I’d like to thank the PRACE SoHPC for the opportunity to work in CINECA’s super computing center. It is a very important experience for me and I am fully enjoying my presence here. The working environment is incredible, regarding both the people working here and the facilities. I have been looking forward to practicing computer science in a high quality organization, such as CINECA, for a long time.
Regarding my tasks during the summer program, our supervisor (Massimiliano Guarassi) has assigned to me and the other student from the HPC program (Vojtech Nikl) to visualize some datasets coming from physics experiments. CINECA’s physics researchers are interested in phenomena occurring around the sun, particularly in the existence of coronal loops on the shell of the sun. Inside of these coronal loops, we can observe high magnetic flux and very high values of
temperature. In Figure 1 you can see a picture of this phenomenon that I found online. While examining the magnetic loops in the sun’s corona, we focus on the time evolution of 4 attributes of the loop: its temperature, density, current and heat. The researchers created a synthetic dataset that simulates the function of a magnetic loop and our task is to extract and visualize the useful information from it. Temperature and density are shown in Figure 2, current and heat are
presented in Figure 3.
During our first part of the program, we had to become familiar with an open-source platform for data analysis and visualization, called Paraview. The above screenshots from Figures 2 and 3 are made in Paraview. After learning the basics of Paraview, we used it to create 4 (static) animations, each one for one attribute. We achieved that by drawing a simple path around the loop, which makes the camera to zoom in and out of the loop for a few seconds. Our next task was to produce time-based (dynamic) animations that reveal the changes of these 4 characteristics of the loop in a time period of 60 seconds. In this last type of video, it is possible to observe for example the increase of the temperature around the loop and see the frequency with which every attribute changes over the time.
In Figure 4, the evolution of the temperature around the loop is shown. We can see the gradual increase of this attribute, as at the beginning the volume is transparent and then its opacity becomes very high, making the color of the volume very intense. In addition, remarkable piece of information is that the range of this attribute is in Million Kelvin degrees.
We developed 4 different time-based animations that correspond to the 4 different components of the loop. We used Paraview for configuring the parts of the loop that we want to show or hide, the type of their representation (e.g. show a part as a surface or as a volume), the range of values applied to the features, the color scales and many other parameters. During the image rendering procedure, Paraview extracted a large set of frames for every feature and then we worked on another open-source platform, called Blender, to combine these sets of images and produce the 4 final animations. Blender is a well-known tool for developing all kinds of animations and movies, so it was very useful for us. It let us configure even more parameters for our videos. During this week, our task is to combine the 4 videos in order to present a complete view of the loop and show all the meaningful and unique information of this phenomenon. In Figure 5 a preview of a combination video is shown. We also started to work on a new dataset, a wind turbine, in order to visualize its structure. In addition, we will try to render our animations by switching from occupying one node of CINECA’s clusters to parallelizing our scripts on many nodes. CINECA’s employees wish to explore the option of performing the rendering using a distributed system and see if there are some benefits from this change.
Although I am not familiar with physics data and I did not study about image processing in the past, I believe that our work will be proved very useful for me. Paraview and Blender have many benefits and countless options for data analysis and of course data visualization. So, I believe that one can use them for many kinds of applications and because I am studying data science for my master’s degree, I see it as a great opportunity for me to learn these frameworks.
Figure 1: “Magnetic Sun”
Figure 2: (a) Temperature and (b) density of the magnetic loop
Figure 3: (a) Current -electric charge- and (b) heat of the magnetic loop
Figure 4: Time evolution of the temperature of the magnetic loop
Figure 5: Left: current, right: heat, center: temperature
The SoHPC Experience
I would like to provide you with some further information about the second part of the summer program at CINECA. As I wrote at my first article, we focused our work on experiments concerning solar coronal loops, using basically 2 frameworks, Paraview for producing scientific visualizations and Blender for configuring further options on our videos.
This month, we completed our work on the sun’s corona by creating a combination video of the 4 different animations that we created in July (each one for one attribute of the loop). This final video reveals all the characteristics of the coronal loop, how these features evolve in time and how they interact with each other. You can find my version of the combination video in this link https://www.dropbox.com/sh/73ij1dcruqzjbli/AAC6Ipt1Xad8Q0i_Fy_bFzC3a?dl=0 .
In a time period of 60 seconds, the main picture starts with capturing the temperature attribute and changes 4 times ending with the appearance of the density attribute. Every time I plot one feature in the central position of the video, this feature disappears from its position in one of the (4) corners of the screen. In a clockwise direction, you can see in the corners of the screen first the current, then the heat, the temperature and finally the density attribute. This video corresponds to 2000 seconds of physical evolution.
During the first seconds of this movie, we can see that the temperature of the loop is not yet in its highest degree and passing from the temperature to the current we can observe that the electric charge increases around the loop. After this switch, the current is replaced by the heat, which starts to appear only on the basis of the loop and later spreads to the magnetic field lines. Passing from the current to the temperature again, we are now able to distinguish the rise of the temperature around the loop for almost 10 seconds. Finally, the movie terminates by showing us the density attribute, which is perceived only during the last part of the experiment.
Other than the experiments concerning the coronal loops on the shell of the sun, we also developed some videos about a new dataset, a wind turbine simulation. In this dataset, we are interested only in one attribute of the current object, namely its vorticity. We created a movie in which the vorticity evolution is displayed in the center of the screen and in the upper left-hand corner of the screen the grid of the wind turbine is presented. In Figure 1 (a) there is a wind turbine photo and in (b), (c) you can see 2 screenshots of the turbine grid (yellow part of the turbine) made in Paraview. We can also observe the movement of the wheel.
Currently we are arranging all the pending issues for our videos and at the same time we are trying to develop our presentation for the summer program. We have recorded some audio descriptions that will come along with our animations and we are trying to combine all the different components in order to obtain the final presentation.
I would like to thank you again for this amazing opportunity and as I wrote in my previous article, I am totally satisfied and happy about my work at CINECA and my personal experiences in Italy.
(a) Wind turbine design
(b) Showing the grid of the wind turbine (yellow part) -1
(c) Showing the grid of the wind turbine (yellow part) -2