now working for finite exchange frequency (71f005e2) · Commits · NDIP / Tool Sources / Direct-Geometry Spectroscopy / DCA / DCA Main

include/dca/linalg/reshapable_matrix.hpp

+53 −10

Original line number	Diff line number	Diff line
		@@ -100,6 +100,49 @@ public:
		return data_[i + j * leadingDimension()];
		}

		struct Iterator {
		using iterator_category = std::forward_iterator_tag;
		using difference_type = std::ptrdiff_t;
		using value_type = ScalarType;
		using pointer = ScalarType*;
		using reference = ScalarType&;

		Iterator(pointer ptr) : m_ptr(ptr) {}

		reference operator*() const {
		return *m_ptr;
		}
		pointer operator->() {
		return m_ptr;
		}

		Iterator& operator++() {
		m_ptr++;
		return *this;
		}
		Iterator operator++(int) {
		Iterator tmp = *this;
		++(*this);
		return tmp;
		}
		friend bool operator==(const Iterator& a, const Iterator& b) {
		return a.m_ptr == b.m_ptr;
		};
		friend bool operator!=(const Iterator& a, const Iterator& b) {
		return a.m_ptr != b.m_ptr;
		};

		private:
		pointer m_ptr;
		};

		Iterator begin() {
		return Iterator(data_);
		}
		Iterator end() {
		return Iterator(data_ + size_.first * size_.second);
		}

		// Returns the pointer to the (0,0)-th element.
		ValueType* ptr() {
		return data_;
		@@ -221,7 +264,8 @@ template <typename ScalarRhs, DeviceType rhs_device_name, class AllocatorRhs>
		ReshapableMatrix<ScalarType, device_name, Allocator>& ReshapableMatrix<
		ScalarType, device_name,
		Allocator>::operator=(const ReshapableMatrix<ScalarRhs, rhs_device_name, AllocatorRhs>& rhs) {
		static_assert(sizeof(ScalarType) == sizeof(ScalarRhs), "sizeof ScalarType and ScalarRhs are not equal");
		static_assert(sizeof(ScalarType) == sizeof(ScalarRhs),
		"sizeof ScalarType and ScalarRhs are not equal");
		if constexpr (device_name == rhs_device_name)
		if (this != &rhs)
		return *this;
		@@ -257,8 +301,7 @@ ReshapableMatrix<ScalarType, device_name, Allocator>& ReshapableMatrix<
		}

		template <typename ScalarType, DeviceType device_name, class Allocator>
		ReshapableMatrix<
		ScalarType, device_name, Allocator>::ReshapableMatrix(ThisType&& rhs) {
		ReshapableMatrix<ScalarType, device_name, Allocator>::ReshapableMatrix(ThisType&& rhs) {
		swap(rhs);
		}

		@@ -393,8 +436,8 @@ std::size_t ReshapableMatrix<ScalarType, device_name, Allocator>::deviceFingerpr
		}

		template <typename ScalarType, DeviceType device_name, class Allocator>
		std::ostream& operator<<(std::ostream& ostr, const ReshapableMatrix<ScalarType, device_name, Allocator>& rmatrix)
		{
		std::ostream& operator<<(std::ostream& ostr,
		const ReshapableMatrix<ScalarType, device_name, Allocator>& rmatrix) {
		ostr << "{";
		for (std::size_t i = 0; i < rmatrix.size().first; ++i) {
		ostr << "{";

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/g4_helper.cuh

+4 −1

Original line number	Diff line number	Diff line
		@@ -32,9 +32,12 @@ namespace details {

		class G4Helper {
		public:
		/** pass a bunch of information into the G4Helper so it can help./
		* \param[in] extension_offset (WTpExtDmn::dmn_size() - WTpDmn::dmn_size()) / 2
		*/
		static void set(int nb, int nk, int nw,
		const std::vector<int>& delta_k, const std::vector<int>& delta_w,
		const int* add_k, int lda, const int* sub_k, int lds);
		const int extension_offset, const int* add_k, int lda, const int* sub_k, int lds);

		__device__ auto get_bands() const {
		return nb_;

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/tp_accumulator_base.hpp

+2 −0

Original line number	Diff line number	Diff line
		@@ -197,6 +197,8 @@ TpAccumulatorBase<Parameters, DT>::TpAccumulatorBase(
		if (WDmn::dmn_size() < WTpExtDmn::dmn_size())
		throw(std::logic_error("The number of single particle frequencies is too small."));

		std::cout << "WTpExtDmn Elements: " << vectorToString(WTpExtDmn::get_elements()) << '\n';

		initializeG0();

		// Reserve storage in advance such that we don't have to copy elements when we fill the vector.

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/tp_accumulator_cpu.hpp

+4 −0

Original line number	Diff line number	Diff line
		@@ -456,6 +456,8 @@ double TpAccumulator<Parameters, DT, linalg::CPU>::updateG4(const int channel_id
		// = -1/2 sum_s G(k2+k_ex, k1+k_ex, s) G(k1, k2, -s)
		for (int w_ex_idx = 0; w_ex_idx < exchange_frq.size(); ++w_ex_idx) {
		const int w_ex = exchange_frq[w_ex_idx];
		if(w_ex != 0)
		std::cout << "cpu w_ex: " << w_ex << '\n';
		for (int k_ex_idx = 0; k_ex_idx < exchange_mom.size(); ++k_ex_idx) {
		const int k_ex = exchange_mom[k_ex_idx];
		for (int w2 = 0; w2 < WTpDmn::dmn_size(); ++w2)
		@@ -481,6 +483,8 @@ double TpAccumulator<Parameters, DT, linalg::CPU>::updateG4(const int channel_id
		// - (s1 == s2) G(k2+k_ex, k1+k_ex, s1) G(k1, k2, s1)]
		for (int w_ex_idx = 0; w_ex_idx < exchange_frq.size(); ++w_ex_idx) {
		const int w_ex = exchange_frq[w_ex_idx];
		if(w_ex != 0)
		std::cout << "cpu w_ex: " << w_ex << '\n';
		for (int k_ex_idx = 0; k_ex_idx < exchange_mom.size(); ++k_ex_idx) {
		const int k_ex = exchange_mom[k_ex_idx];
		for (int w2 = 0; w2 < WTpDmn::dmn_size(); ++w2)

include/dca/phys/dca_step/cluster_solver/shared_tools/accumulation/tp/tp_accumulator_gpu.hpp

+1 −0

Original line number	Diff line number	Diff line
		@@ -394,6 +394,7 @@ void TpAccumulator<Parameters, DT, linalg::GPU>::computeGSingleband(const int s)

		template <class Parameters, DistType DT>
		void TpAccumulator<Parameters, DT, linalg::GPU>::computeGMultiband(const int s) {
		std::cout << "WTpExtDmn::dmn_size(): " << WTpExtDmn::dmn_size() << '\n';
		details::computeGMultiband(G_[s].ptr(), G_[s].leadingDimension(), get_G0()[s].ptr(),
		get_G0()[s].leadingDimension(), n_bands_, KDmn::dmn_size(),
		WTpExtDmn::dmn_size(), beta_, queues_[s]);