CDS implementation in Java of the HEALPix tesselation.
For informations on HEALPix in general, see:
- The official web site
- The Wikipedia page
- The two main related papers: Gorsky (2005) and Calabretta (2007)
See a much cleaner version of the code ported in Rust here and the project to use the Rust code from Python here.
This software is released under the 3-Clause BSD license. See the license file.
- The libray in still in the testing phase, especially for the exact cell-in-cone function.
- Logical operations on BMOC are not yet debugged (they already are in the Rust project).
- The external edge method is not yet debuuged (it is in the Rust project)
- The bilinear interpolation has not been tested yet (it has been tested in the Rust project)
- ...
- Implement exact cone solution
- Implement exact polygon solution
- Add tests from the Rust code into the Java code
- Add regression tests on
newConeComputerApprox
- Debug approximate elliptical cone
- Debug BMOC logical operation (see Rust code)
- Debug external edge method (see Rust code)
- Test bilinear interpolation (see Rust code)
- Add RING scheme? (see Rust code)
- Compressed MOCs: test M.Rein. solution
- Clean the code!!
- ...
You need ant to be installed on your machine to test and build the project.
Then, in the root of the project (the directory containing the build.xml file, type
> ant -p
to list possible actions.
You should get:
CDS Healpix library build.xml
Main targets:
make.all Compile all the software, run tests, and create the .jar
make.build Compile main classes, to test if the soft compiles
make.doc Compile all the software and create the .jar
make.jar Compile main classes and create both the jar and the fatjar
make.test Compile all classes (main an test), and run tests
Default target: print_properties
To build the jar, simply type
> ant make.jar
To build the javadoc:
> ant make.doc
Then open it in you favourite browser, e.g.:
> firefox target/docs/apidocs/index.html
The test contains performances tests, run all of them with:
> ant make.test
Firs build the JAR file
ant make.jar
and then execute it:
java -jar cdshealpix.x.x_x.jar
Or direclty
ant exec
Look here.
The library entry point is the class Healpix
in the package cds.healpix
.
// Inputs
int depth = 8;
double ra = Math.toRadians(10.509734);
double dec = Math.toRadians(21.657381);
// Get the cell number
HealpixNested hn = Healpix.getNested(depth);
long cellNumber = hn.hash(ra, dec);
// Or
HealpixNestedFast hnf = Healpix.getNestedFast(depth);
cellNumber = hnf.hash(ra, dec);
// In a multi-threaded environment, use one HashComputer per thread
HashComputer hc = hn.newHashComputer();
cellNumber = hc.hash(ra, dec);
// Static imports
import static cds.healpix.VerticesAndPathComputer.LON_INDEX;
import static cds.healpix.VerticesAndPathComputer.LAT_INDEX;
// Inputs
int depth = 8;
long cellNumber = 12394L;
// Get the center coordinates
HealpixNested hn = Healpix.getNested(depth);
double[] centerCoos = hn.center(cellNumber);
double raDeg = Math.toDegrees(centerCoos[LON_INDEX]);
double decDeg = Math.toDegrees(centerCoos[LAT_INDEX]);
// Again, in a multi-threaded environment, use one VerticesAndPathComputer per thread
VerticesAndPathComputer vpc = hn.newVerticesAndPathComputer();
centerCoos = vpc.center(cellNumber);
raDeg = Math.toDegrees(centerCoos[LON_INDEX]);
decDeg = Math.toDegrees(centerCoos[LAT_INDEX]);
// Static import
import static cds.healpix.VerticesAndPathComputer.ALL_CARDINAL_POINTS;
// Inputs
int depth = 8;
long cellNumber = 12394L;
// Get vertices
HealpixNested hn = Healpix.getNested(depth);
EnumMap<Cardinal, double[]> vertices = hn.vertices(cellNumber, ALL_CARDINAL_POINTS);
// and then
for (double[] vertexCoos : vertices) {
...
}
// or
double[] northVertexCoos = vertices.get(Cardinal.N);
double[] eastVertexCoos = vertices.get(Cardinal.E);
double[] southVertexCoos = vertices.get(Cardinal.S);
double[] westVertexCoos = vertices.get(Cardinal.W);
// If you are only interested in a single vertex
northVertexCoos = hn.vertex(Cardinal.N);
// If you are only interested in East and West vertices
vertices = vertices(cellNumber, EnumSet.of(Cardinal.E, Cardinal.W));
// Inputs
int depth = 8;
long cellNumber = 12394L;
// Get the neighbours list
HealpixNested hn = Healpix.getNested(depth);
NeighbourList neigList = hn.neighbours(cellNumber);
// Iterates
FlatHashIterator hIt = neigList.iterator();
while (hIt.hasNext()) {
System.out.println("Neighbour cell number: " + hIt.next());
}
// Or access by MainWinds
long northeastCellNumber = neigList.get(MainWind.NE);
WARNING: the exact method is still in testing phase, so far the approx method is more robust!
// Inputs
int depth = 8;
double coneCenterRa = Math.toRadians(10.509734);
double coneCenterDec = Math.toRadians(21.657381);
double coneRadiusRad = Math.toRadians(1.0);
// Resulting BMOC (MOC + flag telling if a cell is fully or partially overlapped by the cone)
HealpixNested hn = Healpix.getNested(depth);
// Choose one of the two following lines of code:
HealpixNestedFixedRadiusConeComputer cc = hn.newConeComputer(coneRadiusRad); // beta code!!
HealpixNestedFixedRadiusConeComputer cc = hn.newConeComputerApprox(coneRadiusRad); // robust code
HealpixNestedBMOC bmoc = cc.overlappingCells(coneCenterLonRad, coneCenterLatRad);
// BMoc view
for (HealpixNestedBMOC.CurrentValueAccessor cell : bmoc) {
System.out.println("Cell number: " + cell.getHash()
+ "; cell depth:" + cell.getDepth()
+ "; fully covered: " + cell.isFull());
}
// Flat view at depth 8:
FlatHashIterator hIt = bmoc.flatHashIterator();
while (hIt.hasNext()) {
System.out.println("Cell number at depth 8: " + hIt.next());
}
// Inputs
int depth = 8;
double xmatchRadius = Math.toRadians(1.0 / 3600.0); // 1 arcsec
// Results
HealpixNested hn = Healpix.getNested(depth);
HealpixNestedFixedRadiusCone4XMatch xm = hn.newConeComputer4Xmatch(xmatchRadius);
long[] cellsNum = new long[9];
int indexMax = 0;
for (double[] coosInRadians : listOfPosToXmatch) {
indexMax = xm.overlappingCells(coosInRadians[0], coosInRadians[1], cellsNum);
for (int ic = 0; i < indexMax; i++) {
// look at cellsNum[ic], ...
}
}