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)

The current @Citation is automatically inherited. If beast.gss.GeneralisedSteppingStoneStep, beast.inference.PairedPathSamplingStep, beast.inference.PathSamplingStep do not have citations, then they will use BEAST 2 paper as the citation because they are inherited from MCMC.

See https://groups.google.com/g/beast-users/c/qrEeWy3h8fY for an example.

It confuses users that want to do a stepping stone analysis, but calculate scores using the PathSampleAnalyser. Even though by default it is a stepping stone analysis that is being done.

Regarding the BFD* tutorial for beast 2.6 which suggests you add this block of code to the xml:

<run spec=‘beast.inference.PathSampler’
chainLength="1000"
alpha=‘0.3’
rootdir=‘/home/desktop/BFD*-tutorial/runA/’
burnInPercentage=‘0’
preBurnin="0"
deleteOldLogs=‘true’
nrOfSteps=‘24’>
cd $(dir)
java -cp $(java.class.path) beast.app.beastapp.BeastMain $(resume/overwrite) -java -seed $(seed) beast.xml

If you fail to correct beast.inference.PathSampler to modelselection.inference.PathSampler (which I just did and I suspect most users will also do), then it would make sense to see an error which reads "cannot find beast.inference.PathSampler, did you mean modelselection.inference.PathSampler?"

However, instead of this I got the error message:

java.io.FileNotFoundException: /home/desktop/BFD*-tutorial/runA/step0/mcmc.xml (No such file or directory)

which made the problem much more difficult to debug.

I just wonder if there is a way of making this issue easier for futrue users to fix when they inevitably encounter it

e.g. instead of multiple lines of this:

cp /tmp/analysis/step0/beast.xml.state /tmp/analysis/step1
cd "/tmp/analysis/step1"
java -cp "/Applications/Phylogenetics/BEAST 2.4.0/BEAUti.app/Contents/Java/Classes:/Applications/Phylogenetics/BEAST 2.4.0/BEAUti.app/Contents/Java/launcher.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEAST/lib/beast.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEAST/lib/beast.src.jar:/Applications/Phylogenetics/BEAST 2.4.0/lib/beast.src.jar:/Users/simon/Library/Application Support/BEAST/2.4/Babel/lib/Babel.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/Babel/lib/itextpdf-5.5.9.jar:/Users/simon/Library/Application Support/BEAST/2.4/BDSKY/lib/BDSSM.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEAST_CLASSIC/lib/BEAST_CLASSIC.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEAST_CLASSIC/lib/mtj.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEASTLabs/lib/BEASTlabs.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/MODEL_SELECTION/lib/MODEL_SELECTION.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/morph-models/lib/MM.addon.jar:/Users/simon/Library/Application Support/BEAST/2.4/subst-bma/lib/SubstBMA.addon.jar:/Applications/Phylogenetics/BEAST 2.4.0/lib/beast.src.jar:/Users/simon/Library/Application Support/BEAST/2.4/BEAST/lib/beast.src.jar" beast.app.beastapp.BeastMain -resume -java -seed 815032149 beast.xml

could become:

SEED=815032149
CWD="/tmp/analysis/"
CLASSPATH="/Applications/Phylogenetics/BEAST ....etc"

cp $(CWD)step0/beast.xml.state $(CWD)step1
cd "$(CWD)step1"
java -cp $(CLASSPATH) beast.app.beastapp.BeastMain -resume -java -seed $(SEED) beast.xml

...which would then make the subsequent analysis scripts easier to move between machines (say from laptop to a cluster, where e.g. CWD and CLASSPATH is different).

Not sure how .bat files handle variables though, so this might be tricky to get working cross-platform.

Even better would be a loop, as then you could just increment $MAX_STEPS and rerun, but again cross-platform compatibility could be nightmare (e.g. from memory for loop syntax isn't consistent even across POSIX).

I get the following error when trying to run the PathSampling app:

beast.util.XMLParserException:
Error 1006 parsing the xml input file

nullCannot instantiate class. Please check the spec attribute.

Error detected about here:
  <beast>
      <subtemplate id='HyperPrior'>
          <connect>

    at beast.util.XMLParser.createBeastObject(Unknown Source)
    at beast.util.XMLParser.createObject(Unknown Source)
    at beast.util.XMLParser.parseInputs(Unknown Source)
    at beast.util.XMLParser.createObject(Unknown Source)
    at beast.util.XMLParser.parseFragment(Unknown Source)
    at beast.util.XMLParser.parseBareFragment(Unknown Source)
    at beast.app.beauti.BeautiConfig.initAndValidate(Unknown Source)
    at beast.app.tools.PathSampler.main(Unknown Source)

... this happens if I use the GUI or try to run it via appstore PathSampler.

only posterior(prior,likelihood) is expected, but cladeage can add another CompoundDistribution that (if not used) messes up the pattern by adding posterior(prior,likelihood,fossilcalibrations). If this is empty, the path sampler should not stop for it.

If you start a PS analysis without complete path, the run files do not get absolute paths, and possibly won't start.

To fix this, the java.class.path should be expanded to include absolute paths.