The Numerical Algorithms Group... from 0 - 40 in a flurry of achievements
Four decades of numerical software achievements
In May 2010 NAG celebrated 40 years since the inception of NAG Ltd - quite an achievement in the fast moving and uber-competitive software market. NAG's commitment to the research and development of high quality numerical and statistical algorithms began in 1970, six years before its incorporation as a company, and this commitment continues right through to the present day.
During the four decades that followed NAG's birth, many great product and project triumphs have been witnessed by the experts that develop NAG's products, our customers in industry and academia, and the media. Here are just a few of them:
The first decade 1970 - 1980
The first decade of NAG as a company saw a successful transformation from a collaborative university-based activity, dependent on UK public funding during the 1970s, into a successful, financially-independent, trading company, operating worldwide with an impressive list of major academic, commercial and governmental clients within its growing customer base. Crucially, however, within this period of transformation, the NAG "ethos" - the unending quest for technical excellence, the in-grained belief in the value of collaboration - was strongly preserved. Examples of that ethos in action appear in the summary of the first decade of NAG given below: involvement in UK, European and trans-Atlantic collaborative projects, early advances in the development of High Performance Computing software, portable graphics software etc.
- The NAG project, led by Brian Ford, launched in the UK as a collaborative, inter-university activity in May 1970, coordinated from Nottingham University
- Funding support came from the Computer Board, a UK government body.
- Ambitious standards set for designing, documenting and testing numerical software
- NAG Algol 60 and Fortran Libraries, Mark 1, released in October 1971, for specific ICL mainframe series
- Steve Hague joins as first, full-time employee in October 1971
- The Library continues to grow - three more Marks (major editions) over next five years
- First non-ICL Library implementations appear - a ground-breaking event in portable numerical software
- Interest in use of Library grows from industry and outside UK
- Pioneering use made of software tools in numerical software development
- NAG moves from Nottingham to Oxford in August 1973 (and "N" in "NAG" becomes "Numerical", rather than "Nottingham")
- NAG Ltd incorporated on 18th March 1976 in order to meet growing demand beyond academia for access to NAG software.
- Formed as a not-for-profit company (a status that it still retains) to encourage spirit of collaboration to continue.
- NAG Inc established in 1978 to meet growing demand in North America
- Because of pioneering experience in use of software tools, NAG coordinates US government-funded Toolpack project (1979-1982).
1980 - 1990... the second decade
- In 1980, NAG Ltd becomes financially self-supporting after the end of the ten-year period of grant support from the Computer Board.
- The first meeting of the NAG Users Association was held in 1981.
- Another portability first ... the multi-platform, multi-package, NAG Graphics Library launched in 1982.
- Release of the first partially vectorized implementation of the NAG Fortran Library (Cray-1) in 1983
- NAG Technical Policy Committee established and held first meeting in 1985
- Pilot implementations in Ada project begins in 1985 funded by European Union (EU)
- Fortran software tools project begins in 1984, funded by UK Alvey initiative
- DIAMOND (accurate arithmetic) project begins in 1985, funded by EU ESPRIT II.
- GLIMPSE (statistical expert system) project begins in 1986, funded by UK SERC
- Brian Ford, Founder Director of NAG, awarded OBE in 1989 in "recognition of outstanding services to British industry and research
- Collaboration in the US-funded LAPACK linear algebra project begins 1987
- Funded by EU ESPRIT II, the FOCUS (intelligent front-ends) and Supernode 2 (high performance computing) projects begin.
1990 - 2000... collaborations and development
Along with a continuing emphasis on quality and collaboration, the principal feature of the third decade of NAG was the broadening of the NAG product range to include new versions of the principal Library products, Fortran tools and the world's first Fortran 90 compiler, the advanced visualization application building system, IRIS Explorer, and involvement in symbolic/numeric software development. The world of scientific computing was undergoing major transformation during this decade, with the advent of the PC and the internet, and the range of NAG product and research activities sought to reflect and anticipate this continuing process of change.
- NAG GmbH is established in Germany in1990.
- NAG's first set of routines in C is launched in 1990, The NAG C Library, Mark 1.
- New NAG product launches of:
- NAGWare Fortran Tools, Release 1
- NAGWare f90 Compiler, Release 1
- IRIS Explorer, Release 3
- The NAG Parallel Library, Release 1
- AXIOM (symbolic solver system), Release 1
- Major new Marks of NAG's principal Library products continue to be released - in keeping with NAG's commitment to continued enhancement
- GRASPARC (computational visualisation) project begins in 1990, funded by the Department of Trade and Industry
- First availability of LAPACK routines in 1992
- NAG Web site established in 1993
- Commencement of PINEAPL (parallel numerical software) and FRISCO (polynomial solvers) projects begin, both funded by EU Fourth Framework
- Nihon NAG KK (formerly IRIS Explorer Center Japan) formed in Tokyo, 1996.
2000 - 2010
With the NAG ethos still very much to the fore, the fourth decade of NAG as a company has seen the organization building on the success of the past three decades. This decade has seen much upheaval in the world of scientific/technical/analytical computing, and as with all other organizations in the computing business, NAG has not been immune from those changes. The organization has adapted to those changing times, and words such as "agility" and "flexibility" figure much more prominently in recent incarnations of business plans than they did in the past! The specifics of business, organizational and technical issues may be in some respects quite different to their counterparts in 1970, 1980 or even 1990, but the NAG of 2010 stands proudly and recognizably with the same values and aspirations that have characterized the organization throughout those previous decades.
- NAG involved in EU Fourth Framework projects, STABLE (visual statistics), DECISION (design optimization), OpenMath (mathematics e-publishing) and JULIUS (high performance problem solving)
- New NAG product launches of:
- The NAG SMP Library, Release 1 (world first)
- NAGWare f95 Compiler, Release 1 (world first)
- NAG Statistical Add-Ins for Excel, Release 1
- NAG Data Mining Components, Release 1
- NIKE uses IRIS Explorer to design hi-tech footwear
- EU Fifth Framework commences with projects EUREDIT (data editing/cleansing), MONET (NET-based mathematics) and MKM (mathematical knowledge)
- NAG and AMD partnership formed with release of the co-developed AMD Core Math Library
- In July 2004, Brian Ford retires as Director of NAG after over 30 years of service. President of NAG Inc, Rob Meyer, becomes CEO of the NAG Group and Steve Hague of NAG Ltd becomes COO of NAG Oxford and CTO of NAG Group.
- NAG develops visualization component for the world's largest climate change experiment in 2005
- Mark 8 of the C Library released in January 2006
- NAG Oxford awarded Investors in People standard in March 2006 (retains this status in 2009
- NAG signs the largest deal in its history to provide CSE support to the supercomputing facility, HECToR
- Steve Hague, NAG's first employee, retires from his position as COO of NAG Ltd in June 2007
- NAG launches new connector product, the NAG Toolbox for MATLAB in April 2008
- Two new prototype products launched in 2009 - the NAG Library for .NET and NAG Numerical Routines for GPUs
- Serving the needs of Greater Chinia, NAG opens office in Taiwan in November 2009
- Mark 22 of the Fortran Library released in 2010
- Mark 22 of the NAG Library for SMP & multicore released in 2010
So what does the future hold for NAG? Predicting the future is a risky enterprise but there are a number of things about which we are confident. The need for sophisticated mathematical and statistical software and developer tools will continue to grow. The amount of data as well as the complexity of problems and applications continues to grow. Despite the promises, the world and the challenges we face are not growing simpler.
In ten year's time we still expect to be adding computational software to our code base with the same emphasis on accuracy and robustness while employing software engineering to make it accessible where it is needed. We also expect to see a continuing evolution of the customers and users of NAG software. Increasingly we see two groups emerging; professional software developers and researchers, analysts and modelers.
Professional developers will continue to use NAG code to build applications for internal users and external customers. Even in companies whose business is to provide a service, software is increasing becoming the means to deliver sophisticated services. Increasingly, the users of these applications, internal users or customers of an ISV, are "ordinary mortals" using sophisticated software, with NAG inside, to make better decisions.
To gain a sense for this, think about the car you drive. It almost certainly has a processor with software constantly computing and adjusting the air-fuel mixture to provide acceptable performance while reducing emissions. More software is keeping the driver informed of instantaneous and average fuel economy as well as "miles to empty". Drivers use this information to adjust their driving habits and plan refuelling stops. Very few of us think of ourselves in these cases as users of computers and software but in fact we are. End users are already forecasting the economics of pharmaceuticals, calculating retail promotional prices, and modeling exposure to hazardous materials without ever knowing that there is sophisticated NAG code doing the computation.
Increasingly, we also expect to see sophisticated analysts, modelers and researchers using NAG code via various desktop software and without programming in conventional languages such as Fortran, C, or Java. In many respects these NAG users will be like those that fuelled NAG's success in the 80's and 90's but they will be able to focus more energy on their models and less on the mechanics of programming. NAG will serve these users by building connections to the platforms where these users need NAG functionality.
NAG was founded on the principle that robust, accurate, and portable software was critical to the success of researchers. Throughout the decades we have built the expertise and maintained the emphasis on portability for a very good reason: hardware and software environments continue to change at a rapid pace. We see no letup in the pace of change for hardware. In little more than a year we can expect to see eight processors on a single, low-cost chip. Linux and Windows are likely to dominate the market but each will continue to evolve as will the compilers and run-time environments (Java, .Net) that support code portability. These evolutions in the hardware and software environment re-affirm the need for NAG's tradition of quality supported by our processes for algorithm selection, software development, testing and quality assurance.
NAG's success over the past forty years and our plans for the decade ahead hinge upon the quality of our people (staff, members and advisors) and the institutionalized "ethos" that carries on even as pioneers retire and new people step into their shoes. Certainly, an important element of that ethos is an emphasis on quality. NAG has emphasized a pragmatic balance of speed, accuracy, robustness and documentation in its products. "Getting the wrong answer faster" has never been part of our culture and never will be. NAG's success in serving customers is based, in equal parts, on the expertise of our people as well as their honesty and integrity. When we say "Results matter, trust NAG", we're talking as much about the people our customers encounter as our products. Through all of the changes we have seen and expect to see in hardware and software, customers and applications, we believe that transmitting these values from generation to generation will make us as successful in serving customers in the next four decades as in the previous four.