Merge pull request #2266 from peastman/vector

Fixed bug in vector operations in CustomIntegrator on CUDA

platforms/cuda/include/CudaExpressionUtilities.h

@@ -123,6 +123,7 @@ private:
     void findRelatedPowers(const Lepton::ExpressionTreeNode& node, const Lepton::ExpressionTreeNode& searchNode,
             std::map<int, const Lepton::ExpressionTreeNode*>& powers);
     void callFunction(std::stringstream& out, std::string singleFn, std::string doubleFn, const std::string& arg, const std::string& tempType);
+    void callFunction2(std::stringstream& out, std::string fn, const std::string& arg1, const std::string& arg2, const std::string& tempType);
     std::vector<std::vector<double> > computeFunctionParameters(const std::vector<const TabulatedFunction*>& functions);
     CudaContext& context;
     FunctionPlaceholder fp1, fp2, fp3, periodicDistance;

platforms/cuda/src/CudaExpressionUtilities.cpp

@@ -547,6 +547,15 @@ void CudaExpressionUtilities::processExpression(stringstream& out, const Express
             out << "RECIP(" << getTempName(node.getChildren()[0], temps) << ")";
             break;
         case Operation::ADD_CONSTANT:
+            if (isVecType) {
+                string val = context.doubleToString(dynamic_cast<const Operation::AddConstant*>(&node.getOperation())->getValue());
+                string arg = getTempName(node.getChildren()[0], temps);
+                out << "make_" << tempType << "(";
+                out << val << "+" << arg << ".x, ";
+                out << val << "+" << arg << ".y, ";
+                out << val << "+" << arg << ".z)";
+            }
+            else
                 out << context.doubleToString(dynamic_cast<const Operation::AddConstant*>(&node.getOperation())->getValue()) << "+" << getTempName(node.getChildren()[0], temps);
             break;
         case Operation::MULTIPLY_CONSTANT:
@@ -610,10 +619,10 @@ void CudaExpressionUtilities::processExpression(stringstream& out, const Express
             break;
         }
         case Operation::MIN:
-            out << "min((" << tempType << ") " << getTempName(node.getChildren()[0], temps) << ", (" << tempType << ") " << getTempName(node.getChildren()[1], temps) << ")";
+            callFunction2(out, "min", getTempName(node.getChildren()[0], temps), getTempName(node.getChildren()[1], temps), tempType);
             break;
         case Operation::MAX:
-            out << "max((" << tempType << ") " << getTempName(node.getChildren()[0], temps) << ", (" << tempType << ") " << getTempName(node.getChildren()[1], temps) << ")";
+            callFunction2(out, "max", getTempName(node.getChildren()[0], temps), getTempName(node.getChildren()[1], temps), tempType);
             break;
         case Operation::ABS:
             callFunction(out, "fabs", "fabs", getTempName(node.getChildren()[0], temps), tempType);
@@ -927,3 +936,15 @@ void CudaExpressionUtilities::callFunction(stringstream& out, string singleFn, s
             out<<singleFn<<"("<<arg<<")";
     }
 }
+
+void CudaExpressionUtilities::callFunction2(stringstream& out, string fn, const string& arg1, const string& arg2, const string& tempType) {
+    bool isVector = (tempType[tempType.size()-1] == '3');
+    if (isVector) {
+        out<<"make_"<<tempType<<"(";
+        out<<fn<<"("<<arg1<<".x, "<<arg2<<".x), ";
+        out<<fn<<"("<<arg1<<".y, "<<arg2<<".y), ";
+        out<<fn<<"("<<arg1<<".z, "<<arg2<<".z))";
+    }
+    else
+        out<<fn<<"(("<<tempType<<") "<<arg1<<", ("<<tempType<<") "<<arg2<<")";
+}

tests/TestCustomIntegrator.h

@@ -1042,10 +1042,12 @@ void testVectorFunctions() {
     integrator.addPerDofVariable("angular", 0.0);
     integrator.addPerDofVariable("shuffle", 0.0);
     integrator.addPerDofVariable("multicross", 0.0);
+    integrator.addPerDofVariable("maxplus", 0.0);
     integrator.addComputeSum("sumy", "x*vector(0, 1, 0)");
     integrator.addComputePerDof("angular", "cross(v, x)");
     integrator.addComputePerDof("shuffle", "dot(vector(_z(x), _x(x), _y(x)), v)");
     integrator.addComputePerDof("multicross", "cross(vector(1, 0, 0), cross(vector(0, 0, 1), vector(1, 0, 0)))");
+    integrator.addComputePerDof("maxplus", "max(x, 0.1)+0.5");
     OpenMM_SFMT::SFMT sfmt;
     init_gen_rand(0, sfmt);
     vector<Vec3> positions(numParticles);
@@ -1063,14 +1065,16 @@ void testVectorFunctions() {
     // See if the expressions were computed correctly.
     
     double sumy = 0;
-    vector<Vec3> angular, shuffle, multicross;
+    vector<Vec3> angular, shuffle, multicross, maxplus;
     integrator.getPerDofVariable(0, angular);
     integrator.getPerDofVariable(1, shuffle);
     integrator.getPerDofVariable(2, multicross);
+    integrator.getPerDofVariable(3, maxplus);
     for (int i = 0; i < numParticles; i++) {
         ASSERT_EQUAL_VEC(velocities[i].cross(positions[i]), angular[i], 1e-5);
         ASSERT_EQUAL_VEC(Vec3(1, 1, 1)*velocities[i].dot(Vec3(positions[i][2], positions[i][0], positions[i][1])), shuffle[i], 1e-5);
         ASSERT_EQUAL_VEC(Vec3(0, 0, 1), multicross[i], 1e-5);
+        ASSERT_EQUAL_VEC(Vec3(max(positions[i][0], 0.1)+0.5, max(positions[i][1], 0.1)+0.5, max(positions[i][2], 0.1)+0.5), maxplus[i], 1e-5);
         sumy += positions[i][1];
     }
     ASSERT_EQUAL_TOL(sumy, integrator.getGlobalVariable(0), 1e-5);