Academic supervisor : Ioane Muni Toke

● Pricing financial derivatives using GPUs (Undergraduate level)

Undergraduate level, scientific project
Date : September 2008 – June 2009
Scientific working group of 4 students.
Tutor : Ioane Muni Toke

● Solving a chemical master equation using GPUs

Undergraduate level, scientific project
Date : September 2008 – June 2009
Scientific working group of 4 students.
Tutor : Florian De Vuyst, Professor

3. Teaching background

The Applied Mathematics Department of Ecole Centrale Paris has been known for many years as
a center of excellence in educating and training managers for banks and insurance
companies. Future traders, quantitative researchers, financial engineers and structurers
receive at the Ecole Centrale Paris a highly scientific and technical education and are
recruited among the largest banks worldwide.
JP Morgan, Lehman Brothers, Deutsche Bank, BNP-Paribas, Société Générale, Natexis,
Calyon, Reuters and AXA are some of the direct sponsors of our M.Sc. in Applied
Mathematics for Finance.

4. Research background

The laboratory of Mathematics Applied to Systems (MAS) of Ecole Centrale Paris has been
very actively involved in parallel and distributed scientific computing for finance. In 2003
and 2004, its research were twice awarded the IBM PhD fellowship for Grid Computing for
Finance. These research in algorithmic were then pursued in one European and one French
Grid projects :
● BeinGrid : “Risk management in Finance”
● ANR-GCPMF : “Solving high-dimensional parabolic equations in Finance”.
Ecole Centrale Paris is also an active member of :
● “Finance and Innovation”, competitive cluster for finance
● Ter@tec, European cluster for high performance computing
Finally, a BNP-Paribas – Ecole Centrale Paris Chair of Quantitative Finance was created in
2007. Professor Frédéric Abergel is the current holder of the chair.

5. Publications

In the field of scientific computing for finance, we can mention the following publications :
• J.Y. Girard, I. Muni Toke, J.L. Lepesant and J.P. Prost, « Deploy a C Application as a Grid
Service : How to easily implement a grid service from an existing application. », IBM
developerWorks, May 2004.
• Muni Toke and J.Y. Girard, «Monte Carlo Valuation of Multidimensional American
Options through Grid Computing », In: Large-Scale Scientific Computing: 5th
International Conference, LSSC 2005. Volume 3743 of Lecture Notes in Computer
Science, Springer Verlag (2006) 462-469
• Muni Toke , « Parareal Web Service on a Grid Architecture for Fast PDE Solving –
Application to Option Pricing », In: Proceedings of the 2007 International Conference
on Parallel and Distributed Processing Techniques and Applications, PDPTA'07, CSREA
Press, 2007.
• C. Audouze, F. Daim, F. De Vuyst, P. Laurent and I. Muni Toke, « The Parareal
Algorithm as time domain decomposition method. Applications in Finance and
Biology », Book chapter in: « Domain Decomposition Methods: Algorithms and
Practice », F.Magoulès ed., Sax-Coburg Publications, 2008.

6. Industrial collaboration

Some of the CUDA research projects and teaching course will be done in collaboration with
the company Arbitragis, a proprietary trading firm specialized in quantitative derivatives
trading strategies.

7. Contacts

Applied Maths Department – MAS Laboratory
Ecole Centrale Paris
Grande Voie des Vignes
92295 Chatenay-Malabry CEDEX

Ioane Muni Toke

Assistant Professor
Tel :+33 (0)1 41 13 18 91
Fax : +33 (0)1 41 13 17 35
Email : ioane.muni-toke_AT_ecp.fr

Lionel Gabet

Professor
Dean of the Department of Mathematics
Email : lionel.gabet_AT_ecp.fr

Christian Saguez

Professor
Director of the MAS Laboratory
Email : christian.saguez_AT_ecp.fr