CITIC

Financial mathematics and supercomputing

26/09/2018

The scientific seminar on “Financial mathematics and supercomputing”, sponsored by RES (the Spanish Supercomputing Network), is organized by Appentra Solutions, CITIC (IT Research Center), ITMATI (Technological Institute of Industrial mathematics) and UDC (University of A Coruña).

Improving the productivity, quality and sustainability of scientific software in the field of engineering and computational science is a challenge of growing importance in the field of supercomputing, particularly relevant due to the increasing complexity of computational models, the architecture of computers and the software projects needed in future exascale systems. It is of vital importance to create an ecosystem of methodologies, good practices, processes and tools that allow for the development of better scientific software for HPC.

The valuation of financial products and processes is based on the numerical simulation of the mathematical models that govern the operation thereof. Among others, Monte Carlo techniques, numerical resolution of differential equations and numerical optimization methods are used. The size of the financial portfolios, the increasing complexity of the models (stochastic volatility, early exercise, etc.) and the number of underlying (for example in derivatives of types) demands a large number of hours of computation. The need for answers in real time or within a reasonable time frame often requires the use of the resources of high-performance computing systems, sometimes using very heterogeneous systems(CPUs, GPUs, etc.). For a long time, the RES has provided these resources to the research groups that need it. However, it is necessary to provide researchers with the necessary knowledge to efficiently take advantage of RES resources, taking into account the criteria of productivity, quality and sustainability of scientific software.

Financial mathematics and supercomputing” is a seminar with the aim to present experiences of researchers in the field of quantitative finance, in which they will share their own experiences using supercomputing, as well as the challenges they face in the development of quality numerical simulation programs allowing them to minimize the effort and cost of software development and maintenance, maximizing the results of their R&D projects.

The seminar will be held in conjunction with CESGAHack 3, a 5-day hackathon that will help participates accelerate the execution of their scientific applications.

AGENDA

09:00-09:30 Welcome and Opening Adress [Organizing Committee, Marta Renato (RES)]
09:30-10:20 “An Asset-Liability Management (ALM) model for insurance companies and its implementation in GPUs “[PhD José Antonio (ITMATI / UDC)]
10:20-11:10 “GPU acceleration in early-exercise option valuation” [PhD Álvaro Leitao (CRM Barcelona)]
11:10-11:40 Break (coffee break)
11:40-12:30 “Efficient regression Monte Carlos schemes for semilinear PDEs and its application to problems arising in mathematical finance” [PhD José Germán López-Salas (École Polytechnique Paris)]
12:30-13:20 “Supercomputing: A View from Financial Mathematics” [PhD Carlos Vázquez-Cendón (ITMATI / UDC)]
13:20-14:30 Break (lunch)
14: 30-15: 00 The Spanish Supercomputing Network [PhD Marta Renato (BSC/RES)]
15:00-16:00 “The quality of HPC numerical simulation software” [PhD Manuel Arenaz (Appentra / UDC)]
16:00-17:00 “The problem of the development of quality software in HPC” [Panel Discussion]

Registration