Skip to content

Release 0.2.0
Compare
Choose a tag to compare
@mlxd mlxd released this 22 Jul 12:36
· 23 commits to main since this release
0.2.0
3589078
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

New features since last release

  • The Jet interpreter for XIR scripts now accepts a dimension option for CV circuits. (#47)

  • The Jet interpreter for XIR programs now supports a probabilities output. (#44)

  • The Circuit class now supports taking the expected value of an observable. (#46)

  • The Jet interpreter for XIR scripts now handles gate definitions. (#37)

  • CudaTensor now supports Transpose calls. (#42)

  • Full and sliced contractions can now be run with TaskBasedContractor on the GPU using the CudaTensor class. (#29)

  • The TaskBasedCpuContractor class has been renamed to TaskBasedContractor. (#29)

  • An XIRProgram which declares the gates supported by Jet is now bundled with the jet Python package. (#34)

  • The jet Python package now includes an interpreter for XIR programs. (#24)

  • Gates may now be instantiated by name using the GateFactory Python class. (#23)

  • Quantum circuit and state models have been added to the jet Python package. (#21)

  • Quantum gate models have been added to the jet Python package. (#16)

  • Python bindings are now available for the TaskBasedCpuContractor class. (#19)

  • Python bindings now include a factory method which accepts a dtype parameter. (#18)

  • Running make build from the python directory now creates a Python distribution package. (#13)

  • A new intermediate representation (IR) is added, including a parser, IR representation program, and a Strawberry Fields interface. (#11)

  • Python bindings are now available for the TensorNetworkSerializer class. (#5)

  • Python bindings are now available for the TensorNetwork and PathInfo classes (#7)

  • Python bindings are now available for the Tensor class. (#2)

  • Running CMake with -DBUILD_PYTHON=ON now generates Python bindings within a jet package. (#1)

Improvements

  • The Jet interpreter for XIR programs now uses the task-based contractor. (#44)

  • CudaTensor SliceIndex calls now avoid intermediate CPU transfer operations. (#42)

  • A Python (Ubuntu) test report is now generated alongside the C++ (Ubuntu) test report. (#39)

  • Tensor transposes are now significantly faster when all the dimensions are powers of two. (#12)

  • Use camel case for type aliases. (#17)

  • Exceptions are now favoured in place of std::terminate with Exception being the new base type for all exceptions thrown by Jet. (#3)

  • TaskBasedCpuContractor now stores Tensor results. (#8)

  • Tensor class now checks data type at compile-time. (#4)

Breaking Changes

  • The Jet interpreter for XIR scripts now requires an amplitude state to be specified as an array. (#45)

  • The Jet interpreter for XIR scripts is now case-sensitive with respect to gate names. (#36)

  • Python bindings for complex<float> and complex<double> specializations are now suffixed with C64 and C128, respectively. (#15)

  • Indices are now specified in row-major order. (#10)

Bug Fixes

  • The Jet and XIR Python packages can now be installed from a source distribution. (#38)

  • An issue with the CudaTensor indices was fixed when converting between the CPU Tensor class. (#29)

  • The Jet versions returned by Jet::Version() (C++) and jet.Version() (Python) are now correct. (#26)

  • The documentation build no longer emits any Doxygen warnings. (#25)

  • Running make build in the python directory now correctly uses the virtual environment. (#31)

  • The output of TensorNetwork::Contract() and TaskBasedCpuContractor::Contract() now agree with external packages. (#12)

  • TaskBasedCpuContractor::AddReductionTask() now handles the reduction of non-scalar tensors. (#19)

  • The output of TensorNetwork::Contract() and TaskBasedCpuContractor::Contract() now agree with one another. (#6)

  • PathInfo now correctly names intermediary tensors in a sliced tensor network (#22).

Documentation

  • The "Using Jet" section of the Sphinx documentation now has Python code snippets. (#43)

  • The Sphinx documentation now includes API documentation for the jet Python package. (#40)

  • The "Using Jet" section of the Sphinx documentation website now compiles with the latest Jet headers. (#26)

  • The license comment headers at the top of the IR source files have been removed. (#14)

Contributors

This release contains contributions from (in alphabetical order):

Mikhail Andrenkov, Jack Brown, Theodor Isacsson, Josh Izaac, Lee J. O'Riordan, Antal Száva, Trevor Vincent.

Assets 8
Loading