Skip to content

Resolve "Saving and reading states"

Closes #41 (closed)

  • result architecture is solved through composition, each result contains a ResultInfo object and an ExperimentConfig object
  • every result is automatically saved, together with a report.json that fully describes the experiment, material, machine, etc. (such that it can be repeated based on that and possible mesh data)
  • results are (almost completely) immutable and contain no mutable references
  • each result can have a label that can be assigned when running the experiment, e.g. `eigenmodes(..., label="backward_volume")```
  • added get_mode_local() to EigenResult
  • added AbsorptionResult and PerturbationResult
  • Perturbation is implemented (produces its own result) but does not store individual stiffness fields yet

Example


%matplotlib notebook
import matplotlib.pyplot as plt
from tetrax import Sample, geometries, experiments

layer = Sample(geometries.layer.monolayer(50,1), name="monolayer50nm")
layer.mag = (1, 0, 0)


Bext = 20e-3

layer.external_field = (Bext, 0, 0)
experiments.relax(layer)


interactions=[layer.get_interaction(name) for name in ["zeemann", "exchange"]]

exchange_spectrum = experiments.eigenmodes(layer, num_cpus=-1, interactions=interactions, label="exchange")
perturbed_spectrum = exchange_spectrum.perturb_with(layer.get_interaction("dipole"), label="dipole")
exact_spectrum = experiments.eigenmodes(layer, num_cpus=-1, label="exact")

fig, ax = plt.subplots()

for i in range(2):
    plt.plot(perturbed_spectrum.k,perturbed_spectrum.frequencies(i), ls="--")
    plt.plot(perturbed_spectrum.k,exact_spectrum.frequencies(i))

produces

disp

And the directory looks like

monolayer50nm/
    eigen_exact
        external_field.vtk
        eigenvalue_dataframe.csv
        mode_profiles
            ...
        equilibrium.vtk
        report.json 
    eigen_exchange
        eigenvalue_dataframe.csv
        mode_profiles
           ...
        equilibrium.vtk
        perturbation_dipole
            perturbed_frequency_dataframe.csv
            report.json
        external_field.vtk
        report.json
    references.bib
    relax
        energy_mag_log.csv
        final_state.vtk
        report.json
        external_field.vtk
        initial_state.vtk

for example, eigen_exchange/report.json looks like

{
    "result_type": "eigen",
    "config": {
        "km_limits": [
            null,
            -40000000.0,
            40000000.0,
            81,
            null,
            -5,
            5
        ],
        "save_modes": true,
        "optional_output": [],
        "parallel_backend": "loky",
        "num_cpus": 8,
        "frequency_tolerance": 0.001,
        "inverse_tolerance": 0.0001,
        "maxiter_inverse": 1000,
        "interactions": [
            "zeemann",
            "exchange"
        ],
        "_comment": "Toothless interactions (with relevant material parameters equal zero) are marked with '*'."
    },
    "outcome": {
        "has_profiles": true,
        "has_linewidths": false,
        "k (rad/m)": "-40000000.0, -39000000.0, ..., 39000000.0, 40000000.0",
        "m": "0.0",
        "modes_per_km": 20,
        "files": [
            "external_field.vtk",
            "equilibrium.vtk",
            "eigenvalue_dataframe.csv",
            "mode_profiles/"
        ]
    },
    "result_info": {
        "type": "eigen",
        "label": "exchange",
        "path": "monolayer50nm/eigen_exchange",
        "created": "2024-03-31T16:08:20.234266",
        "environment": {
            "tetrax_version": "2.0.0dev0",
            "platform": "Darwin",
            "cpu": "Intel(R) Core(TM) i5-8259U CPU @ 2.30GHz"
        },
        "sample_snapshot": {
            "name": "monolayer50nm",
            "geometry_type": "layer",
            "magnetic_order": "FM",
            "material": {
                "Msat": 796000.0,
                "Aex": 1.3e-11,
                "gamma": 176085964400.0,
                "alpha": 0.008,
                "Ku1": 0,
                "e_u": [
                    0.0,
                    0.0,
                    1.0
                ],
                "Dbulk": 0,
                "Didmi": 0,
                "e_d": [
                    0.0,
                    1.0,
                    0.0
                ],
                "Kc1": 0,
                "e_c1": [
                    1.0,
                    0.0,
                    0.0
                ],
                "e_c2": [
                    0.0,
                    1.0,
                    0.0
                ]
            }
        }
    }
}
Edited by Koerber, Lukas (FWIN-C) - 108045

Merge request reports

Loading