Single peak and multi BG (multiple workspace) Fitting

[Nick Draper]

Need to get more detail from Toby on this but this appears to be when you have several datasets and you want to optimise a single peak on each simultaneously while allowing separate background functions.

[Nick Draper]

Bulk move of tickets to iteration 27, if your ticket is essential for Iteration 26 then move it back.

[Nick Draper]

Bulk move of tickets at the end of iteration 27

[Nick Draper]

Bulk move of tickets at the end of iteration 28

[Nick Draper]

"New" tickets moved at the code freeze of iteration 29

[Roman Tolchenov]

In [14319]:

Added MultiBG function for doing a multispectral fit. Updated FitPropertyBrowser to be able to use it. re #966

[Roman Tolchenov]

In [14320]:

Trying to make rhel5 and macs happy. re #966

[Roman Tolchenov]

In [14327]:

Trying to fix the tests. re #966

[Roman Tolchenov]

In [14410]:

Implemented derivative calculation for MultiBG, changed many int to size_t in functions. re #966. Removed references to Workspace2D from API/CompositFunctionTest and API/FunctionTest, fixes #3653.

[Roman Tolchenov]

In [14411]:

Fixing the linux. re #966

[Roman Tolchenov]

In [14416]:

Hide the workspace index property if a function is fitted to all workspaces, re #966

[Mathieu Doucet]

In [14430]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14431]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14432]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14433]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14435]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14437]:

Fixing warnings Re #966 and #3653

[Mathieu Doucet]

In [14439]:

Fixing warnings Re #966 and #3653

[Roman Tolchenov]

In [14443]:

Fixed parameter tying in MultiBG. re #966

[Mathieu Doucet]

In [14448]:

Fixing warnings Re #966 and #3653

[Roman Tolchenov]

In [14521]:

Added a dialog for easier function creation. re #966

[Roman Tolchenov]

In [14522]:

Fixing the test. re #966

[Mathieu Doucet]

In [14559]:

Fix warnings... Re #966

[Mathieu Doucet]

In [14560]:

Fix warnings... Re #966

[Russell Taylor]

In [14641]:

Remove member variable (and calls to it) as it wasn't doing anything anyway and was causing Mac compiler warnings. Re #966.

[Russell Taylor]

In [14642]:

Clear Mac compiler warnings. Re #966.

[Vickie Lynch]

Can't get this to work

[Roman Tolchenov]

To testers. It's easy:

1. Load ~10 spectra from MUSR00015189.nxs
2. Rebin to MUSR00015189_1 0,0.1,29
3. Open the fit browser in MantidPlot.
4. Select MUSR00015189_1 for fitting.
5. Change function type from CompositeFunctionMW to MultiBG
6. Add ExpDecay to Functions.
7. Set the initial values: Height=400, Lifetime=2
8. Right click on Functions and select Setup multifit
9. Check the box against Lifetime in the dialog. This parameter will be global - the same for all spectra
10. Press OK. The fit browser will be populated with copies of ExpDecay, one for each spectrum in the WS. All Lifetimes will be tied to f0.Lifetime.
11. Click on Fit. Created output: MUSR00015189_1_NormalisedCovarianceMatrix, MUSR00015189_1_Parameters, MUSR00015189_1_Workspace, MUSR00015189_1_Param_series.

MUSR00015189_1_Workspace is a WorkspaceGroup, each member is similar to an output from a simple Fit

[Martyn Gigg]

Observed a crash after following the above instructions which is related to the the algorithm notifications and threading issues.

[Roman Tolchenov]

In [14845]:

Process multifit results in gui thread. fixes #966

[Martyn Gigg]

Tried the same again and that looks much better and more responsive.

Also tried with larger and smaller data sets and it looked good.

I saw another crash though:

• If you perform steps 1-5
• Click addFunction and select MultiBG as the function type instead of ExpDecay
• When you click okay you get a crash. Unfortunately it is the default so this is probably worth looking at. The stack trace is below. Basically it seems to be a problem in Mantid::CurveFitting::MultiBG::getWorkspace

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff6d71b04 in shared_ptr<Mantid::API::MatrixWorkspace const> (this=0x7fffffffbf20, r=...)
    at /usr/include/boost/smart_ptr/shared_ptr.hpp:238
238	    : px( r.px ), pn( r.pn ) // never throws
(gdb) bt
#0  0x00007ffff6d71b04 in shared_ptr<Mantid::API::MatrixWorkspace const> (this=0x7fffffffbf20, r=...)
    at /usr/include/boost/smart_ptr/shared_ptr.hpp:238
#1  0x00007fffd56f5961 in Mantid::CurveFitting::MultiBG::getWorkspace (this=0x279b480)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/Framework/CurveFitting/inc/MantidCurveFitting/MultiBG.h:62
#2  0x00007ffff65dccf4 in MantidQt::MantidWidgets::PropertyHandler::initWorkspace (this=0x29d45a0)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/MantidQt/MantidWidgets/src/PropertyHandler.cpp:311
#3  0x00007ffff65db9cd in MantidQt::MantidWidgets::PropertyHandler::init (this=0x29d45a0)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/MantidQt/MantidWidgets/src/PropertyHandler.cpp:127
#4  0x00007ffff6d5f683 in Mantid::API::IFitFunction::setHandler (this=0x279b590, handler=0x29d45b0)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/Framework/API/src/IFitFunction.cpp:261
#5  0x00007ffff65dd927 in MantidQt::MantidWidgets::PropertyHandler::addFunction (this=0x174aa00, fnName=...)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/MantidQt/MantidWidgets/src/PropertyHandler.cpp:435
#6  0x00007ffff65b2881 in MantidQt::MantidWidgets::FitPropertyBrowser::addFunction (this=0x1442040)
    at /home/dmn58364/mantidproject/svn/trunk/Code/Mantid/MantidQt/MantidWidgets/src/FitPropertyBrowser.cpp:539

[Roman Tolchenov]

In [14848]:

Removed MultiBG from the function lists. fixes #966

[Martyn Gigg]

Confirmed that the function has been removed, as it wasn't supposed to be there anyway, so the crash can no longer occur.

[Stuart Campbell]

This ticket has been transferred to github issue ​1814