I did that test long ago and forgot some details. I tried it again this morning. Neither with ns nor without ns work. The test code is listed below.
the xpath method should be selectNodeList or selectSingleNode from org.apache.xpath.XPathAPI. It operates on w3c node.
public static void main(String[] args) throws Exception
{
Unmarshaller um = new MicromagSchema.ObjectFactory().createUnmarshaller();
Marshaller m = new MicromagSchema.ObjectFactory().createMarshaller();
Object o = um.unmarshal(new URL("
http://jacob.chem.uno.edu/schemas/newSample2.xml"));
DocumentBuilder db = DocumentBuilderFactory.newInstance().newDocumentBuilder();
Document doc = db.newDocument();
org.w3c.dom.Element root = doc.createElement("Root");
m.marshal(o, root);
root = (org.w3c.dom.Element)root.getFirstChild();
String uri = root.getNamespaceURI(); //correctly returns my schema location
String name = root.getTagName(); //returns ns1:Task where Task is name of my root element
NodeList nl = root.getElementsByTagName(name); //getLength() returns zero
for (int i=0; i<nl.getLength(); i++) System.out.println(nl.item(i));
nl = root.getElementsByTagNameNS(name, uri); //nl is null from this call
for (int i=0; i<nl.getLength(); i++) System.out.println(nl.item(i));
}