sn.get_bibliography() is added

Author: ehsanhaghighat

examples/example-fitting-fourier.py

@@ -31,7 +31,7 @@
 # Each network is defined by Functional.
 y1 = sn.Field('y1', 10)
 y2 = sn.Field('y2', 10)
-y1, y2 = sn.Functional([y1,y2], xf, [10, 10, 10], 'tanh', output_activation='tanh')
+y1, y2 = sn.Functional([y1,y2], xf, [10, 10, 10], 'l-tanh', output_activation='tanh')
 
 y = sn.Functional('y', [xf*y1, xf*y2])
 
@@ -59,3 +59,4 @@
 y_pred = y.eval(model, x_true)
 d_pred = d.eval(model, x_true)
 
+sn.get_bibliography(format="bibtex")  #bibtexml

sciann/__init__.py

@@ -25,6 +25,7 @@
 from .utils.utilities import set_default_log_path, get_default_log_path
 from .utils.utilities import set_random_seed
 from .utils.utilities import set_floatx
+from .utils.utilities import get_bibliography
 
 # SciANN.
 __author__ = "Ehsan Haghighat"
@@ -33,17 +34,15 @@
 __credits__ = []
 __url__ = "http://github.com/sciann/sciann]"
 __license__ = "MIT"
-__version__ = "0.6.0.0"
+__version__ = "0.6.0.1"
 __cite__ = \
-    '@misc{haghighat2019sciann, \n' +\
-    '    title={SciANN: A Keras/Tensorflow wrapper for scientific computations and physics-informed deep learning using artificial neural networks}, \n' +\
-    '    author={Ehsan Haghighat and Ruben Juanes}, \n' +\
-    '    year={2020}, \n' +\
-    '    eprint={2005.08803}, \n' +\
-    '    archivePrefix={arXiv}, \n' +\
-    '    primaryClass={cs.OH}, \n' +\
-    '    url = {https://arxiv.org/abs/2005.08803}' +\
-    '    howpublished={https://github.com/sciann/sciann.git}' +\
+    '@article{haghighat2021sciann, \n' +\
+    '    title={SciANN: A Keras/TensorFlow wrapper for scientific computations and physics-informed deep learning using artificial neural networks}, \n' +\
+    '    author={Haghighat, Ehsan and Juanes, Ruben}, \n' +\
+    '    journal={Computer Methods in Applied Mechanics and Engineering}, \n' +\
+    '    year={2021}, \n' +\
+    '    url = {https://doi.org/10.1016/j.cma.2020.113552}, \n' +\
+    '    howpublished={https://github.com/sciann/sciann.git}, \n' +\
     '}'
 
 # Import message.
@@ -67,3 +66,4 @@
 
 # set default logging directory.
 set_default_log_path(os.path.join(os.getcwd(), "logs"))
+initialize_bib(os.path.join(os.path.dirname(__file__), 'references', 'bibliography'))

sciann/references/bibliography

@@ -0,0 +1,66 @@
+@article{haghighat2021sciann,
+  title={SciANN: A Keras/TensorFlow wrapper for scientific computations and physics-informed deep learning using artificial neural networks},
+  author={Haghighat, Ehsan and Juanes, Ruben},
+  journal={Computer Methods in Applied Mechanics and Engineering},
+  volume={373},
+  pages={113552},
+  year={2021},
+  publisher={Elsevier},
+  url={https://www.sciencedirect.com/science/article/pii/S0045782520307374}
+}
+
+@article{raissi2019physics,
+  title={Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations},
+  author={Raissi, Maziar and Perdikaris, Paris and Karniadakis, George E},
+  journal={Journal of Computational Physics},
+  volume={378},
+  pages={686--707},
+  year={2019},
+  publisher={Elsevier},
+  url={https://www.sciencedirect.com/science/article/pii/S0021999118307125}
+}
+
+@article{wang2020gp,
+  title={Understanding and mitigating gradient pathologies in physics-informed neural networks},
+  author={Wang, Sifan and Teng, Yujun and Perdikaris, Paris},
+  journal={arXiv preprint arXiv:2001.04536},
+  year={2020},
+  url={https://arxiv.org/abs/2001.04536}
+}
+
+@article{wang2020ntk,
+  title={When and why PINNs fail to train: A neural tangent kernel perspective},
+  author={Wang, Sifan and Yu, Xinling and Perdikaris, Paris},
+  journal={arXiv preprint arXiv:2007.14527},
+  year={2020},
+  url={https://arxiv.org/abs/2007.14527}
+}
+
+@article{li2017visualizing,
+  title={Visualizing the loss landscape of neural nets},
+  author={Li, Hao and Xu, Zheng and Taylor, Gavin and Studer, Christoph and Goldstein, Tom},
+  journal={arXiv preprint arXiv:1712.09913},
+  year={2017},
+  url={https://arxiv.org/abs/1712.09913}
+}
+
+@article{wang2020eigenvector,
+  title={On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks},
+  author={Wang, Sifan and Wang, Hanwen and Perdikaris, Paris},
+  journal={arXiv preprint arXiv:2012.10047},
+  year={2020},
+  url={https://arxiv.org/abs/2012.10047}
+}
+
+@article{jagtap2020locally,
+  title={Locally adaptive activation functions with slope recovery for deep and physics-informed neural networks},
+  author={Jagtap, Ameya D and Kawaguchi, Kenji and Em Karniadakis, George},
+  journal={Proceedings of the Royal Society A},
+  volume={476},
+  number={2239},
+  pages={20200334},
+  year={2020},
+  publisher={The Royal Society},
+  url={https://royalsocietypublishing.org/doi/abs/10.1098/rspa.2020.0334}
+}
+

sciann/utils/callbacks.py

@@ -14,6 +14,7 @@
 import numpy as np
 
 from .utilities import unpack_singleton, to_list, get_log_path
+from .utilities import append_to_bib
 from .math import tf_gradients
 
 
@@ -60,6 +61,7 @@ def __init__(self, model, inputs, targets, weights,
                  beta=0.1, freq=100, log_freq=None,
                  hessian=False, types=None, **kwargs):
         super(GradientPathologyLossWeight, self).__init__()
+        append_to_bib("wang2020gp")
         # limit number of samples for performance concerns.
         if inputs[0].shape[0] > 20000:
             sample_ids = np.random.choice(inputs[0].shape[0], 20000, replace=False)
@@ -358,6 +360,7 @@ def __init__(self, model, inputs, targets, weights,
                  norm=2, resolution=11, layer_wise=True,
                  path=None, trials=1):
         super(LossLandscapeHistory, self).__init__()
+        append_to_bib("li2017visualizing")
         self._model = model
         self._inputs = inputs
         self._layers = [layer for layer in model._layers if layer.weights]
@@ -491,6 +494,7 @@ def __init__(self, model, inputs, targets, weights=None,
                  beta=0.1, freq=100, log_freq=None,
                  hessian=False, types=None, **kwargs):
         super(NTKLossWeight, self).__init__()
+        append_to_bib("wang2020ntk")
         if weights is None:
             weights = [np.ones(inputs[0].shape[0]) for i in range(len(model.outputs))]
         # limit number of samples for performance concerns.

sciann/utils/math.py

@@ -35,6 +35,7 @@ def fourier(f, w=10):
         A Functional.
     """
     validate_variable(f)
+    append_to_bib("wang2020eigenvector")
     layers = []
     outputs = []
     for fi in f.outputs:

sciann/utils/utilities.py

@@ -29,7 +29,13 @@
 from .initializers import SciBiasInitializer as BInitializer
 from .activations import get_activation, SciActivation, SciActivationLayer
 
+from pybtex.database.input import bibtex
+
+
 _DEFAULT_LOG_PATH = ""
+_BIBLIOGRAPHY = None
+_BIBLIOGRAPHY_TO_OUTPUT = []
+
 
 def _is_tf_1():
     return tf.__version__.startswith('1.')
@@ -70,6 +76,45 @@ def get_default_log_path():
     return _DEFAULT_LOG_PATH
 
 
+def initialize_bib(bib_file):
+    global _BIBLIOGRAPHY
+    global _BIBLIOGRAPHY_TO_OUTPUT
+    _BIBLIOGRAPHY = bibtex.Parser().parse_file(bib_file)
+    _BIBLIOGRAPHY_TO_OUTPUT.append(_BIBLIOGRAPHY.entries['haghighat2021sciann'])
+    _BIBLIOGRAPHY_TO_OUTPUT.append(_BIBLIOGRAPHY.entries['raissi2019physics'])
+
+
+def append_to_bib(bib_entery):
+    global _BIBLIOGRAPHY
+    global _BIBLIOGRAPHY_TO_OUTPUT
+    bib = _BIBLIOGRAPHY.entries[bib_entery]
+    if bib not in _BIBLIOGRAPHY_TO_OUTPUT:
+        _BIBLIOGRAPHY_TO_OUTPUT.append(bib)
+
+
+def get_bibliography(format='bibtex', file_name=None):
+    """Returns the bibliography based on the feastures you used in your model.
+
+    # Argument
+        format: 'bibtex', 'bibtextml', 'yaml', ...
+            check `pybtex` documentation for other options.
+            default: 'bibtex'
+        file_name: path to a file.
+            default: None. This results in printing the bib file in the outputs.
+    """
+    global _BIBLIOGRAPHY_TO_OUTPUT
+    bib = ""
+    for b in _BIBLIOGRAPHY_TO_OUTPUT:
+        bib += str(b.to_string(format)) + '\n'
+
+    if file_name is None:
+        print(bib)
+    else:
+        with open(file_name,'w') as f:
+            f.write(bib)
+            f.close()
+
+
 def get_log_path(path=None, prefix=None):
     file_path = _DEFAULT_LOG_PATH if path is None else path
     if not os.path.isdir(file_path):
@@ -169,6 +214,7 @@ def prepare_default_activations_and_initializers(actfs, seed=None):
         if len(lay_actf) == 1:
             activations.append(SciActivation(w, f))
         else:
+            append_to_bib("jagtap2020locally")
             activations.append(SciActivationLayer(w, f))
 
     return activations, bias_initializer, kernel_initializer

setup.py

@@ -15,7 +15,7 @@
 
 setup(
     name='SciANN',
-    version='0.6.0.0',
+    version='0.6.0.1',
     description='A Keras/Tensorflow wrapper for scientific computations and physics-informed deep learning using artificial neural networks.',
     long_description=long_description,
     author='Ehsan Haghighat',
@@ -28,6 +28,7 @@
                       'pyyaml',
                       'h5py',
                       'sklearn',
+                      'pybtex',
                       'tensorflow>=2.1.0',
                       ],
     extras_require={