https://github.com/londogard/londogard-nlp-toolkit that the one, @Hampus Londögård, it is great. I haven't dived too deep into it, but could it be possible to make it compatible with tokenisation required in BERT?

This week on Friday (14.05.2021, 15:30, Moscow time) we will have presentation by Andrey Kislitsyn, Aleksandra Novikova, Roland Grinis (@Ролан) on they work with tensor API and implementation in KMath during spring internship. Title: Multiplatform library for tensors Abstract: Within the KMath library, we present a multiplatfom implementation of some of the basic functionality for tensors, as one would find in PyTorch's ATen library for example. In the talk, we shall discuss the algorithms we used, including broadcasting routines and numerical power methods for SVD. We will also touch upon applications to machine learning, with examples for OLS, PCA and NEURAL NETWORKS. Calendar link Multiplatform library for tensors Friday, May 14 · 15:30 – 16:30 Google Meet joining info Video call link: https://meet.google.com/vvn-dzkz-umb Or dial: ‪(US) +1 346-702-3560‬ PIN: ‪487 432 197‬#

@breandan I've sent a PR to Kotlingrad https://github.com/breandan/kotlingrad/pull/22

Such code doesn't fail, but BigInt breaks the convention by

if (other is BigInt) compareTo(other) == 0 else error("Can't compare KBigInteger to a different type")

Multiplication of BigInt and Int.MIN_VALUE doesn't work fine:

public operator fun times(other: Int): BigInt = if (other > 0)
        this * kotlin.math.abs(other).toUInt()
    else
        -this * kotlin.math.abs(other).toUInt()

Incorrect parsing to BigInt of "" and "0x". Both are expected to return null, but first fails with an exception and second returns 0.

@altavir Let the discussion about the core change be here. I suggested inventing fast power method for Rings.

Jupyter notebooks by Andrei Kislitsin demonstrating examples of using tensors with kmath from the talk are available here: https://github.com/AndreiKingsley/kmath-tensors-example-jupyters

What do you think about the following math-related kotlin lib: https://plus.qconferences.com/plus2021/presentation/differentiable-programming-kotlin?

Having visualization capabilities is a game-changer for developing mathematic software. This example shows the difference between cubic spline interpolation and original function (KMath, fully multiplatform as always).

A nice feature would be a nice way to build defined matrices. Like in python ([[1,2,3],[4,5,6],[7,8,9]]) or better like matlab by omitting the comma and using line returns. There is nice syntax for HTML and SVG... I don't understand why we wouldn't have something nice and clean also...

(Something like Mat { Vec {1,2,3}; Vec {4,5,6}; Vec {7,8,9}} could be close enough)

@Zhelenskiy concerning equality, it is not an obvious thing. First, equality should be symmetrical, it means you can't override it without thinking thirst. Also, there is a problem with performace. For example, if you take larget matrices, the content equality could be greatly time-consuming. So overriding equals could lead to improper use. For now I decided not to use equals for comparison for structures. One of the reason is that we in some cases use inline classes that lack identity.

Hi team, Glad I found this channel. Very basic question though I'd be happy if someone could help please. What is/are number/numbers? To visualise it , I am finding it difficult and the current symbols does not provide the meaning. Also if number are symbols, then symbols come from maths (arch, line etc....) Which is the superset or holds the number.

Does

multik

support sparse arrays?

Looking a little at Kmath, how am I supposed to use the

LinearSpace

like

EjmlLinearSpaceFSCC

. More exactly the extension-functions like

Matrix<Float>.toEjml()

I've tried to use the

with (EjmlLinearSpaceFSCC) { code... }

syntax but it doesn't work. Not sure how else I'd get those to work?

The latest PR brings to

KMath

finalized optimization and interpolation API as well as unique optimization technique called quasi-optimal weights method (https://arxiv.org/abs/physics/0604127). https://github.com/mipt-npm/kmath/pull/404. This feature finalizes the feature set for 0.3.0.

I was benchmarking some of my own kotlin code vs BigInteger to test performance. In doing so I eventually copied the BigInteger.java source to my own project so I could edit some things. Without any changes that was running about 2-3x slower at least for multiplication (same speed as my kotlin algorithms). Is there any explanation on why there is a speed difference using the library directly vs source code I imported from it?

message has been deleted

@Pavel Gorgulov this seems to be wrong if I remember correctly:

@Pavel Gorgulov I am making a KMath adapter for Mutik (so one can use Multik internals with KMath API). There are some questions.

Finally made Multik wrapper work. Performance for linear add operations is of the same order for KMath-specialized buffer, Viktor and Multik (about 10-20% difference which is not relevant). It is much faster for @Ролан's tensor in-place operations (like 300%). I guess that memory allocation is the most expensive thing for simple operations. In KMath it is possible to do memory buffer pooling inside fixed size algebra context. I wonder if it makes sense to do.

Interesting library containing features including geometry functions like clothoid, fresnel etc https://github.com/tum-gis/rtron

I have an idea (and not only as IDE). There are a lot of scientific and mathematic libraries in Kotlin so it would be nice to have a semi-regular seminar on those libraries, where authors present what they are doing. We can arrange the virtual meetings under the JetBrains research. Please write to me if you want to present something. cc @roman.belov

I was looking for example to try out some of new features in Multik e.g. https://blog.jetbrains.com/kotlin/2021/10/multik-0-1-is-out/#Solving-linear-systems ....

How do we round only if the number is 10.0 Ex 10.0 = 10 10.3 = 10.3 10.7 = 10. 7 Is there any easier way

How to round double input to two decimal places? Given the output is Decimal something like

toFixed(2)

in JS

Some interesting task for those who wants to implement some low-level maths: https://twitter.com/noraltavir/status/1457991689897975812?s=20