v0.12.0

• Improved:
  • Divisions in atan2_polyApprox() are speeded-up under TMS320C6748 DSP environment using RCPSP or RCPDP instructions.

v0.11.0

• Added
  • fast floating point number reciprocal computation using TMS320C6748 DSP instructions RCPSP and RCPDP.
  • fast floating point number reciprocal square root using TMS320C6748 DSP instructions RSQRSP and RSQRDP.

v0.10.0

• Fixed
  • missing doxygen comment for LinAlg::fillLowTri function
• Improved
  • automatic data type promotion for some functions in LinAlg namespace
• Added
  • LinAlg::copyLowTri function
  • LinAlg::copyConjLowTri function

v0.9.0

• Added
  • More data type flexibility of addDiag(); Different data types are allowed for vector and matrix.

v0.8.0

• Fixed
  • Fixed a issue in CMakeLists.txt: disable address sanitizer on Windows (GCC doesn't support it).
  • outdated version macro constants MATH_LIB_VER_...
• Improved
  • Improved CMakeLists.txt: cancel build on unknown platform.

v0.7.0

• Added setReIm, conjProd function.

v0.6.0

• Rename "Bug Hunting Mode" to "Canary Mode".
• Added SqrtTable.

improved performance and new functions

• Added MATH_LIB_INLINE_AGGRESSIVELY option.
• Speeded up following functions:
  • Analysis::
    • conj, sqAbs
  • LinAlg::
    • scaleMat, scaleMatEachRow, innerProd, hermitianInnerProduct, vecSelfOuterProd, mulMat, ldlDecomp, solveLinearEquation, solveLinEqHermitian, hermitianInnerProduct
  • SigProc::
    • convolve, convolve_type2
• Improved include-guard: replaced old-style guard using #ifndef with #pragma once.
• Added following functions:
  • Analysis::
    • addConjProd
  • LinAlg::
    • fillLowTri, addSqMat, scaleMat
• Refactored polynomial approximation of arc tangent:
  • Added atan_polyApprox_deg7 function.
  • Renamed atan_polyApprox to atan_polyApprox_deg9.
  • Add template function atan_polyApprox which is reduced to atan_polyApprox_deg7 or atan_polyApprox_deg9 according to its template parameter.
  • Add template function atan2_polyApprox whose polynomial degree is determined at compile-time
• Made the following small functions target of aggressive inline-expansion:
  • LinAlg::
    • isEqualMat, complexMat, conjugateMat, addMat, scaleMat
• Forcibly inline-expand the following small functions:
  • LinAlg::
    • isEqualVec, complexVec, conjugateVec, addVec, scaleVec, l2Norm
• Added compiler options for secure programming
  • -Wextra, -Wold-style-cast, -Wredundant-decls, -Wshadow, -Wswitch-default, -Wswitch-enum (debug/release build)
  • -fstack-protector, -ftrapv (debug build)

bug fix and refactor

• Fixed missing header inclusion in sigProc.hpp
• Improved include-guard: replaced old-style guard using #ifndef with #pragma once.

added new features

• Renamed project: MotchyMathLib -> MathLib.
• Optimized atan2_polyApprox function to speed up on Texas Instruments DSP TMS320C6748.
  • There is no effect on modern PC processors because the compilers for them are clever enough to perform above easy hand-optimization.
• Added vecSelfOuterProd function.
• Added bug hunting mode.
• Added dropSubMat function.
• Added setDiag function.
• Added addDiag function.
• Speeded up ldlDecomp function, and consequently solveLinEqHermitian function is speeded up.
• Add in-place version of conjugateMat.
• Add innerProd function
• Rename addMat_inplace, addVec_inplace to addMat, addVec