Curry 2.0

[Stefan Stoll]

We are planning a major rewrite of curry for the next major ES release (summer 2018). This will not be backwards-compatible with the current release (5.2). The goal is to make this function more flexible and simpler to use.

If you are an ES user that is working heavily with SQUID magnetometry data, we would like to have your feedback about the following points:

1.
One of the principal planned changes will be that you can directly specify the unit system that should be used: Opt.Units = 'SI' (SI units; default) or Opt.Units = 'CGS' (CGS-emu units). Do you think this is useful?

2.
Another change will be that the outputs of chili can be completely customized. For example, Opt.Output = 'chimolT mumol' tells curry to return the molar susceptibility times temperature as first output, and the molar magnetic moment as second output. Currently, we are envisioning the following possible outputs:

• 'mu': single-center magnetic moment (SI unit: J/T; CGS-emu unit: erg/G or emu)
• 'muBM': single-center magnetic moment in Bohr magnetons (unitless and numerically identical in SI and CGS-emu)
• 'mumol': molar magnetic moment (SI unit: J/T/mol; CGS-emu unit: erg/G/mol or emu/mol)
• 'chi': single-center magnetic susceptibility (SI unit: m^3; CGS-emu unit: cm^3)
• 'chimol': molar magnetic susceptibility (SI unit: m^3/mol; CGS-emu unit: cm^3/mol)
• 'chimolT': molar magnetic susceptibility times temperature (SI unit: K m^3/mol; CGS-emu unit: K cm^3/mol)
• 'mueff': effective Bohr magneton number, also called effective magnetic moment (unitless and numerically identical in both SI and CGS-emu)

Do you think we should add anything?

3.
Currently, curry calculates the differential magnetic susceptibility chi = dmu/dB, but in experimental practice the susceptibility is often approximated as mu/B, for some value of B. Should curry also provide capabilities for this approximate susceptibility? How?

[thanasis]

Dear Stefan,

I personally welcome this overhaul of curry. Regarding your questions:

1. This is a useful feature, although with a simple line users should be able to modify the input/output to the desired units.
But since you are making the effort, you might consider allowing users to define both the input AND output units (e.g. input as cgs, output as SI).

2. You might want to add:
-1/chi, which also is used some times.
-reduced magnetisation (M/T), which is also a useful feature for isothermal magnetisation experiments

3. I have come to notice that the two approaches do not yield identical results, in particular for strong fields. Since experimental data are obtained as emu and the transformed to chi by dividing by the field, probably the latter approach could be the default. Although you might want to allow users to switch to dmu/dB when required.

4. Also, as an experimentalist, I wish there was support for fits of multiple data sets. For instance, I outline a problem I am dealing with in a previous post (viewtopic.php?f=3&t=446). It would be great if two data files could be treated simultaneously. E.g.:
-sus.dat for isofield chi vs T experiments, with specification (T, B, chi). There, more than one B should be handled.
-mag.dat for isothermal M vs B experiments, with specification (T, B, mag). Similarly, more than one T should be handled.

5. Another request would be the treatment of more than one systems. E.g. paramagnetic impurities are the first case that comes to mind, but ES could treat multiple systems in a very powerful way by extending its approach for EPR experiments. Many packages for magn. susceptibility simulations just request the spin of the impurity and fit the percentage byt treating it within the simple Curie law framework. However, if the field is strong, and/or the percentage is high, and/or there is strong zfs, this approach is deficient. ES could make a significant contribution there.

Thanks for the good work!

[chemshd]

if the curry could include the fitting of the SH parameters such as coupling constant, ZFS parameters based on MH or MT data, it would be excellent!!
what I am doing now is to write my own fitting function based on easyspin, simply minimizing the different between simulation results and the simulation ones.

[Stefan Stoll]

You might want to look into esfit, which can be used with curry.

[esalerno]

Hello,

I was wondering if you had an ETA for this release, as I would like to start using the program but would prefer to wait for the new release before doing so.

Thanks!