Molecule Types Linear Molecules <Prev Next>

Multiple State Fitting Example - The A-X transition in C3

In this example we will show some of the considerations required when fitting multiple states simultaneously, particularly with regards to the putting the line lists together for fitting. We start where the single state fit in Fitting example - the A-X transition in C3 finished by adding another state. The experimental spectrum does not seem to show the presence of another Q branch, suggesting a Σ-Σ transition should be added, rather than another Π-Σ transition like the first. If we assume the lower state is the same for both transitions, then this can easily be generated by right clicking on the upper state, selecting "Copy with linked items" and then "Paste". (There is more than one way to structure the addition of a state - see the "Different Manifolds" section below for a discussion of this.) Setting the symmetry of the newly created state to Σ+ and manually adjusting the origin of the new state to 27166.8 gives a promising simulation:


To obtain this appearance the colour of the new state has been set to "Navy" and the Π state to "Green"; the file is saved as  multc3fitinital.pgo, and then clicking the "Show Parts" () and "Show Sum" () buttons. Now, while the simulation is quite good, note that the indicating marks at the top show a very poor fit, and fitting with the old line list will fail. Deleting the line list and starting again would work fine, but there are better ways to do this which we we will now explore.

The origin of the problem introduced is best demonstrated by looking in the Energy Level Plot Window. Select "View, Levels", and for the best plot put the upper state rotational constant in the "Subtract" box, select the upper state manifold at "State/Manifold/Molecule", select line under "Plot Type" and both the buttons under "Observed". Pressing "All" should give a view like this:

If you have trouble setting this up, the settings are saved in multc3fitinital.pgo.This is a plot of energy against J, but with 0.408J(J+1) subtracted from the energy, so simple states will give horizontal lines. The two excited states are clearly visible on this plot as the green and blue horizontal lines, and the circles give the observed energy levels calculated from the line list by adding the calculated lower state energy to the observed transition energy. The vertical lines join the calculated to the observed energy levels, and it is clear from this that about half of the assignments are to the wrong energy level. This is because the line list constructed specifies states in terms of J, symmetry (+ or -) and eigenvalue number, and we have just added a state below the previous one, increasing the eigenvalue number. There are various ways of dealing with this and we work through several possibilities here, all of which work for this case, but may not in more complicated cases.

Different Manifolds

The first possibility is to put the new state in a different manifold from the initial Π state. The original line list will then work, as the manifold is specified for each state. The disadvantage of this is that perturbations between states can't be simulated, and states close together in energy as these are may well interact. The required set in is shown in the right hand side below, with the original set up on the left:

Setup with two states (Sigma, Pi) in a single excited state manifold (A)
Setup with each state in its own manifold (A for Pi, Asigma for Sigma)

Converting between the two is straightforward:

Note that a Transition Moment from the ground state is required for both states (here <Pi|T(1)|v=0> and <Sigma|T(1)|v=0>), and a separate Transition Moments object (here <A|mu|X> and <Asigma|mu|X>) is required for each connected pair of manifolds. When the name of a manifold is changed the line list will need adjustment using one of the methods described below.

Using a separate line list file

For the most flexible handling, the line list should be put in a separate text file and edited with a separate text editor, the Tabslave Editor included with PGOPHER or other text editor of your choice. Use of a separate editor is more flexible than editing in the line list window (undo/redo, search and replace are available, and comments can be added), and it is easy to combine separate sources of data into a combined fit. To generate a file use "More, Save to File" in the Line List Window, and then use the Log Window to select the file and fit. The choice of formats is significant, and all three options are explained below. - for the the next few sections "In Standard format" is assumed though, as discussed below, other choices may be better. 

Line List file in Simple Format

The "Simple" format uses the conventional quantum numbers to specify the transitions. Such a file can be generated from the Line List Window using "More, Save to File, In Simple Format" and choosing a suitable file name, say APisimple.lin.  Now go to the Log Window, check than the file you have just created is shown and press "Edit" to edit the file. A line:

UpperState Pi

needs to be added to the start of the file to identify which of the two possible upper vibronic states is involved in the observed transition. The file then needs to be saved - if you use the Tabslave Editor included with PGOPHER this step is not necessary as it a save is forced on each PGOPHER fit or observation load. The file should look like this:

NQN 2
UpperState Pi
1 e 0 e 27179.774 1 .03172 : LDS751.dr b - C3AX.ovr
3 e 2 e 27181.277 1 .05494 : LDS751.dr b - C3AX.ovr
5 e 4 e 27182.625 1 .0589 : LDS751.dr b - C3AX.ovr
...
You can check this is right by hitting the reload button in the Energy Level Plot Window and the upper state should now show no long vertical lines, indicating a correct assignment:
    Only two of many possible directives have been used here. Others include defaultmolecule, uppermanifold, lowermanifold or lowerstate which might also be required if there is more than one possibility for each. For a complete list of directives Line List Input File Format.
    The quantum numbers here are J' parity' J" parity"; see  Simple Format for the possible formats of the quantum numbers. Note than NQN, the number of quantum numbers present for each state, must be specified. The value after the observed frequency (all 1 here) is the relative estimated standard deviation of the observation. The remainder of the line is only a comment; the procedure above puts a calculated intensity and source of the observation in this by default, but it can be left blank or other text added.

Branch Format

An alternative way of specifying transitions is to use traditional branch notation, such as P(3), implying J' = 2, J" = 3. Such a file can be generated from the Line List Window using "More, Save to File, In Branch Format". The resulting file (in APibranch.lin) looks like this:

UpperState Pi
R(0) 27179.774 1 .03172 : LDS751.dr b - C3AX.ovr
R(2) 27181.277 1 .05494 : LDS751.dr b - C3AX.ovr
R(4) 27182.625 1 .0589 : LDS751.dr b - C3AX.ovr
...
As for the simple format above, the "UpperState Pi" is required to identify which of the two possible upper vibronic states is required. The branch needs to appear without spaces at the beginning of the line - see Branch Format for more details. The rest of the line is as for simple format.

State Renumbering

The final option in fixing the line list is simply to change the eigenvalue numbers that are wrong in the new simulation set up. In the line list window this will be the #' column, and in this case the values for the odd J levels need increasing by 1. This change could be done manually, but a further change may need making if a state is added or removed, so we use the indexoffsets directive. To use this we put the line list in a separate file. Select "More, Save to File, In Standard Format" in the Line List Window window and choose a suitable filename, say APi.lin. To add the required offsets add:

upperindexoffsets 1 0

at the start of the file and then save the changed file. You can check this is right by hitting the reload button in the Energy Level Plot Window and the upper state should now show no long vertical lines, indicating a correct assignment:

State Searching

As an alternative the quantum numbers given in the comments at the end of each line can also be used to work out the eigenvalue number . In the Line List Window this can be forced for an individual line by clearing the eigenvalue number (#' and #") columns or in a line list file by setting the eigenvalue number to 0. The symmetry (S', S") can also be similarly left unset, though a negative value, rather than 0 must be used in the line list file to flag this. To force this for the entire file use:

upperindexoffsets search

to work out the eigenvalue numbers or

upperindexoffsets searchall

to work out both symmetry and eigenvalue number. Alternatively:

UpperState Pi

can be used to indicate that the eigenvalue numbers are to be taken within the state. Any of these methods will work fine in many cases, though where the quantum numbers are ambiguous, typically in the presence of strong state mixing, the quantum numbers assigned can change as the parameters change. In fitting the quantum number assignments for any given state can flip between successive fit cycles.

Notes on Standard Line List Format

A few lines of the line list file are shown below; see Line List Input File Format for a detailed explanation of each field.

LinearMolecule   A   1  1  1   X   0  0  1    27179.774 1 .03172      0 :     R(0) : A Pi  1 e - X v=0  0 e : LDS751.dr b - C3AX.ovr
LinearMolecule A 3 1 1 X 2 0 1 27181.277 1 .05494 0 : R(2) : A Pi 3 e - X v=0 2 e : LDS751.dr b - C3AX.ovr
LinearMolecule A 5 1 1 X 4 0 1 27182.625 1 .0589 0 : R(4) : A Pi 5 e - X v=0 4 e : LDS751.dr b - C3AX.ovr
...

Notes on this file:

UpperManifold A
1 1 1 0 0 1 27179.774 1
3 1 1 2 0 1 27181.277 1
5 1 1 4 0 1 27182.625 1
...

Multiple State Fit

With the Π state line list saved to a separate file, the line list window can be cleared and an independent fit of the Σ-Σ transition can be performed using the standard method. If trying this for yourself, you will need to fix all the Π state parameters, and remove q from the Σ state. When right clicking on the simulation you may end up selecting both Σ and Π state lines; you will have to check for any Π state lines manually and delete them. The resulting file is available in multc3fitsigma.pgo. A fit to both states simultaneously can now be done by saving the line list for the Σ state to a separate file in, say simple format (ASigma.lin) looking like this:
NQN 2
1 e 2 e 27164.765 1
3 e 4 e 27163.026 1
...

To use this file with one of the previous Π state files set up a separate file, say Aboth.lin, with the following format:

Colour Green
Include APisimple.lin

Colour Navy
UpperState Sigma
Include ASigma.lin

The include statements simply read the contents of the given file. All the information could be put in one file, or split in any other way as convenient. The colour directives set the colour of the points in the Residuals Window and the indicating marks at the top of the main window. The final fit gives an excellent overall simulation: