NSF program

DMREF reporting, and where the year actually goes

Designing Materials to Revolutionize and Engineer our Future

DMREF awards are smaller and sharper than a MRSEC. The reporting obligations barely shrink to match, and there is usually nobody whose job it is to meet them.

What the report leans on

The data-driven materials design story, and the collaboration between experimentalists and theorists.

The sections that cost the most time

Research products beyond papers: Datasets, software and codes are first-class outputs in a DMREF and are reported as such. They rarely have DOIs, owners or version records until somebody goes looking.
Experiment and theory collaboration: The design loop is the whole premise of the program, so the report has to show it closing. That is a narrative built from evidence spread across two groups.
Personnel on a small team: The same role and demographic breakdowns as a large center, on a team small enough that one graduate student changes a percentage.
Leveraged funding attribution: Tracing a new award back to the collaboration or seed that produced it, without a system that recorded the link when it happened.

In more detail

What a DMREF is

DMREF is NSF's materials-by-design program, part of the Materials Genome Initiative. A DMREF award is smaller and more tightly scoped than a MRSEC: a focused team pairing experimentalists with theorists and computational groups, iterating on a design loop rather than running a facility.

What the annual report asks for

Broadly the same sections as any NSF center award. The narrative emphasis is different: reviewers want to see the design loop actually closing, prediction informing experiment and experiment correcting prediction, rather than two workstreams that share a cover page.

Research products matter more here than the section title suggests. Datasets, software, codes and workflows are outputs a DMREF is expected to produce and report, alongside publications.

Where the time actually goes

Into being small.

A MRSEC has a coordinator whose job includes the report. A DMREF often has a PI, a co-PI and a graduate student who drew the short straw. The reporting obligations do not scale down proportionally, so the same eleven sections land on people with no administrative support and no system.

The specific trap is research products. Papers have DOIs and arrive in a bibliography. A dataset on a group server and a repository with four contributors do not, and by the time somebody asks what shipped this year, the answer has to be reassembled from memory and commit logs.

The record the report comes out of

A DMREF carries a center's obligations on a small team's staffing. Anything that needs a person to run it does not get run.

We run the NSF program sites for MRSEC, PREM, NeuroNex and the Materials Innovation Platforms, close to a decade inside the Division of Materials Research, and the same platform runs for individual awards.

The part that pays for itself fastest on a small team is publications. DOIs import through CrossRef, the funding acknowledgment is matched against the award number automatically, and anything that cannot be verified is flagged rather than quietly counted. People carry roles with start and end dates, so the reporting-period question has an answer without a reconstruction.

Small teams do not need more process. They need the record to be right without anyone maintaining it.

Other programs

Reporting season should be a review, not a reconstruction.

Thirty minutes, no deck. Tell us which sections cost you the most last year and we will tell you whether they are worth automating.

Start a conversation