Removes Var Kind constructors

apps/tensor_times_vector/tensor_times_vector.cpp

@@ -26,7 +26,7 @@ int main(int argc, char* argv[]) {
   c.pack();
 
   // Form a tensor-vector multiplication expression
-  Var i, j, k(Var::Sum);
+  Var i, j, k;
   A(i,j) = B(i,j,k) * c(k);
 
   // Compile the expression

include/taco/expr.h

@@ -23,35 +23,15 @@ class ExprVisitorStrict;
 /// represent iteration over a tensor dimension.
 class Var : public util::Comparable<Var> {
 public:
-  enum Kind { Free, Sum };
+  Var();
+  Var(const std::string& name);
 
-private:
-  struct Content {
-    Var::Kind   kind;
-    std::string name;
-  };
-
-public:
-  Var(Kind kind = Kind::Free);
-  Var(const std::string& name, Kind kind = Kind::Free);
-
-  std::string getName() const {return content->name;}
-
-  Kind getKind() const {return content->kind;}
-
-  bool isFree() const {return content->kind == Free;}
-
-  bool isReduction() const {return content->kind != Free;}
-
-  friend bool operator==(const Var& l, const Var& r) {
-    return l.content == r.content;
-  }
-
-  friend bool operator<(const Var& l, const Var& r) {
-    return l.content < r.content;
-  }
+  std::string getName() const;
+  friend bool operator==(const Var&, const Var&);
+  friend bool operator<(const Var&, const Var&);
 
 private:
+  struct Content;
   std::shared_ptr<Content> content;
 };
 

src/expr.cpp

@@ -8,12 +8,26 @@ using namespace std;
 namespace taco {
 
 // class Var
-Var::Var(const std::string& name, Kind kind) : content(new Content) {
+struct Var::Content {
+  std::string name;
+};
+
+Var::Var() : Var(util::uniqueName('i')) {}
+
+Var::Var(const std::string& name) : content(new Content) {
   content->name = name;
-  content->kind = kind;
 }
 
-Var::Var(Kind kind) : Var(util::uniqueName('i'), kind) {
+std::string Var::getName() const {
+  return content->name;
+}
+
+bool operator==(const Var& a, const Var& b) {
+  return a.content == b.content;
+}
+
+bool operator<(const Var& a, const Var& b) {
+  return a.content < b.content;
 }
 
 std::ostream& operator<<(std::ostream& os, const Var& var) {

src/parser/parser.cpp

@@ -313,7 +313,7 @@ bool Parser::hasIndexVar(std::string name) const {
 Var Parser::getIndexVar(string name) const {
   taco_iassert(name != "");
   if (!hasIndexVar(name)) {
-    Var var(name, (content->parsingLhs ? Var::Free : Var::Sum));
+    Var var(name);
     content->indexVars.insert({name, var});
 
     // dimensionSizes can also store index var sizes

src/tensor.cpp

@@ -536,12 +536,6 @@ void TensorBase::setExpr(const vector<taco::Var>& indexVars, taco::Expr expr) {
     })
   );
 
-  // Check that the index variables on the left-hand-side are free
-  for (auto& indexVar : indexVars) {
-    taco_uassert(indexVar.getKind() == Var::Free) <<
-        "Can only use free index variables to index the left-hand-side";
-  }
-
   // The following are index expressions we don't currently support, but that
   // are planned for the future.
   // We don't yet support distributing tensors. That is, every free variable

test/expr_factory.cpp

@@ -40,7 +40,7 @@ MatrixMultiplyFactory::operator()(Tensors& operands, Format outFormat) {
   Tensor<double> A({operands[0].getDimensions()[0],
                     operands[1].getDimensions()[1]}, outFormat);
 
-  Var i("i"), j("j"), k("k", Var::Sum);
+  Var i("i"), j("j"), k("k");
   A(i,j) = operands[0](i,k) * operands[1](k,j);
 
   return A;
@@ -53,7 +53,7 @@ MatrixTransposeMultiplyFactory::operator()(Tensors& operands,
 
   Tensor<double> A(operands[0].getDimensions(), outFormat);
 
-  Var i("i"), j("j"), k("k", Var::Sum);
+  Var i("i"), j("j"), k("k");
   A(i,j) = operands[0](k,i) * operands[0](k,j);
 
   return A;
@@ -66,7 +66,7 @@ MatrixColumnSquaredNormFactory::operator()(Tensors& operands,
 
   Tensor<double> A({operands[0].getDimensions()[1]}, outFormat);
 
-  Var i("i"), j("j", Var::Sum);
+  Var i("i"), j("j");
   A(i) = operands[0](j,i) * operands[0](j,i);
 
   return A;
@@ -91,7 +91,7 @@ MTTKRP1Factory::operator()(Tensors& operands, Format outFormat) {
   Tensor<double> A({operands[0].getDimensions()[0],
                     operands[1].getDimensions()[1]}, outFormat);
 
-  Var i("i"), j("j"), k("k", Var::Sum), l("l", Var::Sum);
+  Var i("i"), j("j"), k("k"), l("l");
   A(i,j) = operands[0](i,k,l) * operands[2](l,j) * operands[1](k,j);
 
   return A;
@@ -104,7 +104,7 @@ MTTKRP2Factory::operator()(Tensors& operands, Format outFormat) {
   Tensor<double> A({operands[0].getDimensions()[1],
                     operands[1].getDimensions()[1]}, outFormat);
 
-  Var i("i"), j("j"), k("k", Var::Sum), l("l", Var::Sum);
+  Var i("i"), j("j"), k("k"), l("l");
   A(i,j) = operands[0](k,i,l) * operands[2](l,j) * operands[1](k,j);
 
   return A;
@@ -117,7 +117,7 @@ MTTKRP3Factory::operator()(Tensors& operands, Format outFormat) {
   Tensor<double> A({operands[0].getDimensions()[2],
                     operands[1].getDimensions()[1]}, outFormat);
 
-  Var i("i"), j("j"), k("k", Var::Sum), l("l", Var::Sum);
+  Var i("i"), j("j"), k("k"), l("l");
   A(i,j) = operands[0](k,l,i) * operands[2](l,j) * operands[1](k,j);
 
   return A;
@@ -129,7 +129,7 @@ TensorSquaredNormFactory::operator()(Tensors& operands, Format outFormat) {
 
   Tensor<double> A({}, outFormat);
 
-  Var i("i", Var::Sum), j("j", Var::Sum), k("k", Var::Sum);
+  Var i("i"), j("j"), k("k");
   A() = operands[0](i,j,k) * operands[0](i,j,k);
 
   return A;
@@ -142,7 +142,7 @@ FactorizedTensorSquaredNormFactory::operator()(Tensors& operands,
 
   Tensor<double> A({}, outFormat);
 
-  Var i("i", Var::Sum), j("j", Var::Sum);
+  Var i("i"), j("j");
   A() = operands[0](i) * operands[0](j) * operands[1](i,j) * 
         operands[2](i,j) * operands[3](i,j);
 
@@ -156,7 +156,7 @@ FactorizedTensorInnerProductFactory::operator()(Tensors& operands,
 
   Tensor<double> A({}, outFormat);
 
-  Var i("i", Var::Sum), j("j", Var::Sum), k("k", Var::Sum), r("r", Var::Sum);
+  Var i("i"), j("j"), k("k"), r("r");
   A() = operands[0](i,j,k) * operands[1](r) * operands[2](i,r) * 
         operands[3](j,r) * operands[4](k,r);
 

test/expr_storage-tests.cpp

@@ -16,7 +16,7 @@ typedef std::vector<IndexArray> Index;      // [0,2] index arrays per Index
 typedef std::vector<Index>      Indices;    // One Index per level
 
 Var i("i"), j("j"), m("m"), n("n");
-Var k("k", Var::Sum), l("l", Var::Sum);
+Var k("k"), l("l");
 
 
 struct TestData {

test/storage_alloc-tests.cpp

@@ -52,8 +52,7 @@ TEST_P(alloc, storage) {
   ASSERT_STORAGE_EQUALS(expectedIndices, expectedValues, tensor);
 }
 
-Var i("i"), j("j"), m("m"), n("n");
-Var k("k", Var::Sum), l("l", Var::Sum);
+Var i("i"), j("j"), m("m"), n("n"), k("k"), l("l");
 
 IndexArray dlab_indices() {
   IndexArray indices;