In the U.S., has developed a method for predicting extreme events around the world
Researchers at the Massachusetts Institute of technology (USA) have created an algorithm that can identify patterns that lead to extreme events.
“Currently, there is no way to explain when there are these extreme events. But if we can predict where the occurrence of these extreme events, we hope that we will be able to apply some control strategy to avoid them”, – said Themistocles Sapsis, associate Professor of mechanical and ocean engineering at the Massachusetts Institute of technology.
The scientist tried to predict extreme events in such systems, using a variety of complex mathematical formulas-called dynamic equations. After solving these equations can predict the state of a complex system, taking into account the passage of time. By including a set of “initial conditions” into the equation, scientists determined whether these conditions are warning signs for a specific event.
But, according to researchers from the Institute of Massachusets, this method often contains errors. In particular, because it uses more information than you need, and this information gives not one solution but many. And as you know from the school course of mathematics, many of the solutions shows that the solutions to this problem there.
In the study, a team at mit have developed a computer algorithm that uses equations and available data to determine warning signs of extreme events in the real world.
The algorithm acts as a sieve to weed out the previous events that in the real system in real time go unnoticed by the people.
“We look at the equations that predict potential events in which the tension is growing, and so they become extreme events. But there is also a large uncertainty in these projections because any event that under the influence of different factors may be extreme, can develop into PE, which will be massive, but so exotic, impossible for these terms, its probability will be reduced to zero,” said Sapsis.
The researchers tested their prediction algorithm for plume of cigarette smoke, water circulation in the ocean and air flow around a jet engine.
“We used the equations describing the system, and some basic properties of the system, expressed using data obtained from a small number of numerical calculations, and we came up with the name of the phenomena of life that are characteristic signals reporting the probability of extreme events,” said Sapsis.
The researchers found that previous events predicted by their method, developed in extreme events in 75-99% of cases.
Accordingly, the method can be applied to all types of systems in the real world.
“It happens in random places around the world, and the question is how to predict where they will occur, these vortices or hot spots of extreme events. If you can predict where these things take place, you may be able to develop some control methods to suppress them”, – summed up Professor Sapsis.
© 2017 – 2019, paradox. All rights reserved.
