Further miscellaneous options

ALPHAPROPSD (double) The Metropolis-Hastings update step for $\alpha$ involves picking a value $\alpha'$ from a Normal with mean $\alpha$ and standard deviation ALPHAPROPSD$>0$. The value of ALPHAPROPSD does not affect the asymptotic behaviour of the Markov chain, but may have a substantial impact on the rate of convergence. If there is a lot of information about $\alpha$, small values of ALPHAPROPSD are preferable to obtain a reasonable acceptance rate. If there's not much information about $\alpha$, larger values produce better mixing.

FPROPSD (double) The standard deviation of the proposal for the allele frequency correlation $f^{(l)}$. Analogous to ALPHAPROPSD. Only relevant when FREQSCORR = 1.

STARTATPOPINFO (Boolean) Use given populations as the initial condition for population origins. (Need POPDATA==1). This option is here to provide another check that the Markov chain is converging adequately. This option only affects the starting point of the Markov chain and, in theory, it should not affect the final answers, provided that the BURNIN is long enough. It is assumed that the PopData in the input file are between 1 and $k$ where $k\leq$MAXPOPS. Individuals for whom the PopData are not in this range are initialized at random.

RANDOMIZE (Boolean) Use a different random number seed for each run (this is taken from the system clock).

METROFREQ (int) Frequency of using a Metropolis-Hastings step to update $Q$ under the admixture model. When this is used, a new proposal $q^{(i)}{'}$ is chosen for each $q^{(i)}$. This proposal is sampled from the prior (ie $q^{(i)}{'}\sim{\cal D}(\alpha,\alpha,...,\alpha)$). The rationale for having this update is that it may improve mixing when alpha is quite small, by making it easier for individuals to jump between populations. The Metropolis-Hastings move is used once every METROFREQ iterations. If METROFREQ is set to 0, it is never used.

REPORTHITRATE (Boolean) Report acceptance rate of Metropolis update for $q^{(i)}$ (see METROFREQ).