07-07-2015

# New in this release

* Removed all tensor-related classes (they will be a deprecated section of the soon-to-come companion library vmml::tensor with these functionalities)
* Added serialization functions read_from_stream and write_to_stream to matrix.hpp
* Moved sections from RELNOTES.md to relevant files
* Removed leading and trailing underscores from header guards
* Changed class names to CamelCase. Created typedefs for backwards compatibility (they can be deactivated, but are on by default --preprocessor variable VMMLIB_OLD_TYPEDEFS)
* Updated copyright notes
* Merged updates from Eyescale fork
* Disabled (by default) old tests

05-02-2013

# New in this release

###CMake
* use now only CMake to create a Makefile

###Use Tensor Approximation Classes (based on LAPACK, BLAS)
* define VMMLIB_USE_LAPACK in Makefile (changed from VMMLIB_BASIC_ONLY)

### Matrix
* added matrix validators

###BLAS Wrapper
* added BLAS DAXPY wrapper

###Filtering
* added intersection, lowpass_filter class
* updates in frustum_culler

###Unit Tests
* boolean global results of every test are now consistent with every subtest

## Bug Fixes

vmmlib 1.6 includes several fixes over the last release, such as:

* various test fixes
* precision bug of CP3 tensor reconstruction error measurement
* tensor4 "<<" operator
* computation of residual norm in HOOI
* corrected formula for quaternion slerp

## Known Issues

* memory mapping for windows to be tested
* test for slerp and matrix validators are not yet implemented
* tests that depend on rand() are deactivated: they may break with different stdlib versions
* tests that depend on LAPACK and BLAS are not fully supported for Windows


