Doubts about Onion Doc (Trac #1107)

[sasview-bot]

quite generally:
to facilitate citation and discussion, I'd suggest to number equations
in doc pages that contain lots of them.

concerning http://marketplace.sasview.org/models/73:

The introduction announces that the SLD of each shell may be »described
by an exponential, linear, or constant function«. In the following,
however, only only the exponential and the constant case are explicitly
covered. For a linear SLD, the reader is left with the hint that
the exponential function with small A is a good approximation.

rho_shell(r) is defined in terms of six parameters. One of them is redundant
since rho_in=B+C. The parameters B and C have no obvious physical meaning,
and they do not appear in the parameter list of Iq. Two equation blocks later,
the paremeter B is defined a second time, now in terms of rho_in, rho_out, and A.
This makes the logic very obscure.

I would suggest that rho_shell(r) be defined in terms of the parameters
A, rho_in, rho_out, r_in, r_out. Then one would introduce Delta t_shell,
B, and C as abbreviations, which makes them categorically distinct from
the API parameters.

• Joachim

Migrated from http://trac.sasview.org/ticket/1107

{
    "status": "new",
    "changetime": "2019-03-28T23:55:42",
    "_ts": "2019-03-28 23:55:42.664130+00:00",
    "description": "quite generally:\nto facilitate citation and discussion, I'd suggest to number equations\nin doc pages that contain lots of them.\n\nconcerning http://marketplace.sasview.org/models/73:\n\nThe introduction announces that the SLD of each shell may be \u00bbdescribed\nby an exponential, linear, or constant function\u00ab. In the following,\nhowever, only only the exponential and the constant case are explicitly\ncovered. For a linear SLD, the reader is left with the hint that\nthe exponential function with small A is a good approximation.\n\nrho_shell(r) is defined in terms of six parameters. One of them is redundant\nsince rho_in=B+C. The parameters B and C have no obvious physical meaning,\nand they do not appear in the parameter list of Iq. Two equation blocks later,\nthe paremeter B is defined a second time, now in terms of rho_in, rho_out, and A.\nThis makes the logic very obscure.\n\nI would suggest that rho_shell(r) be defined in terms of the parameters\nA, rho_in, rho_out, r_in, r_out. Then one would introduce Delta t_shell,\nB, and C as abbreviations, which makes them categorically distinct from\nthe API parameters.\n\n- Joachim",
    "reporter": "Joachim Wuttke",
    "cc": "",
    "resolution": "",
    "workpackage": "SasModels Model Issues",
    "time": "2018-06-07T09:07:58",
    "component": "sasmodels",
    "summary": "Doubts about Onion Doc",
    "priority": "minor",
    "keywords": "",
    "milestone": "SasView 4.3.0",
    "owner": "",
    "type": "defect"
}

[RichardHeenan]

Trac update at 2018/06/08 14:21:23: richardh commented:

I did work out the equations for sld linear increasing or decreasing spherical shells, so one day we could add them properly.

Richard

[smk78]

Trac update at 2019/03/28 18:07:49: smk78 commented:

Joachim's comment about B & C not appearing in the parameter table is invalid because they are defined by parameters that do; as the docs clearly state.

But I have reworked the docs a little to try and make the different conditions a bit clearer.

His remaining comments are essentially suggesting a reparameterisation of fthe model which is not a documentation issue.

[smk78]

Trac update at 2019/03/28 18:09:00: smk78 changed workpackage from "SasView Bug Fixing" to "SasModels Model Issues"

[pkienzle]

Trac update at 2019/03/28 23:55:42: pkienzle commented:

There are equations for linear shells in spherical_sld:

https://github.com/SasView/sasmodels/blob/62dc8895694e8dcacf8cb65a82b543b1efa56f42/sasmodels/models/spherical_sld.c#L46

I suspect there is a Taylor series for low qr that improves accuracy and handles q=0.

Don't know if they are valid for large steps, or if they are only an approximation.