Plasma State Component
Software Status
• Current Version: 1.001; no recent updates.
• Implemented as 3 libraries layered over the NTCC library set: xplasma, pspline, …
• Location of plasma state libraries– SWIM svn repository– NTCC distribution (formal module is planned).
• New visualization application: cstate– Compare 2 or 3 states or view a single state.
Near Term Upgrades
• Add Torques (angular momentum sources).
• Reduced Species List:– Many components don’t need full impurity
charge state spectrum.– Reduced list for Z>2 elements, with 2 proxy
elements at low & high end of impurity Z range• Zeff and Quasineutrality conserved.
– Need to identify software component to maintain proxy element densities…
• Plasma State could trigger off updates to do it.
Likely Future Upgrades
• C/C++ get/set access to state elements.– For FACETS; would help Python access also.
• Representation of L. Zakharov ESI data:– Bicubic Hermite representation.– Existing internal spline representation can be
converted without affecting applications.– New interface can be added to enable external
control of Hermite node derivatives.
Issue: Update Propagation
• Long running jobs need stable set of binary executables:– Each run owns a private set of binaries?– This is done in TRANSP production system.– Avoid constraint on code development.
• Many things will need to be added to the state over time:– How to minimize disruption to component
developers?
Issue: Machine Description
• Add “Machine Description” section to plasma state?– Make loadable from database of namelists:
“DIII_rev1”, “DIII_rev2”, “CMOD_rev1”, …– Leave elements modifiable by applications?
• But prevent changes “after initialization”?
– Who defines contents? Try for compatibility with EU ITM / CPO effort?
– Components need this now.
Issue: Fast Species Components
• Support multiple co-existing fast specie components: “FP” & “FPMC” for example?– This gets ugly...– Separate non-Maxwellian species lists? – Different grids for each individual specie?– Two copies of summed quantities like PBE
because state can’t know if FP or FPMC will compute it? This is hard on EPA.
– Further discussion needed.
Issue: Large State Data
• Use of state for large data, 3d or more?– Such as fast particle distribution functions…– …and fast particle source distributions.
• State interpolation tools limited to 3d at present.
• File I/O with large data elements will be expensive: may want to split state into 2d and 3d+ subsets to manage performance.
Other issues?
• You tell me…