Resilient Extreme-Scale Solvers ("RX-Solvers")
Funding Opportunity Announcements (FOA) DE-FOA-0000742 and Program Announcement LAB12-742
Q1:What is the expected size of awards?
A1: Under item E. of PART II - AWARD INFORMATION, it is stated that "The award size will depend on
the number of meritorious applications and the availability of appropriated funds. The total project size is
anticipated to range from $150,000 up to $4,500,000 per year."
Q2: Does the collaboration need to be led by a lab?
A2: No, collaborations can be led by a University or Industry based PI. PIs need to keep in mind
that only projects relevant to DOE missions will be competitive.
Q3: The FOA states, "Applications whose research plans do not include numerical algorithms
research in a substantive manner will be deemed nonresponsive. Applications whose research plans do not include
work in resilience or fault tolerance in a substantive manner will be deemed nonresponsive. Applications whose
research plans do not include characterization and demonstration of the performance of proposed algorithms in a
substantive manner will be deemed nonresponsive." What are you looking for?
A3: We are seeking projects that contain at least three elements: numerical algorithms, fault
tolerance or resilience, and demonstration of the stated performance of the proposed numerical algorithms. Please
see Merit Review Criteria.
Q4: I have not been funded on DOE Office of Science projects before. Should I obtain a letter of
support from my colleagues funded by DOE Office of Science to attest to the DOE mission relevance of my work?
A4: No! Please note that in Part IV, Section C, subsection 2 of DE-FOA-0000742 (and the
corresponding section within the companion announcement Lab-12-742), it is clearly stated:
"Please do not submit general letters of support as these are not used in making funding decisions and can
interfere with the selection of peer reviewers."
Q5: How many awards overall are targeted?
A5: There is no target number of awards. Until we review the proposals and make selection
decisions, we will not know how many awards we will make.
Q6: The FOA states, "Each project is expected to have a management plan appropriate for the
size and complexity of the project." I am proposing a small, single PI project. Do I have to have a
management plan?
A6: You need to include a plan that makes sense for your project. The FOA further states,
"The roles of senior and key personnel must be clearly articulated. A communication plan, both for
communication within the project and for communication with the broader community, is expected from each project.
The Principle Investigator is expected to commit sufficient time to be accountable for the successful execution of
the project's plans. Other senior project personnel are expected to commit sufficient time in order to make
significant technical contributions to the project. The project’s management and communication plans and the
levels of commitment of effort of lead project investigator, key personnel, and other senior personnel will be
considered in Merit Review."
Q7: The FOA states that general letters of support should not be included. Is it permissible to
include quotations from them within the project narrative?
A7: Other than page limits, there are no restrictions on how you describe your collaborations in
your narrative.
Q8: I would like to propose a small project that only addresses one of the three elements. Will
that be responsive to this call?
A8: No. Please note that, in addition to A3 above, the FOA also states that all
projects will be subject to the same Merit Review Criteria.
Q9: Page 7 states: Large team awards: awards involving multiple investigators from two or more
institutions. The funding level for these awards will be roughly around $1M - $2.5M per year for up to three
years. Small team awards: awards involving single investigators or small teams. The funding level for these
awards will be up to $250K per year for up to three years. Must a 'large team' have two or more institutions, or is
a large team from a single institution permissible? And, is it anticipated that only these two sizes will be
funded, or is it possible that a project in the $250k - $1M range (say $700k) or larger than $2.5M would be
funded?
A9: In general, it is unlikely that one institution would possess sufficient capabilities to
satisfy the three requirements of the call (algorithms, resilience, and performance – please see Merit Review
Criteria); however, if one single institution can in fact meet all these requirements, there is no further
requirement that the research team must include another institution.
The figures for the Large team awards are estimates of what it would take to satisfy all three requirements of the
call successfully. Award - or non-award - decisions will not be based solely on the dollar amount requested. You
may find that the proposed Large Team project could satisfy the requirements successfully and meritoriously for
less than $1M per year for up to three years. You may also find that you may need more than $2.5M per year up to
three years to carry out the proposed technical work successfully. The total amount available for this FOA is $4.5M
per year for up to three years.
Conventional wisdom suggests that it is quite unlikely that a project for up to $250K per year for three years
would be able to meet all three requirements of the call; however, we do not wish to exclude the possibility that
there may be some revolutionary research ideas out there that could address all three requirements meritoriously in
a single-PI setting. Be sure to consult Merit Review Criteria before proposing these Small Team projects!
Q10:Is there a citizenship requirement for grants? Can foreign institution apply?
A10: Any university or other institution of higher education or other non-profit or for-profit
organization, non-Federal agency or entity, is eligible for a grant or cooperative agreement. An unaffiliated
individual also is eligible for a grant or cooperative agreement. For more information go to:
http://science.doe.gov/grants/policy.asp#rules
Q11: Would you please clarify the eligibility of FFRDCs for us? Based on verbiage in Parts I and III, our understanding is that an FFRDC can be a part of a collaborative proposal but may not submit as a lead institution. Is that correct?
A11: Our intent is to allow FFRDC's to submit individually, as well as being part of a collaborative proposal - thus FFRDC's may submit as a lead institution. FFRDC's may apply to DE-FOA-0000742.
Q12: From the call, I read the following: "Small team awards: awards involving single investigators or small teams. The funding level for these awards will be up to $250K per year for up to three years". Does it mean the small team has to have a single PI or it allows 2-3 PIs?
A12: The small team awards are intended for small single-PI or small collaborative projects. Please note that projects of all sizes will be evaluated using the same Merit Review criteria, which include whether there is sufficient time commitment from key project personnel. Should a small team project be judged meritorious and if funds are available, it may be considered regardless of whether it's a single PI project or if it has 2-3 PIs on the project.
Q13: Can you please clarify the following instructions for collaborative proposals:
"Collaborative proposals submitted from different institutions should clearly indicate they are part of a proposed collaboration and contain the same title, Abstract, and Narrative for that research project. In addition, such proposals must describe the work and the associated budget for the research effort being performed under the leadership of the Principal Investigator at that participating institution."
So, for the lead institution proposal:
1) Does the proposal include the FWP and budget pages/justifications for all institutions or just the lead institution (including the roles and responsibilities of all personnel at all institutions)?
2) Does the proposal include the biographical sketches for all personnel at all institutions?
3) Does the proposal include facilities and resources for all institutions?
4) Does the current and pending support correspond to personnel at all institutions?
For the collaborating institutions:
1) Where do they indicate that this is a collaboration? On the cover page?
2) Presumably, they only provide FWP, budget pages, bios, facilities, and support for their institution, correct?
3) Where are they to describe the work specific to that institution? In the FWP, in the budget explanation, or are you looking for more specific narrative in addition to the common narrative? If the latter, is there a page limit?
A13: The easiest way to do this would be
(a) For everyone in the collaboration: include the same narrative, use the same proposal title
(b) For the lead institution: include all budget pages, all bio sketches, all facilities & resources, and all current & pending for all institutions
(c) For the collaborating institutions: provide budget, bio, facilities & resources, current & pending for personnel at the submitting institution only.
Institution-specific narratives in addition to the common narrative is not required - the roles and responsibilities of each collaborator should be part of the common narrative already.
Q14: We know you are looking for three things: algorithms, resilience, and performance demonstration. How much computer science should I include in my research plans? How little must I have? Since this is a math research call, is there a limit on how much "computer science" research I can propose?
A14: We are looking for "basic research in Resilient Extreme-Scale Solvers ("RX-Solvers") that demonstrably advances the state of science and practice for scalable, resilient, extreme-scale numerical algorithms, to enable scientific discovery on the supercomputers expected to come online in the next 5-10 years and lay the foundation for research in numerical algorithms for extreme-scale scientific computing." The main focus of this FOA is numerical algorithms, that is, "solvers" research. This is not an FOA for research in resilience, programming model, abstract machines, OS, performance modeling, etc., topics that are usually the focus of computer science research. You will need to propose a balance of applied mathematics and computer science that makes sense for your algorithmic research project - and make a case for the balance. For example, it may make sense for you to propose to tailor an existing computer science tool or to develop a specific tool for your particular solvers research; either could be acceptable if the proposed solvers research logically calls for the tool.
Q15: For the "Performance of Proposed Algorithm", is it sufficient to implement my algorithm into an existing application code and run the code on a DOE Leadership machine such as Mira?
A15: According to the FOA, "Researchers are expected to devise a suitable suite of performance metrics and demonstrate, beyond "toy problems', "linearized analysis', or "back-of-the-envelope estimates', measurable progress towards conquering the extreme-scale challenges." It is entirely up to you to determine what a suitable suite of performance metrics ought to be.
Q16: Do I have to build my project around an existing DOE code?
A16: No.
Q17: Do I have to implement my algorithm for a particular DOE application?
A17: No. While DOE relevance is important, and a DOE mission-relevant application may provide a good context for the solvers research, we are seeking basic research in solvers. Having your research solver used by a DOE application in the course of the research is not a required outcome of this FOA.