PyDPF returning arrays with wrong zero values

[fceyhuns]

Hello,

I have been using the PyDPF to extract data from rst files and then sometimes write new data in rst files to export to vtk files later. I will share parts of my code below.

I have noticed that what is represented in vtk files are very meaningful where what is extracted from the rst files are full of zero values. Although I am updating the rst file with these full of zero arrays, the final vtk is always a good one. I am quite perplexed here. Let me try to detail my case:

The peice of code I am struggling with:

def readFromDPF(rstFile, elt):
    dpfresult = dpf.Model(rstFile)
    rstSteps = dpfresult.metadata.time_freq_support.time_frequencies.data.tolist()
    noOfres = len(rstSteps)
    ansysresults = dpfresult.results
    inResultDict = {}
    inResults = []
    scanSteps = []
    ### in my example elt = [['displacement']]
    for ep in elt:
        try:
            inResultDict[ep[-1]].append(ep[0])
        except:
            inResultDict[ep[-1]] = [ep[0]]

    for ep in inResultDict.keys():
        inResults.append(ep)
        rstsearch = inResultDict[ep]
        subSt = []
        for i in rstsearch:
            if not 'rst' in i:
                subSt = range(noOfres)
            else:
                subSt.append(rstSteps.index(float(i.split('_')[-1])))
        scanSteps.append(subSt)
    for x in inResults:
        try: ## timeorfreqDPF is a function which checks if there are time steps or frequencies
            timeVecrange, fr = timeorfreqDPF(ansysresults.__getattribute__(x), noOfres)
            break
        except Exception as e:
            print(e)
            continue

    fullNodalresults = []
    fullNodalcoords = []
    for tAr in range(len(inResults)):
        tempAttri = inResults[tAr]
        rrSets = []
        rrSetsF = []
        thisResult = ansysresults.__getattribute__(tempAttri)  ## in this example tempAttri = 'displacement'
        try:
            tempEval = thisResult.on_all_time_freqs.eval()
        except Exception as e:
            print(e)
            continue
        tempShape = tempEval[0].data.shape
        for rr in scanSteps[tAr]:
            rstInPrev = [x['rst'] for x in fullNodalresults]
            if str(rstSteps[rr]) in rstInPrev:
                slctInd = rstInPrev.index(str(rstSteps[rr]))
                readyresults = fullNodalresults[slctInd]
            else:
                readyresults = {'rst': str(rstSteps[rr])}
                slctInd = len(fullNodalresults)
                fullNodalresults.append(readyresults)
            currentKey = tempAttri
            readyresults[currentKey] = []
            if fr == 1:
                readyresults[currentKey] = tempEval[timeVecrange[rr]].data_as_list + tempEval[timeVecrange[rr]+1].data_as_list
            else:
                readyresults[currentKey] = tempEval[rr].data_as_list  ### the error seems to be here. tempEval[rr].data_as_list  returns an array with a lot of zero's although the exported vtk shows something else. 

            fullNodalresults[slctInd] = readyresults

    ncoord_f = dpfresult.metadata.meshed_region.nodes.coordinates_field
    xcStr = ['x', 'y', 'z']
    coordFullList = ncoord_f.data_as_list
    # readyresults = {'nodes': 'coordinates'}
    for xc in range(3):
        coordList = [coordFullList[x] for x in range(xc, len(coordFullList), 3)]
        readyresults = {xcStr[xc]: coordList}
        fullNodalcoords.append(readyresults)

    return fullNodalresults, fullNodalcoords


def writeWithDPF(rstFile, outputDict, vtuNameCand):
    dpfresult = dpf.Model(rstFile)  ## this is a reference rst file
    rstSteps = dpfresult.metadata.time_freq_support.time_frequencies.data.tolist()
    meshed_region = dpfresult.metadata.meshed_region
    noOfres = len(rstSteps)
    ansysresults = dpfresult.results
    fc = dpf.FieldsContainer()
    fieldcounter = 1
    for newOut in outputDict.keys(): ## outputDict simply contains the values read by the readFromDPF, processed by other functions for some name convention 
        elt = newOut.split('|')
        newRes = outputDict[newOut]
        if 'rst' in elt[0]:
            timeSt = [list(rstSteps).index(float(elt[0].split('_')[-1]))]
        else:
            timeSt = list(range(noOfres))

        tempAttri = \
        [y for y in [xx._result_info.name for xx in ansysresults] if y[:-1] in '_'.join(newOut.lower().rsplit())][0]
        thisResult = ansysresults.__getattribute__(tempAttri)
        timeVecrange, fr = timeorfreqDPF(thisResult, noOfres)
        tempEval = thisResult.on_all_time_freqs.eval()
        tempShape = tempEval[0].data.shape
        if len(tempShape) > 1:
            subLim = tempShape[-1]
        else:
            subLim = 1  # tempShape[1]
        for rr in range(len(outputDict[newOut][0])):
            reShaped = []
            reShaped2 = []
            strtNo = 0
            ndCn = tempShape[0]
            for ii in range(ndCn):
                unfolded = [x[rr] for x in outputDict[newOut][strtNo:strtNo + subLim]]
                unfolded2 = None
                if fr ==1:
                    unfolded2 = unfolded2 = [x[rr] for x in outputDict[newOut][strtNo+ndCn*subLim:strtNo + subLim+ndCn*subLim]]
                if unfolded:
                    reShaped.append(np.array(list(unfolded)))
                else:
                    reShaped.append(np.array([0] * subLim))
                if unfolded2:
                    reShaped2.append(np.array(list(unfolded2)))
                else:
                    reShaped2.append(np.array([0] * subLim))                    
                strtNo += subLim
            reShaped = np.array(reShaped)
            reShaped2 = np.array(reShaped2)
            pickedRR = timeSt[rr]
            tempEval[timeVecrange[pickedRR]].data = reShaped
            fc.add_field_by_time_id(tempEval[timeVecrange[pickedRR]], fieldcounter)  ## writing values from outputDict
            fieldcounter += 1
            if fr ==1:
                tempEval[timeVecrange[pickedRR]+1].data = reShaped2
                fc.add_field_by_time_id(tempEval[timeVecrange[pickedRR]+1], fieldcounter)
                fieldcounter += 1

    filename2 = vtuNameCand #newRSTbaseName + '.vtu'
    op = dpf.operators.serialization.vtk_export(file_path=filename2, mesh = meshed_region, fields1=fc)
    op.run()
    return ansysresults

I am using the readFromDPF function to create dictionaries with nodal values and nodal coordinates from an rst file. The nodal values normally go under some process and they are updated by out internal tool. Then I want to update the rst file with these new values. I use writeWithDPF function. Updated rst file cannot really be saved so I export rst file with new results to vtk before releasing it.

When I open the vtk file in Paraview, the results are meaningful. When I update the rst with some other values and export again to vtk, that new vtk has new values, so there is an update process.

However, when I plot the values read in readFromDPF function, these values make no sense. When I try to plot them, I see that there are a lot of 0 values, which are not in the vtk files.

Could you please let me know if there is any known bug in PyDPF which may cause this problem. How on earth is it possible to write meaningful values to vtk by passing arrays full of wrong zero values?

Here is the list of Python package versions that we are using:

pyvista==0.39.0

ansys-dpf-core==0.8.1

And if you are curious to see the difference in plots, the image compareplots shows Paraview result and node cloud in our internal tool . Sorry for the low quality of the png but I believe it will give the idea.

[fceyhuns]

It is very likely that the order of nodal values reported and the nodal coordinates reported are different. Is there any way to trace back the order of nodal values, i.e. the nodal id's for each output type?