feat: add initializer list init for wrt in derivatives

include/suboptimal/autodiff/derivatives.h

@@ -22,6 +22,13 @@ class Gradient {
    */
   Gradient(const Variable& var, const Eigen::Ref<const VectorXv>& wrt);
 
+  /**
+   * Constructs a gradient object
+   * @param var the variable to compute the gradient of
+   * @param wrt the list of variables to compute the gradient with respect to
+   */
+  Gradient(const Variable& var, std::initializer_list<Variable> wrt);
+
   /**
    * Gets the value of the gradient based on the current value of wrt
    */
@@ -82,6 +89,13 @@ class Jacobian {
    */
   Jacobian(const Eigen::Ref<const VectorXv>& vars, const Eigen::Ref<const VectorXv>& wrt);
 
+  /**
+   * Constructs a Jacobian object
+   * @param vars the variable to compute the gradient of
+   * @param wrt the list of variables to compute the gradient with respect to
+   */
+  Jacobian(const Eigen::Ref<const VectorXv>& vars, std::initializer_list<Variable> wrt);
+
   /**
    * Gets the value of the Jacobian based on the current value of wrt
    */
@@ -115,6 +129,13 @@ class Hessian {
    */
   Hessian(const Variable& var, const Eigen::Ref<const VectorXv>& wrt);
 
+  /**
+   * Constructs a Hessian object
+   * @param var the variable to compute the gradient of
+   * @param wrt the list of variables to compute the Hessian with respect to
+   */
+  Hessian(const Variable& var, std::initializer_list<Variable> wrt);
+
   /**
    * Gets the value of the Hessian based on the current value of wrt
    */

src/autodiff/derivatives.cpp

@@ -17,6 +17,8 @@ Gradient::Gradient(const Variable& var, const Eigen::Ref<const VectorXv>& wrt) :
   }
 }
 
+Gradient::Gradient(const Variable& var, const std::initializer_list<Variable> wrt) : Gradient{var, VectorXv{wrt}} {}
+
 const Eigen::SparseVector<double>& Gradient::getValue() {
   if (var.getLinearity() > Linearity::Linear) {
     std::ranges::for_each(wrt, [](const Variable& v) { v.expr->adjoint = 0.0; });
@@ -112,6 +114,9 @@ Jacobian::Jacobian(const Eigen::Ref<const VectorXv>& vars, const Eigen::Ref<cons
   value.setFromTriplets(triplets.begin(), triplets.end());
 }
 
+Jacobian::Jacobian(const Eigen::Ref<const VectorXv>& vars, const std::initializer_list<Variable> wrt)
+    : Jacobian{vars, VectorXv{wrt}} {}
+
 const Eigen::SparseMatrix<double>& Jacobian::getValue() {
   if (nonlinear_rows.empty()) {
     return value;
@@ -139,6 +144,8 @@ MatrixXv Jacobian::getExpr() {
 Hessian::Hessian(const Variable& var, const Eigen::Ref<const VectorXv>& wrt)
     : jacobian(Gradient{var, wrt}.getExpr(), wrt) {}
 
+Hessian::Hessian(const Variable& var, const std::initializer_list<Variable> wrt) : Hessian{var, VectorXv{wrt}} {}
+
 const Eigen::SparseMatrix<double>& Hessian::getValue() {
   return jacobian.getValue();
 }

test/autodiff/gradient_test.cpp

@@ -21,7 +21,7 @@ TEST_CASE("Autodiff - Basic gradient", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = (x * x - 2 * x * y + y * y) / (5 * x);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-100.0, 100.0)));
   const double y_val = GENERATE(take(10, random(-100.0, 100.0)));
@@ -37,7 +37,7 @@ TEST_CASE("Autodiff - Basic gradient", "[autodiff]") {
 TEST_CASE("Autodiff - Gradient of abs", "[autodiff]") {
   Variable x{};
   const Variable f = suboptimal::abs(x);
-  Gradient grad{f, VectorXv{{x}}};
+  Gradient grad{f, {x}};
 
   const double x_val = GENERATE(take(10, random(-100.0, 100.0)));
   x.setValue(x_val);
@@ -72,7 +72,7 @@ TEST_CASE("Autodiff - Gradient of exp", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = suboptimal::exp(x * y);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-10.0, 10.0)));
   const double y_val = GENERATE(take(10, random(-10.0, 10.0)));
@@ -89,7 +89,7 @@ TEST_CASE("Autodiff - Gradient of log", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = suboptimal::log(x * y);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(0.1, 100.0)));
   const double y_val = GENERATE(take(10, random(0.1, 100.0)));
@@ -106,7 +106,7 @@ TEST_CASE("Autodiff - Gradient of pow", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = suboptimal::pow(x, y);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(0.0, 100.0)));
   const double y_val = GENERATE(take(10, random(-100.0, 100.0)));
@@ -123,7 +123,7 @@ TEST_CASE("Autodiff - Gradient of hypot", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = suboptimal::hypot(x, y);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-100.0, 100.0)));
   const double y_val = GENERATE(take(10, random(-100.0, 100.0)));
@@ -141,7 +141,7 @@ TEST_CASE("Autodiff - Gradients of trig functions", "[autodiff]") {
   Variable y{};
   Variable z{};
   const Variable f = suboptimal::sin(x) * suboptimal::cos(y) * suboptimal::tan(z);
-  Gradient grad{f, Vector3v{x, y, z}};
+  Gradient grad{f, {x, y, z}};
 
   const double x_val = GENERATE(take(5, random(-100.0, 100.0)));
   const double y_val = GENERATE(take(5, random(-100.0, 100.0)));
@@ -162,7 +162,7 @@ TEST_CASE("Autodiff - Gradients of inverse trig functions", "[autodiff]") {
   Variable y{};
   Variable z{};
   const Variable f = suboptimal::asin(x) * suboptimal::acos(y) * suboptimal::atan(z);
-  Gradient grad{f, Vector3v{x, y, z}};
+  Gradient grad{f, {x, y, z}};
 
   const double x_val = GENERATE(take(5, random(-1.0, 1.0)));
   const double y_val = GENERATE(take(5, random(-1.0, 1.0)));
@@ -182,7 +182,7 @@ TEST_CASE("Autodiff - Gradient of atan2", "[autodiff]") {
   Variable x{};
   Variable y{};
   const Variable f = suboptimal::atan2(y, x);
-  Gradient grad{f, Vector2v{x, y}};
+  Gradient grad{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-100.0, 100.0)));
   const double y_val = GENERATE(take(10, random(-100.0, 100.0)));

test/autodiff/hessian_test.cpp

@@ -19,7 +19,7 @@ TEST_CASE("Autodiff - Basic Hessian") {
   Variable x{};
   Variable y{};
   const Variable f = (x * x - 2 * x * y + y * y);
-  Hessian h{f, Vector2v{x, y}};
+  Hessian h{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-100, 100)));
   const double y_val = GENERATE(take(10, random(-100, 100)));
@@ -38,7 +38,7 @@ TEST_CASE("Autodiff - Complicated Hessian") {
   Variable y{};
   Variable z{};
   const Variable f = x * y * z + y * suboptimal::sin(x) + z * suboptimal::cos(y);
-  Hessian h{f, Vector3v{x, y, z}};
+  Hessian h{f, {x, y, z}};
 
   const double x_val = GENERATE(take(5, random(-100, 100)));
   const double y_val = GENERATE(take(5, random(-100, 100)));

test/autodiff/jacobian_test.cpp

@@ -21,7 +21,7 @@ TEST_CASE("Autodiff - Basic Jacobian", "[autodiff]") {
   Variable y{};
   const Vector2v f{4 * suboptimal::pow(x, 2) * y,  //
                    x - suboptimal::pow(y, 2)};
-  Jacobian j{f, Vector2v{x, y}};
+  Jacobian j{f, {x, y}};
 
   const double x_val = GENERATE(take(10, random(-100, 100)));
   const double y_val = GENERATE(take(10, random(-100, 100)));
@@ -42,7 +42,7 @@ TEST_CASE("Autodiff - Complicated Jacobian", "[autodiff]") {
   const Vector3v f{x * y * z,               //
                    y * suboptimal::sin(x),  //
                    z * suboptimal::cos(y)};
-  Jacobian j{f, Vector3v{x, y, z}};
+  Jacobian j{f, {x, y, z}};
 
   const double x_val = GENERATE(take(5, random(-100.0, 100.0)));
   const double y_val = GENERATE(take(5, random(-100.0, 100.0)));