Six Reasons to Learn Julia in 2024
This year, the Exponent together with the development environment Engee became information partners of the free Julia Summer School. This is a unique event for Russia, because no one has ever conducted online courses on this modern and very promising programming language. We are confident that in today's political and economic realities, this language has a great future, especially in the scientific and engineering environments, and we invite everyone to take part in this important event.
And to whet your interest, we have prepared a short article (based on this materialthanks to the authors, and added many theses based on our experience with Julia) describing the main reasons why you should urgently get interested in Julia and apply for the Summer School. Enjoy reading!
The Julia language is a new star in the programming universe, recognized for its versatility. It is fast, has a convenient syntax, and is great for scientific computing in a variety of fields, from data analysis to calculating the strength of buildings.
Julia recently made her debut in TIOBE Index – a ranking of programming language popularity based on search engine results, discussion forums, and other resources. In this article, we’ll tell you six reasons to add a Julia certificate to your resume, give a quick comparison of Julia with Python and MATLAB, and speculate about the future of this new language.
Julia's rise in popularity can be seen in this graph:
Source: https://www.tiobe.com/tiobe-index/julia/
3 Reasons Julia Is Popular
Combines the speed of C execution and the dynamic typing of MATLAB/Python. Reason: Julia is a compiled programming language, unlike MATLAB and Python.
Allows you to bridge the gap between high-level and low-level programming with a simple syntax similar to Python and MATLAB.
Implements modern programming technologies. For example, the concept of function overloading has been expanded to multimethodsand the code is processed as a data structurewhich makes metaprogramming a lot easier. Oh yeah, and it allows you to use any unicode characters in your code, not just ASCII…
In addition, the community has developed libraries for any area of calculations, from neural networks to quantum physics, and the package manager of this language is easy to use, like conda and pip.
Julia is speed
One of Julia's greatest strengths is its speed. Designed to be as fast as languages like C and Fortran, Julia provides the high-performance computing capabilities that are so important in today's data-intensive world.
Julia's high speed is primarily due to its Just-In-Time (JIT) compiler. This allows Julia to compile efficient native code, making it a good choice for running complex algorithms on real hardware.
In the context of machine learning and deep learning, Julia's speed is especially beneficial. It allows you to quickly process large data sets and efficiently perform computationally heavy tasks, thereby accelerating the pace of AI development.
Source: An article comparing the speed of different languages: https://modelsandrisk.org/appendix/speed_2022/
Julia is user-friendly
Julia syntax is friendly to Python and MATLAB users, meaning it is convenient to program in for those who are already familiar with these languages. Read on – if you know Python/MATLAB, then you practically know Julia, you just need to dive in. Also, if you compare the syntax directly with MATLAB, then significant differences can be counted on the fingers, they are described here.
In addition, the high-level syntax makes it easy to express complex algorithms, making the language more accessible and giving it significant expressive power.
Julia's syntax is intuitive and easy to learn. Variables can be assigned without declaring their type, the language supports all common algorithmic structures (loops, conditions), all common data structures (multidimensional matrices, dictionaries), and there are plenty of free libraries for complex data types.
Julia has a strong package ecosystem
Julia provides a powerful ecosystem of packages that extend its functionality. The Flux, MLJ, and Knet packages provide powerful tools for machine learning and deep learning.
> At Exponent we use Julia for heavy engineering tasks, and we'd like to highlight a few of these packages to make it easier for everyone to find references and start solving problems in control systems, signal processing, etc.
DataFrames.jl is a Julia package that works with data in a similar way to Python's pandas. It provides a set of tools for data management and statistical analysis, making it a valuable companion for data scientists.
Flux.jl And Lux.jl – are quite powerful packages for deep learning. They are written entirely in Julia, which allows you to create custom layers and models directly in Julia.
Plastic bag JuliaControl – a variety of libraries for developing automatic control systems, which contain:
ControlSystems.jl – a functional package for automatic control systems. procedures for the analysis and synthesis of control systems (primarily linear);
RobustAndOptimalControl.jl contains more sophisticated functions for LCG design, deep analysis and synthesis, fuzzy modeling, as well as named systems and an interface for interacting with DescriptorSystems.jl;
SymbolicControlSystems.jl contains basic capabilities for generating C code for linear systems;
ControlSystemIdentification.jl— is a toolkit for identifying linear systems with constant parameters (LSP) using time or frequency domain data. This package allows data-based evaluation of state space models, transfer function models, and Kalman filters that can be used to calculate control systems;
DiscretePIDs.jl Contains a reference implementation of a discrete PID controller in Julia, including weighting setpoints, integrator windup limiting, differential filtering, and soft transfer.
JuliaImages – a high-performance set of open-source libraries for working with images, which also includes many packages for loading, binarizing, and transforming images with any number of channels.
DSP.jl – is a package that provides a wide range of digital signal processing (DSP) capabilities, making it suitable for use in a variety of signal analysis applications.
Polynomials.jl is a package designed to work with polynomials. It offers functions such as arithmetic operations, integration, differentiation, coefficient estimation, root finding, and data approximation.
Jump.jl is a specialized modeling language for mathematical optimization integrated into the Julia programming language. It allows users to formulate and solve a variety of optimization problems, including linear, mixed integer, and nonlinear programming.
We have described the libraries from our own point of view, but they can be applied in many other areas at once. As for quality plug-in libraries, in fact, their hundreds for completely different tasks.
Julia in Scientific Computing and Modeling
The largest modeling and simulation package in the Julia ecosystem is called SciML. It combines powerful tools for scientific computing (Sci) and machine learning (ML) – everything you need to work with mathematical models. Let's see what libraries are included in this project.
DifferentialEquations.jl – a package for numerical solution of differential equations, also written in Julia. DifferentialEquations allows you to solve ordinary differential equations, partial differential equations, stochastic differential equations, and others.
ModelingToolkit.jl (MTK) is a whole language (a superstructure over Julia) for high-performance symbolic-numerical computations. What does it allow you to do? Solve algorithmic and acausal problems defined using symbolic description, in general – easily extensible models. The MTK toolkit allows you to analyze them, transform them – for example, by reducing the order – and pass them to DifferentialEquations for solution.
SciML includes a lot more modeling tools (finance, astrophysics, agents…), analytical toolsand, of course, tools for machine learning (everything from differentiable surrogate models to neural networks). But we remember that machine learning also works with models, it just makes them from data.
Thus, Julia, having all these libraries in its arsenal, provides a sea of possibilities for building complex models in engineering and science-intensive areas.
Julia could play a key role in the IT transformation of science
Julia's high-level syntax and efficient JIT compiler allow Julia to play an important role in rapidly growing areas of science, such as quantum computing.
In one of recent articles Julia was proposed as a tool for developing algorithms for quantum computers. The paper presented Yao.jl, an extensible and efficient framework for developing quantum algorithms, highlighting the potential of Julia in the field of quantum computing.
Yao.jl enables quantum programmers to develop and test quantum algorithms with features such as GPU support and an automatic differentiation engine. The package promises state-of-the-art performance, further highlighting Julia’s potential in quantum computing.
Julia has great potential for AI innovation
Julia's capabilities are well-suited to the requirements of AI development. It has the computing power needed to process large data sets, the flexibility to implement complex AI algorithms, and the speed to deliver results in real time.
Packages like Flux.jl, SciML, and Knet expand the scope for innovation in your field with Julia. These packages provide powerful tools for machine learning and deep learning, helping you pave the way to cutting-edge AI solutions.
Julia is a developed community, including in Russia
Julia has a strong community that offers support and resources for both new and experienced users. From online forums to documentation and tutorials from companies and universities, you'll find helpful people working to promote the language, making it a valuable resource for learning and mastering Julia.
When diving into development, sometimes you'll find yourself in areas where there's no reference or documentation. Fortunately, when working with Julia, we can often rely on the code and knowledge of existing libraries and projects, as there are plenty of open-source repositories written in the language, including the aforementioned application packages. This collaborative approach in turn leads to the emergence of numerous packages and tools that extend the functionality and usability of Julia.
Julia or Python: How to Choose?
There will always be tasks where other programming languages have their advantages or have simply become established as de facto leaders. Maybe because no one has tried to work differently. Julia is a reliable and evolving language, but there are plenty of tasks with specific requirements where Python works better. But, for example, for developing streaming data processing systems or communication systems, Julia is already preferable due to its speed advantages.
Python is a general-purpose language, easy to learn, with a large ecosystem and a large community. But if your project involves heavy numerical and scientific computations, requires high performance, and you are happy with a small but specialized community, Julia is a better choice.
In many cases, the optimal solution may be to use both languages in tandem. Performance-critical components can be written in Julia, and Python can be used for other aspects of the project, since this syntax is supported directly from Julia thanks to the PyCall and MATLAB.jl libraries, which allow you to insert code in the above-mentioned languages and fully interact with them directly from Julia.
Both Python and Julia are powerful languages, each excelling in different areas. But it’s worth noting that knowledge-intensive tasks typically rely on well-written libraries in other languages (anything from C to Julia) that are used via a wrapper, and this aspect works equally well in Julia and Python. Your choice should be driven by your specific project needs and goals. Remember that the goal is not to find the best language overall, but to choose the one that’s most suitable for your specific project or use case.
The Future of Julia
Given Julia’s unique features and capabilities, the demand for Julia programmers is likely to grow in the coming years. Therefore, learning Julia now can open up lucrative and exciting opportunities in the future. In terms of the job market for programmers, due to various circumstances, we may see a shift to Julia by MATLAB specialists, leading to an explosive growth in the community and demand in the market.
Being an open-source language, Julia is constantly evolving, regularly adding new features and improvements. And keeping up with the latest developments is interesting for both students and teachers. For some, it is important that Julia constantly implements the latest requirements of the programming world, for others – that new courses will always be in demand.
Whether you’re an experienced programmer or a beginner, Julia is a language worth learning. Its combination of speed, ease of use, expressive power, and community-recognized high potential, even in emerging areas like quantum computing, make it an attractive choice for those looking to improve their programming skills or drive IT transformation in new application areas.
Bonus Reason: Julia is the best language to handle with ChatGPT
According to one study, “ChatGPT was best at generating code in Julia. For a fairly large set of tasks, the generated code in Julia was successful 81.5% of the time. The worst was in C++, where only 7.3% of the code was successfully executed. In particular, this model generally worked better on high-level dynamically typed languages (Javascript, Julia, Perl, Python, R, Ruby, Smalltalk) than on low-level statically typed languages (C, C++, Go).”
That's right, of all the languages, “ChatGPT is the best at writing Julia code.” What's the point? 🙂
You can read more about this study at the link:
FAQ
How did Julia come about?
Julia was developed at MIT and first appeared in 2012, but only began to gain traction with the stable release of Julia 1.0 in 2018.
Is Julia used often in general?
Julia is used in scientific computing and numerical modeling, high-performance computing (HPC), data analysis and visualization (Plots.jl), machine learning and deep learning (Flux.jl), and parallel and distributed computing.
Who or what is Julia named after?
According to the developers, it was not named after anyone.
