CONTENT

1.  Introduction
2. DYON output
   1. Files provided
   2. Running the scripts
   3. The main files (functions 'put_to_IMAS', 'get_from_IMAS' and test scripts)
   4. IDS core_profiles
   5. IDS core_sources
   6. IDS equilibrium
   7. IDS gas_injection
   8. IDS magnetics
   9. IDS radiation
   10. IDS wall
   11. Service functions
3. DYON inputquit
4. To be done

1. Introduction

This page contains a description of MATLAB scripts supporting the storage of DYON input and output data files in IMAS. 

The repository folder of the DYON output wrapper is ~g2yyakov/public/dyon/stable.

This document can be found on the ACH-PSNC Confluence page (DYON IMASification - Scientific Worfklows - PCSS Confluence (psnc.pl), https://docs.psnc.pl/display/WFMS/DYON+IMASification).

2.  DYON output

The storage of the DYON output in IMAS is supported by two MATLAB functions, put_to_IMAS and get_from_IMAS. Depending on the situation, they can be integrated to the code or used separately as wrappers providing the connection between DYON and IMAS. The scripts test_put_to_IMAS and test_get_from_IMAS are provided, which illustrate how the functions can be called as wrappers.

The available IMAS Data Dictionary (DD) does not satisfy all requirements of DYON. Therefore, the functions put_to_IMAS and get_from_IMAS are to be extended after the planned extension of the DD. 

Now the output data are placed into 7 IMAS IDS's (Interface Data Structures - sections of the IMAS DD describing tokamak subsystems or tokamak physics chapters): 'core_profiles', 'core_sources', 'equilibrium', 'gas_injection', 'magnetics', 'radiation', and 'wall' (this list is to be extended). The allocation of the data in the IDS's and the functions supporting it are described below.

2.1.  Files provided

2.2. Running the scripts

To launch the test scripts, do the following:

1. Enter:

module load imasenv/<IMASversion> 

Here <IMASversion> is 3.37.0 or higher.

2. The scripts in its present form use the IMAS database 'step'. If you are planning to use it and this database is non-existent, enter the following command to create the database:
imasdb step

You can replace 'step' with another IMAS database name, but do not forget to create it and change the variable 'machine' in test_put_to_IMAS.m and test_get_from_IMAS.m.

3. Enter either 
     matlab
  or
     matlab -nodesktop -nosplash

In the nodesktop mode, MATLAB should work faster (I did not notice a significant difference - YY). When using the nodesktop mode, do not forget to run the command 'quit()' to end the session (otherwise, a dangling MATLAB session may remain).

4. Run test_put_to_IMAS.m in order to save in IMAS the data from an available DYON output MAT-file (you can change shot, run, IDS comments etc. by changing the parameters in this script).

5. Type 'viz' (on a console with imasenv loaded) to see the content of the IMAS database.

6. To get back the DYON output data, run test_get_from_IMAS.m. 

2.3. The main files (functions 'put_to_IMAS', 'get_from_IMAS' and test scripts)

The script test_put_to_IMAS.m loads a MAT-file containing the DYON results (the name of the file is defined in the script). It is assumed that there is a structure with the name 'output' among the variables loaded from the MAT-file and that this structure holds all the data that must be saved.

The script then defines the rest of arguments of the function 'put_to_IMAS' and invokes it. The dummy arguments of this function are as follows:

The function creates an IMAS data entry for the given 'shot' and 'run' parameters (note that the database with the required name must exist; otherwise, an error is raised). Then it invokes several functions that put data to separate IDS's ('put_core_profiles', 'put_core_sources', 'put_equilibrium', 'put_gas_injection', 'put_magnetics', 'put_wall', and 'put_radiation') and closes the database. After that, the content of the 'output' structure is put to IMAS. The DD fields to which the data are placed can be found in the table IMAS interface with DYON - Google Documents. Some more comments will be given in next sections of this document for each IDS separately.

The script test_get_from_IMAS.m defines the input parameters of the function get_from_IMAS and invokes it. The dummy arguments of this function are as follows:

The function calls several functions that extract data from separate IDS's ('get_core_profiles', 'get_core_sources', 'get_equilibrium', 'get_gas_injection', 'get_magnetics', 'get_wall', and 'get_radiation') and closes the database. The function returns the 'output' structure, which is to be equivalent (after the script is complete) to the 'output' structure that was initially put to IMAS. In addition, the function prints to the screen the comments extracted from all IDS's.

The script in its current form contains some parts that are used for verification and debugging. It imports the original DYON output and compares it with the one recovered from IMAS. These part should be removed in the future when such comparison is not required.

2.4. IDS core_profiles

The quantities stored in this IDS include electron and ion temperatures, densities of electrons, several ion species and several neutral species, loop voltage, and plasma current. Although all these quantities are represented by single values, most of them are stored as 1D-profiles on the formal grid consisting of a single point. When the ion and neutral densities are recovered from the IDS, the ion species are selected in the array of structures by their charge numbers (not by their places in the array). This is done to permit more flexibility if the set of the ions simulated by the code changes. At present, the storage of ion quantities in the IDS is organized as follows:

Here 'ion label' and 'ion index' are the label and the index of the species in the core_profiles.profiles1d.ion AoS (array of structures), respectively; 'neutral label' and 'neutral index' are the label and the index of the species in the core_profiles.profiles1d.neutral AoS (array of structures), respectively. 

For all ions except for H, several ion states (charge states) are defined, the index of the 'state' array element being equal to the ion charge number.

The following functions support I/O with the 'core_profiles' IDS:

function put_core_profiles (output, H_mass, idx, comment, data_source, imas_version, access_layer)

Write a suitable part of the 'output' structure to the 'core_profiles' IDS. 

This function invokes the functions 'arrange_ion_densities' and 'arrange_neutral_density'.

function [output, comment, data_source, imas_version, access_layer] = get_core_profiles (output, idx, get_time)

Add the data from the 'core_profiles' IDS to the 'output' structure. 

This function invokes the functi on ' extract_ion_densities '.

2.5. IDS core_sources

The source densities are stored as 1D-profiles on the formal grid consisting of a single point although they are represented by single values. 

When the data are recovered from the IDS, the sources are sought in the array of structures by their DD indices (not by their places in the array). This is done to permit more flexibility if the set of the sources calculated by the code changes. At present, the storage of sources in the IDS is organized as follows:

Here 'IDS index' is the location of the source in the core_sources.source AoS (which can be changed if required). The 'IDS name' is the label of the source (a part of the source identifier), which is not used when the data are extracted from the IDS. The source 901 (ECH + EBW power absorbed to electrons) is a custom source missing in the DD.

The data are put to the 'core_sources' IDS by the function 'put_core_sources'. Its interface is as follows:

function put_core_sources (output, idx, comment, data_source, imas_version, access_layer)

The data are extracted from the 'core_sources' IDS by the function 'get_core_sources'. Its interface is identical to that of 'get_core_profiles' (Section 2.4). 

2.6. IDS equilibrium

Data exchange with the 'equilibrium' IDS is carried out by the functions 'put_equilibrium' and 'get_equilibrium'. The interface of 'get_equilibrium' is identical to that of 'get_core_profiles' (Section 2.4). The interface of 'put_equilibrium' is identical to that of 'put_core_sources' (Section 2.5).

2.7. IDS gas_injection

Data exchange with the 'gas_injection' IDS is carried out by the functions 'put_gas_injection' and 'get_gas_injection'. The interface of 'get_gas_injection' is identical to that of 'get_core_profiles' (Section 2.4). The interface of 'put_gas_injection' is identical to that of 'put_core_sources' (Section 2.5).

The gas puffing units are electron/s in DYON and Pa*m^3/s in IMAS. The unit conversion is made on the assumption that the gas is H2 (D2) molecules with the temperature of 300 K.

2.8. IDS magnetics

Data exchange with the 'magnetics' IDS is carried out by the functions 'put_magnetics' and 'get_magnetics'. The interface of 'get_magnetics' is identical to that of 'get_core_profiles' (Section 2.4). The interface of 'put_magnetics' is identical to that of 'put_core_sources' (Section 2.5).

2.9. IDS radiation

The emission densities are stored as 1D-profiles on the formal grid consisting of a single point although they are represented by single values. 

When the data are recovered from the IDS, the emission processes are sought in the array of structures by their DD indices. At present, the storage of process in the IDS is organized as follows:

Here 'IDS index' is the location of the source in the 'radiation.process' AoS (which can be changed if required). The 'IDS name' is the label of the process (a part of the process identifier), which is not used when the data are extracted from the IDS. The processes with DD indices 901 and 902 are custom processes missing in the DD.

The radiating species are sought in the AoS's 'radiation.process(:).profiles1d(1).ion' and 'radiation.process(:).profiles1d(1).neutral' by their labels, which are the standard chemical element notations. 

Data exchange with the 'radiation' IDS is carried out by the functions 'put_radiation' and 'get_radiation'. The interface of 'get_radiation' is identical to that of 'get_core_profiles' (Section 2.4). The interface of 'put_radiation' is identical to that of 'put_core_sources' (Section 2.5). 

2.10. IDS wall

The 'wall' IDS is used to store sputtering coefficients. When the data are extracted from the IDS, each coefficient is found in the AoS by the labels of the sputtered and incident species, which are the standard chemical element notations.

Data exchange with the 'wall' IDS is carried out by the functions 'put_wall' and 'get_wall'. The interface of 'get_wall' is identical to that of 'get_core_profiles' (Section 2.4). The interface of 'put_wall' is identical to that of 'put_core_sources' (Section 2.5). 

2.11. Service functions

function [ion] = arrange_ion_densities (mass, Z_n, label, neutral_index, densities)

Prepare a structure containing the densities of all ion charge states of an element for the core_profiles IDS and other element characteristics.

function [neutral] = arrange_neutral_density (mass, Z_n, label, ion_index, density)

Prepare a structure containing the density of neutral atoms of an element for the core_profiles IDS.

function [neutral_sort] = arrange_radiation (label, mass, Z, power_density)

Arrange the structure holding the data about emissivity and other characteristics of a neutral species.

function [density, ion_found] = extract_ion_densities (profiles, Z)

Extract densities of all charge states of an ion species from the core_profiles IDS.

function [AOS_index] = find_source (sources, dictionary_index)

Find location of the element with a given data dictionary index in an AOS (in particular, the source in core_sources.source and the process in radiation.process).

function index = find_species (species_list, label)

Find location of the element with a given label in AoS. The function is used to find the required spruttering coefficient in the 'wall' IDS. The label is the chemical element label; it is hold in the 'label' leaf of each AoS element.

3.  DYON input

Description of the scripts processing the DYON input (to be done).

4. To be done

• Extending the input script (pending the planned DD extension)
• Adding code descriptions to all IDS's.
• Switching to the HDF5 IMAS backend (pending HDF5 debugging on Gateway).