Broyden

Broyden
Name	Broyden
Birth date	1923
Death date	2018
Nationality	British
Fields	Mathematics, Numerical Analysis, Optimization
Known for	Quasi-Newton methods, Broyden's methods

Broyden was a British numerical analyst and mathematician notable for seminal contributions to iterative methods for solving nonlinear systems and optimization. His work influenced computational practices in applied mathematics, engineering, physics, and computer science, shaping algorithms used in organizations such as Argonne National Laboratory, Los Alamos National Laboratory, NASA, and industrial firms. Broyden collaborated with contemporaries connected to institutions like University of Cambridge, Imperial College London, Bell Labs, and the National Physical Laboratory during a career spanning mid-20th century numerical linear algebra and optimization developments.

Early life and education

Born in the United Kingdom, Broyden received formative training in mathematics and applied analysis, studying at institutions associated with University of Cambridge and perhaps interacting with scholars from University of Oxford and King's College London. His early mentors and colleagues included figures from the British numerical analysis community connected to C. B. Moler-era influences and researchers tied to National Physical Laboratory computing projects. Broyden's education coincided with postwar expansions in computational resources at centers such as Manchester University and University of Edinburgh, where developments in electronic computing and numerical methods guided his interests toward practical algorithms for nonlinear problems.

Mathematical contributions

Broyden developed theoretical and algorithmic frameworks for updating approximations to Jacobians and inverse Jacobians in iterative solvers for systems of nonlinear equations and unconstrained optimization. His analyses intersected with work by John von Neumann, Alan Turing, George Dantzig, Richard Bellman, and Magnus Hestenes on computational methods and iterative optimization. He provided rigorous derivations of rank-one and rank-two update formulas linked to quasi-Newton theory advanced by researchers at IBM Research and Bell Labs and related to matrix identities used in numerical linear algebra at Courant Institute and Princeton University. His contributions clarified convergence properties, stability, and efficiency criteria, connecting to theoretical frameworks elaborated by Walter Murray and researchers at MIT and Stanford University.

Broyden's work addressed challenges in high-dimensional nonlinear systems appearing in contexts studied at Lawrence Livermore National Laboratory and Argonne National Laboratory, offering alternatives to direct Jacobian evaluation used in approaches from S. J. Wright-type theories. He analyzed trade-offs between storage, computational complexity, and robustness, contributing to the maturation of iterative solvers used alongside preconditioning strategies developed at ETH Zurich and Technical University of Denmark.

Broyden's methods

He introduced families of quasi-Newton update rules that produce approximations to Jacobian matrices or their inverses without full derivative evaluations, now known collectively as a set of methods bearing his name. These include updates that minimize changes under specific norm conditions, align with secant conditions, and maintain low-rank corrections akin to techniques from Kronecker product-based formulations seen in research at University of California, Berkeley and University of Illinois Urbana-Champaign. Broyden's "good" and "bad" update variants were evaluated relative to contemporaneous schemes like the Davidon–Fletcher–Powell formula and the BFGS family developed within optimization groups at Rutgers University and University of Michigan.

The methods are characterized by low per-iteration cost and moderate storage, making them suitable for large-scale problems tackled by research groups at Los Alamos National Laboratory and industrial modeling teams at Shell Oil Company and General Electric. Broyden's updates satisfy secant equations analogous to those used in quasi-Newton frameworks advanced at University of California, Los Angeles and integrate with line-search and trust-region globalization strategies studied by scholars at University of British Columbia and University of Maryland.

Applications and impact

Broyden's algorithms found rapid adoption across disciplines: in computational fluid dynamics projects at NASA Ames Research Center and European Space Agency, in electronic structure calculations pursued at CERN and Brookhaven National Laboratory, and in econometric modeling linked to research at London School of Economics and Harvard University. Engineering teams at Rolls-Royce and Siemens used his updates for large-scale simulation and design optimization, while chemists at Max Planck Institute and California Institute of Technology employed them within self-consistent field procedures.

The impact extended to software ecosystems: numerical libraries at Netlib, packages developed in environments influenced by AT&T Bell Laboratories and Microsoft Research, and implementations in scientific computing frameworks originating from Oak Ridge National Laboratory incorporated Broyden-style routines. His methods informed algorithmic choices in electronic design automation at Intel and in inverse problem workflows at Columbia University. Citation trails link his work to later advances by practitioners at University of Cambridge Department of Applied Mathematics and Theoretical Physics and research groups at Princeton Plasma Physics Laboratory.

Selected publications and honors

Key publications include original papers presenting his update formulas and analyses in journals and conference proceedings circulated among numerically focused institutions like SIAM and IEEE. He received recognition from professional societies connected to Institute of Mathematics and its Applications and was cited in award contexts alongside laureates from Fields Medal-adjacent mathematical communities and optimization prizes sponsored by organizations such as INFORMS.

Representative writings were disseminated through proceedings associated with International Conference on Numerical Analysis and monographs used by researchers at Delft University of Technology and University of Tokyo. Honors reflected engagement with the international numerical analysis community, linking him to contemporaries affiliated with Royal Society-associated networks and optimization groups at Centre National de la Recherche Scientifique.

Category:Numerical analysts Category:British mathematicians