The use of MI is gaining ground today in the field of solar and plasma physics and space weather research. Recently, a new model was developed from which, based on previous scientific studies, it was hoped that it would be able to reproduce the Sun in almost perfect detail at approx. A magnetic map of its 5400-degree surface, the photosphere. For the MI, the observations made by the NASA SDO satellite in the Sun were given as input parameters.
An accurate assessment of the Sun’s atmosphere would be an important advance in the field of plasma astrophysics, as the so-called the magnetic field of our central star plays a very important role in the formation of solar magnetic active regions. To map it, Róbert Erdélyi, professor of astronomy at the University of Sheffield and ELTE, chairman of the board of the Hungarian Solar Physics Foundation, established the Gyula-based SAMNet (Solar Activity Monitor Network) international space weather observation network. Use SAMNet’s proprietary magnetic field measuring instruments to study the lower atmosphere of the Sun between the photosphere and the chromosphere.
In this dynamically changing active region, high-energy fleas (light flares) and also plasma pulses (corona ejections), which can cause very serious space-weather disturbances. Space weather is a summary name for the disturbances caused by the Sun that can be observed in the area around the Earth. More serious anomalies in space weather, such as storms, can severely damage, for example, our GPS and telecommunications satellite systems, create surges in our high-voltage power lines, interrupting uninterruptible power supplies even in continental areas.
Artificial intelligence can be achieved with fantasy but as Róbert Erdélyi said, their research showed that without a mathematical analysis of the physical phenomena characterizing high-energy solar flares, the results of MI can easily be misleading. “ Artificial intelligence should not be seen as an omniscient sphere of spell; if used incorrectly, we can draw the wrong conclusions. Mathematical and physical modeling is essential in research ,” said the professor.
“ The research was aimed at verifying our results obtained during space weather forecasting with the help of artificial intelligence ” – added Marianna Korsós, Department of Astronomy, Eötvös Loránd University postdoctoral researcher, member of the international research group. “ This is an exciting and rapidly evolving interdisciplinary field, the results of which, however, should be treated with caution. I am very proud to have been part of this excellent international collaboration as a young researcher, learning a great deal about the extent to which new techniques can be used. “
Jiajia Liu (University of Sheffield and Queen’s Belfast), Yimin Wang (University of Sheffield), Xin Huang (Chinese Academy of Sciences), Marianna Korsós (ELTE and Aberystwyth University) , Ye Jiang (University of Sheffield), Yuming Wang (University of Science and Technology of China) and Robert Erdélyi, after introducing innovations, proved that
criticism and very carefully use the data of the MI model to predict the structure of the solar magnetic field.
“ We noticed that the previously thought it was perfect Our MI model performs much worse than expected “,” Jia explained jia Liu and Yimin Wang. “ Artificial intelligence is not yet able to adequately reproduce completely unsigned solar atmospheric magnetic flux values or other important physical parameters, such as the value of net magnetic flux or the number of neutral lines separating the magnetic field, which are essential. “
The researchers added that their results are supported by currently known physical models, as the theory of magnetohydrodynamics states that the chromosphere and the observations of the crown do not provide sufficient information on the detailed photospherical magnetic field structures
“ MI is a rapidly evolving discipline that is indeed widely applicable and is becoming more common in our everyday lives. However, users should be aware l also have its limitations, especially in terms of applicability in science “- warned Róbert Erdélyi. “ In the absence of basic mathematical and physical models, MI often generates erroneous models or data, even when using the most advanced artificial intelligence or machine learning techniques. ”
The research was carried out at Eötvös Loránd University within the framework and support of the astro- and particle physics theme of the Higher Education Institutional Excellence Program. The breakthrough result was reported by researchers in the prestigious journal Nature Astronomy
Hardware, software, tests, curiosities and colorful news from the IT world by clicking here