# Examples

These short examples showcase the features of the `pykeops.numpy`

and `pykeops.torch`

modules,
from automatic differentiation to block-sparse reductions.

Note

If you run a KeOps script for the first time,
the internal engine may take a **few minutes** to compile all the
relevant formulas. Do not worry: this work is done **once and for all**
as KeOps stores the resulting *shared object* files (‘.so’) in a
cache directory.

## Numpy API

Sum reduction

GPU Selection

Block-sparse reductions

Arg-K-Min reduction

KernelSolve reduction

KernelSolve reduction (with LazyTensors)

KernelSolve reduction (with LazyTensors)

SumSoftMaxWeight reduction

SumSoftMaxWeight reduction (with LazyTensors)

SumSoftMaxWeight reduction (with LazyTensors)

## Pytorch API

Advanced syntax in formulas

Anisotropic kernels

Sum reduction

LogSumExp reduction

Vectorial LogSumExp reduction

Multi GPU

Block-sparse reductions

GPU Selection

KernelSolve reduction

KernelSolve reduction (with LazyTensors)

KernelSolve reduction (with LazyTensors)

SumSoftMaxWeight reduction