SOAPify.analysis

Module that contains various analysis routines

Functions

`getTimeSOAPSimple`(soapDataset[, window, ...])	Shortcut to extract the timeSOAP from large datasets.
`listNeighboursAlongTrajectory`(inputUniverse, ...)	produce a per frame list of the neighbours, atom per atom
`neighbourChangeInTime`(nnListPerFrame)	return, listed per each atoms the parameters used in the LENS analysis
`timeSOAP`(SOAPTrajectory[, window, stride, ...])	performs the 'timeSOAP' analysis on the given SOAP trajectory
`timeSOAPsimple`(SOAPTrajectory[, window, ...])	performs the 'timeSOAP' analysis on the given normalized SOAP trajectory

SOAPify.analysis.getTimeSOAPSimple(soapDataset, window=1, stride=None, backward=False)[source]

Shortcut to extract the timeSOAP from large datasets.

This function is the equivalent to:

loading a chunk of the trajectory from a h5py.Dataset with a SOAP fingerprints trajectory

filling the vector with SOAPify.utils.fillSOAPVectorFromdscribe()

normalizing it with SOAPify.utils.normalizeArray()

calculating the timeSOAP with timeSOAPsimple()

and then returning timeSOAP and the derivative

Parameters

soapDataset (h5py.Dataset) – the dataset with the SOAP fingerprints
window (int) – the dimension of the windows between each state confrontations. See timeSOAPsimple() Defaults to 1.
stride (int) – the stride in frames between each state confrontation. See timeSOAPsimple() Defaults to None.
backward (bool) – If true the soap distance is referred to the previous frame. See timeSOAPsimple() . Defaulst to True.

Returns

timedSOAP the timeSOAP values, shape(frames-1,natoms)
deltaTimedSOAP the derivatives of timeSOAP, shape(natoms, frames-2)

Return type

tuple[numpy.ndarray,numpy.ndarray]

SOAPify.analysis.listNeighboursAlongTrajectory(inputUniverse, cutOff, trajSlice=slice(None, None, None))[source]

produce a per frame list of the neighbours, atom per atom

Original author: Martina Crippa

Mantainer: Daniele Rapetti

Parameters

inputUniverse (Universe) – the universe, or the atomgroup containing the trajectory
cutOff (float) – the maximum neighbour distance
trajSlice (slice, optional) – the slice of the trajectory to consider. Defaults to slice(None).

Returns

list of AtomGroup wint the neighbours of each atom for each frame

Return type

list[list[AtomGroup]]

SOAPify.analysis.neighbourChangeInTime(nnListPerFrame)[source]

return, listed per each atoms the parameters used in the LENS analysis

Original author: Martina Crippa

Mantainer: Daniele Rapetti

Parameters

nnListPerFrame (list[list[AtomGroup]]) – a frame by frame list of the neighbours of each atom: output of :func:`listNeighboursAlongTrajectory

Returns

lensArray The calculated LENS parameter
numberOfNeighs the count of neighbours per frame
lensNumerators the numerators used for calculating LENS parameter
lensDenominators the denominators used for calculating LENS parameter

Return type

tuple[numpy.ndarray,numpy.ndarray,numpy.ndarray,numpy.ndarray]

SOAPify.analysis.timeSOAP(SOAPTrajectory, window=1, stride=None, backward=False, returnDiff=True, distanceFunction=<function simpleSOAPdistance>)[source]

performs the ‘timeSOAP’ analysis on the given SOAP trajectory

Original author: Cristina Caruso

Mantainer: Daniele Rapetti

Parameters

SOAPTrajectory (int) – a trajectory of SOAP fingerprints, should have shape (nFrames,nAtoms,SOAPlenght)
window (int) – the dimension of the windows between each state confrontations. Defaults to 1.
stride (int) – the stride in frames between each state confrontation. NOT IN USE. Defaults to None.
backward (bool) –

If true the soap distance is referred to the previous frame.
NOT IN USE. Defaulst to True.
returnDiff (bool) – If true returns also the first derivative of timeSOAP. Defaults to True.
distanceFunction (callable, optional) – the function that define the distance. Defaults to SOAPify.distances.simpleSOAPdistance().

Returns

timedSOAP the timeSOAP values, shape(frames-1,natoms)
deltaTimedSOAP the derivatives of timeSOAP, shape(natoms, frames-2)

Return type

tuple[numpy.ndarray,numpy.ndarray]

SOAPify.analysis.timeSOAPsimple(SOAPTrajectory, window=1, stride=None, backward=False, returnDiff=True)[source]

performs the ‘timeSOAP’ analysis on the given normalized SOAP trajectory

this is optimized to use SOAPify.distances.simpleSOAPdistance(), without calling it.

Warning

this function works only with normalized numpy.float64 soap vectors!

The SOAP distance is calculated with

\[d(\vec{a},\vec{b})=\sqrt{2-2\frac{\vec{a}\cdot\vec{b}}{\left\|\vec{a}\right\|\left\|\vec{b}\right\|}}\]

That is equivalent to

\[ \begin{align}\begin{aligned}d(\vec{a},\vec{b})=\sqrt{2-2\hat{a}\cdot\hat{b}} = \sqrt{\hat{a}\cdot\hat{a}+\hat{b}\cdot\hat{b}-2\hat{a}\cdot\hat{b}} =\\\sqrt{(\hat{a}-\hat{b})\cdot(\hat{a}-\hat{b})}\end{aligned}\end{align} \]

That is the euclidean distance between the versors

Original author: Cristina Caruso

Mantainer: Daniele Rapetti

Parameters

SOAPTrajectory (int) – a **normalize ** trajectory of SOAP fingerprints, should have shape (nFrames,nAtoms,SOAPlenght)
window (int) – the dimension of the windows between each state confrontations. Defaults to 1.
stride (int) – the stride in frames between each state confrontation. NOT IN USE. Defaults to None.
backward (bool) –

If true the soap distance is referred to the previous frame.
NOT IN USE. Defaulst to True.
returnDiff (bool) – If true returns also the first derivative of timeSOAP. Defaults to True.

Returns

timedSOAP the timeSOAP values, shape(frames-1,natoms)
deltaTimedSOAP the derivatives of timeSOAP, shape(natoms, frames-2)

Return type

tuple[numpy.ndarray,numpy.ndarray]