At least for now, function wmeanandsig.pro is in /data/psrdata/rmgalprojs
If you create an array of RM and an equal-sized array of sig_RM (with no ∞ sig_RM’s), then:
wmean_and_sdm=wmeanandsig(RM,
will put the weighted mean and weighted sdm into a two-element array called wmean_and_sdm
(and the way functions work, you can replace the variable that above i call “wmean_and_sdm”, with
a variable of any name you wish)
Right now there is a “driver” program in the same directory, called, “wmeanandsigDRIVER.pro” . It
is a module that tests wmeanandsig.pro . I used it extensively to do so, to the point where I can now
call wmeanandsig.pro “ready to go” . (Luckily the NIST web pages give sample input/output for weighted
mean and for weighted SDM, which I endeavored to match.)
Caution: In 2018 April, I changed the calculation of the weighted SDM. Previously, we had used the NIST definition, which changed the data weights so as to achieve reduced χ^2 to 1. (The NIST webpage gave a formula for wgtd SD but did not note that it included this readjustment. ) In 2018 April, I changed the calculation to be
σ^2_wgtdmean = 1 / [ Σ 1/σ^2_j]. Note that this calculation depends entirely on the σ_j’s of the data points, and nothing else (such as deviation from mean).
Be First to Comment