pyforc

FORC analysis in Python.

Installation

Install from PyPI

The easiest way to get started is by installing via pip:

pip install pyforc

This will grab the latest published release on the Python Package Index and install it to your current python environment.

Installation from source

Install from source by doing

pip install git+https://github.com/peytondmurray/pyforc

Alternatively you can clone this repo and run

pip install .

Contributions

Contributions are welcome - open an issue or create a pull request. I'm trying to stick to PEP8 as much as I can, except I'm using line lengths of 100 characters. I'm using numpydoc formatting for the documentation as well. Don't worry too much about this stuff though, we can work together to integrate your code.

Pre-commit hooks

This project makes use of pre-commit hooks for linting and style checking. If you haven't used pre-commit hooks before, first install pre commit:

pip install pre-commit

Then inside the repository install the hooks themselves:

pre-commit install

Now, pre-commit hooks will run automatically any time you type git commit.

Usage

Let's analyze a FORC dataset. Currently pyforc support data in from Princeton Measurement Corporation (PMC) vibrating sample magnetometers (VSM). The data comes in two formats.

Hc/Hb Format

The first component is a header containing metadata about the dataset, including measurement configuration details, temperature, number of curves and data points.

The reversal curves come after this, with each curve containing

1. A single drift measurement, which is a measurement of the magnetization carried out at some magnetic field $H$ above the field at which the magnet is known to be saturated, $H_{sat}$.
2. A series of magnetization measurements at a monotonically increasing magnetic field $H$, with a constant¹ spacing between data points.

MicroMag 2900/3900 Data File (Series 0015)
First-order reversal curves
Configuration   :  VSM
Hardware version:  0004
Software version:  09/27/2005 D
Units of measure:  cgs
Temperature in  :  Kelvin
01/22/2015  17:35
"Delaware FeCo D 20nm wires OOP FORC"

Averaging time = +2.000000E+00
Hb1            =  0.000000E+00
Hb2            = +2.500000E+03
Hc1            = +1.500000E+03
Hc2            = +4.000000E+03
HCal           = +6.804196E+03
HNcr           = +3.500000E+01
HSat           = +1.500000E+04
NCrv           = 152
PauseCal       = +1.000000E+00
PauseNtl       = +3.000000E+00
PauseSat       = +1.000000E+00
SlewRate       = +1.000000E+04
Smoothing      = 3

Field range    = +6.804196E+03
Moment range   = +2.000000E-03
Temperature    = N/A
Orientation    = -9.000001E+01
Elapsed time   = +7.236108E+04
Slope corr.    = N/A
Saturation     = N/A
NData          = 30628

+6.808018E+03,+1.282651E-03            <---- Drift measurement

+1.140678E+03,+1.220527E-03            <---- Start of a reversal curve
+1.174292E+03,+1.220665E-03
+1.211422E+03,+1.221466E-03
+1.244758E+03,+1.222031E-03
+1.281733E+03,+1.221932E-03
+1.315307E+03,+1.222621E-03
...

H/Hr Format

Other FORC software

• FORCinel: Uses LOESS to caclulate the FORC distribution. Freeware, but requires IGOR Pro, so I've never used it, and I don't know how it handles the reversible ridge. Smoothing is done by increasing the LOESS radius.
• Quantum Design has FORC software that can calculate the distribution in real time, apparently. I've never used this as I don't have a PPMS, and there's no publicly available source, so there's no way to know how the distribution is calculated, how the reversible ridge is handled, or how smoothing is done.
• FORCIT (originally hosted at the UC Davis paleomag group site which is no longer hosted). Uses a finite-difference method for calculating the FORC distribution, and requires the Generic Mapping Tools to run. This can take an extremely long time to run (~15 min was not uncommon in my lab, despite being written in Fortran). Handles the reversible ridge by using a constant extension for $H<Hr$. Smoothing is done by applying a 2D gaussian filter before differentiation.
• Chris Pike's original Mathematica script. Nigh unusable at this point, this is the original Mathematica script Chris Pike used to calculate FORC distributions.

Footnotes

1. Not actually constant, but we account for this in the data processing by interpolating the dataset over a constant field step. ↩