Merge pull request #92 from joglekara/paper

JOSS paper updates

.circleci/config.yml

@@ -7,7 +7,7 @@ jobs:
   build:
     # Run the steps with Docker
     docker:
-      - image: cimg/python:3.7
+      - image: cimg/python:3.8
 
       # Specify service dependencies here if necessary
       # CircleCI maintains a library of pre-built images

paper/paper.bib

@@ -197,16 +197,21 @@ @article{Hoyer2017
 year = {2017}
 }
 
-@article{Zaharia2018,
-  title={Accelerating the Machine Learning Lifecycle with MLflow.},
-  author={Zaharia, Matei and Chen, Andrew and Davidson, Aaron and Ghodsi, Ali and Hong, Sue Ann and Konwinski, Andy and Murching, Siddharth and Nykodym, Tomas and Ogilvie, Paul and Parkhe, Mani and others},
-  journal={IEEE Data Eng. Bull.},
-  volume={41},
-  number={4},
-  pages={39--45},
-  year={2018}
+@inproceedings{Chen2020,
+abstract = {MLflow is a popular open source platform for managing ML development, including experiment tracking, reproducibility, and deployment. In this paper, we discuss user feedback collected since MLflow was launched in 2018, as well as three major features we have introduced in response to this feedback: a Model Registry for collaborative model management and review, tools for simplifying ML code instrumentation, and experiment analytics functions for extracting insights from millions of ML experiments.},
+address = {New York, NY, USA},
+author = {Chen, Andrew and Chow, Andy and Davidson, Aaron and DCunha, Arjun and Ghodsi, Ali and Hong, Sue Ann and Konwinski, Andy and Mewald, Clemens and Murching, Siddharth and Nykodym, Tomas and Ogilvie, Paul and Parkhe, Mani and Singh, Avesh and Xie, Fen and Zaharia, Matei and Zang, Richard and Zheng, Juntai and Zumar, Corey},
+booktitle = {Proceedings of the Fourth International Workshop on Data Management for End-to-End Machine Learning},
+doi = {10.1145/3399579.3399867},
+isbn = {9781450380232},
+publisher = {Association for Computing Machinery},
+series = {DEEM'20},
+title = {{Developments in MLflow: A System to Accelerate the Machine Learning Lifecycle}},
+url = {https://doi.org/10.1145/3399579.3399867},
+year = {2020}
 }
 
+
 @article{Fahlen2009,
 author = {Fahlen, J. E. and Winjum, B. J. and Grismayer, T. and Mori, W. B.},
 doi = {10.1103/PhysRevLett.102.245002},

paper/paper.md

@@ -175,15 +175,92 @@ In the ``VlaPy`` simulation code, we have verified that the known damping rates
 We include validation against this phenomenon as an automated integrated test. The tests can be found in 
 `tests/test_landau_damping.py`
 
-Below, we also illustrate a manual validation of this phenomenon through the fully integrated workflow. After running a properly initialized simulation, we show that the damping rate of an electron plasma wave with $k=0.3$ is reproduced accurately through the UI. This can also be computed manually (please see the testing code for details).
+Below, we also illustrate a manual validation of this phenomenon through the fully integrated workflow of running a simulation on a local machine and sending the results to the MLFlow-driven logging mechanism. After running a properly initialized simulation, we show that the damping rate of an electron plasma wave with $k=0.3$ is reproduced accurately through the UI. This can also be computed manually (please see the testing code for details).
 
 ![](../notebooks/screenshots_for_example/ui.png)
 ![](../notebooks/screenshots_for_example/damping.png)
 
 To run the entire testing suite, make sure `pytest` is installed, and call `pytest` from the root folder for the repository. Individual files can also be run by calling `pytest tests/<test_filename>.py`.
 
+# Example Run Script For Landau Damping
+    import numpy as np
+    from vlapy import manager, initializers
+    from vlapy.infrastructure import mlflow_helpers, print_to_screen
+    from vlapy.diagnostics import landau_damping
+
+    if __name__ == "__main__":
+        # Pick a random wavenumber
+        k0 = np.random.uniform(0.3, 0.4, 1)[0]
+        
+        # This is a collisionless simulation. Provide float value if collisions should be simulated
+        log_nu_over_nu_ld = None
+        
+        # This initializes the default parameters
+        all_params_dict = initializers.make_default_params_dictionary()
+        
+        # This calculates the roots to the EPW dispersion relation given the wavenumber
+        all_params_dict = initializers.specify_epw_params_to_dict(
+            k0=k0, all_params_dict=all_params_dict
+        )
+        
+        # This specifies the collision frequency given nu_ld
+        all_params_dict = initializers.specify_collisions_to_dict(
+            log_nu_over_nu_ld=log_nu_over_nu_ld, all_params_dict=all_params_dict
+        )
+
+        # The solvers can be chosen here
+        all_params_dict["vlasov-poisson"]["time"] = "leapfrog"
+        all_params_dict["vlasov-poisson"]["edfdv"] = "exponential"
+        all_params_dict["vlasov-poisson"]["vdfdx"] = "exponential"
+
+        all_params_dict["fokker-planck"]["type"] = "lb"
+        all_params_dict["fokker-planck"]["solver"] = "batched_tridiagonal"
+        
+        # The pulse shape can be chosen here
+        pulse_dictionary = {
+            "first pulse": {
+                "start_time": 0,
+                "t_L": 6,
+                "t_wL": 2.5,
+                "t_R": 20,
+                "t_wR": 2.5,
+                "w0": all_params_dict["w_epw"],
+                "a0": 1e-7,
+                "k0": k0,
+            }
+        }
+        
+        # Mlflow experiment name and location
+        mlflow_exp_name = "landau-damping"
+        
+        # Either an IP address for your MLflow server or "local" if no server specified
+        uris = {
+            "tracking": "local",
+        }
+        
+       
+        # Start!
+        that_run = manager.start_run(
+            all_params=all_params_dict,
+            pulse_dictionary=pulse_dictionary,
+            diagnostics=landau_damping.LandauDamping(
+                vph=all_params_dict["v_ph"],
+                wepw=all_params_dict["w_epw"],
+            ),
+            uris=uris,
+            name=mlflow_exp_name,
+        )
+        
+        # Assess if the simulation results match the actual damping rate
+        print(
+            mlflow_helpers.get_this_metric_of_this_run("damping_rate", that_run),
+            all_params_dict["nu_ld"],
+        )
+
+
+
 # Acknowledgements
-We use xarray [@Hoyer2017] for file storage and MLFlow [@Zaharia2018] for experiment management.
+We use xarray [@Hoyer2017] for file storage and MLFlow [@Chen2020] for experiment management.
 
 We acknowledge valuable discussions with Pierre Navarro on the implementation of the Vlasov equation.
 

run_debug.py

@@ -42,12 +42,8 @@
     all_params_dict["vlasov-poisson"]["edfdv"] = "exponential"
     all_params_dict["vlasov-poisson"]["vdfdx"] = "exponential"
 
-    all_params_dict["backend"]["core"] = "numpy"
-    all_params_dict["backend"]["max_GB_for_device"] = 1
-
     all_params_dict["fokker-planck"]["type"] = "lb"
-    all_params_dict["fokker-planck"]["type"] = "batched_tridiagonal"
-
+    all_params_dict["fokker-planck"]["solver"] = "batched_tridiagonal"
 
     pulse_dictionary = {
         "first pulse": {
@@ -68,10 +64,6 @@
         "tracking": "local",
     }
 
-    print_to_screen.print_startup_message(
-        mlflow_exp_name, all_params_dict, pulse_dictionary
-    )
-
     that_run = manager.start_run(
         all_params=all_params_dict,
         pulse_dictionary=pulse_dictionary,

run_nlepw.py

@@ -23,7 +23,6 @@
 import numpy as np
 
 from vlapy import manager, initializers
-from vlapy.infrastructure import mlflow_helpers, print_to_screen
 from vlapy.diagnostics import nlepw
 
 if __name__ == "__main__":
@@ -47,13 +46,6 @@
     all_params_dict["vlasov-poisson"]["edfdv"] = "exponential"
     all_params_dict["vlasov-poisson"]["vdfdx"] = "exponential"
 
-    all_params_dict["backend"]["core"] = "numpy"
-    all_params_dict["backend"]["max_GB_for_device"] = 0.25
-
-    all_params_dict["fokker-planck"]["type"] = "lb"
-    all_params_dict["fokker-planck"]["solver"] = "batched_tridiagonal"
-
-
     pulse_dictionary = {
         "first pulse": {
             "start_time": 0,
@@ -73,10 +65,6 @@
         "tracking": "local",
     }
 
-    print_to_screen.print_startup_message(
-        mlflow_exp_name, all_params_dict, pulse_dictionary
-    )
-
     that_run = manager.start_run(
         all_params=all_params_dict,
         pulse_dictionary=pulse_dictionary,
@@ -87,4 +75,3 @@
         uris=uris,
         name=mlflow_exp_name,
     )
-

run_vlapy.py

@@ -23,7 +23,7 @@
 import numpy as np
 
 from vlapy import manager, initializers
-from vlapy.infrastructure import mlflow_helpers, print_to_screen
+from vlapy.infrastructure import mlflow_helpers
 from vlapy.diagnostics import landau_damping
 
 if __name__ == "__main__":
@@ -42,12 +42,8 @@
     all_params_dict["vlasov-poisson"]["edfdv"] = "exponential"
     all_params_dict["vlasov-poisson"]["vdfdx"] = "exponential"
 
-    all_params_dict["backend"]["core"] = "numpy"
-    all_params_dict["backend"]["max_GB_for_device"] = 1
-
     all_params_dict["fokker-planck"]["type"] = "lb"
-    all_params_dict["fokker-planck"]["type"] = "batched_tridiagonal"
-
+    all_params_dict["fokker-planck"]["solver"] = "batched_tridiagonal"
 
     pulse_dictionary = {
         "first pulse": {
@@ -68,10 +64,6 @@
         "tracking": "local",
     }
 
-    print_to_screen.print_startup_message(
-        mlflow_exp_name, all_params_dict, pulse_dictionary
-    )
-
     that_run = manager.start_run(
         all_params=all_params_dict,
         pulse_dictionary=pulse_dictionary,

setup.py

@@ -36,7 +36,7 @@
 
 setup(
     name="vlapy",
-    version="1.0",
+    version="0.1",
     packages=find_packages(),
     url="",
     license="MIT",

tests/test_landau_damping.py

@@ -84,10 +84,6 @@ def __run_integrated_landau_damping_test_and_return_damping_rate__(
             "tracking": "local",
         }
 
-        print_to_screen.print_startup_message(
-            mlflow_exp_name, all_params_dict, pulse_dictionary
-        )
-
         that_run = manager.start_run(
             all_params=all_params_dict,
             pulse_dictionary=pulse_dictionary,

vlapy/infrastructure/print_to_screen.py

@@ -24,11 +24,12 @@
 from datetime import datetime
 
 
-def print_startup_message(exp_name, all_params_dict, pulse_dictionary):
+def print_startup_message(exp_name, all_params_dict, pulse_dictionary, uri):
     print("Starting VlaPy at " + datetime.now().strftime("%m/%d/%Y %H:%M:%S"))
     print("MLFlow experiment name: " + exp_name)
     print("Run parameters: ")
     pprint.pprint(all_params_dict)
     print("Driver parameters: ")
     pprint.pprint(pulse_dictionary)
-    print()
+    print("MLFlow server: ")
+    pprint.pprint(uri)

vlapy/manager.py

@@ -29,6 +29,7 @@
 import numpy as np
 
 from vlapy import storage, outer_loop
+from vlapy.infrastructure import print_to_screen
 
 
 def start_run(all_params, pulse_dictionary, diagnostics, uris, name="test"):
@@ -46,8 +47,9 @@ def start_run(all_params, pulse_dictionary, diagnostics, uris, name="test"):
     :param name: (string) the name of the MLFlow experiment
     :return: (Mlflow.Run) returns the completed Run object
     """
-    t0 = time()
 
+    t0 = time()
+    print_to_screen.print_startup_message(name, all_params, pulse_dictionary, uris)
     if "local" not in uris["tracking"].casefold():
         mlflow.set_tracking_uri(uris["tracking"])