 So, let's talk a bit about what you find once you get there in the system. So the most important thing, you're not using the cluster for the power of SSH just to play with the SSH and the command line, you're going there to do some analysis or some science with some applications. And you're hoping that you have the applications there and you hope that you get the resources like machines, accelerators, so forth, so that you can run these applications in a more powerful way. So basically, you want to do some stuff with some stuff with some programs. And here, we have tried to list in this page. And over here, we can first go through the basic ways of how do you find if you already have the possibility, well, if you already have the applications that you need in the system. And if you don't, what kind of checklist should you go through? And I can say that most of our, maybe precisely most of CMOS time is spent trying to install applications onto the cluster, because there's such a wide variety of things. And so many of them are not installed and packaged very well. So it makes difficulties, let's say. Yes. So that's why we usually try to solve the issue for a larger group of people at the same time. We don't want, we don't, so we try to, of course, we try to help individuals, but we also try to help the community as a whole, so that even those people who are those people who didn't send the ticket about certain applications, they can benefit from getting the application. So first, we have our applications page. Like you can go there and see if there's a software already available, and if there's some training examples there. Some of these examples, like if you're in other university, the software might be a bit different. Like some of the parts might be different, something like that. But if you find like one good submission script somewhere in the internet, you usually can run it on another cluster as well if you modify it a bit. So I highly recommend also checking, well, just googling for wider internet where they have similar kinds of HBC clusters have similar kinds of application pages. But first off, I would highly recommend looking at your own local clusters applications page. If you have the software, if the software is available, it's usually loaded via this module system, we will talk about the module system in a more detail later on. But basically, they are a way of loading these different software, because we have such a variety of software, we can't set defaults, we can't play favorites in this game, we have to be able to enable many of these software simultaneously. You should also check, of course, these tutorials. And same like if you're in auto, some of the information might be, like if you're working at the workstation or something like that, it might be in the workstation tutorial. If you can't find the software you need, you might want to check the usual channels where we do the communication with our users. So the issue tracker, the Zulip chat, other from with other universities, you might contact the admins. In our case, also join the garage, ask us directly, because usually that you don't have a tutorial base doesn't necessarily mean that we haven't seen the program before. It might just say that there's so few users that we haven't yet prioritized creating the tutorials page or updating it up to the higher standards, because we just have to prioritize time often in this case. And then, well, we might need to adapt that based on the world users. So if the users ask for or something, we put it by the priority. And throughout this process, there's few things that you should remember. Did you write these things to remember, because they look really clever and I didn't remember seeing them before? Yeah, I wrote them yesterday evening. But basically, I tried to distill what what's what a good thing to remember throughout this process is that like, scientific software, like the process itself is collaborative. So try to work with other users and share knowledge and see, try to seek knowledge from other users as well, because they might have the answer that you need. Another thing is that interesting problems draw people together independently. So if you're working on a certain type of a problem, it's most more quite likely that some other scientist is also working on a similar kind of problem. So it's a good idea to check who are these people so you can like, collaborate, use the same tools, work like you get like free free lunch, basically, if you work with the tools that somebody else has already created. Another thing is that you should try to form connections between users of similar software. So many software suites are like across vastly different fields, they're using the same kind of software. So this is especially like common in a statistical modeling or machine learning and stuff like that because that's that can be used in neuroscience, it can be used in in economics, it can be used in physics, all kinds of fields. So if your group and your field doesn't have people using the same software as you do, they might be somebody that works in a completely different field, but they are also using the same software. So it might be a good idea to collaborate among fields and among like look beyond your research group for people of the same software. We are trying to like get this kind of a user group being working in the altar. So basically like people who use the same kind of software can meet and we have a common discussion channel, let's say in the Zulit so that you can share information and problems that you encounter with the software or encounter with the applications. And the last thing is that if you don't if you find something interesting, tell people about it because like if you don't share it, it's basically only on you and it doesn't benefit the community. And like if you don't know who to share it with, share it with us because like we try to then disseminate the same information to our users. So basically if you encounter a problem or if you find some really nice tool that works for you, if you let us know about it, we can tell the next person who asked for similar kind of a thing about this tool. And like just a good example of this whole thing is the question we put on the icebreaker. And if you look at the icebreaker list here, like you can see that quite a lot of people are using Python and Matlab and similar kinds of tools like independently. Like I'm pretty sure a lot of you aren't working in the same fields. So if you have problems related to these packages, if you find some tools that are used in these kinds of fields, it's very good to communicate and share their knowledge with us, with other users and stuff like that because then it will make your job easier as well. Like I said, you don't want to use your own words, you might want to use the same words that everybody else is using, similarly with the software. And as an example, like many of the most common software is already installed in our systems, this might be a bit site specific. There are some common modules that I'm currently in the process of updating, of trying to get these even more like all of our software stack basically available on other sites as well. But most common modules like Python, R, Matlab, Julia, they are available. Yeah, there's an interesting question here. Can should you render Blender scenes on Triton? Yeah, well, I think maybe the more interesting question is not, is that possible? But what kind of things do you have to think about when doing something like that? Yeah, these kinds of like, yeah, this is actually a very nice question because these are the kinds of questions that basically you like, I would say it's like an 80 to any kind of a thing where like 80% of my time is trying to solve these kinds of questions because Blender for example is this kind of a graphical rendering software and it's meant for usually to be run in the systems with graphical hardware or graphical like displays basically. So it's very hard to get them working. And that's why it's even more important to basically have this kind of a collaborative process to get it working because like the first person who gets it working, it's very valuable information because it can take like weeks or months to get it working. And if somebody has to break through of how to get it working, it's amazing because then well, other sites are also interested in it. Some of these are possible to be done in try them as well. Yeah. And if you do get something like this working, go and post it to the issue tracker. And even saying, okay, these are the commands I ran and this is what I had to figure out. And now it works. Or you know, if you have a blog or things like that, go make a blog post about it or whatever, you know, just get that information out there somewhere that it can be found and then reused. So we are at 1243 here. What should we well, yeah, I would quickly meant we could quickly mention like, so when you're working with scientific software, it's good idea to make it reusable and share it with other people. And of course, like, you don't want to share it if you're in the middle of a publication or something. It's very secret. But like, if it's possible to get more people working with your project, do it because that helps a lot of people like vast majority of problems I'm solving are solved by other people. I'm just finding the solutions. So basically, I'm just googling stack overflow, you know, issue tracker, like, crawling through GitHub issues, some public request and trying to solve like this kind of stuff. So it's, that's that's the workflow that you need to like, if you go to that workflow, and you come, you put your software so that it's available for other people, you will get a lot more like collaboration, and mark, well, somebody might even use your software in their publication, you get a free citation. So that's really nice. Yeah. But okay, so so we might want to do a one exercise here. So so basically, on the exercises, the second exercise might be the most helpful, actually, the first one is, is not if you have time to the first one as well. But but the second one is that go to the application space of our application space or your own clusters application page and try to find if you find those software that you are actually using and try to find it if there's like documentation on that. And and see if you if you find it and try to like, look through the strategies, like that I posted that are at the top of this page and see if you can like, what how could you work with that? Like, so basically, do you find the software, do you find the documentation? If you don't, where could you ask for help on this? Okay, so how about we have, say, four minutes or so to discuss this and then break until one o'clock. So that's 14 minutes for a combination of chatting about stuff and breaks and all that. Is that good to everyone? Let's hope so. I think that will be good. But yeah, don't worry too much about this exercise. It's mainly more of a discussion than something you need to do. So yes, we'll start with the hands on modules at yeah, one o'clock then. Yeah, okay. See you later. So welcome back. There's a little bit, well, looking at the hack and D from applications, because if there's one final thing I can say, tell me you have found that things go wrong, like information is out of date, whatever. Unfortunately, this happens. And in that case, it's good to ask us rather than struggling too much. So we try to keep all of the documentation nice and up to date, but it doesn't always work so well. Okay, so are we anything else? Any other comments? Yeah, well, I guess let's go on. So there was in the hack and D there was lots of questions related to some of the exercises that we did previously about the applications, and especially on the TensorFlow exercise over here. And this is a bit of like this, like a really good way of demonstrating like how we want to like work with the users or work with the feedback that we get. So basically, this is pretty badly formulated question. It's most likely, or like last year, I wrote this because, well, this is kind of a thing that usually happens with users, they want to use some framework, and then they don't know how to use it. And for example, TensorFlow is one of them. And it might be a bit complicated to find from a documentation because their documentation isn't necessarily up to date. Now that I looked at it, the solution that was supposed to be found was this kind of a like a if you're in in Triton, you would do module load anaconda to load the Python environment. And then you would do something like Python for TensorFlow. And this would load you the TensorFlow. So this was the solution. But it really shows that the documentation necessarily or the exercise here aren't necessarily the most well, best thought out. So we'll have to add up. But this is the kind of thing that is very important to like, that we get the feedback. Because then we like the like what the other one talked previously about how to formulate your question and how to think about the problem when you're supplying the question. In our case, it's a bit of other way around. So we are dealing with these programs all the time. So for us, we constantly have to work to think about how somebody who has never seen these programs would see the problem. And that can be as complicated, almost as complicated as the other way around. So basically, do give us feedback on what is complicated, what is not clear, and we will try to adapt it because then we know that, okay, like we thought about this problem differently than you did. And that is very important. But okay, let's we can now move forward in our schedule. So we will talk about these modules and software loading later on. But we first want to work with some other topics. But we'll return to this TensorFlow thing in a different kind of way later on.