Huawei cloud has always had the dream of "heaven"
The Starry Sea is the most infinite dream of mankind
Chinese researchers have never stopped looking at the universe. In this distant attempt and expectation, Huawei cloud has always had the dream of going to heaven
if you, like Huawei cloud, have looked up at the deep sky and thought about the vast starry sky at night, it is very necessary to understand ska, one of the largest scientific projects in human history
The ska square kilometer array, a foul level artifact for exploring the universe, is the world's largest synthetic aperture radio telescope planned by international astronomers and a new generation Radio Telescope project to be built in Australia and South Africaska for what
it may be used to detect the dawn and dark periods of the universe, study galaxy evolution, cosmology and dark energy, verify whether there are other life in the universe, search for extraterrestrial civilization, test the strong gravitational field with pulsars and black holes, and study the origin and evolution of the cosmic magnetic field.
in short, as Philip diamond, director general of ska, introduced:
the purpose of this project is to observe the universe and understand our position in the universe, From the big bang to the local region of our galaxy and the sun. But its value, in addition to being used in astronomical research, will also bring us many technological and computational innovations
em~~ then the problem comes. What problems will you encounter when you use this artifact to chase and explore the sea of stars
the data veil that blocks the truth of the universe
every truth is often covered with layers of veil, and ska is the finger that wants to lift the veil of the universe. With ultra-high sensitivity, large field of view, ultra-fast Sky Survey speed and ultra-high resolution, ska will map the sky several times faster than the existing facilities, and the observation of the sky will show unprecedented details
but the resulting problem is that unprecedented data will also be generated! Massive data is not terrible. What's terrible is getting lost in the ocean of data
the data shows that after the completion of ska, about 600pb of data will be archived every year, and the data flow is 10 times that of the current global interconnection. Data storage and data analysis are the major challenges for the project to significantly improve the green manufacturing level of relevant enterprises. Moreover, ska facilities will be spread across the northern and southern hemispheres and three continents. To share these data with other research institutions, data transmission is also a big problem
data cannot be effectively processed one day, and the truth of the universe is hidden behind the veil one day. It's like we finally came to a mine full of treasures, but we couldn't find the entrance in the fog. Knowing that the treasure is in front of you, you can't go in and carry it away
then, how to mine the value information of data and lift the veil
Shanghai Observatory gave the answer: Huawei cloud AI
we all know the significance of cloud computing for various industries: coordinate the deployment of data across space, flexibly purchase resources according to their own needs, and share the update of software capabilities in the cloud at any time. To solve the storage, transportation and analysis of massive data, cloud +ai+ full computing power can be said to be the most effective way at present cloud + AI + full computing power adds infinite possibilities to exploration. relying on Huawei cloud's elastic storage service, massive data can be stored quickly and flexibly, and at the same time, the access threshold of data is reduced. For massive data analysis, the current world's fastest AI training cluster and Huawei cloud AI shengteng cluster service provide surging computing power on demand, greatly shortening the process of ska astronomical research 01 positioning and classification of stars in the universe in the past, when using traditional methods to process the sky observation data of ska pilot project, it took astronomers 169 days to locate and identify about 200000 stars in the southern hemisphere sky only once now, the Shanghai Observatory and Huawei jointly check whether the cooling medium activity in the cooling tank is good, train the AI model through deep learning methods, and use the world's top AI powerful computing power of Huawei cloud AI shengteng cluster service. It takes only 10 seconds to complete this task. At the same time, the average accuracy of AI model can reach 81%, which greatly improves the efficiency of scientific research 02 carry out scientific research on stars in the universe in addition to positioning and classification of stars in the universe, under the full stack AI capability of Huawei cloud, Shanghai Observatory has also been able to bring more AI capabilities into scientific research, such as pulsar signal recognition research this research is to train and establish an AI pulsar recognition model in a short time based on hundreds of thousands of pulsar signal data, using Huawei cloud AI shengteng cluster service, and analyze whether the image is a pulsar signal. The recognition accuracy and recall rate are above 98% large scale neural network training for the application prospect of cloud and AI in astronomical research projects, as an Tao, the Chinese representative of the steering committee of ska regional center and the head of ska project at Shanghai Observatory, said: for the flexible resource allocation of computing and storage, cloud provides a good solution, which is a trend of ska scientific computing in the future people should have dreams, so should clouds we want to go to heaven Hello,World! Hello,Bigger World!
LINK
Copyright © 2011 JIN SHI