Update version and add pip installation instructions, adjust test tolerance for new NumPy

Author: mgreminger

README.md

@@ -2,6 +2,10 @@
 
 trust-constr optimization algorithm from the SciPy project that was originally implemented by [Antonio Horta Ribeiro](https://github.com/antonior92). This is a version of the trust-constr algorithm that does not depend on the rest of SciPy. The only dependency is NumPy. The goal is to have a version of the trust-constr algorithm that can run within the Pyodide environment.
 
+# Installation
+
+`pip install trust-constr`
+
 # Examples Using trust-constr
 
 Since the trust-constr algorithm was extracted from the `scipy.optimize` library, it uses the [same interface](https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.minimize.html#scipy.optimize.minimize) as `scipy.optimize.minimize`. The main different is that everything is imported from `trust_constr` rather than from `scipy.optimize`. The other difference is that the only optimization method available is 'trust-const'. The examples below show how to use trust-constr with a variety of different types of constraints.

README.tex.md

@@ -2,6 +2,10 @@
 
 trust-constr optimization algorithm from the SciPy project that was originally implemented by [Antonio Horta Ribeiro](https://github.com/antonior92). This is a version of the trust-constr algorithm that does not depend on the rest of SciPy. The only dependency is NumPy. The goal is to have a version of the trust-constr algorithm that can run within the Pyodide environment.
 
+# Installation
+
+`pip install trust-constr`
+
 # Examples Using trust-constr
 
 Since the trust-constr algorithm was extracted from the `scipy.optimize` library, it uses the [same interface](https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.minimize.html#scipy.optimize.minimize) as `scipy.optimize.minimize`. The main different is that everything is imported from `trust_constr` rather than from `scipy.optimize`. The other difference is that the only optimization method available is 'trust-const'. The examples below show how to use trust-constr with a variety of different types of constraints.

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name="trust-constr",
-    version="0.0.1",
+    version="1.0.0",
     description="trust-constr optimization algorithm from the SciPy project.",
     long_description=long_description,
     long_description_content_type="text/markdown",

tests/test_response_surface.py

@@ -68,17 +68,17 @@ def evaluate_response_surface_grad(
 
 
 @pytest.mark.parametrize(
-    "data,factorization_method,rtol",
+    "data,factorization_method,rtol,iter_buffer",
     [
-        (bottle_data_set, "SVDFactorization", 1e-5),
-        (bottle_data_set, "AugmentedSystem", 1e-5),
-        (panel_data_set, "SVDFactorization", 1e-5),
-        (panel_data_set, "AugmentedSystem", 1e-5),
-        (brake_data_set, "SVDFactorization", 2e-4),
-        (brake_data_set, "AugmentedSystem", 2e-4),
+        (bottle_data_set, "SVDFactorization", 1e-5, 10),
+        (bottle_data_set, "AugmentedSystem", 1e-5, 10),
+        (panel_data_set, "SVDFactorization", 1e-5, 40), # 10->40 NumPy 1.19.2 -> 1.19.5
+        (panel_data_set, "AugmentedSystem", 1e-5, 20),  # 10->20 NumPy 1.19.2 -> 1.19.5
+        (brake_data_set, "SVDFactorization", 2e-4, 10),
+        (brake_data_set, "AugmentedSystem", 2e-4, 10), 
     ],
 )
-def test_generate_pareto_data(data, factorization_method, rtol):
+def test_generate_pareto_data(data, factorization_method, rtol, iter_buffer):
     inputs = np.array(data["inputs"])
     outputs = np.array(data["outputs"])
     y_axis_index = data["y_axis_index"]
@@ -213,7 +213,7 @@ def objective_func_grad(x, index, sign=1.0):
 
     pareto_output_values = np.array(pareto_output_values)
 
-    assert total_iterations < baseline_total_iterations + 10
+    assert total_iterations < baseline_total_iterations + iter_buffer
     assert_allclose(pareto_output_values, baseline_pareto_points, rtol=rtol)