Saving and loading a demagnetized collection

Demagnetization can be expensive to compute. Once the result is obtained it is useful to persist the demagnetized state to disk so that further field analyses can be run later without repeating the simulation.

to_json serializes any supported magpylib collection to a JSON string; from_json reconstructs the objects from that string. All geometry (position, orientation, paths), magnetic polarizations and susceptibilities are preserved exactly.

Define sources, mesh and apply demagnetization

from pathlib import Path
import time

import magpylib as magpy
import numpy as np
from magpylib_material_response import to_json
from magpylib_material_response.demag import apply_demag
from magpylib_material_response.meshing import mesh_all

magpy.defaults.display.backend = "plotly"

# soft magnet in the centre
cube = magpy.magnet.Cuboid(polarization=(0, 0, 0), dimension=(0.002, 0.002, 0.004))
cube.susceptibility = 3999
cube.style.label = "Iron core"

# current loop around it
loop = magpy.current.Circle(current=10, diameter=0.003)
loop.style.label = "Current loop"

coll = magpy.Collection(cube, loop, style_label="setup")

# mesh and apply demagnetization — clock this
t0 = time.perf_counter()
coll_demag = mesh_all(coll, target_elems=200)
apply_demag(coll_demag, inplace=True)
t_demag = time.perf_counter() - t0
print(f"Demag computation : {t_demag * 1e3:.0f} ms")

Demag computation : 1482 ms

Save to disk

json_path = Path("demag_collection.json")
json_path.write_text(to_json(coll_demag, indent=2))
print(f"Saved {json_path.stat().st_size / 1024:.1f} kB → {json_path.name}")

Saved 235.0 kB → demag_collection.json

Reload and display

import time

from magpylib_material_response import from_json

# reload — no simulation needed
t0 = time.perf_counter()
coll_loaded = from_json(json_path.read_text())[0]
t_load = time.perf_counter() - t0

print(f"Load from JSON : {t_load * 1e3:.0f} ms")
print(f"Full demag     : {t_demag * 1e3:.0f} ms  ({t_demag / t_load:.0f}× slower)")

magpy.show(*coll_loaded)

Load from JSON : 40 ms
Full demag     : 1482 ms  (38× slower)