Merge pull request #129 from marty1885/apichange

more indexing method

Etaler/Core/Tensor.hpp

@@ -211,15 +211,15 @@ struct ETALER_EXPORT Tensor
 	Tensor log() const { return backend()->log(pimpl()); }
 	Tensor logical_not() const { return backend()->logical_not(pimpl()); }
 
-	Tensor add(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->add(a(), b()); }
-	Tensor subtract(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->subtract(a(), b()); }
-	Tensor mul(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->mul(a(), b()); }
-	Tensor div(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->div(a(), b()); }
-	Tensor equal(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->equal(a(), b()); }
-	Tensor greater(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->greater(a(), b()); }
-	Tensor lesser(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->lesser(a(), b()); }
-	Tensor logical_and(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->logical_and(a(), b()); }
-	Tensor logical_or(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->logical_or(a(), b()); }
+	Tensor add(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->add(a.pimpl(), b.pimpl()); }
+	Tensor subtract(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->subtract(a.pimpl(), b.pimpl()); }
+	Tensor mul(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->mul(a.pimpl(), b.pimpl()); }
+	Tensor div(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->div(a.pimpl(), b.pimpl()); }
+	Tensor equal(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->equal(a.pimpl(), b.pimpl()); }
+	Tensor greater(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->greater(a.pimpl(), b.pimpl()); }
+	Tensor lesser(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->lesser(a.pimpl(), b.pimpl()); }
+	Tensor logical_and(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->logical_and(a.pimpl(), b.pimpl()); }
+	Tensor logical_or(const Tensor& other) const { auto [a, b] = brodcast(other); return backend()->logical_or(a.pimpl(), b.pimpl()); }
 
 	inline bool any() const { return cast(DType::Bool).sum(std::nullopt, DType::Bool).item<uint8_t>(); }
 	inline bool all() const { return cast(DType::Bool).sum(std::nullopt).item<int32_t>() == int32_t(size()); }
@@ -252,14 +252,14 @@ struct ETALER_EXPORT Tensor
 
 	//Subscription operator
 	Tensor operator [] (const IndexList& r) { return view(r); }
+	template <typename ... Args>
+	Tensor operator () (Args ... args) { return view({args ...}); }
 
 	Tensor sum(std::optional<intmax_t> dim=std::nullopt, DType dtype=DType::Unknown) const;
 	Tensor abs() const { return backend()->abs(pimpl()); }
 	bool isSame (const Tensor& other) const;
 
 	//Utils
-	TensorImpl* operator () () {return pimpl();}
-	const TensorImpl* operator () () const {return pimpl();}
 
 	using iterator = TensorIterator<Tensor>;
 	using const_iterator = TensorIterator<const Tensor>;
@@ -332,18 +332,18 @@ inline Tensor cellActivity(const Tensor& x, const Tensor& connections, const Ten
 			return x;
 		return x.cast(DType::Bool);
 	}();
-	return x.backend()->cellActivity(input(), connections(), permeances(), connected_permeance, active_threshold, has_unconnected_synapse);
+	return x.backend()->cellActivity(input.pimpl(), connections.pimpl(), permeances.pimpl(), connected_permeance, active_threshold, has_unconnected_synapse);
 }
 
 inline void learnCorrilation(const Tensor& x, const Tensor& learn, const Tensor& connection
 	, Tensor& permeances, float perm_inc, float perm_dec, bool has_unconnected_synapse=true)
 {
-	x.backend()->learnCorrilation(x(), learn(), connection(), permeances(), perm_inc, perm_dec, has_unconnected_synapse);
+	x.backend()->learnCorrilation(x.pimpl(), learn.pimpl(), connection.pimpl(), permeances.pimpl(), perm_inc, perm_dec, has_unconnected_synapse);
 }
 
 inline Tensor globalInhibition(const Tensor& x, float fraction)
 {
-	return x.backend()->globalInhibition(x(), fraction);
+	return x.backend()->globalInhibition(x.pimpl(), fraction);
 }
 
 Tensor inline cast(const Tensor& x, DType dtype)
@@ -358,27 +358,27 @@ inline Tensor copy(const Tensor& x)
 
 inline void sortSynapse(Tensor& connection, Tensor& permeances)
 {
-	connection.backend()->sortSynapse(connection(), permeances());
+	connection.backend()->sortSynapse(connection.pimpl(), permeances.pimpl());
 }
 
 inline Tensor burst(const Tensor& x, const Tensor& s)
 {
-	return x.backend()->burst(x(), s());
+	return x.backend()->burst(x.pimpl(), s.pimpl());
 }
 
 inline Tensor reverseBurst(const Tensor& x)
 {
-	return x.backend()->reverseBurst(x());
+	return x.backend()->reverseBurst(x.pimpl());
 }
 
 inline void growSynapses(const Tensor& x, const Tensor& y, Tensor& connections, Tensor& permeances, float init_perm)
 {
-	x.backend()->growSynapses(x(), y(), connections(), permeances(), init_perm);
+	x.backend()->growSynapses(x.pimpl(), y.pimpl(), connections.pimpl(), permeances.pimpl(), init_perm);
 }
 
 inline void decaySynapses(Tensor& connections, Tensor& permeances, float threshold)
 {
-	connections.backend()->decaySynapses(connections(), permeances(), threshold);
+	connections.backend()->decaySynapses(connections.pimpl(), permeances.pimpl(), threshold);
 }
 
 inline void assign(Tensor& x, const Tensor& y)
@@ -420,6 +420,10 @@ inline Tensor logical_or(const Tensor& x1, const Tensor& x2) { return x1.logical
 inline bool all(const Tensor& t) { return t.all(); }
 inline bool any(const Tensor& t) { return t.any(); }
 
+template <typename ... Args>
+inline Tensor view(const Tensor& t, Args... args) { return t.view({args...}); }
+inline Tensor dynamic_view(const Tensor& t, const IndexList& indices) { return t.view(indices); }
+
 inline Tensor zeros_like(const Tensor& x) { return zeros(x.shape(), x.dtype(), x.backend()); }
 inline Tensor ones_like(const Tensor& x) { return ones(x.shape(), x.dtype(), x.backend()); }
 }

docs/source/PythonBindings.md

@@ -1,7 +1,7 @@
 # Python bindings
 
 ## PyEtaler
-[PyEtaler](https://guthub.com/etaler/pyetaler) is the offical binding for Etaler. We try to keep the Python API as close to the C++ one as possible. So you can use the C++ document as the Python document. With that said, some functions are changed in the binding to make it more Pythonic.
+[PyEtaler](https://github.com/etaler/pyetaler) is the offical binding for Etaler. We try to keep the Python API as close to the C++ one as possible. So you can use the C++ document as the Python document. With that said, some functions are changed in the binding to make it more Pythonic.
 
 ```python
 >>> from etaler import et

tests/common_tests.cpp

@@ -9,7 +9,6 @@
 #include <Etaler/Algorithms/SDRClassifer.hpp>
 
 #include <numeric>
-#include <execution>
 
 using namespace et;
 
@@ -320,6 +319,14 @@ TEST_CASE("Testing Tensor", "[Tensor]")
 			CHECK((ones({4,4}) == t).any() == true);
 			CHECK((ones({4,4}) == t).all() == false);
 		}
+
+		SECTION("xtensor style views") {
+			CHECK(view(t, 2).isSame(t.view({2})));
+
+			IndexList lst;
+			lst.push_back(3);
+			CHECK(dynamic_view(t, lst).isSame(t.view({3})));
+		}
 	}
 
 	SECTION("item") {
@@ -994,11 +1001,8 @@ TEST_CASE("Complex Tensor operations")
 		// Test summing along the first dimension. Making sure iterator and sum() works
 		// Tho you should always use the sum() function instead of accumulate or reduce
 		Tensor t = std::accumulate(a.begin(), a.end(), zeros({a.shape()[1]}));
-		Tensor q = std::reduce(std::execution::par, a.begin(), a.end(), zeros({a.shape()[1]}));
 		Tensor a_sum = a.sum(0);
 		CHECK(t.isSame(a_sum));
-		CHECK(q.isSame(a_sum));
-		CHECK(t.isSame(q)); // Should be communicative
 	}
 
 	SECTION("generate") {