SOLE SOURCE: Integration of MPAS atmospheric dynamical core within the existing PyTorch framework for the NWS Office of Modeling and Development in Silver Spring, MD.
Closed Pre-Solicitation Posted
- Solicitation number
- 142085-26-0014
- Agency
-
National Oceanic and Atmospheric Administration
Department of Commerce
- Responses due
- Set-aside
- No set-aside
Opportunity facts
- NAICS code
- 518210 Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services
- Place of performance
- Silver Spring, Maryland 20910, United States
- Points of contact
Notice details come from SAM.gov. Updated .
About this opportunity
NOAA's Acquisition and Grants Office, Eastern Acquisition Division–Kansas City intends to award a purchase order on a sole-source basis to IC Weather LLC for integration and development of the MPAS atmospheric dynamical core within an existing PyTorch framework for the NWS Office of Modeling and Development in Silver Spring, Maryland, with a response deadline of August 27, 2026.
The sole-source award is justified under FAR 6.103-1 on the grounds that only IC Weather LLC is capable of meeting the requirement at the required quality level. The vendor's proprietary MPAS-PyTorch integration is the only existing capability that satisfies three linked technical objectives: adapting the MPAS dynamical core to leverage GPU acceleration for improved computational efficiency, integrating into an infrastructure that enables machine learning–based model parameterizations, and providing direct access to parameter gradients for extending NOAA's data assimilation capabilities to 4-Dimensional Variational algorithms (4DVAR) and enabling Forecast Sensitivity to Observation Impact (FSOI) techniques. IC Weather LLC has demonstrated via MPAS-PyTorch a substantial decrease in computational time to run a forecast compared with the current FORTRAN version of MPAS. The integration approach allows for testing at higher spatial resolutions required for NWS operations and supports future development of PyTorch or machine learning versions of physics schemes to further improve computational efficiency. NAICS 518210 (Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services); PSC 7A20 (IT and Telecom—Application Development Software, Perpetual License Software).
Notice text
The National Oceanic and Atmospheric Administration’s (NOAA) Acquisition and Grants Office, Eastern Acquisition Division-Kansas City intends to negotiate and award a purchase order, on a sole source basis, under authority of RFO FAR 6.103-1, with small business entity IC Weather LLC for integration and development of MPAS atmospheric dynamical core within the existing PyTorch framework for the NWS Office of Modeling and Development in Silver Spring, MD.
Sole Source Justification:
The sole source determination is based on NOAA’s Sole Source Justification. Only one source is capable of providing the supplies or services required at the level of quality required because the supplies or services ordered are unique or highly specialized. The specific requirement is to adapt the MPAS dynamical core to leverage GPU acceleration for improved compute efficiency, integration into an infrastructure that enables future work on model parameterizations based on machine learning (ML) algorithms, and provides direct access to parameter gradients which may be used to extend NOAA’s data assimilation (DA) capabilities to
4-Dimensional Varational algorithms (4DVAR) as well as enable advanced observation impact studies including Forecast Sensitivity to Observation Impact (FSOI) techniques. The first of these requirements is capable of being met through a proprietary software package that has integrated MPAS into the PyTorch framework. The vendor has demonstrated via the “MPAS-PyTorch” a substantial decrease in the computational time to run a forecast compared with the current FORTRAN version of MPAS. The nature of integration will allow for further testing at higher spatial resolutions that would meet NWS operational requirements, in conjunction with PyTorch / ML versions of physics schemes used in MPAS to further improve computational efficiency, and use of the native gradients computed by PyTorch for 4DVAR and FSOI capabilities.
Since this initial capability exists, it is not in the interest of the Government to open a competitive contract acquisition process to duplicate the effort or risk exposure to intellectual property infringement. However, any derivatives of the current software will become open source per NOAA policy.
This pre-solicitation notice is not a request for competitive proposals and no solicitation package is available. However, firms that believe they can fully meet the government’s requirements may submit substantiating documentation or a request for additional information in writing to the (Contract Specialist) within 15 days after publication of this pre-solicitation notice. Such requests and documentation will be evaluated solely for the purpose of determining whether or not to conduct this procurement on a competitive basis. A determination by the government not to conduct a competitive procurement, based upon responses to this notice, is solely
within the discretion of the government. Oral communications are not acceptable in response to this pre-solicitation notice.
Similar open opportunities
| Opportunity | Agency | Set-aside | Responses due |
|---|---|---|---|
| NOAAPort Replacement Capabilities | National Oceanic and Atmospheric Administration | No set-aside | |
| Generation and Delivery of Meteorological and Hydrological Forecasts for the Western Washington and Northwest Montana River Basins | ENDIST Seattle | No set-aside |
On GovTribe
Work this opportunity
- Track it in your pipeline
- Find teaming partners
- Similar opportunities
- Ask GovTribe AI about this opportunity