A month and a half ago I wrote that we had resurrected ECCE, the Extensible Computational Chemistry Environment that PNNL stopped developing years ago. Version 8 built and ran on a modern Debian, which was already more than anyone had managed for a long time. Version 9, which came out this week (9.0.1 at the time of writing), is a much bigger step, and since it changes how ECCE is installed and used I thought it deserved a post of its own.
ECCE is a graphical front end for computational chemistry: you build or import a structure, set up a calculation for one of several codes, launch it on your own computer or a cluster, and look at the results (orbitals, vibrations, spectra, charges, geometry steps) in the same program. It was always good at this, and what killed it was not the idea but the dependencies. Most of the work on 8 and 9 has gone into replacing what had rotted. As with 8, a large part of the code was written with the help of LLMs, with me testing and deciding what goes in.
What is new in 9
It runs on Windows and macOS, not just Linux. There is an installer for Windows 10 and 11, and an app for Macs with Apple silicon and Intel processors. The macOS version is still experimental.
No server is needed. ECCE used to require a data server and a message broker, which made sense at PNNL in 2005 and makes very little sense on a laptop in 2026. You can now keep your calculations in an ordinary folder. The central server is still there, because that is how it can be used it for teaching: students run the client and connect to one server.
It has its own quantum chemistry program. ECCE-QM is a small Hartree-Fock/DFT program built into ECCE, so you can run a calculation without installing anything else. It does energies, geometry optimisations, frequencies with IR and (on main, in the next release) Raman intensities, thermochemistry, orbitals and the usual population analyses. It is meant for small molecules and teaching, not for production work, and I have checked it against ORCA throughout.
The codes it drives are ORCA, Gaussian 16, NWChem, MOPAC, and experimentall, Quantum ESPRESSO and GROMACS. Jobs can run locally or through Slurm and the other common queueing systems, and ECCE shows the progress of the SCF, optimisation and frequency steps while they run.
Things I wanted for teaching: molecular orbital correlation diagrams (experimental; they can be wrong for larger molecules, so check them against the list of orbitals), geometry constraints and dummy atoms that are easier to set up, and a set of example calculations that are imported the first time you start ECCE, so there is something to look at straight away.
Installing it
Everything is on the releases page.
Debian 13 (the only distribution I test on; packages for Ubuntu 24.04, RHEL/Rocky 9 and Fedora are built but untested). Download the client and server packages and install both:
sudo apt install ./ecce-client_9.0.1_amd64.deb ./ecce-server_9.0.1_amd64.deb ecce
The ./ matters, since it tells apt that these are local files, and apt then pulls in the dependencies itself. The first time you start ECCE it asks whether to keep your data on this computer or connect to a server, and if you pick a data server on your own computer it offers to create the account.
Windows: run ECCE-windows-9.0.1.msi. It installs for your own user, without administrator rights. Windows will show a SmartScreen warning because the program is not signed; choose More info and then Run anyway.
macOS: take ECCE-9.0.1-arm64.dmg if About This Mac says "Chip: Apple M..." and ECCE-9.0.1-x86_64.dmg if it says Intel. Getting the wrong one is easy, and one user already has. Drag ECCE to Applications. Since the app is not notarised by Apple, macOS will say that it cannot check it for malware; either run
xattr -dr com.apple.quarantine /Applications/ECCE.app
in a terminal, or open ECCE once, click Done, and then choose System Settings > Privacy & Security > Open Anyway.
There are two short videos showing a first calculation, one with ECCE-QM and one with ORCA.
What does not work yet
The Windows and macOS clients have had much less use than the Linux one, the MO diagrams are experimental, and GROMACS support is at an early stage. If something breaks, please open an issue on GitHub; starting ECCE with ecce --bug, reproducing the problem and attaching the archive it writes helps a lot.
