 Very curious to understand. What kind of engines do we use as the solvers in your programs? Is it like heuristic, something hill climbing algorithm? Or what is it? Or simplex or whatever? What is the core? What is the engine of your solving? What are you using? I think he answered you. It depends on the project. Historically we used to use C-Plex, but we switched to Express. I don't know the exact reason why we switched, but I know internally we used Express for the MIT Solver. We used GuroB, so it doesn't matter. We've done studies and for the most part they're roughly the same. But I think Express, I think we chose for the solutions were consistently better and then they also offered better support. So that's also a big aspect of it. Honestly I'm not familiar with these things because this is not my major thinner power. I'm just interested to understand which are the base algorithms that are behind your engines? You guys just tried to answer basically mixing the jet programming first. So mixing the jet programming is nothing but the enhanced version of linear programming. You might not know basically where to solve linear programming. It's basically just here mathematical solving. But then some of the constraints, actually telling you that some of these cannot be fractioned. Because linear programming will only give you fraction number. But you cannot do fraction all the time. Some of them have to be exact number. What is a fleetside? It cannot be 2.5, it has to be 3 or 2. This is how linear programming come into play, that they do some kind of heuristic. I think they call branch and bow. I'm not an expert there, but do branch and bow. To try to see that what integer actually closer to real solution, while they're still maintaining optimality of the solution. Also MIT actually have three major components. One is objectives, right? Basically you want to minimize cost. So that is your objective. Then the next thing is constraint. Basically what have to be integer? What actually have to be in a certain range? 1 to 5, 2 to 10, you actually exactly. That's a constraint. The last one is decision variable. It means that what decision you have to give. You want the fleetside a3, a0 for 5, you want a3, 3, 0 for 10. Those become your decision at the end. And they have to be integer. So that's done. So you're going through iteration. Until you say, hey, I'm very close to optimality here. And I can no longer improve anymore. I'm done. And it keeps some kind of heuristic improvement. Until it reaches a certain limit, a certain time, then they give up. It becomes a solution of MIT. So reasonable? Yeah, sure. The combination and the heuristic is also playing a role. Another thing that Chairman explained earlier, neighborhood search. Neighborhood search. Neighborhood search when he said network optimization. That's also another key and important heuristic approach that we use in a company. Yeah, construct some initial solution. After that try to improve it directly until we're close to optimality. Do you have experience to work with local companies here in Armenia? Or there were no cases of mainly focused for international markets? Because our solution is really required from before. It requires, let's say, large companies or large-scale companies. So I think, therefore, we are focused for this market. But basically we don't have any means for cooperation. There is interest there, so we can discuss and to see whether we can find some mutual interest there. So, I mean, for instance, our metropolitan would apply or offer some. I mean, we need some optimization. You know, we are currently discussing with our ministry also. I think basically we need to evaluate how efficient it will be. Because e-scale is really very small. Maybe the solution will not be required. Maybe by hand it will be much easier just to implement such kind of really complex solution. Or maybe some really similar things can be provided. We are discussing to see whether it will be feasible to implement something for Armenia. Because also it's also in our class that we are currently discussing. So if it will be feasible, it will be interesting for Armenia. Definitely, I think also our company is ready to support. Because we are here, we have some presence here. And I think also our management would like to support Armenian government with this initiative. I mean, we're always looking for interesting business cases to solve. But solving these problems is not cheap. And again, we're not free. And it's a good thing we aren't. Because then we wouldn't be able to pay for the hackathon and the internship program. Okay? Thank you. Okay, so thank you once again. So just on behalf of us here in the College of Zabin and Sonya, in the College of Science and Engineering. Thanks again for coming, taking time out of your travels. And for those of you who do it here. At your work day to come here and share this with our community here. So we're growing as a science and engineering community here. We, as many know, we have an industrial engineering master's program, a computer science bachelor's and master's. And pending accreditation will be opening a new undergraduate degree in engineering sciences this fall. Kind of an electron mechanical. Maybe we'll help build some of these devices that you guys can then optimize. So, spread the word, because it's just a pretty recent decision on our particular met direction with the undergraduate degree. And in general, if a bachelor's or master's degree in these fields, they're interesting to you or your colleagues, brothers, sisters, folks, please spread the word. And many of our students do work at the same time. So if we find a good accommodation for, let's say, part-time studies and full-time work or something like that, we do a more pretty class. So thanks again, guys, and look forward to collaboration. And the hackathon should be, I think, a lot of interest to our students and maybe alumni and folks who are in the field. So thanks again.