Department of Energy FY 2019 Request: Good for Computing but with Some Asterisks


Last month, President Trump released his budget request for Fiscal Year 2019. As we have done in years past, the CRA Policy Blog will be doing a series of posts on the assorted agency budgets that are important to the computing research community. In this post we highlight the Department of Energy (DOE).

The two key parts of DOE for the computing community are the Office of Science (SC), home of most of the agency’s basic research support, and ARPA-E, or the Advanced Research Projects Agency-Energy. For SC, the President’s FY 2019 is flat-funded at FY17’s number of $5.39 billion (remember, we are still waiting on FY18 to be settled by Congress, which has until March 23 to do so). While that doesn’t look good, the Advanced Scientific Computing Research program, which is within the Office of Science, and where most of the computing research at the agency is located, would receive a significant increase over FY17 levels; the program would be funded at $899 million, an increase of $252 million or 39 percent. As for ARPA-E, it would once again be zeroed out, as it was in the President’s budget request last year. Let’s get into the details.

Within ASCR, the program is divided into three parts: Mathematical, Computational, & Computer Sciences Research; High Performance Computing & Networking Facilities; and the Exascale Computing Project (ECP). Our regular readers will remember that in last year’s request, the Administration slated DOE to make one exascale system operational by 2021 and a second one a year later, but those plans had hinted at cannibalizing the research budget to accomplish it.

That goal is still in place but there is now at least some emphasis on the research side as well. Mathematical, Computational, & Computer Sciences Research would receive a plus up of $33 million, going from $114 million in FY17 to $147 million in FY19 (a 29 percent increase). But the majority of the increase would go to High Performance Computing & Networking Facilities (up $150 million, going from $370 million in FY17 to $520 million in FY19, a 41 percent increase) and ECP (up $69 million, going from $164 million in FY17 to $233 million in FY19, an increase of 42 percent). Both of these accounts are on the computer construction and facilities side of ASCR, rather than the research side, with large increases going to the Leadership Computing Facilities (LCF) subaccount to prepare for an exascale system.

Within the budget justification document, DOE says the department is funding computing research to, “reassert(ing) U.S. leadership in this critical area.” The justification goes on to state that, “this Request also increases support for ASCR’s fundamental research in Applied Mathematics and Computational Partnerships with a focus on advanced technologies such as quantum information science, including quantum computing and networking, and on new methods, software and tools for scientific machine learning for discovery and decision support.” However, exascale is shown to be DOE’s main goal for the program, as the document says the increases for the LCFs are to, “continue site preparations and non-recurring engineering investment with their vendors that will allow them to deploy at least one exascale-capable system as rapidly as possible.”

With regard to ARPA-E, there isn’t language to justify the elimination of the program. To give some history, some policymakers have never embraced the agency, particularly those who view the applied research the agency performs as best done by industry; the Administration seems to embrace this viewpoint. And ARPA-E has been used as a bargaining chip in Congress, between those who champion the agency’s mission and those who want to eliminate it; that’s expected to continue. The agency’s ultimate fate is far from certain though.

With that in mind, what are the prospects in Congress for the agency’s budget request? Probably pretty good. Computing research enjoys broad support in both parties in Congress. However, the sticking points will be ARPA-E and the other research projects within the Office of Science. Those research areas within SC, such as Fusion, Biological & Environmental Research, and Basic Energy Sciences, are slated for reductions in this request; ASCR is the only one with an increase. Congress is likely to see that as robbing Peter to pay Paul; the question then becomes, where does any extra money for those other research areas come from? Given the budget deal that was reached last month, DOE SC could receive a boast that is shared equally. But that’s not a given and only time will tell what happens. We’ll keep tracking the budget as it moves through the process, so check back for updates.

Department of Energy FY 2019 Request: Good for Computing but with Some Asterisks