mx3-lab

# Simulating a measurement **mx3-match** compares against data you already took. This is the other direction: the measurement *protocol* itself, turned into a field sequence. Getting the sequence wrong makes the comparison meaningless however good the physics is — a minor loop is not a major loop with a smaller range, and a switching field at 45 degrees is not the same experiment as one at 0. ```bash P=${CLAUDE_PLUGIN_ROOT}/skills/mx3-lab/scripts/protocol.py python3 $P major --Ms 800e3 --A 13e-12 --size 200e-9 --thick 20e-9 --out loop.mx3 python3 $P minor ... --Breverse 25 --out minor.mx3 python3 $P astroid ... --angles 0 15 30 45 60 75 90 --out astroid.mx3 python3 $P fmr ... --Bbias 200 --out fmr.mx3 ``` Fields are given in **mT** and angles in **degrees**, because that is what the instrument reads. The script converts to SI. Always gate the result before running it: ```bash ${CLAUDE_PLUGIN_ROOT}/skills/mx3-authoring/scripts/vet.sh loop.mx3 ``` ## The four protocols | | What it is | Why it is separate | |---|---|---| | `major` | saturate, sweep down, sweep back | the baseline VSM/MOKE loop | | `minor` | turn round at `--Breverse`, before saturation | the loop does **not** close; it probes the reversal in progress, and the remanent state depends on where you turned | | `astroid` | one loop per field angle | a far stronger test of shape and reversal mode than coercivity alone, and much less sensitive to defects | | `fmr` | bias field, small transverse step, record the decay | frequencies come from the spectrum of the ringdown | `astroid` writes a template with the angle on one line. Generate one file per angle and batch them: ```bash for a in 0 15 30 45 60 75 90; do sed "s/^angle_deg := .*/angle_deg := $a/" astroid.mx3 > astroid_$a.mx3 done ${CLAUDE_PLUGIN_ROOT}/lib/mx3 batch astroid_*.mx3 --jobs 3 ``` ## Reading the result ```bash mx3 loop sweep.out --field B_extx --moment mx # Hc, Mr/Ms, squareness mx3 view sweep.out --mode kerr # what a polar MOKE would see ``` For the FMR ringdown, the frequencies are in the spectrum of the recorded trace — `observe.peak_frequency` in the shared library, and compare against `mx3 physics` for the Kittel prediction. ## What this cannot mirror Say these plainly when reporting, because they are why a simulated loop and a measured one differ even when everything else is right: - **Sweep rate.** A quasi-static simulation relaxes fully at each step; a real sweep does not. Thermally activated reversal is rate-dependent and is not modelled at all at zero temperature. - **Defects and edge roughness.** A clean simulated element reverses coherently and overestimates coercivity, often several-fold. See `mx3-match`. - **Temperature.** Unless `Temp` is set, this is a 0 K measurement. - **Ensemble.** A VSM measures many elements at once; this is one. Shape, remanence and angular dependence transfer well. Absolute coercivity does not. ## Related skills - **mx3-match** — compare the result against the measured data - **mx3-run** — run it and extract the loop metrics - **mx3-view** — render it the way the microscope would show it - **mx3-plan** — if the material regime is not settled yet