AiiDA Team Meeting 2024-09-19

# AiiDA Team Meeting 2024-09-19 ###### tags: `team meetings` ###### time: 14:00 CET [TOC] ### Present * Julian * Ali * Miki * Edan * Alex * Kristjan * Xing * Jason * Giovanni * ### Catch-up round *Max. 3 minutes each* Julian * [LinkedIn post here](https://www.linkedin.com/posts/nccr-marvel_computational-marathon-matches-the-efficiency-activity-7242157122199662595-RLk8?utm_source=share&utm_medium=member_desktop) * Write on Discourse announcement for the new page * Will delete the [AiiDA team LinkedIn profile](https://www.linkedin.com/in/aiida-team-1190b5324/), only keep [AiiDA institutional page](https://www.linkedin.com/company/aiidateam/about/?viewAsMember=true); announcement [here](https://www.linkedin.com/posts/aiida-team-1190b5324_aiida-linkedin-activity-7242114031916843008-jXjY?utm_source=share&utm_medium=member_desktop) (actually, AiiDA institutional page was referenced by MARVEL hero run post, way to go!) * Two blog posts ready for publishing: * ["Simplifications to the installation of AiiDA since v2.0"](https://github.com/aiidateam/aiida-website/pull/69): `verdi presto` and `verdi profile setup` (including different storage backends). If anybody wants to give it a read, you can do so [here](https://aiida-website--69.org.readthedocs.build/news/posts/2024-09-20-simpler-installation.html). Otherwise, I'm planning to merge today and make announcement on LinkedIn tomorrow. * ["Improvements in the ways to get your data out of AiiDA"](https://github.com/aiidateam/aiida-website/pull/72): `verdi process dump` and new QB syntax. Scheduled for 2024-10-04 (biweekly posts on Fridays). @EB wanted to still give a review. * Further planning: https://hackmd.io/_Ic1IR6pQ6yZYnqZUflMjg (let's check later on) * [[WG #299] Add `node_value` & `get_node_value` to `TaskSocket`](https://github.com/aiidateam/aiida-workgraph/pull/299)] Avoids having to use `.value.value` constructs to get actual Python value from `Socket` that has AiiDA ORM type attached * [[#6565] Refactor: Add the `_prepare_yml` method to `AbstractCode`](https://github.com/aiidateam/aiida-core/pull/6565) Ali * `disk-objectstore` PR on progress bar was [merged](https://github.com/aiidateam/disk-objectstore/pull/171) * `aiida-core`: PR on progress bar at [verdi storage maintain](https://github.com/aiidateam/aiida-core/pull/6562) depends on: * `disk-objectstore-feedstock`: [PR to be merged](https://github.com/conda-forge/disk-objectstore-feedstock) * if somebody can have a look at [PR verdi group delete](https://github.com/aiidateam/aiida-core/pull/6556) * URGENT: google results. Xing: - As planned, we targeted a September release to fix the API. And AiiDA WorkGraph 0.4.0 is released last week. Project [WorkGraph 0.4](https://github.com/orgs/aiidateam/projects/9) - I updated the Project [WorkGrah 0.5](https://github.com/orgs/aiidateam/projects/10), which served as the plan for next half year. Miki: - Summary of the discussion (September 16) about `StructureData` is [here](https://github.com/aiidateam/aiida-atomistic/discussions/43) - Mainly working on how to store `StructureData` in the db (to avoid duplication of information - see the discussion) - Next meeting focused on `StructureData`: October 2, afternoon? Jason: - NumFOCUS collaboration meeting tomorow (Friday, at 10:30am CT (3:30pm UTC)). Topic: Project Funding Models. [zoom link](https://www.google.com/url?q=https://numfocus-org.zoom.us/j/5909875714?pwd%3Dvzg23zjPDhctwXv3gMyPWjk3AJ16Gt.1%26omn%3D83552316997&sa=D&source=calendar&ust=1727165374160353&usg=AOvVaw3JUvHBmXd8dvPXiDgUh_KX) password: 607129 - Send me the news of (is it published in AiiDA linkin, already?): 1. Hero run results -> [LinkedIn post here](https://www.linkedin.com/posts/nccr-marvel_computational-marathon-matches-the-efficiency-activity-7242157122199662595-RLk8?utm_source=share&utm_medium=member_desktop) 2. release summary of v2.6.2 - I know it is not proper, but I did a reimplemntation of disk-objectstore in Rust. Without any tricks on detail designs, the performance improved to ~2x (I did very detail benchmarks compare both loose/pack/losse2pack with/without compression, one different size of targeting files). It also reveal some abstractions on how the disk-objectstore can be extented and further improve the performance. Anyway, I'll discuss with Giovanni first. Kristjan * Cecam machine learning interatomic potentials (MLIP) workshop: * uk & france both are building similar MLIP-related infrastructures and both integrate AiiDA * e.g.: https://github.com/stfc/aiida-mlip * Gave a talk on recent AiiDA usability features, MC3D pbesol-v1, archive-OPTIMADE integration. Edan: - Testing out AiiDA Workgraph on an [Empa project](https://github.com/edan-bainglass/tb-hamiltonian) - Invited Davide TISI and Lorenzo Piersante (of EPFL) to give it a go on their projects - Met with reps from [Constructor tech](https://constructor.tech/) regarding possible integration of / collaboration with AiiDA - More to come down the road - Planning to dedicate the coding week to revamping the AiiDA website - Great if we can collect thoughts on the current site / wish list for new site - Will create a document somewhere Giovanni: - disk-objectstore feedstock: how does it work? Should I and Ali be assigned as maintainers? - workgraphs: good to have a plan for how often to release, what should be included and what not, etc. shared with the whole team, as we plan to make this a supported feature. Let's also plan some time to test the most recent version and collect feedback (does not need to be a coding day, maybe just distribute the work between Xing, Alex and Julian, and each interacts directly with ~2 people for a workflow they have, to have faster turnaround feedback) - I wanted to have >99% coverage in disk-objecstore - I know it's annoying, but the work it's already done, there are not major changes planned, and it's a matter of copying how tests are done elsewhere. and most importantly, this is a critical part of AiiDA, that should not break otherwise users lose data. If we go to 95%, then the next contributions will have no tests, and then we'll complain that 95% is too much. - complex workgraph: maybe reproduce what we did with Sebastiaan on testing RPN (reverse polish notation), it's still one of the tests. - [XW] Thanks, I will try. Alex: - aiida-tutorials: [#474](https://github.com/aiidateam/aiida-tutorials/pull/474) Integrating aiida-workgraph tutorials using qe (and fixed CI) - aiida-workgraph: rework various doc examples: - [#277](https://github.com/aiidateam/aiida-workgraph/pull/277) Nested workflow - [#288](https://github.com/aiidateam/aiida-workgraph/pull/288) Parallel workflow - [#289](https://github.com/aiidateam/aiida-workgraph/pull/277) Caching with Fibonacci - [#287](https://github.com/aiidateam/aiida-workgraph/pull/287) Aggregation - aiida-core: [#6563](https://github.com/aiidateam/aiida-core/pull/6563) Allows specifiy ssh key for tests. Not sure if useful ### Specific Agenda Items Julian * Blog post planning: https://hackmd.io/_Ic1IR6pQ6yZYnqZUflMjg volunteers welcome!