Merge pull request #587 from ModECI/experimental

Addition of simple network example for testing conditions

Author: pgleeson

.github/workflows/ci_test_all.yml

@@ -1,4 +1,4 @@
-name: CI Test script
+name: CI test_all.sh script
 
 on:
   push:
@@ -14,7 +14,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: [ "3.12"]
+        python-version: [ "3.13" ]
         runs-on: [ubuntu-latest]
 
     steps:

.gitignore

@@ -317,3 +317,4 @@ examples/TensorFlow/Keras/keras_to_MDF
 /docs/sphinx/source/api/export_format/MDF/images/newton.png
 /docs/sphinx/source/api/export_format/MDF/images/newton_plot.png
 /examples/MDF/networks/Net_acyclical
+/examples/MDF/networks/Net_network2

examples/MDF/Simple.bson

@@ -0,0 +1,50 @@
+{
+    "Net_network2": {
+        "format": "ModECI MDF v0.4",
+        "generating_application": "Python modeci-mdf v0.4.14",
+        "graphs": {
+            "Net_network2": {
+                "nodes": {
+                    "A": {
+                        "input_ports": {
+                            "in_port_0": {
+                                "default_value": 5
+                            }
+                        },
+                        "parameters": {
+                            "total_inputs": {
+                                "value": "total_inputs + in_port_0"
+                            },
+                            "execute_count": {
+                                "value": "execute_count + 1"
+                            }
+                        },
+                        "output_ports": {
+                            "out_port_0": {
+                                "value": "total_inputs"
+                            }
+                        }
+                    },
+                    "B": {
+                        "input_ports": {
+                            "in_port_0": {}
+                        },
+                        "parameters": {
+                            "total_inputs": {
+                                "value": "total_inputs + in_port_0"
+                            },
+                            "execute_count": {
+                                "value": "execute_count + 1"
+                            }
+                        },
+                        "output_ports": {
+                            "out_port_0": {
+                                "value": "total_inputs"
+                            }
+                        }
+                    }
+                }
+            }
+        }
+    }
+}

examples/MDF/networks/Net_network2.json

@@ -0,0 +1,29 @@
+Net_network2:
+    format: ModECI MDF v0.4
+    generating_application: Python modeci-mdf v0.4.14
+    graphs:
+        Net_network2:
+            nodes:
+                A:
+                    input_ports:
+                        in_port_0:
+                            default_value: 5
+                    parameters:
+                        total_inputs:
+                            value: total_inputs + in_port_0
+                        execute_count:
+                            value: execute_count + 1
+                    output_ports:
+                        out_port_0:
+                            value: total_inputs
+                B:
+                    input_ports:
+                        in_port_0: {}
+                    parameters:
+                        total_inputs:
+                            value: total_inputs + in_port_0
+                        execute_count:
+                            value: execute_count + 1
+                    output_ports:
+                        out_port_0:
+                            value: total_inputs

examples/MDF/networks/Net_network2.png

@@ -0,0 +1,123 @@
+"""
+    Example of ModECI MDF - Testing networks
+"""
+
+from modeci_mdf.mdf import *
+import sys
+import os
+from modeci_mdf.utils import simple_connect
+
+
+def main(ref="network2"):
+
+    mod = Model(id="Net_%s" % ref)
+
+    mod_graph = Graph(id="Net_%s" % ref)
+    mod.graphs.append(mod_graph)
+
+    for id in ["A", "B"]:
+
+        node = Node(id=id)
+
+        if id == "A":
+            ip1 = InputPort(id="in_port_0", default_value=5)
+            node.input_ports.append(ip1)
+
+        if id == "B":
+            ip1 = InputPort(id="in_port_0")
+            node.input_ports.append(ip1)
+
+        p1 = Parameter(id="total_inputs", value="total_inputs + in_port_0")
+        node.parameters.append(p1)
+
+        p2 = Parameter(id="execute_count", value="execute_count + 1")
+        node.parameters.append(p2)
+
+        op1 = OutputPort(id="out_port_0", value=p1.id)
+        node.output_ports.append(op1)
+
+        mod_graph.nodes.append(node)
+
+    # simple_connect(mod_graph.get_node("A"), mod_graph.get_node("B"), mod_graph)
+    # simple_connect(mod_graph.get_node("B"), mod_graph.get_node("A"), mod_graph)
+
+    new_file = mod.to_json_file("%s.json" % mod.id)
+    new_file = mod.to_yaml_file("%s.yaml" % mod.id)
+
+    eg = None
+
+    if "-run" in sys.argv:
+        verbose = True
+        # verbose = False
+        from modeci_mdf.utils import load_mdf, print_summary
+
+        from modeci_mdf.execution_engine import EvaluableGraph
+
+        eg = EvaluableGraph(mod_graph, verbose)
+        dt = 1
+
+        duration = 2
+        t = 0
+        recorded = {}
+        times = []
+        ai = []
+        bi = []
+        at = []
+        bt = []
+        ao = []
+        bo = []
+        while t <= duration:
+            times.append(t)
+            print("======   Evaluating at t = %s  ======" % (t))
+            if t == 0:
+                eg.evaluate()  # replace with initialize?
+            else:
+                eg.evaluate(time_increment=dt)
+
+            ai.append(eg.enodes["A"].evaluable_parameters["execute_count"].curr_value)
+            bi.append(eg.enodes["B"].evaluable_parameters["execute_count"].curr_value)
+            at.append(eg.enodes["A"].evaluable_parameters["total_inputs"].curr_value)
+            bt.append(eg.enodes["B"].evaluable_parameters["total_inputs"].curr_value)
+            ao.append(eg.enodes["A"].evaluable_outputs["out_port_0"].curr_value)
+            bo.append(eg.enodes["B"].evaluable_outputs["out_port_0"].curr_value)
+            t += dt
+
+        if "-nogui" not in sys.argv:
+            import matplotlib.pyplot as plt
+
+            plt.plot(times, ai, marker="x", label="A exe count")
+            plt.plot(times, bi, label="B exe count")
+            plt.plot(times, at, marker="D", label="A total_inputs")
+            plt.plot(times, bt, label="B total_inputs")
+
+            plt.legend()
+            plt.show()
+
+    if "-graph" in sys.argv:
+        mod.to_graph_image(
+            engine="dot",
+            output_format="png",
+            view_on_render=False,
+            level=3,
+            filename_root=mod_graph.id,
+            only_warn_on_fail=(
+                os.name == "nt"
+            ),  # Makes sure test of this doesn't fail on Windows on GitHub Actions
+        )
+
+    return mod_graph, eg
+
+
+if __name__ == "__main__":
+    g, eg = main()
+
+    print("--------------------")
+    print(g.to_yaml())
+    print("--------------------")
+    if eg is not None:
+        for n in eg.enodes:
+            enode = eg.enodes[n]
+            print(enode)
+            print([enode.evaluable_outputs for enode in eg.enodes.values()])
+    else:
+        print("No execution engine created. Run with: -run")

examples/MDF/networks/Net_network2.yaml

@@ -2,11 +2,11 @@
     {
         "model_name": "squeezenet1_1",
         "model_type": "SqueezeNet",
-        "pytorch_time": 0.0839,
+        "pytorch_time": 0.07,
         "pytorch_predictions": 10,
-        "mdf_time": 4.4275,
+        "mdf_time": 5.949,
         "mdf_predictions": 10,
         "node density": 66,
-        "mdf : pytorch ratio": "52.79"
+        "mdf : pytorch ratio": "84.98"
     }
 ]

examples/MDF/networks/network2.py

@@ -11,7 +11,7 @@ name = "modeci-mdf"
 dynamic = ["version"]
 description = "ModECI (Model Exchange and Convergence Initiative) Model Description Format"
 readme = "README.md"
-requires-python = ">=3.9"
+requires-python = ">=3.10"
 license = {text = "LGPLv3"}
 authors = [
     {name = "ModECI contributors", email = "p.gleeson@gmail.com"},
@@ -25,23 +25,21 @@ classifiers = [
     "License :: OSI Approved :: GNU Lesser General Public License v3 (LGPLv3)",
     "Natural Language :: English",
     "Operating System :: OS Independent",
-    "Programming Language :: Python :: 3.8",
     "Topic :: Scientific/Engineering",
-    "Intended Audience :: Science/Research",
     "Programming Language :: Python",
     "Programming Language :: Python :: 3",
     "Programming Language :: Python :: 3 :: Only",
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
-    "Topic :: Scientific/Engineering",
     "Topic :: Software Development",
     "Typing :: Typed",
 ]
 dependencies = [
     "graph_scheduler>=1.1.1",
     "numpy",
+    "scipy",
     "matplotlib",
     "graphviz",
     "h5py",
@@ -92,7 +90,6 @@ optional = [
     "netron",
     "torch>=1.11.0",
     "torchvision",
-    "h5py",
 ]
 
 all_except_psyneulink = [
@@ -144,9 +141,6 @@ exclude = ["tests", "examples"]
 [tool.setuptools.dynamic]
 version = {attr = "modeci_mdf.__version__"}
 
-#[tool.setuptools_scm]
-#write_to = "src/modeci_mdf/version.py"
-
 [tool.pytest.ini_options]
 junit_family = "xunit2"
 addopts = "--benchmark-disable -Wd --strict-markers"
@@ -156,7 +150,7 @@ required_plugins = ["pytest-benchmark"]
 markers = [
     "coremdf: marks tests which are core to the MDF specification",
     "actr: marks tests which require ACT-R to be installed.",
-    "pytorch: marks test which require PyTorch to be installed.')",
+    "pytorch: marks tests which require PyTorch to be installed.",
     "psyneulink: marks tests which require PsyNueLink to be installed.",
     "neuroml: marks tests which require NeuroML to be installed.",
     "tensorflow: marks tests which require TensorFlow to be installed.",
@@ -182,7 +176,7 @@ per-file-ignores = [
 
 [tool.mypy]
 warn_unused_configs = true
-python_version = "3.6"
+python_version = "3.10"
 files = ["src"]
 disallow_any_generics = true
 disallow_subclassing_any = true

examples/PyTorch/PyTorch_MDF/benchmark_script/squeezenet1_1_benchmark.json

@@ -679,6 +679,9 @@ def evaluate(
             )
         return self.curr_value
 
+    def __repr__(self):
+        return f"EvaluableOutput({self.output_port.id} = {_val_info(self.curr_value)} )"
+
 
 class EvaluableInput:
     """Evaluates input value at the :class:`~modeci_mdf.mdf.InputPort` of the node during MDF graph execution.
@@ -766,7 +769,7 @@ def evaluate(
                 )
             )
 
-        self.curr_value = None
+        self.curr_value = None  # to be tested for conditions...
         return final_val