PEST

1. Code Name: PEST
2. Code Category: Linear ideal MHD stability
3. Primary Developer: Ray Grimm (originally), J. Manickam (currently)
4. Other Developers and Users: R. L. Dewar, M. Chance, J. Greene and J. Johnson; Berkery, Betti, Boyle, Breslau, Chance, Diallo, Feibush, Foley, Gerhardt, Hudson, Kessel, Maingi, Paoli, Sontag
5. Short description (one line if possible): Determines linear stability of fast growing ideal MHD modes
6. Type of input required (including files and/or output from other codes): Requires equilibrium files from EFIT, LRDFIT, JSOLVER, QSOLVER, CHEASE etc
7. Type of output produced (including files that are read by other codes and size of files if large and synthetic diagnostics): ASCII and binary data files used for mode analysis and post-processing
8. Describe any postprocessors: A general post-processor GOPLOTTER computes perturbed field from the displacement vector. A major code to include kinetic effects MISK was developed by Betti and Hu. Several synthetic diagnostics are written in IDL, these include, ECE. MSE, SXR, and Mirnoov.Chance has an extensive suite of VACUUM and external fields code.
9. Status and location of code input/output documentation: Primitive manuals are available on demand
10. Code web site? No dedicated web site
11. Is code under version control? What system? Is automated regression analysis performed? At the present time Manickam is the sole developer and maintains the code.
12. One to two paragraph description of equations solved and functionality including what discretizations are used in space and time: The code is based on the energy principle. The practical application is based on a linearized Galerkin procedure, where we represent linearized perturbations of the equilibrium to be represented by a subset of a complete set of expansion functions. The representation uses Fourier modes in the poloidal and toroidal directions and lowest order finite elements in the radial direction. The variational calculation is reduced to the determination of the eigenvalues and eigenvectors of a matrix eigenvalue problem. In most cases we are seeking only the unstable modes; An inverse vector iteration technique is used. Typical expansion parameters for NSTX plasmas include 200 radial elements and upto 120 fourier modes, the matrix rank is 72,000 or greater.
13. What modes of operation of code are there (eg: linear, nonlinear, reduced models, etc ): The standard code is a full 3D axisymmetric version with a proper kinetic energy. A related version, PEST-2 uses a scalar form of delta-W with a model kinetic energy.
14. Journal references describing code: Grimm, R. C, Greene, J., and Johnson, J. L., in Meth. Comp. Physics 9 (1976) 253. Grimm, R. C., Dewar R. L., and Manickam J., in j. Comp. Phys. 49 (1983) 94.
15. Codes it is similar to and differences (public version): The PEST code was one of the earliest delta-W codes in full toroidal geometry. Since then several other codes were developed and were benchmarked against PEST. These include, ERATO, GATO, DCON, KINKX. These codes use different choices of expansion function or solution methods. A major difference is the use of 2D finite elements as opposed to the Fourier-finite element combination in PEST.
16. Results of code verification and convergence studies (with references): The code has been extensively benchmarked Chance et al. J. Comp. Phys. 28 (1978) 1.
17. Present and recent applications and validation exercises (with references as available): There have been about 100 publications with theoretical models and comparison with expt.
18. Limitations of code parameter regime (dimensionless parameters accessible):
19. What third party software is used ? (eg. Meshing software, PETSc, ...): Nag
20. Description of scalability: N/A
21. Major serial and parallel bottlenecks: None
22. Are there smaller codes contained in the larger code? Describe: Related codes are the BALLOON and VACUUM codes.
23. Supported platforms and portability: The code is mostly in FORTRAN and portable. The only unusual feature is the use of direct access binary I/O used to optimize the matrix solver. It is presently on Portal. Eliott Feibush has developed an ELVIS app of PEST
24. Illustrations of time-to-solution on different platforms and for different complexity of physics, if applicable: The runtime depends on the q-profile and the toroidal mode number of interest. For q-edge ~5 and n=1 it takes about 10-15 secs. For q-edge ~ 15 and n-5 it takes up to 10 minutes, on portal.