Wrote seven unit tests to identify source of how/why I can get $ $’s to be larger in the parrotfish group than wrasses (when only $alpha $ varies across regimes) on the same trait for which $ $’s are larger (when only they vary). Tests successively restrict pattern into the likelihood equation, some treatment of the regimes must be wrong.
Unit tests.
Nov 29, 2011
cboettig 2011-11-29 17:29:04
cboettig authored 2011-11-29 17:27:26
cboettig authored 2011-11-29 15:23:54
cboettig authored 2011-11-29 15:22:50
cboettig authored 2011-11-29 14:36:49
cboettig authored 2011-11-29 14:27:31
cboettig authored 2011-11-29 14:27:17
cboettig authored 2011-11-29 14:27:07
cboettig authored 2011-11-29 14:26:41
cboettig authored 2011-11-29 14:26:22
cboettig authored 2011-11-29 10:26:01
cboettig authored 2011-11-29 10:25:46
cboettig authored 2011-11-29 10:24:44
cboettig authored 2011-11-29 09:09:02
Time to check these
Mean \[ E(X_t) = \exp \left( - \sum \alpha_i \Delta t_i \right) \left( X_{t_i} + \sum \theta_i \left( e^{\alpha_i t_i}-e^{\alpha_i t_{i-1} } \right) \right) \]
Covariances \[ \omega = \frac{\sigma^2}{2 \alpha } \left( e^{2\alpha t_i} - e^{2 \alpha t_{i-1}} \right) \] \[ \langle x_i, x_i \rangle = \omega \exp( \alpha_i t_i - \alpha_j t_j) \]
(Note - dates pulled out of html with vim gsub, note that vim needs the parentheses for grouping to be escaped (backwhacked!)
:%s/<time.*title="$.*$".*time>/\1/g