Hi, I'm a backend developer, have been developing financial (banking, cards, e-transfer, e-commerce) software since 2006 in Java, currently mostly writing Kotlin and a bit of Java 8 (90/10% K/☕ ). Joined this channel just because I'm interested in science (popular science level, mind you) and got curious about Kotlin used in the field. Regarding graphing, I haven't heard about anything Kotlin specific, but Java offers many possibilities and all of them can of course be used from Kotlin. There's also kotlin-native which opens the world of native libraries and tools, although native is a bit rough around the edges still, still JNA can be used instead, same as from java.

For native, we plan to add iOs to the rendering platforms. But I still find Kotlin/Native not efficient enough to work with. So I wait a little bit before doing that part.

Another thing that I would like to see is a better browser than webview. I don't know how much it evolved this last year, but it was still somewhat limited

@Pavel Apisov It highly depends which country you are in. You're talking about US? I've seen both approaches, entry level jobs and master degrees, being successful (in US) for people… Some took the entry level jobs and then did a master degree in parallel. So there are a lot of ways.

@Pavel Apisov you might find courses here - https://www.kadenze.com/ useful

Hi @altavir, I’m working on #data2viz and charts.kt libraries. What is your target platform? We planned to make charts.kt free for non commercial organizations and of course, schools and scientist research. Our biggest problem now is that we need to earn money for living, 😉. Even, if our main goal is to provide tools to help people understand data, at some point we need a little money. So the idea is to have data2viz open source and free for all usages, and charts.kt, a higher level tool, provided in exchange of a commercial licence for commercial usages (see highcharts.js). We think that the multiplatform nature of kotlin is a big asset and hope that a lot of developments will move to kotlin. Commercial licences could pay for the work on the library.

- We want at least one package to follow Python conventions so it will be easy for Python guys to switch to Kotlin. It probably should still follow common API somewhere on lower level.

Those more or less sum it up. I propose to create two Keep-like repositories to track Design solutions for mathematics and visualization (I can do it right now).

Have you heard about vega-lite? https://vega.github.io/vega-lite/

We have a first pull request. Badly formed though. @Zelenyi wants multi-dimensional histograms which are used quite frequently in particle physics. Does someone have the similar experience?

This is really nice. @thomasnield you can try to do your simulation in 3d now 😉

But this warms me up to continuous modeling a little more, and reminds me that there is a time and place for both discrete and continuous.

@altavir I'll keep that in mind. Thanks for doing this. Brilliant discussion

@altavir why are you using a Poisson distribution for elapsed times between customer arrivals. Shouldn't it be

Exponential

? like this?

Anyone know of any kotlin libs for quaternion & versor functions?

Does anyone know actual use-case for integer nd-structures?

We just had a short discussion with @Vsevolod Tolstopyatov [JB] about coroutines new scheduler and its possible applications for heavy load scientific computations. My experiments with coroutines for lazy nd-structure transformations shows that making simple numeric operations with them won't work, the overhead for switching contexts is too large, but there is one application my group was working on for some time: Monte-Carlo simulation in particle physics. Simulation looks like follows: particle interacts (not very expensive but also not trivial computation). Upon interaction it could die or produce one or several particles that propagates further. Now we want to move computations to other threads to utilize multiple cores or even cluster. I can't create a thread for each computation, it is too expensive. Current engines use different types of really ugly and ineffective queuing. The idea is to use coroutines and channels for that. We have a prototype engine, but it requires further testing.

I'm also working on a Tensorflow Kotlin idiomatic wrapper. Tensorflow is porting gradients to the C++ core so they are becoming available on other languages, making it posible to define the computation graph with a DSL and execute the training all in Kotlin. I'll be sharing that soon as well.

@altavir I would like that a lot. I may not include some of the machine learning stuff in 2.0 at first, other than the naive Bayes classifier. But I do need to port all the core functionality that uses statistics distributions.

I realize kotlin-statistics has

xxxBy

methods, but it'd be handy to have extensions that'd work on the result of kotlin's

groupBy

. For example, it'd allow solving the "find sums by name length" example like so:

val sumsByLengths = sequence
                .groupBy { it.name.length }
                .sumBy { it.value }

@dalexander without digging into it, I highly doubt it. So you want JVM implementation to not use my own interface? That's ok.

@altavir it's been awhile since I've used Spark with Kotlin. I've been out if the loop. Did you look at Sparklin and who is actively developing that?

My poster from ongoing ACAT-2019 conference

Had to work on Julia for a bit with my student. For some time I was wondering if Kotlin is the right choice for scientific applications and whether we should consider Julia. Well, now I am sure. Julia maybe good as an idea, but on-demand compilation, dynamic typing and almost no tooling at all - it is terrible.

I'm bringing in k-math for a personal project I'm messing with.

Python at least has PyCharm and good plugin for VSCode.

That it does... but I still find PyCharm can only do so much in its meta analysis of the code.

Guys, I hope you will like it. Here is the brand new plotly wrapper in pure kotlin+multiplatform (multiplatform is not yet fully capable): https://github.com/mipt-npm/plotly.kt It supports both static html export and dynamic server which handles all updates in the model. The documentation is quite thin for now. I am too tired today to expand it, but we will update it as soon as we can. We plan to add full dash functionality in future.

Been messing around with kmath. Really cool, I like where this is going.

@altavir A lot of JVM-based linear algebra libraries like ojAlgo don't have

pow()

functions. Not sure about Numpy. I don't know if it's inherently clear whether that a "dot pow" either, or just an element-wise pow() operation.