Use JSON for input and output

Running simulations from JSON input.

In addition to the usual way of running simulations using a Config and SimParams, HallThruster supports the use of JSON for input and output. This can be achieved by calling run_simulation("input.json"), where "input.json" could be any json file. The contents of this file should be laid out in one of two ways. The first way has up to three keys – two mandatory ("config" and "simulation") and one optional ("postprocess"). These exactly mirror the Config, SimParams, and Postprocess structs that are normally passed to run_simulation and should be relatively self-explanatory.

Below, we show an example of this type of JSON input.

{
    "config": {
        "thruster": {
            "name": "SPT-100",
            "geometry": { "inner_radius": 0.0345, "outer_radius": 0.05, "channel_length": 0.025 },
            "magnetic_field": { file = "bfield_spt100.csv" }
        },
        "propellant": "Krypton",
        "discharge_voltage": 300.0,
        "anode_mass_flow_rate": 5e-6,
        "domain": [0.0, 0.08],
        "anom_model": {
            "type": "TwoZoneBohm",
            "c1": 0.00625,
            "c2": 0.0625
        }
    },
    "simulation": {
        "adaptive": true,
        "dt": 5e-9,
        "grid": {
            "type": "EvenGrid",
            "num_cells": 200,
        },
        "duration": 1e-3,
        "num_save": 1000,
    },
    "postprocess": {
        "output_file": "output.json"
        "save_time_resolved": false,
        "average_start_time": 5e-4
    }
}

In the second way, the JSON file has a top level field input, with members config, simulation, and (optionally) postprocess, e.g.

{
    "input": {
        "config": {...},
        "simulation": {...},
        "postprocess": {...}
    }
}

where the contents of these keys are exactly the same as in the first method.

In both cases, the field names and types of these inputs are exactly the same as in the corresponding HallThruster types. Note the "type" field for config.anom_model and simulation.grid, which precedes the other fields for that type. This pattern is also used for config.conductivity_model and config.wall_loss_model, should you wish to provide those. Note that custom anomalous transport models and propellants are not supported using the JSON interface at this time.

Writing output files

If postprocess is provided and postprocess.output_file is not empty, HallThruster will write an output JSON file to that file. The output file contains two top-level fields: input and output. The input field reproduces the inputs used to run the simulation, exactly as described above. The output field has at most four fields: retcode, error, fields, and average, e.g.

{
    "input": {...},
    "output": {
        "retcode": "success",
        "error": "",
        "average": {...},
        "frames": [
            {...}, {...}, ...
        ]
    }
}

The output_field somewhat mirrors the Solution object. The retcode gives the simulation status (one of "success", "error", or "failure"). If retcode is error, the error will contain a string with the error that occurred. If postprocess.average_start_time is greater than or equal to zero, the average field contains the output of HallThruster.time_average(sol, average_start_time). Finally, if postprocess.save_time_resolved is true, then output.frames contains simulation.num_save frames.

You can also manually write a Solution to a JSON file using the write_to_json function.

Frame format

The frames in the output JSON file are laid out similarly to those in the Solution struct, with some additions. In addition to plasma properties, each frame also stores its time, as well as thrust, discharge current, and component efficiencies.

Restarts