Numerical Algorithms Group

Bronze Sponsor

Information

The Numerical Algorithms Group (NAG) are experts in numerical algorithms, software engineering and high-performance and cloud computing. NAG has served the finance industry over five decades with outstanding product quality and technical support. Specifically relevant to the finance industry, the NAG Library pioneers with numerical, statistical and machine learning components ideal for building Quant Libraries, Risk Applications and the like.

NAG also provides best-in-class C++ operator-overloading AD tools for CPU and GPU called dco (derivative computation through overloading) and dco/map (dco meta adjoint programming). The NAG Library and AD tools are embedded in Quant Libraries and XVA applications by many of the largest investment banks.

As a not-for-profit company, NAG reinvests surpluses into the research and development of its products, services, staff and collaborations.

Social media

Contact details

+1 630 971 2337
infodesk@nag.com
www.nag.com
60532, Lisle, United States

Events

Products

Services

Projects

Documents & Links

Nonlinear Optimization: A Comparison of Two Competing ApproachesThis poster compares SQP and IPM solvers, taking a look at the effect of their different approaches to handling a problem's constraints.
Interior Point Method for Nonlinear OptimizationSome key features of NAG's nonlinear optimization solver based on an IPM, along with advice on what solver to choose, are the subject of this mini-article.
Semidefinite Programming (SDP)This article is a very brief overview of a linear and nonlinear semidefinite programming (SDP) solver and its applications.
Unscented Kalman FilterAn Unscented Kalman filter is applied to a simple illustration of a nonlinear state space model, the movement of a hypothetical robot.
dco/c++: Derivative Code by Overloading in C++The dco/c++ tool takes C++ AD one step further by focusing on arbitrarily ordered derivatives, shared-memory parallelism, and powerful intuitive interfaces.
Using The NAG Library For Python With Kdb+ and PyQIn this paper, NAG's Christopher Brandt illustrates how to use the NAG Library for Python with kdb+ and PyQ. Examples of accessing NAG routines are included.
Adjoint Algorithmic Differentiation of a GPU Accelerated ApplicationThis poster depicts algorithmic differentiation of a GPU accelerated application.
Algorithmic Differentiation of Nonsmooth and Discontinuous FunctionsNAG's dco/c++ tool supports smoothing through extensible adjoint code patterns for nearby nonsmooth and discontinuous functions.
Super-Charged Fixed Pt Iterations Using Anderson Acceleration & the NAG LibraryWithout Anderson Acceleration (AA), a trial Jacobi method took 28734 iterations to converge. With AA, it converged in 313 iterations - a speed-up of over 90x.
Derivative-free Optimization for Data FittingNAG's model-based derivative-free solver calibrates the parameters of complex numerical models to fit real world observations.
Batched Least Squares of Tall Skinny Matrices on GPUsSee how NAG's Batched Least Squares of Tall Skinny Matrices on GPUs code stacks up to the competitors.
Limited-Memory Nonlinear Conjugate Gradient SolverLearn about a novel solver for large-scale unconstrained or bound-constrained nonlinear optimization problems which offers significant performance improvement.
AAD Tool Support for Typical Numerical Patterns in Computational FinanceAdjoint Algorithmic Differentiation computes gradients at computational costs independent of gradient size and similar to the original program running cost.
Interior Point Method for Large Scale Linear Programming (LP) ProblemsThe main characteristic of this solver, compared to active-set methods, is a fast convergence in a low number of iterations.
CVA in the CloudComputing CVA and using Origami sensitivities and AD offers significant performance benefits compared to finite differences and legacy grid execution software.
Impressive results for Scotiabank using NAG Library . . .Scotiabank achieved an impressive runtime speedup of x30 for their risk calculations and derivatives pricing.
Second-order Cone Programming (SOCP)NAG's Second-order cone programming (SOCP) offers a robust and efficient way of solving several types of convex problems based on interior point method (IPM).
Derivative-free Optimization Solver for Calibration ProblemsDerivative-free optimization is absolutely essential when the objective function evaluations are noisy.
Algorithmic Differentiation of the Local Volatility ModelThis poster depicts algorithmic differentiation of the local volatility model.
Stochastic Grid Bundling Method for Bermudan SwaptionsIn a simple LIBOR test case, SGBM is 4x–6x faster than Longstaff–Schwartz for similar accuracy.
Nearest Correlation Algorithms in the NAG LibraryThis poster reviews the nearest correlation algorithms in the NAG Library.
Fast Implied Volatilities using Chebyshev InterpolationImproving performance without losing accuracy on the Black–Scholes formula for the price of a call option.
Cost of Solution for Cloud HPCBranden Moore spoke to the SHPCP about a hypothetical Cloud HPC migration and how to select the “best” cloud resources for that particular HPC workload.
Change Point AnalysisThis whitepaper provides an example of performing a change point analysis on a randomly generated time series.
Accelerating Bayesian inference using Algorithmic DifferentiationRead our recent paper on Markov Chain Monte Carlo (MCMC) for Bayesian uncertainty quantification from time-series data plus accompanying code on NAG's GitHub.

Log in