Description: Created using information available in the FEPS database, the Figure below illustrates the range in empirical age diagnostic as a function of B-V color.
Findings:
- X-ray and probably lithium saturation are apparent.
- We have a few apparently very red stars near the main sequence. If these have had their distances assigned using spectroscopic parallax, someone should cry foul. If they are not pre-ms stars they should not be on our sample as the implied mass is too low.
- For some stars, the fiducial stellar parameters still need revisiting (this is not apparent from the figure perhaps, but a fact).
Description: Age indicators as a function of activity (R'HK) and rotation (vsini).
Findings:
- The empirical age indicators are generally well correlated with one another.
- The dispersion highlights the astrophysical scatter plus the color effects (previous figure).
- Outliers, for example in the Rx vs RHK plot may indicate problem stars.
Description: From application of calibration equations that are tied to open clusters, ages for individual stars can be derived. Shown on the right are the calibrations for 1.0 Msun stars, and on the left the histograms of inferred ages for FEPS stars from the different age diagnostics. Note that not all stars appear in all panels due to lack of data.
Findings:
- Nothing physical should be interpreted from comparison of the histograms. The Figure merely illustrates the overlap in different age ranges of the available age diagnostic data.
- Here is a table with the data. It contains the ages resulting from each different technique plus columns giving the average and sigma among the different age techniques (leaving out the HR diagram ages since this method is only valid for bona fide pre-ms stars) as well as a "best" age which is simply one assigned according to the hierarchy:
We had discussed a switch to use the R'HK age only if vsini < 15km/s and R'HK < -4.4 but that has not yet made it into my code. I am not convinced that "best" ages are better than averaged ages; see below. There are appoximately 33 stars for which the averages can achive lower dispersion through rejection of >1.5 sigma outliers. In the vast majority of cases, the difference in age between straight averaging and rejection averaging is <0.15 dex so I have not implemented this.
- cluster (114 stars)
- hrd if younger than 20 myr and this is supported by another activity/rotation/lithium diagnostic (33 stars)
- period (77)
- R'HK (88)
- lithium (16)
- xray (0)
- vsini (0)
Description: Scatter plots showing the correlation or lack thereof of each age technique with every other technique. Note that the plots appear here twice, as mirror reflections. Blue, black, and red circles indicate bluer, average, and redder B-V colors based on interquartile ranges.
Findings:
- The HR diagram ages are saturated at 20-30 Myr when ~1 Msun stars hit the main sequence. So ignore the linear feature containing most of data points. At the young end, HR diagram ages seem to correlate well with x-ray ages and perhaps lithium ages. Otherwise they seem pretty independent of other activity/rotation diagnostics, as noted extensively in the literature.
- Other than the HRD panels, overall, the calibrations seem pretty consistent, i.e. reasonably well cross calibrated with some exceptions in certain age ranges. No dramatic offsets are visible by eye.
- The most salient feature is the scatter.
- The x-ray and lithium correlation and the x-ray and vsini correlation have relatively low scatter compared to the other panels.
- Period and vsini ages are also well correlated, though with fewer data.
- Color effects are still present in several of the calibrations.
Description: Dispersion among the different age techniques as a function of the average of these techniques (top right panel) or an assigned "best" technique (bottom left panel). A histogram of the dispersions is shown in the upper right. This histogram is repeated in linear units (middle right) and as a percentage of the age (lower right). Only activity/rotation/lithium ages are included in the averages, i.e. HR diagram ages are excluded.
Findings:
- The dispersion is higher for younger stars and the tail of large dispersion values is dominated by younger stars.
- The mean/median dispersion in the ages derived from different techniques are 0.43/0.38 dex or 0.39/0.34 using 1.5 sigma rejection (almost the size of our nominal age bins!) and the spread in both cases 0.28 dex. However, the distribution has a tail and considering only values at <0.8, the mean/median dispersion are both ~0.35 dex with 0.18 dex spread, and reasonably well fit to a Gaussian though the data have a sharper peak closer to 0.2 dex.
- From the bottom panel, age errors of 25-150% are what we should quote.
Description: Comparison of average ages and "best" ages as described above to those currently in the FEPS database. Error bars indicate the dispersion in age arising from the different techniques. The third panel mimics the first but gives names for outlier identification purposes.
Findings:
- The agreement of the database ages with the new averaged ages is astonishly -- that is
unbelievably
-- good, with some bias around a few hundred Myr. Given all that has transpired and all that is below, honestly, I can't believe how good this actually turns out.- The overall scatter is 0.26 dex, better than the average error we are quoting in the new average ages (see discussion in Figure 5).
- Vertical strings are cluster stars, where the abscissa is the assigned cluster age and the ordinate is the average activity/rotation/lithium age.
- I am not sure I am happy with the "best" age technique, based on examination of individual cases. The "best" ages have larger scatter (0.39 dex), particularly at the low age end. This due to the disagreement between the various rotation/activity/lithium measures and the HR diagram ages which in the case of <20 Myr old stars are assigned as "best" as long as they are corroborated by one other rotation/activity/lithium indicator that the star is this young. There is an explicit "best" vs average plot above as well, where the clusters are now shown as horizontal strings.
- The outliers to the right are all evolved stars and we should not be trying to apply rotation/activity/lithium calibrations to them. Instead, post-ms isochronal ages should be used.
Description: This plot mimics those in the previous figure. Shown now is comparison of the individual age indicators with those currently in the FEPS database. Error bars indicate the systematic error arising from uncertainty in the age calibration equations.
Findings:
- There is significantly more scatter in the individual techniques vs the average ages.
- The R'HK plot exhibits both systematic differences due to use of the revised calibration from Mamajek and Hillenbrand and random differences due to revised fiducial values of the R'HK index in a large number of stars. A question for the group is: how old should we trust this calibration?
- The xray ages (from luminosity, not from Rx) exhibit significantly scatter than the R'HK ages.
- Other than some rapid rotators around 100 Myr which exceed the median rotation at that age, the periods are very well correlated with the heretofore adopted age values.
- The vsini ages exhibit more scatter than the period or activity ages.
- Lithium ages are all over the place, agreeing well in certain limited regimes.
Description: Dispersion among the age techniques plotted against the differences between average new ages and database ages. Blue points are members of ScoCen, Alpha Per, Pleaides, Hyades.
Findings:
- There is no correlation between these quanties.
- There is a nice sea of points connecting the axes between 0.5,0.5.
- We will use the assigned cluster ages for all cluster members. The blue points (cluster members) are no more or less distributed than the black points (field stars) and illustrate the differences/dispersion in classical age dating techniques vs activity/rotation/lithium ages.
- As mentioned above, the stars with largest deviation are in fact evolved, so we should not be trying to apply rotation/activity/lithium calibrations to them. Instead, post-ms isochronal ages should be used.
- Here is a table that is sorted inversely by the size of the difference between the database ages and the new averaged ages. Text comments in the last column suggest which age we ought to adopt. Note from this table that some stars are in need detailed scrutiny in order to make individual decisions.
Description: Average age and dispersion plotted as a function of effective temperature, where the latter is taken directly from the FEPS database.
Findings:
- There is no trend in derived average age with stellar effective temperature. This suggests that any remaining color effects in the age calibrations are not biasing the sample as a whole.
- There is no trend in age "error" with temperature either.