Releases: stecrotti/TensorTrains.jl
Releases · stecrotti/TensorTrains.jl
v0.10.1
TensorTrains v0.10.1
Merged pull requests:
- Use MKL instead of openBLAS for blas and lapack (#40) (@stecrotti)
v0.10.0
TensorTrains v0.10.0
Main changes:
- Now tensor trains contain a scalar
z
as a field. When a tensor train is evaluated:f(x) = 1/Z * A1(x1) * A2(X2) * ... * AL(xL)
- Greatly improved robustness wrt numerical under/over flow
Merged pull requests:
- Untrack test/Manifest.toml (#33) (@stecrotti)
- [WIP] A bunch of updates to allow for negative-valued tensor trains and improve robustness against under/over flow (#36) (@stecrotti)
- Target "HEAD" for benchmarks (#37) (@stecrotti)
- uniform->flat in docs (#38) (@stecrotti)
Closed issues:
- Numerical under/over flow when normalizing (#31)
v0.9.1
TensorTrains v0.9.1
Merged pull requests:
- Remove unused dependencies (#28) (@stecrotti)
- Explicitly import Base.show (#29) (@stecrotti)
- Flat tensor train comes already normalized (#30) (@stecrotti)
v0.9.0
TensorTrains v0.9.0
Merged pull requests:
- Add compat entries for stdlib dependencies (#21) (@stecrotti)
- More flexible TensorCast compat (#23) (@stecrotti)
- Go back to latest version of TensorCast (#24) (@stecrotti)
- Untrack manifest file (#25) (@stecrotti)
- Use julia v1.10 (#26) (@stecrotti)
- Rename uniform_tt to flat_tt (#27) (@stecrotti)
v0.8.0
TensorTrains v0.8.0
Merged pull requests:
- Use generic trace methods when possible (#16) (@abraunst)
- Add benchmarking with PkgBenchmark.jl (#17) (@stecrotti)
- Add run_benchmark.jl file (#18) (@stecrotti)
- only call accumulate_R when debugging on normalize! (#19) (@abraunst)
- Release v0.8 (#20) (@stecrotti)
v0.7.0
v0.6.1
TensorTrains v0.6.1
Merged pull requests:
- More robust threshold-based trunc (#11) (@stecrotti)
- Adhere to style guide for using and import (#12) (@stecrotti)