The more complex the device, the more likely it is to fail. This is one of the harsh laws of technology. There are many other things, often paraphrased by technicians and engineers, and one in particular finds its reality in everyday reality: “Never let a mechanical (or electronic) device understand that you are in a hurry.” Proof is also in one of the most expensive instruments ever built by man, the James Webb Space Telescope (Jwst), a ten-billion-dollar object that, despite its cost, is by no means immune from failures and malfunctions. But like many other things we send into space, if it has a problem it must also figure out how to solve it without a human being able to physically intervene. As happened last August to an item placed inside an infrared device (Merry), which began to deteriorate.

By November, engineers had identified the cause of the problem and began developing a new way to use it, sending only software changes to Webb. Also, in December, the telescope got stuck for two weeks in safe mode and off science observations. Engineers immediately identified the problem: It was a software glitch in the onboard target guidance control system. However, according to a statement from NASA, the observatory returned to normal operation on December 20. Again, on January 15, the Niriss system had a communication delay, that is, it allowed James Webb to “see” certain frequencies, which led to a paralysis of the program he was running and a delay in starting some observations. In this case, too, the problem was resolved and on January 30 the telescope was back in full working condition. But if no one steps forward to fix it, how can it work again? The answer lies both in the existence of redundant systems and self-diagnostics, in the ability of the observatory’s “brain” to manage what in jargon is called a “silent failure”, and through the control logic of multiple subsystems, by sending software packets, but also in the operation of a model The machine learning known as Morpheus created by the researchers assesses captured images, identifies objects against the background of the universe, and evaluates whether or not they are galaxies, and if so what kind they are. In simple words, Morpheus allows cosmic images to be classified based on the type of pixels that develop from the images. Morpheus, in turn, was trained on the Lux supercomputer housed at the University of Santa Cruz, and is a brainchild and dream for computer-loving readers: It’s a computer with only 80 CPU (core unit) compute cores, each containing two Intel Cascade Lake processors. Xeon A 20 along with 28 other CPUs and two Nvidia V100 GPUs each.

second Brant Robertson, a professor of astrophysics at the University of Santa Cruz and one of the principal investigators who helped create Morpheus, the James Webb Space Telescope actually allows us to see the universe like never before. And to reinterpret what we discover. And we wonder how this extraordinary instrument was born, we can answer by saying that the textbook on which Webb’s supercomputer studied before launch was written with 7,600 images taken by NASA’s Hubble Space Telescope from 1991 until a little earlier when Webb was being prepared for launch. And while it is already working, new smart functions are being prepared in California, such as “Deblending”, that is, the ability to distinguish between astronomical objects that appear to be superimposed on each other.