Skip to content

FAQ

Q: How do I pronounce quacc?

A: Like the sound a duck makes: "quack!"

Q: What's the philosophy behind quacc?

A: This is a big question. I'll explain with some points that were on my mind when I made it:

  1. Everyone's computing needs are different, which means not everyone is going to want to use the same workflow engine. This has historically hindered the widespread adoption of prior computational materials science libraries and makes it difficult to share workflows with others. quacc supports several workflow engines out-of-the-box using largely the same syntax, and you can pick your favorite (or use none at all).

  2. We should leverage tools that are widely used in the community. Python packages like ASE and Pymatgen are staples in the computational materials science community. Rather than reinvent the wheel, quacc takes advantage of what is already out there and — crucially — makes them play nicely together.

  3. Developers should be able to focus on the science, not the workflow architecture. Many similar packages have the science closely intertwined with the workflow logic. This makes it difficult to write new workflows without learning the intricacies of the workflow stack. quacc solely relies on the use of function decorators that make it easy to transform typical functions into ones that are workflow-ready.

  4. An overeliance on classes and inheritance can be confusing at the expense of conciseneess. I love classes, but in a computational materials science library, too much class inheritance can make it difficult to track down what parameters are getting set where. Students are also less familiar with classes and may have a difficult time in understanding how they work. The recipes in quacc take a more direct and familiar functional programming approach.

  5. And last but not least, I'm opinionated. quacc represents choices that I think make the most sense for my group and me. Hopefully they make sense for you too.

Q: Which workflow management system should I use?

A: Good question! Check out the Workflow Engines Overview Guide for a comparison of several common options to see which might be best for you.

Q: Why is quacc better than InsertMyFavoriteCode?

A: Better is all in the eye of the beholder. If you are happy with the tools you are using, my goal is not to convince you to switch. That said, if you find that writing high-throughput computational workflows has a huge learning curve or simply takes too much time away from your science, then quacc may be for you.

Q: Do you have any tips for writing new workflows?

A: When developing new workflows, I strongly recommend writing an initial draft based on a low-cost ASE calculator like EMT or LJ, which will allow you to test on your local machine with fast turnaround. Also try to keep things modular. Trying to do too much in one function can make it difficult to debug.

Q: How is quacc different than just using the Atomic Simulation Environment (ASE)

A: quacc is designed to supercharge ASE for the efficient design of high-throughput workflows. Quacc makes it possible to easily design complex ASE workflows that can be run on heterogeneous compute environments with modern workflow managers, such as Covalent and Parsl. Quacc also makes it possible to combine ASE with the powerful Materials Project software stack, such as Custodian for on-the-fly error handling. For the database enthusiasts, quacc is interoperable with Maggma to store calculation results in a readily queryable format like MongoDB.

Q: How is quacc different than Atomate2?

A: Atomate2 is a fantastic code that can be used to run libraries of computational materials science workflows. Quacc was originally inspired by Atomate2 and can be thought of as a "sister code" to it. That said, there are several important differences.

  • In Atomate2, workflows are built around the Materials Project software stack and can be cumbersome to interface with the Atomic Simulation Environment (ASE). In contrast, quacc was built around ASE from the ground up while still making it possible to leverage much of the software underlying the Materials Project. Since quacc is built around ASE, it is also trivial to add new recipes based on any code that has an existing ASE Calculator even if it is not in quacc yet.

  • Atomate2 is built around Jobflow. Quacc is largely workflow engine-agnostic and has out-of-the-box support for numerous workflow codes, including but not limited to Jobflow.

  • Atomate2 is often used as a database-first approach to materials science workflows. Quacc is fully compatible with the use of databases but only emphasizes their use for experts, with the goal of reducing the barrier to getting started.

  • Atomate2 is developed and supported by the Materials Project team at Lawrence Berkeley National Laboratory. Quacc is developed and maintained by the Rosen Research Group at Princeton University.

Q: I want to contribute a new recipe to quacc, but it will require additional dependencies. Is that a problem?

A: Absolutely not a problem! Simply make your dependencies optional in the pyproject.toml file and use a monty.dev.requires decorator to ensure your recipes and tests only run if your necessary dependencies are installed. Refer to quacc.recipes.tblite.core and the [tblite] extras in the pyproject.toml file for an example.