tomoto 0.6.0 → 0.6.2

data/vendor/tomotopy/src/TopicModel/GDMRModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentGDMR);
 
-    IGDMRModel* IGDMRModel::create(TermWeight _weight, const GDMRArgs& args, bool scalarRng)
+	std::unique_ptr<IGDMRModel> IGDMRModel::create(TermWeight _weight, const GDMRArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, GDMRModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/HDP.h

@@ -70,7 +70,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentHDP<TermWeight::one>;
-		static IHDPModel* create(TermWeight _weight, const HDPArgs& args,
+		static std::unique_ptr<IHDPModel> create(TermWeight _weight, const HDPArgs& args,
 			bool scalarRng = false);
 
 		virtual Float getGamma() const = 0;

data/vendor/tomotopy/src/TopicModel/HDPModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentHDP);
 
-    IHDPModel* IHDPModel::create(TermWeight _weight, const HDPArgs& args, bool scalarRng)
+	std::unique_ptr<IHDPModel> IHDPModel::create(TermWeight _weight, const HDPArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, HDPModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/HLDA.h

@@ -34,7 +34,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentHLDA<TermWeight::one>;
-		static IHLDAModel* create(TermWeight _weight, const HLDAArgs& args,
+		static std::unique_ptr<IHLDAModel> create(TermWeight _weight, const HLDAArgs& args,
 			bool scalarRng = false);
 
 		virtual Float getGamma() const = 0;

data/vendor/tomotopy/src/TopicModel/HLDAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentHLDA);
 
-	IHLDAModel* IHLDAModel::create(TermWeight _weight, const HLDAArgs& args, bool scalarRng)
+	std::unique_ptr<IHLDAModel> IHLDAModel::create(TermWeight _weight, const HLDAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, HLDAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/HPA.h

@@ -24,7 +24,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentHPA<TermWeight::one>;
-		static IHPAModel* create(TermWeight _weight, bool _exclusive, const HPAArgs& args,
+		static std::unique_ptr<IHPAModel> create(TermWeight _weight, bool _exclusive, const HPAArgs& args,
 			bool scalarRng = false);
 	};
 }

data/vendor/tomotopy/src/TopicModel/HPAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentHPA);
 
-    IHPAModel* IHPAModel::create(TermWeight _weight, bool _exclusive, const HPAArgs& args, bool scalarRng)
+	std::unique_ptr<IHPAModel> IHPAModel::create(TermWeight _weight, bool _exclusive, const HPAArgs& args, bool scalarRng)
 	{
 		if (_exclusive)
 		{

data/vendor/tomotopy/src/TopicModel/LDA.h

@@ -179,7 +179,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentLDA<TermWeight::one>;
-		static ILDAModel* create(TermWeight _weight, const LDAArgs& args,
+		static std::unique_ptr<ILDAModel> create(TermWeight _weight, const LDAArgs& args,
 			bool scalarRng = false);
 
 		virtual TermWeight getTermWeight() const = 0;

data/vendor/tomotopy/src/TopicModel/LDAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
     TMT_INSTANTIATE_DOC(DocumentLDA);
 
-    ILDAModel* ILDAModel::create(TermWeight _weight, const LDAArgs& args, bool scalarRng)
+    std::unique_ptr<ILDAModel> ILDAModel::create(TermWeight _weight, const LDAArgs& args, bool scalarRng)
     {
         TMT_SWITCH_TW(_weight, scalarRng, LDAModel, args);
     }

data/vendor/tomotopy/src/TopicModel/LDAModel.hpp

@@ -24,11 +24,11 @@ Term Weighting Scheme is based on following paper:
 		{\
 			switch (TW){\
 			case TermWeight::one:\
-				return new MDL<TermWeight::one, ScalarRandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::one, ScalarRandGen>>(__VA_ARGS__);\
 			case TermWeight::idf:\
-				return new MDL<TermWeight::idf, ScalarRandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::idf, ScalarRandGen>>(__VA_ARGS__);\
 			case TermWeight::pmi:\
-				return new MDL<TermWeight::pmi, ScalarRandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::pmi, ScalarRandGen>>(__VA_ARGS__);\
 			}\
 		}\
 		return nullptr; } while(0)
@@ -37,11 +37,11 @@ Term Weighting Scheme is based on following paper:
 		{\
 			switch (TW){\
 			case TermWeight::one:\
-				return new MDL<TermWeight::one, RandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::one, RandGen>>(__VA_ARGS__);\
 			case TermWeight::idf:\
-				return new MDL<TermWeight::idf, RandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::idf, RandGen>>(__VA_ARGS__);\
 			case TermWeight::pmi:\
-				return new MDL<TermWeight::pmi, RandGen>(__VA_ARGS__);\
+				return std::make_unique<MDL<TermWeight::pmi, RandGen>>(__VA_ARGS__);\
 			}\
 		}\
 		return nullptr; } while(0)

data/vendor/tomotopy/src/TopicModel/LLDA.h

@@ -19,7 +19,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentLLDA<TermWeight::one>;
-		static ILLDAModel* create(TermWeight _weight, const LDAArgs& args,
+		static std::unique_ptr<ILLDAModel> create(TermWeight _weight, const LDAArgs& args,
 			bool scalarRng = false);
 
 		virtual const Dictionary& getTopicLabelDict() const = 0;

data/vendor/tomotopy/src/TopicModel/LLDAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentLLDA);
 
-	ILLDAModel* ILLDAModel::create(TermWeight _weight, const LDAArgs& args, bool scalarRng)
+	std::unique_ptr<ILLDAModel> ILLDAModel::create(TermWeight _weight, const LDAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, LLDAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/MGLDA.h

@@ -43,7 +43,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentMGLDA<TermWeight::one>;
-		static IMGLDAModel* create(TermWeight _weight, const MGLDAArgs& args,
+		static std::unique_ptr<IMGLDAModel> create(TermWeight _weight, const MGLDAArgs& args,
 			bool scalarRng = false);
 
 		virtual size_t getKL() const = 0;

data/vendor/tomotopy/src/TopicModel/MGLDAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentMGLDA);
 
-    IMGLDAModel* IMGLDAModel::create(TermWeight _weight, const MGLDAArgs& args, bool scalarRng)
+	std::unique_ptr<IMGLDAModel> IMGLDAModel::create(TermWeight _weight, const MGLDAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, MGLDAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/PA.h

@@ -29,7 +29,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentPA<TermWeight::one>;
-		static IPAModel* create(TermWeight _weight, const PAArgs& args,
+		static std::unique_ptr<IPAModel> create(TermWeight _weight, const PAArgs& args,
 			bool scalarRng = false);
 
 		virtual size_t getDirichletEstIteration() const = 0;

data/vendor/tomotopy/src/TopicModel/PAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentPA);
 
-	IPAModel* IPAModel::create(TermWeight _weight, const PAArgs& args, bool scalarRng)
+	std::unique_ptr<IPAModel> IPAModel::create(TermWeight _weight, const PAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, PAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/PLDA.h

@@ -20,7 +20,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentLLDA<TermWeight::one>;
-		static IPLDAModel* create(TermWeight _weight, const PLDAArgs& args,
+		static std::unique_ptr<IPLDAModel> create(TermWeight _weight, const PLDAArgs& args,
 			bool scalarRng = false);
 
 		virtual size_t getNumLatentTopics() const = 0;

data/vendor/tomotopy/src/TopicModel/PLDAModel.cpp

@@ -2,7 +2,7 @@
 
 namespace tomoto
 {
-	IPLDAModel* IPLDAModel::create(TermWeight _weight, const PLDAArgs& args, bool scalarRng)
+	std::unique_ptr<IPLDAModel> IPLDAModel::create(TermWeight _weight, const PLDAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, PLDAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/PT.h

@@ -26,7 +26,7 @@ namespace tomoto
 	{
 	public:
 		using DefaultDocType = DocumentPT<TermWeight::one>;
-		static IPTModel* create(TermWeight _weight, const PTArgs& args,
+		static std::unique_ptr<IPTModel> create(TermWeight _weight, const PTArgs& args,
 			bool scalarRng = false);
 
 		virtual size_t getP() const = 0;

data/vendor/tomotopy/src/TopicModel/PTModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentPT);
 
-	IPTModel* IPTModel::create(TermWeight _weight, const PTArgs& args, bool scalarRng)
+	std::unique_ptr<IPTModel> IPTModel::create(TermWeight _weight, const PTArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, PTModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/SLDA.h

@@ -33,7 +33,7 @@ namespace tomoto
 		};
 
 		using DefaultDocType = DocumentSLDA<TermWeight::one>;
-		static ISLDAModel* create(TermWeight _weight, const SLDAArgs& args,
+		static std::unique_ptr<ISLDAModel> create(TermWeight _weight, const SLDAArgs& args,
 			bool scalarRng = false);
 
 		virtual size_t getF() const = 0;

data/vendor/tomotopy/src/TopicModel/SLDAModel.cpp

@@ -7,7 +7,7 @@ namespace tomoto
 
 	TMT_INSTANTIATE_DOC(DocumentSLDA);
 
-    ISLDAModel* ISLDAModel::create(TermWeight _weight, const SLDAArgs& args, bool scalarRng)
+	std::unique_ptr<ISLDAModel> ISLDAModel::create(TermWeight _weight, const SLDAArgs& args, bool scalarRng)
 	{
 		TMT_SWITCH_TW(_weight, scalarRng, SLDAModel, args);
 	}

data/vendor/tomotopy/src/TopicModel/TopicModel.hpp

@@ -1,6 +1,7 @@
 #pragma once
 #include <numeric>
 #include <unordered_set>
+#include <variant>
 #include "../Utils/Utils.hpp"
 #include "../Utils/Dictionary.h"
 #include "../Utils/tvector.hpp"
@@ -9,7 +10,6 @@
 #include "../Utils/exception.h"
 #include "../Utils/SharedString.h"
 #include <EigenRand/EigenRand>
-#include <mapbox/variant.hpp>
 
 namespace tomoto
 {
@@ -38,7 +38,7 @@ namespace tomoto
 
 	struct RawDoc : public RawDocKernel
 	{
-		using Var = mapbox::util::variant<
+		using Var = std::variant<
 			std::string, uint32_t, Float,
 			std::vector<std::string>, std::vector<uint32_t>, std::vector<Float>,
 			std::shared_ptr<void>
@@ -63,8 +63,8 @@ namespace tomoto
 		{
 			auto it = misc.find(name);
 			if (it == misc.end()) throw exc::InvalidArgument{ "There is no value named `" + name + "` in misc data" };
-			if (!it->second.template is<_Ty>()) throw exc::InvalidArgument{ "Value named `" + name + "` is not in right type." };
-			return it->second.template get<_Ty>();
+			if (!std::holds_alternative<_Ty>(it->second)) throw exc::InvalidArgument{ "Value named `" + name + "` is not in right type." };
+			return std::get<_Ty>(it->second);
 		}
 
 		template<typename _Ty>
@@ -72,8 +72,8 @@ namespace tomoto
 		{
 			auto it = misc.find(name);
 			if (it == misc.end()) return {};
-			if (!it->second.template is<_Ty>()) throw exc::InvalidArgument{ "Value named `" + name + "` is not in right type." };
-			return it->second.template get<_Ty>();
+			if (!std::holds_alternative<_Ty>(it->second)) throw exc::InvalidArgument{ "Value named `" + name + "` is not in right type." };
+			return std::get<_Ty>(it->second);
 		}
 	};
 

data/vendor/tomotopy/src/Utils/EigenAddonOps.hpp

@@ -56,6 +56,47 @@ namespace Eigen
 	}
 }
 #endif
+#ifdef EIGEN_VECTORIZE_AVX512
+#include <immintrin.h>
+#include "avx512_gamma.h"
+
+namespace Eigen
+{
+	namespace internal
+	{
+		template<> struct to_int_packet<Packet16f>
+		{
+			typedef Packet16i type;
+		};
+
+		template<> struct to_float_packet<Packet16i>
+		{
+			typedef Packet16f type;
+		};
+
+		EIGEN_STRONG_INLINE Packet16f p_to_f32(const Packet16i& a)
+		{
+			return _mm512_cvtepi32_ps(a);
+		}
+
+		EIGEN_STRONG_INLINE Packet16f p_bool2float(const Packet16f& a)
+		{
+			__mmask16 mask = _mm512_cmp_ps_mask(a, _mm512_setzero_ps(), _CMP_NEQ_OQ);
+			return _mm512_maskz_mov_ps(mask, _mm512_set1_ps(1.f));
+		}
+
+		EIGEN_STRONG_INLINE Packet16f p_bool2float(const Packet16i& a)
+		{
+			return p_bool2float(_mm512_castsi512_ps(a));
+		}
+	}
+}
+
+inline __m512 lgamma_subt(__m512 z, __m512 a)
+{
+	return lgamma512_subt(z, a);
+}
+#endif
 #ifdef EIGEN_VECTORIZE_SSE2
 #include <xmmintrin.h>
 #include "sse_gamma.h"

data/vendor/tomotopy/src/Utils/ThreadPool.hpp

@@ -22,11 +22,11 @@ namespace tomoto
 		ThreadPool(size_t threads = 0, size_t maxQueued = 0);
 		template<class F, class... Args>
 		auto enqueue(F&& f, Args&&... args)
-			->std::future<typename std::result_of<F(size_t, Args...)>::type>;
+			->std::future<typename std::invoke_result_t<F, size_t, Args...>>;
 		
 		template<class F, class... Args>
 		auto enqueueToAll(F&& f, Args&&... args)
-			->std::vector<std::future<typename std::result_of<F(size_t, Args...)>::type>>;
+			->std::vector<std::future<typename std::invoke_result_t<F, size_t, Args...>>>;
 
 		~ThreadPool();
 		
@@ -88,9 +88,9 @@ namespace tomoto
 	// add new work item to the pool
 	template<class F, class... Args>
 	auto ThreadPool::enqueue(F&& f, Args&&... args)
-		-> std::future<typename std::result_of<F(size_t, Args...)>::type>
+		-> std::future<typename std::invoke_result_t<F, size_t, Args...>>
 	{
-		using return_type = typename std::result_of<F(size_t, Args...)>::type;
+		using return_type = typename std::invoke_result_t<F, size_t, Args...>;
 
 		auto task = std::make_shared< std::packaged_task<return_type(size_t)> >(
 			std::bind(std::forward<F>(f), std::placeholders::_1, std::forward<Args>(args)...));
@@ -113,9 +113,9 @@ namespace tomoto
 
 	template<class F, class... Args>
 	auto ThreadPool::enqueueToAll(F&& f, Args&&... args)
-		->std::vector<std::future<typename std::result_of<F(size_t, Args...)>::type> >
+		->std::vector<std::future<typename std::invoke_result_t<F, size_t, Args...>> >
 	{
-		using return_type = typename std::result_of<F(size_t, Args...)>::type;
+		using return_type = typename std::invoke_result_t<F, size_t, Args...>;
 
 		std::vector<std::future<return_type> > ret;
 		std::unique_lock<std::mutex> lock(queue_mutex);

data/vendor/tomotopy/src/Utils/Utils.hpp

@@ -296,9 +296,9 @@ namespace tomoto
 	private:
 		_UnaryFunc f;
 	public:
-		using reference = typename std::result_of<
-			const _UnaryFunc(typename std::iterator_traits<_Iterator>::reference)
-		>::type;
+		using reference = std::invoke_result_t<
+			const _UnaryFunc, typename std::iterator_traits<_Iterator>::reference
+		>;
 		using value_type = reference;
 		
 		TransformIter(const _Iterator& _iter = {}, _UnaryFunc _f = {})

data/vendor/tomotopy/src/Utils/avx512_gamma.h

@@ -0,0 +1,46 @@
+#pragma once
+#include "avx512_mathfun.h"
+
+// approximation : lgamma(z) ~= (z+2.5)ln(z+3) - z - 3 + 0.5 ln (2pi) + 1/12/(z + 3) - ln (z(z+1)(z+2))
+inline __m512 lgamma512_ps(__m512 x)
+{
+	__m512 x_3 = _mm512_add_ps(x, _mm512_set1_ps(3));
+	__m512 ret = _mm512_mul_ps(_mm512_add_ps(x_3, _mm512_set1_ps(-0.5f)), log512_ps(x_3));
+	ret = _mm512_sub_ps(ret, x_3);
+	ret = _mm512_add_ps(ret, _mm512_set1_ps(0.91893853f));
+	ret = _mm512_add_ps(ret, _mm512_div_ps(_mm512_set1_ps(1 / 12.f), x_3));
+	ret = _mm512_sub_ps(ret, log512_ps(_mm512_mul_ps(
+		_mm512_mul_ps(_mm512_sub_ps(x_3, _mm512_set1_ps(1)), _mm512_sub_ps(x_3, _mm512_set1_ps(2))), x)));
+	return ret;
+}
+
+// approximation : lgamma(z + a) - lgamma(z) = (z + a + 1.5) * log(z + a + 2) - (z + 1.5) * log(z + 2) - a + (1. / (z + a + 2) - 1. / (z + 2)) / 12. - log(((z + a) * (z + a + 1)) / (z * (z + 1)))
+inline __m512 lgamma512_subt(__m512 z, __m512 a)
+{
+	__m512 _1p5 = _mm512_set1_ps(1.5f);
+	__m512 _2 = _mm512_set1_ps(2.f);
+	__m512 za = _mm512_add_ps(z, a);
+	__m512 ret = _mm512_mul_ps(_mm512_add_ps(za, _1p5), log512_ps(_mm512_add_ps(za, _2)));
+	ret = _mm512_sub_ps(ret, _mm512_mul_ps(_mm512_add_ps(z, _1p5), log512_ps(_mm512_add_ps(z, _2))));
+	ret = _mm512_sub_ps(ret, a);
+	__m512 _1 = _mm512_set1_ps(1.f);
+	__m512 _1_12 = _mm512_set1_ps(1 / 12.f);
+	ret = _mm512_add_ps(ret, _mm512_sub_ps(_mm512_div_ps(_1_12, _mm512_add_ps(za, _2)), _mm512_div_ps(_1_12, _mm512_add_ps(z, _2))));
+	ret = _mm512_sub_ps(ret, log512_ps(_mm512_div_ps(_mm512_div_ps(_mm512_mul_ps(za, _mm512_add_ps(za, _1)), z), _mm512_add_ps(z, _1))));
+	return ret;
+}
+
+
+// approximation : digamma(z) ~= ln(z+4) - 1/2/(z+4) - 1/12/(z+4)^2 - 1/z - 1/(z+1) - 1/(z+2) - 1/(z+3)
+inline __m512 digamma512_ps(__m512 x)
+{
+	__m512 x_4 = _mm512_add_ps(x, _mm512_set1_ps(4));
+	__m512 ret = log512_ps(x_4);
+	ret = _mm512_sub_ps(ret, _mm512_div_ps(_mm512_set1_ps(1 / 2.f), x_4));
+	ret = _mm512_sub_ps(ret, _mm512_div_ps(_mm512_div_ps(_mm512_set1_ps(1 / 12.f), x_4), x_4));
+	ret = _mm512_sub_ps(ret, _mm512_rcp14_ps(_mm512_sub_ps(x_4, _mm512_set1_ps(1))));
+	ret = _mm512_sub_ps(ret, _mm512_rcp14_ps(_mm512_sub_ps(x_4, _mm512_set1_ps(2))));
+	ret = _mm512_sub_ps(ret, _mm512_rcp14_ps(_mm512_sub_ps(x_4, _mm512_set1_ps(3))));
+	ret = _mm512_sub_ps(ret, _mm512_rcp14_ps(_mm512_sub_ps(x_4, _mm512_set1_ps(4))));
+	return ret;
+}

data/vendor/tomotopy/src/Utils/avx512_mathfun.h

@@ -0,0 +1,99 @@
+/*
+AVX-512 implementation of log function
+
+Based on "avx_mathfun.h" by Giovanni Garberoglio
+and "sse_mathfun.h" by Julien Pommier
+*/
+
+#pragma once
+#include <immintrin.h>
+
+#ifdef _MSC_VER
+#define ALIGN64_BEG __declspec(align(64))
+#define ALIGN64_END
+#else
+#define ALIGN64_BEG
+#define ALIGN64_END __attribute__((aligned(64)))
+#endif
+
+typedef __m512  v16sf;
+typedef __m512i v16si;
+
+#define _PS512_CONST(Name, Val) \
+  static const ALIGN64_BEG float _ps512_##Name[16] ALIGN64_END = { Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val }
+#define _PI32_CONST512(Name, Val) \
+  static const ALIGN64_BEG int _pi32_512_##Name[16] ALIGN64_END = { Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val }
+#define _PS512_CONST_TYPE(Name, Type, Val) \
+  static const ALIGN64_BEG Type _ps512_##Name[16] ALIGN64_END = { Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val, Val }
+
+_PS512_CONST(1, 1.0f);
+_PS512_CONST(0p5, 0.5f);
+_PS512_CONST_TYPE(min_norm_pos, unsigned int, 0x00800000u);
+_PS512_CONST_TYPE(inv_mant_mask, unsigned int, ~0x7f800000u);
+
+_PI32_CONST512(0x7f, 0x7f);
+
+_PS512_CONST(cephes_SQRTHF, 0.707106781186547524f);
+_PS512_CONST(cephes_log_p0, 7.0376836292E-2f);
+_PS512_CONST(cephes_log_p1, -1.1514610310E-1f);
+_PS512_CONST(cephes_log_p2, 1.1676998740E-1f);
+_PS512_CONST(cephes_log_p3, -1.2420140846E-1f);
+_PS512_CONST(cephes_log_p4, +1.4249322787E-1f);
+_PS512_CONST(cephes_log_p5, -1.6668057665E-1f);
+_PS512_CONST(cephes_log_p6, +2.0000714765E-1f);
+_PS512_CONST(cephes_log_p7, -2.4999993993E-1f);
+_PS512_CONST(cephes_log_p8, +3.3333331174E-1f);
+_PS512_CONST(cephes_log_q1, -2.12194440e-4f);
+_PS512_CONST(cephes_log_q2, 0.693359375f);
+
+/* natural logarithm computed for 16 simultaneous float */
+inline v16sf log512_ps(v16sf x) {
+	v16si imm0;
+	v16sf one = *(v16sf*)_ps512_1;
+
+	__mmask16 invalid_mask = _mm512_cmp_ps_mask(x, _mm512_setzero_ps(), _CMP_LE_OS);
+
+	x = _mm512_max_ps(x, *(v16sf*)_ps512_min_norm_pos);
+
+	imm0 = _mm512_srli_epi32(_mm512_castps_si512(x), 23);
+
+	x = _mm512_and_ps(x, *(v16sf*)_ps512_inv_mant_mask);
+	x = _mm512_or_ps(x, *(v16sf*)_ps512_0p5);
+
+	imm0 = _mm512_sub_epi32(imm0, *(v16si*)_pi32_512_0x7f);
+	v16sf e = _mm512_cvtepi32_ps(imm0);
+
+	e = _mm512_add_ps(e, one);
+
+	__mmask16 mask = _mm512_cmp_ps_mask(x, *(v16sf*)_ps512_cephes_SQRTHF, _CMP_LT_OS);
+	v16sf tmp = _mm512_maskz_mov_ps(mask, x);
+	x = _mm512_sub_ps(x, one);
+	e = _mm512_mask_sub_ps(e, mask, e, one);
+	x = _mm512_add_ps(x, tmp);
+
+	v16sf z = _mm512_mul_ps(x, x);
+
+	v16sf y = *(v16sf*)_ps512_cephes_log_p0;
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p1);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p2);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p3);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p4);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p5);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p6);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p7);
+	y = _mm512_fmadd_ps(y, x, *(v16sf*)_ps512_cephes_log_p8);
+	y = _mm512_mul_ps(y, x);
+
+	y = _mm512_mul_ps(y, z);
+
+	y = _mm512_fmadd_ps(e, *(v16sf*)_ps512_cephes_log_q1, y);
+
+	tmp = _mm512_mul_ps(z, *(v16sf*)_ps512_0p5);
+	y = _mm512_sub_ps(y, tmp);
+
+	x = _mm512_fmadd_ps(e, *(v16sf*)_ps512_cephes_log_q2, _mm512_add_ps(x, y));
+
+	// negative arg will be NAN
+	x = _mm512_mask_blend_ps(invalid_mask, x, _mm512_set1_ps(__builtin_nanf("")));
+	return x;
+}

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tomoto
 version: !ruby/object:Gem::Version
-  version: 0.6.0
+  version: 0.6.2
 platform: ruby
 authors:
 - Andrew Kane
@@ -64,6 +64,7 @@ files:
 - lib/tomoto/slda.rb
 - lib/tomoto/version.rb
 - vendor/EigenRand/EigenRand/Core.h
+- vendor/EigenRand/EigenRand/CwiseHeteroBinaryOp.h
 - vendor/EigenRand/EigenRand/Dists/Basic.h
 - vendor/EigenRand/EigenRand/Dists/Discrete.h
 - vendor/EigenRand/EigenRand/Dists/GammaPoisson.h
@@ -79,6 +80,9 @@ files:
 - vendor/EigenRand/EigenRand/arch/AVX/MorePacketMath.h
 - vendor/EigenRand/EigenRand/arch/AVX/PacketFilter.h
 - vendor/EigenRand/EigenRand/arch/AVX/RandUtils.h
+- vendor/EigenRand/EigenRand/arch/AVX512/MorePacketMath.h
+- vendor/EigenRand/EigenRand/arch/AVX512/PacketFilter.h
+- vendor/EigenRand/EigenRand/arch/AVX512/RandUtils.h
 - vendor/EigenRand/EigenRand/arch/NEON/MorePacketMath.h
 - vendor/EigenRand/EigenRand/arch/NEON/PacketFilter.h
 - vendor/EigenRand/EigenRand/arch/NEON/RandUtils.h
@@ -447,6 +451,8 @@ files:
 - vendor/tomotopy/LICENSE
 - vendor/tomotopy/README.kr.rst
 - vendor/tomotopy/README.rst
+- vendor/tomotopy/README_pypi.rst
+- vendor/tomotopy/licenses_bundled/EigenRand
 - vendor/tomotopy/src/Labeling/FoRelevance.cpp
 - vendor/tomotopy/src/Labeling/FoRelevance.h
 - vendor/tomotopy/src/Labeling/Labeler.h
@@ -516,6 +522,8 @@ files:
 - vendor/tomotopy/src/Utils/Trie.hpp
 - vendor/tomotopy/src/Utils/TruncMultiNormal.hpp
 - vendor/tomotopy/src/Utils/Utils.hpp
+- vendor/tomotopy/src/Utils/avx512_gamma.h
+- vendor/tomotopy/src/Utils/avx512_mathfun.h
 - vendor/tomotopy/src/Utils/avx_gamma.h
 - vendor/tomotopy/src/Utils/avx_mathfun.h
 - vendor/tomotopy/src/Utils/exception.h
@@ -530,13 +538,6 @@ files:
 - vendor/tomotopy/src/Utils/sse_mathfun.h
 - vendor/tomotopy/src/Utils/text.hpp
 - vendor/tomotopy/src/Utils/tvector.hpp
-- vendor/variant/LICENSE
-- vendor/variant/LICENSE_1_0.txt
-- vendor/variant/README.md
-- vendor/variant/include/mapbox/optional.hpp
-- vendor/variant/include/mapbox/recursive_wrapper.hpp
-- vendor/variant/include/mapbox/variant.hpp
-- vendor/variant/include/mapbox/variant_io.hpp
 homepage: https://github.com/ankane/tomoto-ruby
 licenses:
 - MIT
@@ -555,7 +556,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.9
+rubygems_version: 4.0.3
 specification_version: 4
 summary: High performance topic modeling for Ruby
 test_files: []

data/vendor/variant/LICENSE

@@ -1,25 +0,0 @@
-Copyright (c) MapBox
-All rights reserved.
-
-Redistribution and use in source and binary forms, with or without modification,
-are permitted provided that the following conditions are met:
-
-- Redistributions of source code must retain the above copyright notice, this
-  list of conditions and the following disclaimer.
-- Redistributions in binary form must reproduce the above copyright notice, this
-  list of conditions and the following disclaimer in the documentation and/or
-  other materials provided with the distribution.
-- Neither the name "MapBox" nor the names of its contributors may be
-  used to endorse or promote products derived from this software without
-  specific prior written permission.
-
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
-ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
-WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
-DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR
-ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
-(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
-LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
-ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
-SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

data/vendor/variant/LICENSE_1_0.txt

@@ -1,23 +0,0 @@
-Boost Software License - Version 1.0 - August 17th, 2003
-
-Permission is hereby granted, free of charge, to any person or organization
-obtaining a copy of the software and accompanying documentation covered by
-this license (the "Software") to use, reproduce, display, distribute,
-execute, and transmit the Software, and to prepare derivative works of the
-Software, and to permit third-parties to whom the Software is furnished to
-do so, all subject to the following:
-
-The copyright notices in the Software and this entire statement, including
-the above license grant, this restriction and the following disclaimer,
-must be included in all copies of the Software, in whole or in part, and
-all derivative works of the Software, unless such copies or derivative
-works are solely in the form of machine-executable object code generated by
-a source language processor.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
-SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
-FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
-ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
-DEALINGS IN THE SOFTWARE.