Adjoint Algorithmic Differentiation (AAD) offers an efficient and robust alternative to computing sensitivities compared to the traditional bump-and-revalue approach (finite difference). However its implementation into large existing quant analytics libraries is often considered challenging.

Xcelerit provides training to give clients solid background in the theoretical, software and computational aspects of adjoint algorithmic differentiation (AAD). Xcelerit trainers share their unique expertise to help clients AAD-enable large in-house quantitative finance analytics libraries (typically multi-millions lines of code), while respecting memory constraints, and achieving large performance gains.

Whitepaper: A Guide to Adjoint Algorithmic Differentiation

In this paper we will zoom on Algorithmic Differentiation as an efficient and robust alternative to compute sensitivities. We cover the theory and then focus on practical examples. The paper further gives guidelines on how to cope with the memory requirements, handle parallelisation, and how to incorporate external library functions.

  • Forward and adjoint algorithmic differentiation
  • Checkpointing techniques
  • Handling external function calls
  • Parallelising AD code
  • Differentiation of large code bases

Learn More About AAD

  • AAD Background
  • Basic and advanced AAD code
  • Applying AAD to real-world code
  • Develop clean maintainable AAD code
  • AAD & Monte-Carlo applications
  • AAD & PDE solvers
  • AAD & model calibration
  • Reducing memory in adjoint code
  • Maximising AAD code performance

About Xcelerit

Xcelerit is a leading provider of acceleration solutions for Quantitative Finance, engineering, and research. Our portfolio of solutions addresses a range of acceleration challenges from algorithmic optimisations to software acceleration. Xcelerit is the maker of the award-winning toolkit that allows domain-specialists to unlock the performance of accelerators (GPUs and Xeon Phi), and optimally deploy the advanced features of conventional multi-core processors. All of this is achieved with minor modifications to the existing source code.

Xcelerit extensive experience enables the firm to deliver a full solution from expert consultancy, bespoke development, training, and software acceleration. Our distinct competitive advantage derives from our unique combination of domain specialist knowledge and High Performance Computing expertise. This allows us to forge the most efficient solutions to better address our clients’ computational challenges.

Xcelerit has received recognition as a finalist in the Red Herring Europe Top 100 award, the Red Herring Top 100 Global award, and a two-time winner of HPC Wire’s “Best use of High Performance Computing in Financial Services” award. Our satisfied customers include the leading firms in investment banking, asset management, and insurance.

top