Class Experiment

Return tuples representing all the scale factors in this experiment.

A tuple will contain multiple entries if several variables share a
scale factor.

Set the type of prior to place on a given group of scalefactors.

The group contains a collection of variables that are sharing a given
scale factor which may be just one variable. You can see what the
current groups are with expt.get_sf_groups().

Currently implemented prior types are:
    'uniform in sf': This is a uniform prior over scale factors. This
    is simplest and fastest to compute, but it tends to weight
    parameter sets that yield large scale factors heavily. It takes no
    parameters.

    'gaussian in log sf': This is a Gaussian prior over the logarithm
    of the scale factor. This should avoid the problem of weighting
    large factors heavily. It takes two parameters: the mean of the
    normal distribution, and it's standard deviation. For example,
    parameters (log(3.0), log(10)), will place a prior that holds 95%
    of the probability between 3 / 10**2 and 3 * 10**2.

Constrain the period of the oscillations to a value (period)
with the error (sigma). The period is found using the maximum
to maximum distance of the first two maxima found between
startTime and two periods after the startTime.

Turn on applying a constraint that the integrated
area in two different parts of the plot should be the
same. startTime and testTime are the starting points to
begin the integration for the period-long each.

Add an integral data set to the experiment.

calcKey      The id of the calculation this data corresponds to
traj         The trajectory to compare against
uncert_traj  A trajectory of data uncertainties
vars         What variables to fit against
interval     The time interval to fit over, defaults to the entire traj