JuliaPro is all set for a huge makeover post Julia 1.0 release, this talk is all about revealing the new features that will be included in the next generation of JuliaPro.
Thanks to Julia’s speed and expressiveness, our very small team (1 developer!) was quickly able to create an ecosystem of Julia libraries for state-of-the-art automatic analysis of drug mixtures using nuclear magnetic resonance (NMR) spectroscopy data.
Julia is my scientific computing programming language of choice to implement my mathematical model of Burkitt lymphoma, a highly aggressive disease plaguing children in Africa. This talk will appeal to people interested in solving Mathematical Biology applications with Julia.
Fugro
Fugro Roames provides automated extraction of geointelligence at scale. I’ll discuss how we are using Julia in our machine learning pipeline to identify buildings, roads, trees, and other objects in unstructured point cloud data, and how we deliver billions of points without human intervention.
Sorbonne Université
MIToS is a package developed for the analysis of protein sequence and structure. It allows fast and flexible calculation of conservation and coevolution scores and helps to analyze them. It has been used in a large dataset to understand evolutionary signals due to protein structures.
Bayestesting.jl is a Julia package that implements a new Bayesian hypothesis testing procedure that does not suffer from the problems inherent in both the standard Bayesian and frequentist approaches, and is easy to use in practice. Of interest to anyone who does any statistical hypothesis testing.
In this talk I first present the hurdles and challenges associated with teaching a course with a large cohort using Julia while still in development. I then move on to present a variety of unique perspectives on the presentation of elementary statistical concepts via short and concise code snippets.
Eindhoven University of Technology
Scientific modeling concerns a continual search for better models for given data sets. This process can be elegantly captured in a Bayesian inference framework. ForneyLab enables largely automated scientific design loops by deriving fast, analytic algorithms for approximate Bayesian inference.
Some people swear by them. Other, at them. But every language has one. Python has SQLAlchemy, Java has JOOQ, .NET has Entity, Ruby has ActiveRecord. Now Julia has SearchLight. The SearchLight ORM provides a powerful DSL which makes working with SQL databases more productive, secure and fun.
Multi-threading in Julia is an excellent feature to speed up Monte Carlo simulations, e.g. for Asian option pricing. However, if you are not careful how you generate pseudo random numbers you can get wrong results or have your code run slowly. I discuss how one can avoid both problems.
U of Copenhagen
Ecological analysis draws upon many different tools - geographic, phylogenetic, bioinformatic and simulation packages and a wide range of statistics. Most ecologists use R, but its package ecosystem is severely fragmented. EcoJulia is a framework to bring cohesive ecological data analysis to Julia.
Epiq Solutions
Julia has great potential for signal processing, but it’s DSP.jl package is missing a fundamental filter design algorithm: “remez”, also known as the Parks-McClellan algorithm. I’m going to talk about the algorithm, review efforts to implement it in Julia, and compare it with what’s in Scipy.
University of Cologne
I will share my experience on how Julia can improve numerical physics research. This will provide evidence for the claim that Julia can replace Fortran/C++ as workhorses in science. Also I’ll introduce MonteCarlo.jl, a new package for conducting (quantum) Monte Carlo simulations of physical systems.
Domo
At Domo we see a LOT of data. Like, Fortune-500-sized automated pipelines of business-critical data kind of data. And now we’re turning to Julia to get smart about all that data. While deploying pre-1.0 may sound risky, Domo is no stranger to blazing trails in search of the right tool for the job.
Ministry of Social Development, Mexico
Political data is widely available in the internet but non-informative at all. I use Julia capabilities to extract the information from the Mexican Senate, transform it to a temporal network and get insights from the dynamics of the system.
Julia’s basic Logging package is sufficient for console logging but lacks the features necessary for monitoring large, multi-component systems and cloud applications. This talk will attempt to convince you to use Memento instead and demonstrate its strengths.
Warsaw School of Economics
Have you ever wondered what is faster - a supercomputer or a computational cluster in the AWS cloud? Do you want to know what is the easier option to run your massively parallel Julia program? In this presentation you will see how a massively parallel Julia scales with the cluster size increase.
Julia Computing
Harness your superpower for good and advocate for public policy issues that are important to you. I’ll tell the story about how I used a cell phone video and a Julia notebook to become part of a local movement for improved bike safety in the city of Pittsburgh.
The National Institute of Engineering, Mysore
Minecraft is arguably one of the most popular video games. The sandbox game is successful because it promotes building and creating from imagination. The PiCraft package allows manipulation of the Minecraft world. Programming in Julia we build amazing things like Mobius strips and Aztec Temples.
Julia’s JIT-compilation needs to run on each restart, and the compilation delay can slow development of large projects. I will describe a package, Revise.jl, that allows you to do more testing and debugging in a single Julia session.
Sailing craft experience a range of environmental conditions in their voyages across the seas. I show how Julia can be used to model the weather, compare weather scenarios and optimise the route whilst avoiding the structural failure of the craft and to thus reduce cost and crew injury.
JuliaDBMeta’s macros provide a simple syntax to select, filter, map, group and transform JuliaDB data tables for in memory or out-of-core processing. Plots based visualizations can be incorporated in the pipeline. InteractBase provides a graphical interface to define and compose these opera...
The University of Manchester
Modern networks often store the relationship between entities with time stamps. It is difficult to model and study the evolving nature of such network using traditional graph software package. We present EvolvingGraphs.jl, a Julia software package for analysing time-dependent networks.
Parameter estimation is the problem of accurately determining the parameters of a dynamic model. I plan to introduce DiffEqBayes.jl, a package for solving these problems using bayesian techniques.
Big data mass in fields like Bioengineering needs fast computations and simplified operations on complex geometric models. LARLIB.jl is a library for efficient solid modeling operations that works on non-manifold cases with a compact representation that permits fast computations and operations.
Kevala Analytics
For decades, conservationists have captured photographs of whales and their flukes (tails) in the open water. Can these images be used to accurately identify whale species? In this talk, see one implementation of an image recognition tool using Julia that may aid global whale conservation efforts.
Julia Computing
Have you ever wondered why 0.1 > 1//10, 1*pi < pi or 10^16+1 > 1e16+1? I will explain how equalities and inequalities in Julia work across different numeric types.
Research and development pulls together disparate tools and creates seamless product. Julia is an integrated R&D platform: it can command an instrument, e.g. an android phone, call existing C libraries, perform complex scripting and analysis. But best of all, it can unit-test the whole ordeal.
In the proposed talk an efficient root finding algorithm is presented, which can determine whole high-dimensional submanifolds (points, curves, surfaces…) of the roots of implicit non-linear equation systems, even in cases, where the number of unknowns surpasses the number of equations.
Julia’s built-in numeric and datetime types are very good. Sometimes we need better. We introduce SaferIntegers, ArbNumerics and TimesDates for nanosecond resolution.
Etsy
Etsy relies heavily on experimentation to improve our decision-making process. In this talk, I will present how Etsy handles data peeking and how we use Julia to help us investigate and assess the problem on our platform.
Julia Computing, Inc.
Over the past year Julia Computing has released a new version of JuliaBox. It was designed to be not just a hosted notebook service but also to let users deploy, scale and share their julia code. This talk describes how we achieve this and other new features.
U of Toronto
Recent work in machine learning and deep reinforcement learning uses self-tuning optimization methods which utilize higher-order gradients. Higher-order automatic gradients are challenging to implement correctly, even in Tensorflow and PyTorch. I show how to do this using Flux.jl.
U of Cambridge
Our new reaction reader tool is an attempt to automate the boring parts of biochemical modelling (transcribing equations). The user can now spend more time actually analysing those models! Also makes your code prettier.
Gandalf Software, Inc.
String performance is important in many areas, such as parsing text formats such as JSON, CSV, or bioinformatics data, for NLP (natural language processing), and for interfacing with other languages/libraries that use other encodings than UTF-8. This talk will discuss the JuliaString “ecosystem”.
This talk will showcase what is possible with the JuliaGeo related packages, with the aim to get you started if you want to do geospatial analysis in Julia. Making full use of the strengths of Julia, examples are shown of that would be either too slow or too much work in other languages.
Cambridge University
Latexify.jl allows you to create and render LaTeX code from not only simple types, but also arrays and even systems of equations. In this talk, I will introduce what Latexify.jl can do, how Julia’s metaprogramming makes it possible and how the underlying philosophy can be leveraged for other things.
GTRI
The JuliaGraphs ecosystem has expanded this year, with new innovations in abstract representations and modularity. We discuss these improvements, highlighting the effects of changes in Julia 0.6 and 0.7 which affected the design of the JuliaGraphs ecosystem.
Deep neural networks can be highly useful. Nonetheless, some problems require structured programming elements such as variables, conditional execution, and loop clauses. We present a Julia framework that supports gradient based learning even when such programming elements are included in a model.
Julia Computing
Julia Computing and EkStep have launched an initiative to train 500k individuals in the field of AI. Come hear how we’re designing the courses and delivery.
My research group uses Julia to solve large-scale industrial inverse problems in geophysics. These are rich computational problems involving differential equations and optimization. I’ll explain how we’ve used Julia to make fast, modular and scalable production geophysical inversion code.
University of Edinburgh
Can a shiny new language like Julia easily be added to a mature codebase? Yes it can! Come hear about my experience writing Julia code for Madagascar, an open-source software suite for geophysics. Be prepared for repeated abuse of ccall, metaprogramming and pipelines, but little geophysics (phew!).
Optimal dispatch of pumped storage is an important mechanism when optimizing a portfolio of energy assets with regard to its exposure to the electricity market. We present a solution to this problem in Julia and compare its advantages to some existing application that we use in our everyday work.
You just created an awesome new Julia package – congratulations! How can you show it off to potential users (many of whom may not have used Julia yet)? Create an online, interactive demo!
MIT
Earth systems are simulated using numerical models of increasing resolution that play a key role in predicting climate. This talk presents a Julia framework that will help educators and researchers leverage these models via an intuitive and scalable representation of gridded Earth variables.
Lund U
Are you interested in high performance code that is also easy to use? This talk will showcase how to price exotic financial derivatives, without having to compromise between speed and code readability.
Domo
Client and server web interactions are at the heart of most modern applications, Julia and otherwise. Web APIs open doors for cross-language and service interoperability and HTTP.jl aims to provide a robust, modern foundation for Julia programs needing request or server capabilities.
U of Amsterdam
Volatility modeling lies at the heart of much of financial risk management. The workhorse model in this field is the GARCH model, along with its various extensions. The talk describes a package that the author is developing to bring these models to Julia.
In this talk I introduce a new approach to authoring highly polished interactive data visualizations on the web with Julia. Such visualizations historically been difficult to create without writing custom Javascript.
The need for speed: SyMBOL is a model that analyses bank failures. The core of SyMBOL is a Monte Carlo simulation with correlated random shocks and it was written in C. Given the parallel computing capabilities of Julia and a different setup, we were able to reduce the computation time by 50%.
IBM Dublin Research Lab
CombineML main feature is to provide a framework for seamless ensembling of existing machine learning implementations from scikitlearn, caret, and Julia. It supports the following ensembles: stack, voting, and best. It is a heterogeneous ensemble driven by a uniform machine learner API designed f...
Analyzing mechanical testing data can be tricky, especially for those conducting interdisciplinary research (e.g. biomechanics) who may not have in-house code. We are using Julia to develop an open source software package to make this process simpler and more reproducible.
Virtual Reality plays the leading role on the new media revolution with Light Field reconstruction as a common technique for content generation.High industrial interests make this technique hard to understand and implement, I will present a novel method to reconstruct the depth of objects in images.
Caltech/Lawrence Livermore National Lab
How can we make Julia more inclusive of and accessible to new users from all backgrounds and experience levels? We want to create fantastic teaching materials that don’t assume a ton of prior knowledge. I’ll share what we’ve done so far to make Julia more accessible and how you can help!
University of Copenhagen
People doing numerical computing are greedy. They want results now and accurately, and we have been ready to accept loss of readability and extensibility to get it. As a result, the elementary functions we take for granted in paper math are often coded in low-level languages. We can do better.
Universidad de Buenos Aires/CONICET and and
The objective of this talk is to introduce GSReg.jl, a new all-subset-regression package to perform High Performance Computing in econometrics using Julia. GSReg.jl runs 4 to 100 times faster than similar packages and comes with a simplified GUI to allow smooth-transitions for R and Stata users.
Domo
What’s new in the data framework package powering some of the most popular data packages in Julia? Come learn about advancements in flexibly typed schemas, querying functionality directly integrated with IO, and automatic data transfer parallelism.
King's College London
Atomistic simulation needs custom models, to approximate through hard O(N!) scaling of the underlying physical equations. Julia is a natural language to express such physical abstractions in. I will show how you can contribute to materials research with Julia and a fistful of 1960s PhysRevs.
Anyone who works in a technical field has heard something like this from family or friends: “Oh, I was never any good at maths at school.” My job is to set targeted, achievable mathematics homework so that no child ends up feeling this way. Let me tell you how I use Julia to do this.
The Trilinos library features modern iterative solvers for large linear systems. Using the Tpetra library, it can exploit hybrid parallelism and GPUs. We present an overview of the current status of the package that makes this functionality available to Julia.
Highly parallelizable black box combinatorial optimization algorithm that only relies on function evaluations, and which returns locally optimal solutions with high probability.
University College London
Julia’s high-level nature and speed helped me to implement an intricate algorithm for converting a set of high-dimensional data into a very compact representation, allowing me to efficiently solve real-world research problems.
With Joulia.jl we developed a package for electricity sector models in Julia, taking advantage of the possibilities to cover all steps of the modeling workflow from data pre-processing, over the algebraic modeling and solution up to the visualization of results in an open-source environment.
Parameters.jl provides a macro to decorate type-definitions which adds keyword constructors and default values. When working with instances it has macros to unpack their fields. I will demo its core features, the resulting cleaner code patterns, and the improvements to code maintainability.
We develop TIM, a computational engine for large-scale forecasting in energy industry covering load, gas, solar and wind. We share our lessons learned when trying to do AOT compilation of TIM and deploying the binary in the cloud. We cover Linux and Windows scenarios.
Julia Computing
Computational problems can often be represented as specific types of graphs known as DAGs. That allows an execution framework to execute parts in right order, schedule in parallel, and make optimal use of available resources. We present ideas on which we are building a scalable scheduler for Julia.
