Evolution on the Galapagos Islands has helped inspire researchers developing novel ways to tackle some of the trickiest problems in business.
By Mark Ward
BBC News Online technology correspondent
Brighton-based Codefarm is using evolutionary computing to swiftly search through thousands of potential solutions to these problems.
The Galapagos Islands are famous for their tortoises
To speed up the time it takes to find the best answer, the system runs on a cluster, or grid, of computers.
Banks look set to be among the first to use the techniques to advise investors on the best way to balance risks and rewards.
It is well known that Charles Darwin visited the Galapagos Islands and what he saw and studied there helped him put together the theory of evolution.
But the pattern of evolution on the islands is now being put to more commercial uses.
Many firms, said Jeremy Mabbitt, managing director of Codefarm, build numerical models to help them understand what they do and get better at it.
For instance, aircraft designers will produce simulations of plane wings to find the most aerodynamic design.
Similarly, banks often use complicated spreadsheets to map the relative risks and rewards of particular investment strategies.
Cranking through all possible permutations and optimising the models can take huge amounts of time, said Mr Mabbitt.
He said Codefarm had found a way to help businesses search through the space of possible answers by harnessing techniques from evolution and the way life has developed on the Galapagos Islands.
The algorithms used by Codefarm act as the basic genetic material for possible solutions, which are then recombined and mutated to create novel answers.
Urge to merge
"The evolutionary algorithms work on populations," said Mr Mabbitt. "Each one works on different computers and you can run sub-populations on each processor grid."
Many of the banks that are becoming the first customers for Codefarm have dedicated clusters of servers bought to help with computationally intensive problems.
The software produces lots of answers
The result, he said, was discrete islands of possible solutions.
"You can enable a certain amount of migration between nearby islands and that lets the algorithms scale very efficiently," he said.
Currently, he said, many banks were comparing how Codefarm's algorithm toolbox, called Galapagos, performs on scenarios the financial firms have encountered in the past.
The banks were using the software in credit derivative divisions to manage portfolios of money and work out how best to use cash to get the return investors want.
These returns also have to be weighed against the limits that stringent regulations force banks to put on the risks they take with investors' cash.
So far, said Mr Mabbitt, the Galapagos toolkit was producing better results faster than the highly paid staff, who usually do the work, could manage.
Previously it took hours of spreadsheet tweaking to get a good result, said Mr Mabbitt.
In some cases the evolutionary algorithms have produced ingenious solutions that would not have occurred to human investors.
"Initially we are taking that as an undesirable strategy," said Mr Mabbitt. "We only want results that look reasonable, though in the future we may relax this."