Scala/Python vs. Java: SparkContext.map vs. .filter in PI example?

Question

In the Pi example at http://spark.apache.org/examples.html

In the Estimating Pi example, there is a discrepancy in the Python/Scala vs. Java example I don't understand. Python and Scala are both using map and reduce:

Python

def sample(p):
    x, y = random(), random()
    return 1 if x*x + y*y < 1 else 0

count = spark.parallelize(xrange(0, NUM_SAMPLES)).map(sample) \
             .reduce(lambda a, b: a + b)
print "Pi is roughly %f" % (4.0 * count / NUM_SAMPLES)

Scala

val count = spark.parallelize(1 to NUM_SAMPLES).map{i =>
  val x = Math.random()
  val y = Math.random()
  if (x*x + y*y < 1) 1 else 0
}.reduce(_ + _)
println("Pi is roughly " + 4.0 * count / NUM_SAMPLES)

But Java is using filter:

int count = spark.parallelize(makeRange(1, NUM_SAMPLES)).filter(new    
  Function<Integer, Boolean>() {
    public Boolean call(Integer i) {
      double x = Math.random();
      double y = Math.random();
      return x*x + y*y < 1;
   }
}).count();
System.out.println("Pi is roughly " + 4 * count / NUM_SAMPLES);

Is this just a doc typo/bug? Is filter preferable in Java and map/reduce preferred in Scala and Python for some reason?

Answer 1

These approaches are equivalent. Java code simply counts cases where Scala / Python map returns 1. Just to make it a little bit more transparent:

def inside(x, y):
    """Check if point (x, y) is inside a unit circle
    with center in the origin (0, 0)"""
    return x*x + y*y < 1

points = ... 

# Scala / Python code is equivalent to this
sum([1 if inside(x, y) else 0 for (x, y) in points])

# While Java code is equivalent to this
len([(x, y) for (x, y) in points if inside(x, y)])

Finally sum you get is proportional to the fraction of the area of the enclosing square covered by the circle and from the formula we know it is equal π.