Repository navigation
Added
gpjax.xarrayinput transforms.from_xarray(..., transforms=[...])
turns the named inputs into the columns ofX, and theGridSpecapplies the
same fitted transforms to every new grid.Standardisescales inputs with the
training mean and standard deviation.UnitSpheremaps latitude and longitude
to a point on the unit sphere, so a stationary kernel is valid on the whole
globe.Cyclicencodes a periodic input, such as the seasonal cycle, as a
point on a circle.GridSpec.columnsnames the resulting columns.GridSpec.predict: chunked prediction on large grids.
spec.predict(lambda x: posterior(x, covariance="diagonal"), grid)gives the
predictive mean and variance as anxr.Dataset. The model gets at most
chunk_sizecells at a time and is compiled once. If the grid holds Dask
arrays, the result is lazy and is predicted block by block, along the Dask
chunks of the grid.- New example: Infilling Global Surface Temperature. It reads a netCDF
file of the 2024 temperature anomaly from the NCEP-NCAR Reanalysis 1, which is
complete, and removes cells to test the infill against the truth. Cells removed
at random are filled well, and joint samples give a global mean whose interval
holds the truth. Cells removed because they are warm show how a GP is biased,
and overconfident, when data are missing not at random. - The xarray introduction is now Working with Gridded Data, under Getting
started. It usesStandardiseandGridSpec.predict, and shows
UnitSphereandCyclic. Its URL is unchanged.
Fixed
-
TruncatedPeriodicand its state-space SDE no longer return zeros or
NaNs for small Bessel arguments. The Miller downward recurrence in
_stable_scaled_iveformed unnormalised values that grow like
$(2k_{\max}/c)^{k_{\max}}$ , with$c = 1/(4\ell^2)$ . These overflowed in
float64 for a large lengthscale and a high truncation order (for example
$\ell > 12$ withtruncation_order=25), and in float32 already for the
default kernel (lengthscale=1.0,truncation_order=6), whose Gram matrix
was all NaN. The recurrence now runs on the ratios$I_k / I_{k-1}$ , which stay
in$(0, 1)$ . -
PoweredExponential.poweris now constrained to$(0, 2]$
(#813). The kernel is
positive definite in every dimension only in this interval, but any value was
accepted, andpower=3.0gave a Gram matrix with negative eigenvalues. A
float in$(0, 2)$ is now wrapped as a trainableSigmoidBoundedparameter,
sofitlearns the power and cannot move it outside the interval. Before
this change a float power was a plain Python float, whichfitdid not
train.power=2.0(the RBF case) stays fixed, and a value outside the
interval raisesValueError. The type of thepowerleaf changes, so code
that reads it should useval(kernel.power). - Split
BlockDiaginput vectors with static input block sizes, so matrix-vector
products work under JIT and with rectangular blocks. - Read the observation count from
yfor unsupervisedDatasetobjects, so
nandfull_sizework whenXis absent. - Apply the Gaussian covariance transform to each sample for multidimensional
GaussianDistribution.sampleshapes, including shapes with empty axes.
Changed
- Allow JAX and JAXlib 0.11 in downstream environments by removing the
<0.11dependency bounds (#801).
Full diff: v1.0.0...v1.1.0