Veritus Ltd.
all rights reserved

Main Core   |   Technology Core

Contact Veritus   |  Information Core

Veritus technologies logo

MODERN ALGORITHMS

"Despite the dramatic increase in
the speed of hardware
and software, a super-
computer might need centuries..."
New developments in the construction of algorithms are changing entire industries and solving problems that were once thought impossible to solve. Algorithms are step-by-step equations for getting something done. This form of an equation has been around since the Greeks, who used algorithms to build the Parthenon.
    However, today there is a new generation of algorithms. Mathematicians who understand the power of today’s computers have constructed these algorithms. They take advantage of mountains of digital information never before available but can now gleaned from a multitude of databases. Algorithms have become a critical force in the global marketplace because they can find ways to solve problems faster, cheaper and better.
    The equation, which can involve millions of calculations, is fed into a computer in the form of a computer program. That way, you can plug in different variables, run the calculations at high speed and get an optimal result. The problem is, as the number of variables increases, the time it takes to calculate all the possibilities can go up exponentially. Despite the dramatic increase in the speed of hardware and software, a supercomputer might need centuries to do calculations. Especially those calculations involving pattern recognition.
    Modern mathematicians have gone to work on algorithms to make them faster. Quentin Stout, a professor at the University of Michigan has stated, "By rearranging the calculations, you can drastically reduce them. You find a more clever way so you don't do so many. That can be a major win."
    Sometimes, researchers make huge leaps. Until the 1980s, there was only one known kind of algorithm. It was called the simplex algorithm, for doing complex problems with a large number of variables. It essentially would move the calculations from point to point along a Zigzagging imaginary line until it found the optimal point. This process takes a long time when there are many variables. In 1985, Bell Labs researcher Narendra Karmarkar found a way to mathematically leap across boundaries instead of following that line. A simplex algorithm is like driving an interstate from one city to another; a Karmarkar algorithm is like flying there.
    Bill Pulleyblank, who develops algorithms at IBM's Watson Research Centre has stated, "A lot of algorithm development is not breakthrough work, you do better a little at a time." One painstaking method is to run an algorithm on a computer, detail how long each calculation takes, then work on each one to make it faster.
    The advances in Algorithms are having a major impact in what computers can do. IBM proved that in its bid to build a computer system for the Department of Energy that would do a once impossible task: make exact, real-time models of atomic blasts. IBM put together a massive computer to do the job, but the DOE said it cost too much. IBM researchers went to work on the algorithms and eventually cut the processing power needed and the price by half. IBM won a $93 million contract.
    Algorithms also are the key to a new area called pattern recognition. Pattern recognition algorithms are used by Veritus for authentication of paintings. These algorithms sort through massive amounts of data looking for repeated patterns. It will help research the data in the Human Genome Project, looking for obscure patterns among the billions of bits of DNA. IBM plans to supply Monsanto with pattern recognition algorithms, which will be used to match genes to develop better seeds. That kind of research as well as the research conducted by Veritus could have been done before, but it would have taken so long it would not have been cost-effective. Pattern discovery would have taken months, where there are many variables. Now it can be done in hours.
  
"Pattern discovery would have taken months, where there are many variables. Now it can be done in hours."
Technology Core Area

© 1995-2017, All rights reserved

If you have any technical problems with this site, please contact the

This site is being constantly updated and expanded,
it was last modified : 7 May 2016