Ctrl+K

Saving Your Work (working with classes)#

When you have completed your analysis, usually uou will want to save a copy of your work for subsequent. Since everything about your model and results are contained is their own classes (actually a struct for ‘results’), this is very easy. But, there are a couple of pitfalls to be aware of..

Saving to a file#

Suppose you have a model defined in problem, and you have run this through RAT and you want to save your work for later…

RAT input model

So ‘problem’ now contains a copy of the input problem, and results the relevant reflectivities, SLD’s and so on. Because everything is in these two outputs, then saving your work is just a matter of saving these to file. It is good practice to keep everything together, and a good way of doing this is to collect them into a struct. Then, you only need to save the struct to a file to save all of your work:

myResults = struct('problem', problem, 'results', results, 'controls', controls);
save('myResultsFile', 'myResults');

TODO

At a later date, you only need to load back in your struct, split it up into it is components and away you go:

myWork = load('myResultsFile');
myWork.myResults

TODO

myResults =

struct with fields:

    problem: [1x1 projectClass]
    results: [1x1 struct]
    controls: [1x1 controlsClass]

TODO

Copying a project class#

During an analysis, it may be necessary to make a copy of your project, so that you can modify one of them in order to carry out some kind of comparison between them for example. In the example below, problem1 and problem1 are references to the same instance of project class so modifying problem2 will also modify original object problem1 which is not ideal.

>> problem1 = projectClass();
>> problem2 = problem1;
>> problem1.geometry

ans =

    'air/substrate'

>> problem2.setGeometry('substrate/liquid');
>> problem1.geometry

ans =

    'substrate/liquid'

>>> problem1 = RAT.Project()
>>> problem2 = problem1
>>> print(problem1.geometry)

air/substrate

>>> problem2.geometry = "substrate/liquid"
>>> print(problem1.geometry)

substrate/liquid

The proper way to make a copy/clone of the project class is shown in the example below,

>> problem1 = projectClass();
>> problem2 = problem1.clone(); % Copy with clone method
>> problem1.geometry

ans =

    'air/substrate'

>> problem2.setGeometry('substrate/liquid');
>> problem1.geometry

ans =

    'air/substrate'

>>> import copy
>>> problem1 = RAT.Project()
>>> problem2 = copy.deepcopy(problem1) # Copy using deepcopy function in the copy module
>>> print(problem1.geometry)

air/substrate

>>> problem2.geometry = "substrate/liquid"
>>> print(pproblem1.geometry)

air/substrate

Now problem1 and problem2 are seperate instances of project class and changing problem2 no longer changes problem1.

Exporting as a Script#

Although saving a binary version of the class is useful, sometimes it would be better to have a script version which will reproduce the class. This can be done as shown below:

problem = projectClass();
problem.writeScript(script="myProjectScript");

problem = RAT.Project()
problem.write_script(script='myProjectScript')

Then, RAT will create a file containing all the statements needed to re-create your project:

% THIS FILE IS GENERATED FROM RAT VIA THE "WRITESCRIPT" ROUTINE. IT IS NOT PART OF THE RAT CODE.

project = createProject(name='', calcType='non polarised', model='standard layers', geometry='air/substrate', absorption=false);

project.setParameterValue(1, 20);
project.setParameterLimits(1, 1, 5);
project.setParameterFit(1, true);
project.setParameterPrior(1, 'uniform', 0, Inf);


project.removeBulkIn(1);
project.addBulkIn('SLD Air', 0, 0, 0, false, 'uniform', 0, Inf);

project.removeBulkOut(1);
project.addBulkOut('SLD D2O', 6.2e-06, 6.35e-06, 6.35e-06, false, 'uniform', 0, Inf);

project.removeScalefactor(1);
project.addScalefactor('Scalefactor 1', 0.02, 0.23, 0.25, false, 'uniform', 0, Inf);

project.removeQzshift(1);
project.addQzshift('Qz shift 1', -0.0001, 0, 0.0001, false, 'uniform', 0, Inf);

project.removeBackgroundParam(1);
project.addBackgroundParam('Background Param 1', 1e-07, 1e-06, 1e-05, false, 'uniform', 0, Inf);

project.removeResolutionParam(1);
project.addResolutionParam('Resolution par 1', 0.01, 0.03, 0.05, false, 'uniform', 0, Inf);

project.removeBackground(1);
project.removeResolution(1);

project.addBackground('Background 1', 'constant', 'Background Param 1', '', '', '', '');

project.addResolution('Resolution 1', 'constant', 'Resolution par 1', '', '', '', '');

project.removeData(1);
project.addData('Simulation');
project.setData(1, 'simRange', [0.005 0.7]);

# THIS FILE IS GENERATED FROM RAT VIA THE "WRITE_SCRIPT" ROUTINE. IT IS NOT PART OF THE RAT CODE.

import RAT
from RAT.models import *
from numpy import array, inf

problem = RAT.Project(
    name='', calculation='non polarised', model='standard layers', geometry='air/substrate', absorption=False,
    parameters=RAT.ClassList([ProtectedRAT.models.Parameter(name='Substrate Roughness', min=1.0, value=3.0, max=5.0, fit=True, prior_type='uniform', mu=0.0, sigma=inf)]),
    background_parameters=RAT.ClassList([RAT.models.Parameter(name='Background Param 1', min=1e-07, value=1e-06, max=1e-05, fit=False, prior_type='uniform', mu=0.0, sigma=inf)]),
    scalefactors=RAT.ClassList([RAT.models.Parameter(name='Scalefactor 1', min=0.02, value=0.23, max=0.25, fit=False, prior_type='uniform', mu=0.0, sigma=inf)]),
    bulk_in=RAT.ClassList([RAT.models.Parameter(name='SLD Air', min=0.0, value=0.0, max=0.0, fit=False, prior_type='uniform', mu=0.0, sigma=inf)]),
    bulk_out=RAT.ClassList([RAT.models.Parameter(name='SLD D2O', min=6.2e-06, value=6.35e-06, max=6.35e-06, fit=False, prior_type='uniform', mu=0.0, sigma=inf)]),
    resolution_parameters=RAT.ClassList([RAT.models.Parameter(name='Resolution Param 1', min=0.01, value=0.03, max=0.05, fit=False, prior_type='uniform', mu=0.0, sigma=inf)]),
    backgrounds=RAT.ClassList([Background(name='Background 1', type='constant', value_1='Background Param 1', value_2='', value_3='', value_4='', value_5='')]),
    resolutions=RAT.ClassList([Resolution(name='Resolution 1', type='constant', value_1='Resolution Param 1', value_2='', value_3='', value_4='', value_5='')]),
    data=RAT.ClassList([Data(name='Simulation')]),
    )

This is useful because you can then edit this file as you wish, to re-use it as a template for further projects.

On this page