Precision handling and citations (#80)

* type selection when saving the trajectories

* added precision also to saponify

* updated citations

Changelog.md

@@ -10,6 +10,7 @@ Until we get to a release version you may encounter different broken interface p
 - Added `getSOAPSettings()` for getting the SOAP organization from a hdf5 dataset
 - Added `getTimeSOAPSimple()` to give the user a shortcut to solve memory problem with large SOAP datasets
 - Now `createUniverseFromSlice()` returns also types
+- added precision selection for storing the trajectories and the SOAP fingerprints
 
 ## Changes since v0.0.6
 

docs/source/Citation.rst

@@ -6,17 +6,19 @@ Work in progress
 .. code-block::
    :caption: Innate Dynamics and Identity Crisis of a Metal Surface Unveiled by Machine Learning of Atomic Environments
 
-       @misc{CioniCu,
-            title        = {{Innate Dynamics and Identity Crisis of a Metal Surface Unveiled by Machine Learning of Atomic Environments}},
-            author       = {Cioni, Matteo and Polino, Daniela and Rapetti, Daniele and Pesce, Luca and Dell Piane, Massimo and Pavan, Giovanni M.},
-            year         = 2022,
-            archivePrefix = "arXiv",
-            note = {arXiv: 2207.14622},
-            journal      = {arXiv},
-            publisher    = {arXiv},
-            copyright    = {Creative Commons Attribution 4.0 International},
-            keywords     = {Materials Science (cond-mat.mtrl-sci), Computational Physics (physics.comp-ph), FOS: Physical sciences, FOS: Physical sciences}
-        }
+    @article{doi:10.1063/5.0139010,
+        title        = {Innate dynamics and identity crisis of a metal surface unveiled by machine learning of atomic environments},
+        author       = {Cioni,Matteo  and Polino,Daniela  and Rapetti,Daniele  and Pesce,Luca  and Delle Piane,Massimo  and Pavan,Giovanni M.},
+        year         = 2023,
+        journal      = {The Journal of Chemical Physics},
+        volume       = 158,
+        number       = 12,
+        pages        = 124701,
+        doi          = {10.1063/5.0139010},
+        url          = {https://doi.org/10.1063/5.0139010},
+        eprint       = {https://doi.org/10.1063/5.0139010}
+    }
+
 
 .. code-block::
     :caption: Machine Learning of Atomic Dynamics and Statistical Surface Identities in Gold Nanoparticles
@@ -31,4 +33,32 @@ Work in progress
             publisher    = {ChemRxiv},
             copyright    = {Creative Commons Attribution 4.0 International},
             keywords     = {Gold nanoparticles, Machine learning, Atomic environments, Classification, Atomic dynamics, Statistical identities}
+        }
+
+TimeSOAP analysis
+
+.. code-block::
+    :caption: TimeSOAP: Tracking high-dimensional fluctuations in complex molecular systems via time-variations of SOAP spectra
+        
+        @misc{caruso2023timesoap,
+            title        = {TimeSOAP: Tracking high-dimensional fluctuations in complex molecular systems via time-variations of SOAP spectra},
+            author       = {Cristina Caruso and Annalisa Cardellini and Martina Crippa and Daniele Rapetti and Giovanni M. Pavan},
+            year         = 2023,
+            eprint       = {2302.09673},
+            archiveprefix = {arXiv},
+            primaryclass = {physics.chem-ph}
+        }
+
+LENS analysis
+
+.. code-block::
+    :caption: TimeSOAP: Tracking high-dimensional fluctuations in complex molecular systems via time-variations of SOAP spectra
+
+        @misc{crippa2022detecting,
+            title        = {Detecting dynamic domains and local fluctuations in complex molecular systems via timelapse neighbors shuffling},
+            author       = {Martina Crippa and Annalisa Cardellini and Cristina Caruso and Giovanni M. Pavan},
+            year         = 2022,
+            eprint       = {2212.12694},
+            archiveprefix = {arXiv},
+            primaryclass = {physics.chem-ph}
         }

src/SOAPify/HDF5er/ToHDF5.py

@@ -14,6 +14,7 @@ def universe2HDF5(
     trajFolder: h5py.Group,
     trajChunkSize: int = 100,
     trajslice: slice = slice(None),
+    useType="float64",
 ):
     """Uploads an mda.Universe or an mda.AtomGroup to a h5py.Group in an hdf5 file
 
@@ -25,12 +26,14 @@ def universe2HDF5(
         trajChunkSize (int, optional):
             The desired dimension of the chunks of data that are stored in the hdf5 file.
             Defaults to 100.
+        useType (str,optional):
+            The precision used to store the data. Defaults to "float64".
     """
 
     atoms = mdaTrajectory.atoms
     universe = mdaTrajectory.universe
     nat = len(atoms)
-
+    useType = numpy.dtype(useType)
     if "Types" not in list(trajFolder.keys()):
         trajFolder.create_dataset("Types", (nat), compression="gzip", data=atoms.types)
 
@@ -41,7 +44,7 @@ def universe2HDF5(
             compression="gzip",
             chunks=(trajChunkSize, nat, 3),
             maxshape=(None, nat, 3),
-            dtype=numpy.float64,
+            dtype=useType,
         )
 
     if "Box" not in list(trajFolder.keys()):
@@ -51,7 +54,7 @@ def universe2HDF5(
             compression="gzip",
             chunks=True,
             maxshape=(None, 6),
-            dtype=numpy.float64,
+            dtype=useType,
         )
 
     frameNum = 0
@@ -82,6 +85,7 @@ def MDA2HDF5(
     override: bool = False,
     attrs: dict = None,
     trajslice: slice = slice(None),
+    useType="float64",
 ):
     """Creates an HDF5 trajectory groupfrom an mda trajectory
 
@@ -106,6 +110,8 @@ def MDA2HDF5(
         override (bool, optional):
             If true the hdf5 file will be completely overwritten.
             Defaults to False.
+        useType (str,optional):
+            The precision used to store the data. Defaults to "float64".
     """
     with h5py.File(targetHDF5File, "w" if override else "a") as newTraj:
         trajGroup = newTraj.require_group(f"Trajectories/{groupName}")
@@ -114,6 +120,7 @@ def MDA2HDF5(
             trajGroup,
             trajChunkSize=trajChunkSize,
             trajslice=trajslice,
+            useType=useType,
         )
         if attrs:
             for key in attrs.keys():

src/SOAPify/saponify.py

@@ -107,6 +107,7 @@ def _applySOAP(
     SOAPnJobs: int = 1,
     doOverride: bool = False,
     verbose: bool = True,
+    useType="float64",
 ):
     """helper function: applies the soap engine to the given trajectory within the trajContainer
 
@@ -132,7 +133,10 @@ def _applySOAP(
         verbose (bool, optional):
             regulates the verbosity of the step by step operations.
             Defaults to True.
+        useType (str,optional):
+            The precision used to store the data. Defaults to "float64".
     """
+    useType = numpy.dtype(useType)
     nOfFeatures = soapEngine.features
     symbols = trajContainer["Types"].asstr()[:]
     nCenters = (
@@ -157,7 +161,7 @@ def _applySOAP(
             compression_opts=9,
             chunks=(SOAPOutputChunkDim, nCenters, nOfFeatures),
             maxshape=(None, nCenters, nOfFeatures),
-            dtype=numpy.float64,
+            dtype=useType,
         )
     SOAPout = SOAPoutContainer[key]
     SOAPout.resize((len(trajContainer["Trajectory"]), nCenters, nOfFeatures))
@@ -186,6 +190,7 @@ def saponifyMultipleTrajectories(
     useSoapFrom: KNOWNSOAPENGINES = "dscribe",
     doOverride: bool = False,
     verbose: bool = True,
+    useType="float64",
 ):
     """Calculates and stores the SOAP descriptor for all of the trajectories in
     the given group/file
@@ -240,6 +245,8 @@ def saponifyMultipleTrajectories(
         verbose (bool, optional):
             regulates the verbosity of the step by step operations.
             Defaults to True.
+        useType (str,optional):
+            The precision used to store the data. Defaults to "float64".
     """
     for key in trajContainers.keys():
         if isTrajectoryGroup(trajContainers[key]):
@@ -258,6 +265,7 @@ def saponifyMultipleTrajectories(
                 useSoapFrom=useSoapFrom,
                 doOverride=doOverride,
                 verbose=verbose,
+                useType=useType,
             )
 
 
@@ -276,6 +284,7 @@ def saponifyTrajectory(
     useSoapFrom: KNOWNSOAPENGINES = "dscribe",
     doOverride: bool = False,
     verbose: bool = True,
+    useType="float64",
 ):
     """Calculates the SOAP fingerprints for each atom in a given hdf5 trajectory
 
@@ -329,6 +338,8 @@ def saponifyTrajectory(
         verbose (bool, optional):
             regulates the verbosity of the step by step operations.
             Defaults to True.
+        useType (str,optional):
+            The precision used to store the data. Defaults to "float64".
     """
     if isTrajectoryGroup(trajContainer):
         print(f'using "{useSoapFrom}" to calculate SOAP for "{trajContainer.name}"')
@@ -355,6 +366,7 @@ def saponifyTrajectory(
             SOAPnJobs,
             doOverride=doOverride,
             verbose=verbose,
+            useType=useType,
         )
     else:
         raise ValueError("saponify: The input object is not a trajectory group.")

tests/test_HDF5er.py

@@ -1,3 +1,4 @@
+"""Test for HDF5er"""
 import SOAPify.HDF5er as HDF5er
 import h5py
 import numpy