Glotaran Screencast

Screencast Transcript

Hello and welcome to this screencast showing the workings and some of the features of Glotaran, a graphical user interface for the R-package TIMP.

New Project

To start with, let us create a new project. Selecting "a new empty project", clicking next, we name it for instance "Demo" and press finish.

The project now opens in the project tab.

Now we can right click the dataset folder and click "open dataset file".
We locate a sample spectroscopy data file [ref] on our hard disk and press finish.
A new data file node is created under the dataset folder.
Double clicking this node opens the dataeditor for this particular type of data.

Data inspection

The dataeditor allows us to inspect the data by scrolling through it. It also allows us to zoom into various regions of the plot.
This makes it possible to select perhaps a only region of the dataset file for analysis.

Once we have selected a region we are happy with we press "create dataset", specify a name, and a new dataset node will appear as a subnode of the data file node.

Model Specification

Next up is the model specification. We expand the Models folder and open this new empty model.

We start by specifying the number of spectral components we anticipate to find in this dataset.
For this we will look at the singular value decomposition of the dataset.
For this particular dataset it looks like 1, 2, 3, 4 ... 3 components is quite feasible, so we will fill this value in into the model.

We will use some typical starting values for this kind of data, for instance 0.5, 0.05 and 0.01.
We want to use a sequential scheme for our analysis, so we will check the "Sequential analysis" checkbox.
And also we want to restrict these kinetic parameters to be positive, so we will check the "Set Kinetic Paramters Positive" checkbox.

Next is the Instrument Response Function (IRF).
Although it's possible to use a measured IRF here we will use a parameterized IRF function given by the position and width of a gaussian shaped IRF function.
These values we can estimate from our data by looking at the time zero for our excitation wavelength.

Another thing we can see from the data is the dispersion of the IRF.
We can model this by a simple polynomial function specified by 3 parameters starting with the lowest order and entered as comma separated numbers.
For this case "0.3, -0.1, 0.01" will do.

Although possible, for this analysis we won't weigh the data.

We would like to model the coherent artifact. For now only a coherent artifact modeled after the Instrument Response function is supported, so we will enable that.

Finally we save the model and now we are ready to start the analysis.

We drag our dataset into the dataset container and our model into the model container.
Now we can start the analysis by pressing the "Start analysis button from the toolbar.
We will be conservative and use only 1 iteration, then we wait.

Once the analysis has completed it will ask us for the name, and this name will then appear under the Results node.
Press double click to open.

Inspection of results

Here we can see the spectra.
The estimated parameters and total RMS value.
And if we scroll down, the singular value decomposition of the residual matrix.
If we enable a linear-logarithmic scaling we can better inspect the residuals.

On the traces tab we have the original dataset, overlay-ed with the dispersion curve.
By scrolling the sliders we can easily inspect the fitting at various places in the dataset.
Again we can switch to a linear logarithmic scale.
Finally we can take a snap shot of our analysis using the "Auto Select Traces" button.

The last step would be use the estimated parameters to update our model, we can do that by going back to the project tab and selecting our result node.
Then while pressing CTRL we drag the node onto our model node.
A pop-up dialog will appear asking us which parameters in the model we would like to update.
After pressing ok we can see that the corresponding values in the model have been updated with the values of the estimated parameters from the analysis.

With that we conclude this screencast. I hope you enjoyed it and are now able to use Glotaran for your own data analysis. If you find any bugs, or you find you are missing certain features, please contact us and we will get on it as soon as possible.