That paper explains that the Scala engineer performed numerous algorithmic improvements, whereas the Java one "refused" to (he reported somewhere else, on a mailing list? that he wasn't told it was for a paper). The paper also explains that using Scala idiomatic for-comprehensions produced very slow results.
Comments
I found this odd as well. I have always heard that it runs faster. For example: https://days2011.scala-lang.org/sites/days2011/files/ws3-1-H...
That paper explains that the Scala engineer performed numerous algorithmic improvements, whereas the Java one "refused" to (he reported somewhere else, on a mailing list? that he wasn't told it was for a paper). The paper also explains that using Scala idiomatic for-comprehensions produced very slow results.