Day of Russian Science

We congratulate all those people who, in spite of everything, are engaged in a very noble and honorable work. Thanks to them, humanity has gone so far in its development. And February 8 was their professional holiday, the Day of Russian Science.

291 years ago (as time flies), on February 8, 1724, the Russian Senate issued a decree establishing the Academy of Sciences. And since 2000, Russian scientists celebrate their holiday on this day. True, some adhere to the old date: earlier the Day of Science (not Russian, but in general) was celebrated on the third Sunday of April. But this is about how the old New Year. The last decades have been difficult for our science, which must be confessed, but still young people still have an interest in knowledge. Of course, not everyone as a result devotes himself to a scientific career, but the same can be said about any other profession.

Mail.Ru Group, for example, employs several employees with advanced degrees. And on the eve of today's holiday, we talked with them about whether the “weight of scientific knowledge” helps them in life.
')
George Meshkov . Programmer in the game client team (Client team). Candidate of Physical and Mathematical Sciences.

My thesis was called “Electric control of micromagnetic inhomogeneities as a new principle of operation of magnetic electronics devices”. In a nutshell - I wrote about how magnetic dipoles can be moved by a static electric field, which in principle is impossible in a vacuum, but it is possible in some environments. From a practical point of view, this is the development of a new type of non-volatile memory, other than Flash memory.

For me personally, the coolest thing was that we and the supervisor resuscitated a practically dead installation and discovered a new effect that no one had seen before us. It is very valuable for me that we have gone through all the stages of the life cycle of a scientific idea — theoretical prediction, computer simulation, experiment, confirmation of the theory. For the physicist, the most interesting part of my work was the mutual influence of a constant electric and magnetic field. For an ordinary person, a potentially new type of memory and, possibly, an experimental setup capable of analyzing the nanosecond dynamics of processes.

In fact, the candidate’s “crust” itself is quoted only in scientific circles. And also required for the occupation of some government posts. But the experience of scientific work and, at first glance, too fundamental knowledge in their specialty is sometimes very helpful. For example, the day before yesterday I recalled the skills of statistical data processing, when we analyzed the dynamics of FPS growth among players during the Skyforge PTA . When I worked with CCD cameras, to synchronize them with each other, I had to get an oscilloscope and recall radio physics. And the experience of speaking at scientific seminars, the ability to explain complex things in simple words is simply invaluable at meetings and planning meetings;)

Nikolai Anokhin . Head of machine learning group. Candidate of Physical and Mathematical Sciences.

My thesis is devoted to the control of nonlinear dynamical systems. We used methods developed in control theory and stability theory. Although the tasks that I solve in Mail.Ru have points of contact with my scientific work, first of all, academic education gave me the ability to use sources and the habit of diversifying the analysis of any problems.

More about work. Imagine a robotic arm with several “joints”, some of which have motors. The point is to find an algorithm with the specified properties, with which you can control such a manipulator. For example, an algorithm that allows you to spend a minimum of fuel for engines. Or less time to spend on moving the subject. In my thesis, a situation was considered when an exact cast should be guaranteed to occur in a finite time. Particularly interesting was the part devoted to software modeling of various mechanical systems (for example, the same manipulators) using Matlab.

Igor Kretinin . Programmer-researcher in a machine learning group. Candidate of Physical and Mathematical Sciences.

My thesis was called “Dynamic Atomic Polarizability in the Theory of Quantum Defect” and was related to optics and laser spectroscopy. The degree of Candidate of Science allowed me to get a research grant quickly and completely legally exempted me from compulsory military service. I didn’t feel any special influence of my candidate’s degree on my life (my professor once said that a scientific degree is like a tie, something respected in society, but no more than that). However, the very fact of belonging to a professional subset of “people of science” is useful and pleasant.

When I communicate with applicants for vacancies, then, of course, I pay attention to education. Practice shows that people with a good education better understand the subject, know how to work independently, find answers to questions and solve problems. They know how to continuously learn. Moreover, the search for information on the Internet began with the solution of the problems of searching, ranking and structuring scientific information. And this is taught in high school.

Peter Shurkhovetsky . Programmer. Pre-graduate student (that is, my work was approved and allowed to protect). Theme of the dissertation “Effective solution of the problem of resource discovery in globally distributed peering systems”.

My dissertation was devoted to modeling existing peering networks: BitTorrent, Skype, etc. I had a reviewer who developed control systems for space launch vehicles, for example, Proton.

I believe that my "learning" does not affect anything (in the country as a whole and in the company in particular) does not help anything. Unless I began to think more systematically after this training. Again, I can write an article that other people will read, and not drop it at the very beginning.

Vladimir Gulin . The leader of the machine learning group. Candidate of Technical Sciences.

The topic of the thesis is “Research and development of methods and software for text documents classification”. In life, she does not help me. I defended myself in the summer of 2013, but so far there has never been a situation in which the presence of a master's thesis somehow helped me. The only thing worth noting is the increased interest from recruiters from Western IT companies.

The topic of the thesis is closely related to my work in Mail.Ru Group. As soon as I came to the project, I immediately became involved in creating a machine learning system that could be used to rank documents for a specific query. As a result of my work in a few weeks, we had a similar system. Later, it was possible to adapt it for other tasks, such as the classification of requests and the prioritization of documents for their retrieval by the search robot.

Actually, the most significant part of the results of my thesis consists of a description of the algorithms incorporated in this system.

Andrey Kutuzov . Linguist. Candidate of Philology.

The thesis was about computer-mediated communication: how does communication differ from a linguistic point of view?
people through computer networks from real-life communication. I wrote it for a long time, so that the specific content now almost never helps. But help the acquired skills of research, scientific approach to any problem. The correct formulation of experiments, the requirement of reproducibility of results, verifiability, falsifiability - all this can be useful in any intellectual activity. Including in linguistics when developing search engines.

Artyom Libkind . Build manager in the infrastructure team. With the defense of the thesis did not work out, but in the presence of honest three years of graduate school and the development of computational methods for mat deformation modeling.

Matmodeling of deformations of solid media is a very specific thing. This is similar to the strength of materials, only operating not with hardness, but with the state of the system. And one of the most difficult things in it is just the calculation of everything. We are talking about multidimensional matrixes of diffures, which are reduced to a solution iteratively, and all this together is considered to be methods of 20 years ago. As before, the same models were considered on BESM for months, now they are considered days on one core.

In addition to the work on the calculation itself, I first of all was interested in changing the approach to how it is calculated. On the one hand, all Wishlist could not be implemented. It turned out an interesting option, tested on several models, which gives a time saving of 1.5-2 times, if you're lucky.

And then I ran into all the hell of the bureaucracy of the Russian Academy of Sciences, I learned a lot of interesting things about how the defenses are actually being carried out and what is needed for that - and dropped the case. My supervisor did not really understand how what I wrote works, so that all the developments remained for another good math, which just helped with the computational base.

As a result for me, first of all, I was convinced from my own experience that if you have an idea how to do everything better, and it seems good to you, you just need to take and implement it, even if people around you say “don't waste time, much who tried, they did not succeed. "