Scala for/yield runs but doesn't complete - scala

Question

I'm trying to walk through two arrays of potentially different sizes and compose a new array of randomly selected elements from them (for crossover in a genetic algorithm) (childGeneCount is just the length of the longer array).
In the following code snippet, each gene.toString logs, but my code doesn't seem to execute the last log. What dumb thing am I doing?
val genes = for (i <- 0 to childGeneCount) yield {
val gene = if (Random.nextBoolean()) {
if (i < p1genes.length) {
p1genes(i)
} else {
p2genes(i)
}
} else {
if (i < p2genes.length) {
p2genes(i)
} else {
p1genes(i)
}
}
Logger.debug(gene.toString)
gene
}
Logger.debug("crossover finishing - never gets here??")
New to scala, and would be happy for a slap on the wrist accompanied by a "do it this completely different way instead" if appropriate.

Answer 1

You are right, the problem was with "to" should have been "until". I have changed your code a bit to make it more scala like.
val p1genes = "AGTCTC"
val p2genes = "ATG"
val genePair = p1genes.zipAll(p2genes, None, None)
val matchedGene = for (pair <- genePair) yield {
pair match {
case (p1Gene, None) => p1Gene
case (None, p2Gene) => p2Gene
case (p1Gene, p2Gene) => if (Random.nextBoolean()) p1Gene else p2Gene
}
}
println(matchedGene)
The process is:
First zip two dna sequences into one.
Fill the shorter sequence with None.
Now loop over the zipped sequences and populate the new sequence.

Answer 2

Reworked Tawkir's answer, with cleaner None handling:
val p1genes = "AGTCTC"
val p2genes = "ATG"
val genePair = p1genes.map(Some.apply).zipAll(p2genes.map(Some.apply), None, None)
val matchedGene = genePair.map {
case (Some(p1Gene), None) => p1Gene
case (None, Some(p2Gene)) => p2Gene
case (Some(p1Gene), Some(p2Gene)) => if (Random.nextBoolean()) p1Gene else p2Gene
}
println(matchedGene)
If you want to avoid wrapping the sequence with Some, another solution is to use a character known not to appear in the sequence as a "none" marker:
val p1genes = "AGTCTC"
val p2genes = "ATG"
val none = '-'
val genePair = p1genes.zipAll(p2genes, none, none)
val matchedGene = genePair.map {
case (p1Gene, `none`) => p1Gene
case (`none`, p2Gene) => p2Gene
case (p1Gene, p2Gene) => if (Random.nextBoolean()) p1Gene else p2Gene
}
println(matchedGene)

Answer 3

Pretty sure harry0000's answer is correct: I was using "to" like "until", and am so used to exceptions being thrown loudly that I didn't think to look there!
I ended up switching from for/yield to List.tabulate(childGeneCount){ i => {, which fixed the error probably for the same reason.

Answer 4

Since you asked for possible style improvements, here are two suggested implementations. The first one is less idiomatic, but more performant. The second one is prettier but does some more work.
def crossover[E : ClassTag](a: Array[E], b: Array[E]): Array[E] = {
val (larger, smaller) = if(a.length > b.length) (a, b) else (b, a)
val result = Array.ofDim[E](larger.length)
for(i <- smaller.indices)
result(i) = if(Random.nextBoolean()) larger(i) else smaller(i)
for(i <- smaller.length until larger.length)
result(i) = larger(i)
result
}
def crossoverIdiomatic[E : ClassTag](a: Array[E], b: Array[E]): Array[E] = {
val randomPart = (a zip b).map { case (x,y) => if(Random.nextBoolean()) x else y }
val (larger, smaller) = if(a.length > b.length) (a, b) else (b, a)
randomPart ++ larger.drop(smaller.length)
}
val a = Array("1", "2", "3", "4", "5", "6")
val b = Array("one", "two", "three", "four")
// e.g. output: [one,2,three,4,5,6]
println(crossover(a, b).mkString("[", ",", "]"))
println(crossoverIdiomatic(a, b).mkString("[", ",", "]"))
Note that the E : ClassTag are only there to make the compiler happy about using Array[E], if you only need Int for your work, you can drop all the fancy generics.