Releases: RubixML/Tensor
Releases · RubixML/Tensor
2.2.0
- Implemented Singular Value Decomposition (SVD)
- Implemented matrix (Moore-Penrose) pseudoinverse
- Tensor objects now throw namespaced exceptions
- Optimized Reduced Row Echelon Form (RREF)
- Optimize eigendecomposition for symmetric matrices
2.1.4
- Implemented Eigen decomposition in extension
- Optimize row echelon form (REF) of matrix
- Optimize Cholesky decomposition
- Fix memory leak in matrix inverse
- Optimize LU decomposition
2.1.3
- Use BLAS DGEMM for matrix multiplication
- Optimize matrix inverse
2.1.2
- Optimize scalar arithmetic and comparisons
2.1.1
- Zephir redux
- Blanket extension optimizations
- Added n-d array arithmetic and comparison functions
2.1.0
- Extension now compiles directly from C source code
- Remove Zephir extension code
- Update instructions for extension compilation using PHPize
2.0.5
- Percentile method changed to quantile
- Fixed Vector linspace precision
2.0.4
- Individual Arithmetic and Comparison methods now public
- Matrix reductions have their own namespace
- Optimized Matrix structural operations
- Added argmin and argmax custom optimizers
- Added matrix/sub matrix insertion
2.0.3
- Matrix decomp methods now return decomp objects
- Optimized matrix statistical operations
- Optimized vector outer product
- Optimized Matrix flatten
- Cannot instantiate null/empty tensors
2.0.2
- Optimized random number generation