Jan Mandel/Blog/2015 Apr Jun
From CCM
To the current blog page and the archive
HPC Clusters in Cloud computing
Digital Ocean
- How To Set Up And Use DigitalOcean Private Networking
- How To Create a Beowulf Cluster Using Ubuntu 12.04 VPS Instances
- Specs: 1Gb/s neworking, dedicated CPU cores E5-2630L v2 2.40GHz
- Pricing: 1.4 - 1.5c/GB memory/hour, or 0.7 - 4.8c/core/hour calculation
- Largest 20 cores + 64GB $0.952/hr, networking 1Gbs/only
- data transfer few TB included, then $20/TB (do they even measure?)
- New Frankfurt region has 40GbE networking on each hypervisor
Amazon EC2
- Quick MPI Cluster Setup on Amazon EC2, MPI benchmarks
- Some Micro-benchmarks for #HPC: 10Gb Ethernet on #EC2 vs QDR #Infiniband
- Largest SMP c4.8xlarge 36 cores 60GB mem + EBS storage extra $1.856/h, private network 10Gb/s but still slow
- data transfer $90/TB
Apr 6: Assimilation of airborne meteorological data
- https://madis.noaa.gov/madis_acars.shtml
- TAMDAR
- AirDat project: Assimilating TAMDAR Observations Using the WRFDA System https://ams.confex.com/ams/pdfpapers/164854.pdf
- Improving High-Resolution Numerical Weather Simulations by Assimilating Data from an Unmanned Aerial System
- Using Real-Time Weather Data from an Unmanned Aircraft System to Support the Advanced Research Version of the Weather Research and Forecast Model
- TAMDAR overview for the White House
Apr 19: Transfer credit
- Accreditation in the US is a peer review process to assure quality. I was not able to find any relation to transfer credit or course equivalency
- Accepting credits earned at another institution is the prerogative of the receiving institution
- Colorado transfer requirements
- Colorado GT Pathways: gtPathways courses apply to gtPathways requirements, apply to other requirements at the receiving institution’s discretion. Math lists only one, GT-MA1 3 hours general requirement.
- ...although all coursework will transfer, it may not apply
- Articulation and transfer agreements (look in the middle of a long page)
- BS Math Colorado articulation agreement is about transfering a whole community college degree. I did not anything in the document about a requirement to transferring and apply to a degree individual courses.
- It seems that we are required by CO only to apply one GT-MAT1 course to core requirement, or an entire community college degree to the BS Math. The rest is up to the discretion of the institution. I could not find any external requirements to transfer or apply individual courses from any institution, this seems unrelated to accreditation or CO articulation agreements.
- Decisions to transfer individual courses and apply them to requirements seem to be UCD discretion.
- CU Boulder
- Transfer credit general information
- CU System course equivalencies table for Engineering but does not include anything under Calculus
- CU Denver
- Math catalog course descriptions
- English Course Equivalencies
- UCCS catalog
- Metro state has a transfer search tool
Apr 19: Radon-Nikodym derivatives of Gaussian Measures
- Lecture notes from Cornell
- Equivalence and Radon-Nikodym derivatives of Gaussian measures by Henrich
- ref I.E Segal, Distributions in Hilbert space and canonical systems of operators, Trans. Amer. Math. Soc., 88 (1958), pp. 12–41cl
Apr 24: Convergence of Ensemble Kalman Filters in the Large Ensemble Limit and Infinite Dimension
Apr 25: WRF
Air Force to move away from WRF
WRF Release 3.7
- includes vertical nesting!
Apr 28: Live fuel moisture
- National Fuel Moisture Database at WFAS
May 20: Level set methods for fire spread modeling
- Start: Sethian's website on level sets and book now online
- NIST report Rehm-2009-FPU link from the FDS group, advection with flux limiters instead of solving the level set equation. Citing Toro-2009-RSN, see Toro-2009-RSN-ch13 doi ucd page 456.
- Our papers, 1st order upwinding.
- Mandel-2009-DAW doi ucd. The matlab code that was used to develop the propagation scheme in the WRF-Fire code in 2007 is still in the repository in standalone/matlab, it was in wrfv2_fire/phys until moved in 2010. There will be different versions of interest as the code evolved, particularly in 2007-2008, look at the history. See also some files from the old dddas03 CVS repository, now copied to here. There are experiments in Minjeong's thesis.
- Current: Mandel-2011-CAF doi ucd and a supplement with experiments
- Multivac code in C++: Mallet-2009-MWF doi ucd preprint, many methods including high-order ENO
- Implicit method in PetSC Huabsomboon-2007-ILS link
- Randomized Pagnini-2012-RLM link
May 21: WRF-Sfire-Chem in IJWF appeared
- Kochanski, Adam K., Jenkins, Mary Ann, Yedinak, Kara, Mandel, Jan, Beezley, Jonathan, Lamb, Brian, 2015. Toward an integrated system for fire, smoke and air quality simulations, International Journal of Wildland Fire, doi:10.1071/WF14074
May 30: Lorenz 96 model
- Formulation: Lorenz-2006-PPS, originally at ECMWF seminar in Lorenz-1995-PPS.
- The numher of points does matter, and it is not simple scaling.
- Properties Lorenz-1998-OSS, see also Orrell-2003-MEP
- code Lorenz96.m