ChemCompute is free for classes, and you don't need permission to use it. It does help us
to know you're coming, so please email us with your
course and roughly how many students. That way we can make sure there's capacity on lab
day.
We are conducting a research study to learn about how faculty use ChemCompute and their
needs for features, accessibility, or training. You are invited to complete this short,
5-minute survey, even if you have not used ChemCompute before. You may skip questions or
decline without affecting your access to ChemCompute or services. No incentive is offered.
Read the study information and decide whether to participate:
Take the faculty survey
- Have students register with their school email, or sign in with their school
("Login Using Your Institution Email"). Personal addresses (Gmail and so on) can't
submit jobs. Addresses ending in .edu or in .k12.<state>.us work as they are. If
your school uses another domain (common for K-12 districts on .org, .net or .com),
contact us before the lab so we can add it. See
Which email addresses work?
- Password accounts must verify their email before they can submit. Build five minutes
for this into the first session, and tell students to check their spam folder.
- Run the lab yourself first if you can. On
.edu addresses your students get the same
researcher role as you, so
they can do whatever you can: up to 32 cores and 48 hours on the large clusters, or
1 core and 6 hours on Jetstream2. K-12 district addresses get the student role, which
allows up to 8 cores and can't upload Psi4 or PySCF scripts or prebuilt NAMD systems.
Plan K-12 labs around those limits.
- Link to ChemCompute from Canvas or another LMS so that it opens in a new tab. In a
Canvas module, tick Load in a new tab on the External URL item. A page shown inside
the LMS's own frame can't log anyone in, because browsers don't keep a login for a site
displayed inside another site. Students who land there see a notice with a button that
opens ChemCompute in its own tab.
A project is a shared workspace: its members can open each other's jobs filed in it.
This is the way to see all your students' work in one place.
- Go to Profile and Projects (
/profile/) and press + New Project. Name it
after your course. You can nest projects, e.g. one per lab section under a course
project.
- Open workspace, then in Class join code press Create join code. You can
limit it to a number of days or uses.
- Post the 8-character code on your syllabus or a slide.
- Students enter it under Join a project with a code provided by your instructor on
their Profile page.
- Students choose the project in the Project menu on the submission form, or move
jobs into it later from their Dashboard (Move to project).
In the workspace you'll see:
- Members, with roles. owner and admin can manage members; member is
read-only.
- Jobs: every job filed in the project, across all members.
- Usage: jobs, CPU-hours, run time and disk per member and in total.
- Export project records: a zip with each calculation's record and input file, a
combined methods document and the usage summary. Handy for a poster's methods section.
Members can only view each other's jobs. Only a job's owner can cancel or resubmit it,
and job notes stay private to their owner.
Students can also share every job with one person (for example a TA) with Share all your
jobs with on their Profile page. Any job can be made public from the Dashboard.
¶ Guided experiments, background and prelab
Some packages come with guided experiments. These add Background Info and Prelab
tabs, and an Instructions panel beside the submission and results pages that walks
students through each task (Prev / Next, Download Instructions). Students can hide
the panel with Hide Instructions.
Each job's results page also has Export methods record. It downloads a zip with the
calculation record, a methods write-up and the input file, written for a lab report.
- Put your materials in a GitHub repository and use the nbgitpuller link generator on
Jupyter → Clone a github repo to give students a one-click link. Re-running the link
pulls your updates without overwriting their work.
- Tell students to keep their work in the
work folder. Only that folder is kept between
sessions.
- If you need software that isn't installed, ask us early. See
Jupyter notebooks.
- Jobs queue on shared supercomputers. Jetstream2 jobs usually start quickly; the large
clusters can queue for 5–15 minutes or more. The cluster picker shows
current queue estimates. See Clusters, cores and time limits.
- Most failed student jobs come from a handful of mistakes, above all the wrong
charge or multiplicity. Common job errors is worth a skim before the
lab; consider linking it on your handout.
- Ask students to note their job number. It's the fastest way for you or us to look
at a problem.
If ChemCompute contributes to a publication or poster, please cite it; see
How to cite.