PathSampleAnalyser with 14 steps and -alpha 0 fails

[rbouckaert]

With alpha=0, uniform steps are taken, and with 14 steps step1 runs into trouble (see error message below). It looks like there is some numeric instability, which is not encountered with alpha=0.3 (the default).

Fatal exception: Cumulative probability function returned NaN for argument 0.923 p = 0.923
org.apache.commons.math.FunctionEvaluationException: Cumulative probability function returned NaN for argument 0.923 p = 0.923
at org.apache.commons.math.distribution.AbstractContinuousDistribution$1.value(Unknown Source)
at org.apache.commons.math.analysis.solvers.BrentSolver.solve(Unknown Source)
at org.apache.commons.math.analysis.solvers.BrentSolver.solve(Unknown Source)
at org.apache.commons.math.analysis.solvers.UnivariateRealSolverUtils.solve(Unknown Source)
at org.apache.commons.math.distribution.AbstractContinuousDistribution.inverseCumulativeProbability(Unknown Source)
at org.apache.commons.math.distribution.BetaDistributionImpl.inverseCumulativeProbability(Unknown Source)
at beast.inference.PathSampleAnalyser.estimateMarginalLikelihood(Unknown Source)
at beast.inference.PathSampler.analyse(Unknown Source)
at beast.inference.PathSampler.doRuns(Unknown Source)
at beast.inference.PathSampler.run(Unknown Source)
at beast.app.BeastMCMC.run(Unknown Source)
at beast.app.beastapp.BeastMain.(Unknown Source)
at beast.app.beastapp.BeastMain.main(Unknown Source)
Fatal exception: Cumulative probability function returned NaN for argument 0.923 p = 0.923
java.lang.RuntimeException: An error was encounted. Terminating BEAST
at beast.app.util.ErrorLogHandler.publish(Unknown Source)
at java.util.logging.Logger.log(Logger.java:738)
at java.util.logging.Logger.doLog(Logger.java:765)
at java.util.logging.Logger.log(Logger.java:788)
at java.util.logging.Logger.severe(Logger.java:1464)
at beast.app.beastapp.BeastMain.(Unknown Source)
at beast.app.beastapp.BeastMain.main(Unknown Source)