subdyn.py

"""
Tools for SubDyn

- Setup a FEM model, compute Guyan and CB modes
- Get a dataframe with properties
- More todo

"""


import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import copy
import re
# Local 
from pyFAST.io.fast_input_file import FASTInputFile
from pyFAST.tools.tictoc import Timer

idGuyanDamp_None     = 0
idGuyanDamp_Rayleigh = 1
idGuyanDamp_66       = 2 

class SubDyn:

    def __init__(self, sdFilename_or_data=None):
        """ 
        Initialize a SubDyn object either with:
          - sdFilename: a subdyn input file name
          - sdData: an instance of FASTInputFile
        """

        self._graph=None
        self.File=None

        # Read SubDyn file
        if sdFilename_or_data is not None:
            if hasattr(sdFilename_or_data,'startswith'): # if string
                self.File = FASTInputFile(sdFilename_or_data)
            else:
                self.File = sdFilename_or_data

        self.M_tip=None

        # Internal
        self._graph=None
        self._mgraph=None # Member graph
        self._FEM=None

    def __repr__(self):
        s='<{} object>:\n'.format(type(self).__name__)
        s+='|properties:\n'
        s+='|- File: (input file data)\n'
        s+='|* graph: (Nodes/Elements/Members)\n'
        s+='|* pointsMJ, pointsMN, pointsMNout\n'
        s+='|methods:\n'
        s+='|- memberPostPro\n'
        s+='|- setTopMass\n'
        s+='|- beamDataFrame, beamFEM, beamModes\n'
        s+='|- toYAMSData\n'
        return s

    # --------------------------------------------------------------------------------}
    # --- Functions for general FEM model (jacket, flexible floaters)
    # --------------------------------------------------------------------------------{
    def init(self, TP=(0,0,0), gravity = 9.81):
        """
        Initialize SubDyn FEM model 

        TP: position of transition point
        gravity: position of transition point
        """
        import welib.FEM.fem_beam as femb
        import welib.FEM.fem_model as femm
        BC       = 'clamped-free' # TODO Boundary condition: free-free or clamped-free
        element  = 'frame3d'      # Type of element used in FEM

        FEMMod = self.File['FEMMod']
        if FEMMod==1:
            mainElementType='frame3d'
        elif FEMMod==2:
            mainElementType='frame3dlin'
        elif FEMMod==3:
            mainElementType='timoshenko'
        else:
            raise NotImplementedError()
        # Get graph
        graph = self.graph
        #print('>>> graph\n',graph)
        #graph.toJSON('_GRAPH.json')
        # Convert to FEM model
        with Timer('From graph'):
            FEM = femm.FEMModel.from_graph(self.graph, mainElementType=mainElementType, refPoint=TP, gravity=gravity)
        #model.toJSON('_MODEL.json')
        with Timer('Assembly'):
            FEM.assembly()
        with Timer('Internal constraints'):
            FEM.applyInternalConstraints()
            FEM.partition()
        with Timer('BC'):
            FEM.applyFixedBC()
        with Timer('EIG'):
            Q, freq = FEM.eig(normQ='byMax')
        with Timer('CB'):
            FEM.CraigBampton(nModesCB = self.File['Nmodes'])

        with Timer('Modes'):
            FEM.setModes(nModesFEM=30, nModesCB=self.File['Nmodes'])
#             FEM.nodesDisp(Q)

        # --- SubDyn partition/notations
        FEM.MBB = FEM.MM_CB[np.ix_(FEM.DOF_Leader_CB  , FEM.DOF_Leader_CB)]
        FEM.KBB = FEM.KK_CB[np.ix_(FEM.DOF_Leader_CB  , FEM.DOF_Leader_CB)]
        FEM.MBM = FEM.MM_CB[np.ix_(FEM.DOF_Leader_CB  , FEM.DOF_Follower_CB)]
        FEM.KMM = FEM.KK_CB[np.ix_(FEM.DOF_Follower_CB, FEM.DOF_Follower_CB)]
        zeta =self.File['JDampings']/100
        if not hasattr(zeta,'__len__'):
            zeta = [zeta]*FEM.nModesCB
            FEM.CMM = 2*np.array(zeta) * FEM.f_CB * 2 * np.pi

        # --- Matrices wrt TP point
        TI=FEM.T_refPoint
        MBBt = TI.T.dot(FEM.MBB).dot(TI)
        KBBt = TI.T.dot(FEM.KBB).dot(TI)
        MBBt[np.abs(MBBt)<1e-4] =0
        KBBt[np.abs(KBBt)<1e-4] =0
        FEM.MBBt = MBBt
        FEM.KBBt = KBBt

        # --- Set Damping
        dampMod = self.File['GuyanDampMod']  
        alpha_Rayleigh, beta_Rayleigh = None, None
        # 6x6 Guyan Damping matrix
        CC_CB_G = None
        if  dampMod == idGuyanDamp_None:
            FEM.CBBt = np.zeros((6,6))
        elif dampMod == idGuyanDamp_Rayleigh:
            # Rayleigh Damping
            alpha_Rayleigh, beta_Rayleigh = self.File['RayleighDamp']
            FEM.CBBt = alpha_Rayleigh * FEM.MBBt + beta_Rayleigh * FEM.KBBt
        elif dampMod == idGuyanDamp_66: 
            FEM.CBBt = self.File['GuyanDampMatrix']
        else:
            raise Exception()

        # --- Compute rigid body equivalent
        FEM.rigidBodyEquivalent()
        self._FEM = FEM

        return FEM


    def setTopMass(self):
        # TODO
        # Add an optional top mass and ineria
        if TopMass:
            # NOTE: you can use welib.yams.windturbine to compute RNA mass and inertia
            Mtop = 50000  # Top mass [kg]
            M_tip= rigidBodyMassMatrixAtP(m=Mtop, J_G=None, Ref2COG=None)
        else:
            M_tip=None


    def getGraph(self, nDiv=1):
        # See welib.weio.fast_input_file_graph.py to see how SubDyn files are converted to graph 
        # See welib.fem.graph.py for Graph interface
        _graph = self.File.toGraph().divideElements(nDiv,
                    excludeDataKey='Type', excludeDataList=['Cable','Rigid'], method='insert',
                    keysNotToCopy=['IBC','RBC','addedMassMatrix'] # Very important
                    )

        if len(_graph.Members)==0:
            raise Exception('Problem in graph subdivisions, no members found.')
        # Sanitization
        #idBC_Fixed    =  0 # Fixed BC
        #idBC_Internal = 10 # Free/Internal BC
        #idBC_Leader   = 20 # Leader DOF
        MapIBC={0:0, 1:20} # 1 in the input file is leader
        MapRBC={0:10, 1:0} # 1 in the input file is fixed
        for n in _graph.Nodes:
            #print(n)
            if 'IBC' in n.data.keys():
                IBC = n.data['IBC'].copy()
                n.data['IBC'] = [MapIBC[i] for i in IBC[:6]]
            if 'RBC' in n.data.keys():
                RBC = n.data['RBC'].copy()
                n.data['RBC'] = [MapRBC[i] for i in RBC[:6]]
                if any(RBC[:6])==0:
                    print('RBC        ',RBC)
                    print('n.data[RBC]',n.data['RBC'] )
                    print('n          ',n )
                    raise NotImplementedError('SSI')

        return _graph

    @property
    def graph(self):
        if self._graph is None:
            self._graph = self.getGraph(nDiv = self.File['NDiv'])
        return copy.deepcopy(self._graph)


    @property
    def pointsMJ(self):
        """ return a dataframe with the coordinates of all members and joints
        The index corresponds to the SubDyn outputs "M_J_XXX"
        """
        Joints=[]
        labels =[]
        graph = self.graph
        for ie,M in enumerate(graph.Members): # loop on members
            Nodes = M.getNodes(graph)
            for iN,N in enumerate([Nodes[0], Nodes[-1]]):
                s='M{}J{}'.format(ie+1, iN+1)
                Joints.append([N.x,N.y,N.z])
                labels.append(s)
        df =pd.DataFrame(data=np.asarray(Joints), index=labels, columns=['x','y','z'])
        return df

    @property
    def pointsMN(self):
        """ return a dataframe with the coordinates of all members and nodes
        The index would correspond to the SubDyn outputs "M_N_XXX *prior* to the user selection"
        """
        Nodes=[]
        labels =[]
        graph = self.graph
        for im,M in enumerate(graph.Members): # loop on members
            nodes = M.getNodes(graph)
            for iN,N in enumerate(nodes): # Loop on member nodes
                s='M{}N{}'.format(im+1, iN+1)
                Nodes.append([N.x,N.y,N.z])
                labels.append(s)
        df =pd.DataFrame(data=np.asarray(Nodes), index=labels, columns=['x','y','z'])
        return df

    @property
    def pointsMNout(self):
        """ return a dataframe with the coordinates of members and nodes requested by user
        The index corresponds to the SubDyn outputs "M_N_XXX selected by the user"
        """
        Nodes=[]
        labels =[]
        graph = self.graph
        for im, out in enumerate(self.File['MemberOuts']):
            mID = out[0] # Member ID
            iNodes = np.array(out[2:])-1 # Node positions within member (-1 for python indexing)
            nodes = graph.getMemberNodes(mID)
            nodes = np.array(nodes)[iNodes]
            for iN,N in enumerate(nodes): # Loop on selected nodes
                s='M{}N{}'.format(im+1, iN+1)
                Nodes.append([N.x,N.y,N.z])
                labels.append(s)
        df =pd.DataFrame(data=np.asarray(Nodes), index=labels, columns=['x','y','z'])
        return df


    def memberPostPro(self, dfAvg):
        """
        Convert a dataframe of SubDyn/OpenFAST outputs (time-averaged)
        with columns such as: M_N_* and M_J_* 
        into a dataframe that is organized by main channel name and nodal coordinates.
        The scripts taken into account with member ID and node the user requested as outputs channels.
        Discretization (nDIV) is also taken into account.

        For instance:
            dfAvg with columns = ['M1N1MKye_[N*m]', 'M1N2MKye_[N*m]', 'M1N1TDxss_[m]']
        returns:
            MNout with columns ['x', 'y', 'z', 'MKye_[Nm]', 'TDxss_[m]']
               and index    ['M1N1', 'M1N2']
               with x,y,z the undisplaced nodal positions (accounting for discretization)

        INPUTS: 
          -  dfAvg: a dataframe of time-averaged SubDyn/OpenFAST outputs, for instance obtained as:
              df    = FASTInputFile(filename).toDataFrame()
              dfAvg = postpro.averageDF(df, avgMethod=avgMethod ,avgParam=avgParam)
        OUTPUTS
          - MNout: dataframe of members outputs (requested by the user)
          - MJout: dataframe of joints outputs
        """
        import welib.fast.postpro as postpro # Import done here to avoid circular dependency

        # --- Get Points where output are requested
        MJ = self.pointsMJ
        MNo= self.pointsMNout
        MJ.columns = ['x_[m]','y_[m]', 'z_[m]']
        MNo.columns = ['x_[m]','y_[m]', 'z_[m]']

        # --- Requested Member Outputs
        Map={}
        Map['^'+r'M(\d*)N(\d*)TDxss_\[m\]']   = 'TDxss_[m]'
        Map['^'+r'M(\d*)N(\d*)TDyss_\[m\]']   = 'TDyss_[m]'
        Map['^'+r'M(\d*)N(\d*)TDzss_\[m\]']   = 'TDzss_[m]'
        Map['^'+r'M(\d*)N(\d*)RDxe_\[rad\]']  = 'RDxe_[deg]' # NOTE rescale needed
        Map['^'+r'M(\d*)N(\d*)RDye_\[rad\]']  = 'RDye_[deg]' # NOTE rescale needed
        Map['^'+r'M(\d*)N(\d*)RDze_\[rad\]']  = 'RDze_[deg]' # NOTE rescale needed
        Map['^'+r'M(\d*)N(\d*)FKxe_\[N\]'] = 'FKxe_[N]'
        Map['^'+r'M(\d*)N(\d*)FKye_\[N\]'] = 'FKye_[N]'
        Map['^'+r'M(\d*)N(\d*)FKze_\[N\]'] = 'FKze_[N]'
        Map['^'+r'M(\d*)N(\d*)MKxe_\[N\*m\]'] = 'MKxe_[Nm]'
        Map['^'+r'M(\d*)N(\d*)MKye_\[N\*m\]'] = 'MKye_[Nm]'
        Map['^'+r'M(\d*)N(\d*)MKze_\[N\*m\]'] = 'MKze_[Nm]'
        ColsInfo, _ = postpro.find_matching_columns(dfAvg.columns, Map)
        nCols = len(ColsInfo)
        if nCols>0:
            newCols=[c['name'] for c in ColsInfo ]
            ValuesM = pd.DataFrame(index=MNo.index, columns=newCols)
            for ic,c in enumerate(ColsInfo):
                Idx, cols, colname = c['Idx'], c['cols'], c['name']
                labels = [re.match(r'(^M\d*N\d*)', s)[0] for s in cols] 
                ValuesM.loc[labels,colname] = dfAvg[cols].values.flatten()
                if 'deg' in colname and 'rad' in cols[0]:
                    ValuesM[colname] *= 180/np.pi
            # We remove lines that are all NaN
            Values = ValuesM.dropna(axis = 0, how = 'all')
            MNo2 = MNo.loc[Values.index]
            MNout = pd.concat((MNo2, Values), axis=1)
        else:
            MNout = None

        # --- Joint Outputs
        Map={}
        Map['^'+r'M(\d*)J(\d*)FKxe_\[N\]']   ='FKxe_[N]'
        Map['^'+r'M(\d*)J(\d*)FKye_\[N\]']   ='FKye_[N]'
        Map['^'+r'M(\d*)J(\d*)FKze_\[N\]']   ='FKze_[N]'
        Map['^'+r'M(\d*)J(\d*)MKxe_\[N\*m\]']='MKxe_[Nm]'
        Map['^'+r'M(\d*)J(\d*)MKye_\[N\*m\]']='MKye_[Nm]'
        Map['^'+r'M(\d*)J(\d*)MKze_\[N\*m\]']='MKze_[Nm]'
        Map['^'+r'M(\d*)J(\d*)FMxe_\[N\]']   ='FMxe_[N]'
        Map['^'+r'M(\d*)J(\d*)FMye_\[N\]']   ='FMye_[N]'
        Map['^'+r'M(\d*)J(\d*)FMze_\[N\]']   ='FMze_[N]'
        Map['^'+r'M(\d*)J(\d*)MMxe_\[N\*m\]']='MMxe_[Nm]'
        Map['^'+r'M(\d*)J(\d*)MMye_\[N\*m\]']='MMye_[Nm]'
        Map['^'+r'M(\d*)J(\d*)MMze_\[N\*m\]']='MMze_[Nm]'
        ColsInfo, _ = postpro.find_matching_columns(dfAvg.columns, Map)
        nCols = len(ColsInfo)
        if nCols>0:
            newCols=[c['name'] for c in ColsInfo ]
            ValuesJ = pd.DataFrame(index=MJ.index, columns=newCols)
            for ic,c in enumerate(ColsInfo):
                Idx, cols, colname = c['Idx'], c['cols'], c['name']
                labels = [re.match(r'(^M\d*J\d*)', s)[0] for s in cols] 
                ValuesJ.loc[labels,colname] = dfAvg[cols].values.flatten()
            # We remove lines that are all NaN
            Values = ValuesJ.dropna(axis = 0, how = 'all')
            MJ2 = MJ.loc[Values.index]
            MJout = pd.concat((MJ2, Values), axis=1)
        else:
            MJout = None
        return MNout, MJout




    # --------------------------------------------------------------------------------}
    # --- Functions for beam-like structure (Spar, Monopile)
    # --------------------------------------------------------------------------------{
    def beamDataFrame(self, equispacing=False):
        """ """
        # --- Parameters
        UseSubDynModel=True
        TopMass = False


        # Convert to "welib.fem.Graph" class to easily handle the model (overkill for a monopile)
        locgraph = self.graph.sortNodesBy('z')
        # Add nodal properties from propsets (NOTE: Not done anymore by SubDyn because a same node can have different diameters...)
        for e in locgraph.Elements:
            locgraph.setElementNodalProp(e, propset=e.propset, propIDs=e.propIDs)
        df = locgraph.nodalDataFrame()

        if equispacing:
            from welib.tools.pandalib import pd_interp1
            # Interpolate dataframe to equispaced values
            xOld  = df['z']    # NOTE: FEM uses "x" as main axis
            nSpan = len(xOld)
            x = np.linspace(np.min(xOld),np.max(xOld), nSpan)
            df = pd_interp1(x, 'z', df)

        x   = df['z'] # NOTE: FEM uses "x" as main axis
        D   = df['D'] # Diameter [m]
        t   = df['t'] # thickness [m]
        # Derive section properties for a hollow cylinder based on diameter and thickness
        A        = np.pi*( (D/2)**2 - (D/2-t)**2) # Area for annulus [m^2]
        I        = np.pi/64*(D**4-(D-2*t)**4)     # Second moment of area for annulus (m^4)
        Kt       = I                              # Torsion constant, same as I for annulus [m^4]
        Ip       = 2*I                            # Polar second moment of area [m^4]
        df['A']  = A
        df['I']  = I
        df['Kt'] = Kt
        df['Ip'] = Ip
        df['m']  = df['rho'].values*A

        return df

    def beamFEM(self, df=None):
        """ return FEM model for beam-like structures, like Spar/Monopile"""
        import welib.FEM.fem_beam as femb

        BC       = 'clamped-free' # TODO Boundary condition: free-free or clamped-free
        element  = 'frame3d'      # Type of element used in FEM

        if df is None:
            df = self.beamDataFrame()
        x   = df['z']              # NOTE: FEM uses "x" as main axis
        E   = df['E']              # Young modules [N/m^2]
        G   = df['G']              # Shear modules [N/m^2]
        rho = df['rho']            # material density [kg/m^3]
        Ip  = df['Ip']
        I   = df['I']
        A   = df['A']
        Kt  = df['Kt']

        # --- Compute FEM model and mode shapes
        with Timer('Setting up FEM model'):
            FEM=femb.cbeam(x,m=rho*A,EIx=E*Ip,EIy=E*I,EIz=E*I,EA=E*A,A=A,E=E,G=G,Kt=Kt,
                        element=element, BC=BC, M_tip=self.M_tip)
        return FEM

    def beamModes(self, nCB=8, FEM = None):
        """ Returns mode shapes for beam-like structures, like Spar/Monopile """
        import welib.FEM.fem_beam as femb
        element  = 'frame3d'      # Type of element used in FEM
        if FEM is None:
            FEM = self.beamFEM()
        # --- Perform Craig-Bampton reduction, fixing the top node of the beam
        with Timer('FEM eigenvalue analysis'):
            Q_G,_Q_CB, df_G, df_CB, Modes_G, Modes_CB, CB = femb.CB_topNode(FEM, nCB=nCB, element=element, main_axis='x')
        # df_CB.to_csv('_CB.csv',index=False)
        # df_G.to_csv('_Guyan.csv',index=False)
        return  Q_G,_Q_CB, df_G, df_CB, Modes_G, Modes_CB, CB 

    def beamModesPlot(self):
        """ """
        # TODO
        nModesPlot=8
        # --- Show frequencies to screen
        print('Mode   Frequency  Label ')
        for i in np.arange(8):
            print('{:4d} {:10.3f}   {:s}'.format(i+1,FEM['freq'][i],FEM['modeNames'][i]))

        # --- Plot mode components for first few modes
        print(x.shape)
        #Q=FEM['Q'] ; modeNames = FEM['modeNames']
        #Q=Q_CB ;modeNames = names_CB
        Modes=Modes_CB
        nModesPlot=min(len(Modes),nModesPlot)

        fig,axes = plt.subplots(1, nModesPlot, sharey=False, figsize=(12.4,2.5))
        fig.subplots_adjust(left=0.04, right=0.98, top=0.91, bottom=0.11, hspace=0.40, wspace=0.30)
        for i in np.arange(nModesPlot):
            key= list(Modes.keys())[i]

            axes[i].plot(x, Modes[key]['comp'][:,0]  ,'-'  , label='ux')
            axes[i].plot(x, Modes[key]['comp'][:,1]  ,'-'  , label='uy')
            axes[i].plot(x, Modes[key]['comp'][:,2]  ,'-'  , label='uz')
            axes[i].plot(x, Modes[key]['comp'][:,3]  ,':'  , label='vx')
            axes[i].plot(x, Modes[key]['comp'][:,4]  ,':'  , label='vy')
            axes[i].plot(x, Modes[key]['comp'][:,5]  ,':'  , label='vz')
            axes[i].set_xlabel('')
            axes[i].set_ylabel('')
            axes[i].set_title(Modes[key]['label'])
            if i==0:
                axes[i].legend()




    # --------------------------------------------------------------------------------}
    # --- IO/Converters
    # --------------------------------------------------------------------------------{
    def toYAML(self, filename):
        if self._FEM is None:
            raise Exception('Call `initFEM()` before calling `toYAML`')
        subdyntoYAMLSum(self._FEM, filename, more = self.File['OutAll'])


    def toYAMSData(self, shapes=[0,4], main_axis='z'):
        """ 
        Convert to Data needed to setup a Beam Model in YAMS (see bodies.py in yams)
        """
        from welib.mesh.gradient import gradient_regular

        # --- Perform Craig-Bampton reduction, fixing the top node of the beam
        # Get beam data frame
        df = self.beamDataFrame(equispacing=True)
        if np.any(df['y']!=0): 
            raise NotImplementedError('FASTBeamBody for substructure only support monopile, structure not fully vertical in file: {}'.format(self.File.filename))
        if np.any(df['x']!=0): 
            raise NotImplementedError('FASTBeamBody for substructure only support monopile, structure not fully vertical in file: {}'.format(self.File.filename))

        FEM = self.beamFEM(df)
        Q_G,_Q_CB, df_G, df_CB, Modes_G, Modes_CB, CB = self.beamModes(nCB=0, FEM=FEM)

        x     = df['z'].values
        nSpan = len(x)

        # TODO TODO finda way to use these matrices instead of the ones computed with flexibility
        #print('CB MM\n',CB['MM'])
        #print('CB KK\n',CB['KK'])

        # --- Setup shape functions
        if main_axis=='x':
            raise NotImplementedError('')
        else:
            pass
            # we need to swap the CB modes
        nShapes=len(shapes)
        PhiU = np.zeros((nShapes,3,nSpan)) # Shape
        PhiV = np.zeros((nShapes,3,nSpan)) # Shape
        PhiK = np.zeros((nShapes,3,nSpan)) # Shape
        dx=np.unique(np.around(np.diff(x),4))
        if len(dx)>1:
            print(x)
            print(dx)
            raise NotImplementedError()
        for iShape, idShape in enumerate(shapes):
            if idShape==0:
                # shape 0 "ux"  (uz in FEM)
                PhiU[iShape][0,:] = df_G['G3_uz'].values
                PhiV[iShape][0,:] =-df_G['G3_ty'].values
                PhiK[iShape][0,:] = gradient_regular(PhiV[iShape][0,:],dx=dx[0],order=4)
            elif idShape==1:
                # shape 1,  "uy"
                PhiU[iShape][1,:] = df_G['G2_uy'].values
                PhiV[iShape][1,:] = df_G['G2_tz'].values
                PhiK[iShape][1,:] = gradient_regular(PhiV[iShape][1,:],dx=dx[0],order=4)
            elif idShape==4:
                # shape 4,  "vy"  (vz in FEM)
                PhiU[iShape][0,:] = df_G['G6_uy'].values
                PhiV[iShape][0,:] = df_G['G6_tz'].values
                PhiK[iShape][0,:] = gradient_regular(PhiV[iShape][0,:],dx=dx[0],order=4)
            else:
                raise NotImplementedError()

        # --- Dictionary structure for YAMS
        p=dict()
        p['s_span']=x-np.min(x)
        p['s_P0']=np.zeros((3,nSpan))
        if main_axis=='z':
            p['s_P0'][2,:]=x-np.min(x)
            p['r_O']   = (df['x'].values[0], df['y'].values[0], df['z'].values[0])
            p['R_b2g'] = np.eye(3)
        p['m']  = df['m'].values
        p['EI'] = np.zeros((3,nSpan))
        if main_axis=='z':
            p['EI'][0,:]=df['E'].values*df['I'].values
            p['EI'][1,:]=df['E'].values*df['I'].values
        p['jxxG']  = df['rho']*df['Ip']          # TODO verify
        p['s_min'] = p['s_span'][0]
        p['s_max'] = p['s_span'][-1]
        p['PhiU']  = PhiU
        p['PhiV']  = PhiV
        p['PhiK']  = PhiK

        # --- Damping
        damp_zeta     = None
        RayleighCoeff = None
        DampMat       = None
        if self.File['GuyanDampMod']==1:
            # Rayleigh Damping
            RayleighCoeff=self.File['RayleighDamp']
            #if RayleighCoeff[0]==0:
            #    damp_zeta=omega*RayleighCoeff[1]/2. 
        elif self.File['GuyanDampMod']==2:
            # Full matrix
            DampMat = self.File['GuyanDampMatrix']
            DampMat=DampMat[np.ix_(shapes,shapes)]

        return p, damp_zeta, RayleighCoeff, DampMat


# --------------------------------------------------------------------------------}
# --- Export of summary file and Misc FEM variables used by SubDyn
# --------------------------------------------------------------------------------{
def yaml_array(var, M, Fmt='{:15.6e}', comment=''):
    M = np.atleast_2d(M)
    if len(comment)>0:
        s='{}: # {} x {} {}\n'.format(var, M.shape[0], M.shape[1], comment)
    else:
        s='{}: # {} x {}\n'.format(var, M.shape[0], M.shape[1])

    if M.shape[0]==1:
        if M.shape[1]==0:
            s+= '  - [ ]\n'
        else:
            for l in M:
                s+= '  - [' + ','.join([Fmt.format(le) for le in l]) + ',]\n'
    else:
        for l in M:
            s+= '  - [' + ','.join([Fmt.format(le) for le in l]) + ']\n'
    s = s.replace('e+','E+').replace('e-','E-')
    return s



def subdynPartitionVars(model):
    from welib.FEM.fem_elements import idDOF_Leader, idDOF_Fixed, idDOF_Internal
    # --- Count nodes per types
    nNodes    = len(model.Nodes)
    nNodes_I  = len(model.interfaceNodes)
    nNodes_C  = len(model.reactionNodes)
    nNodes_L  = len(model.internalNodes)

    # --- Partition Nodes:  Nodes_L = IAll - NodesR
    Nodes_I = [n.ID for n in model.interfaceNodes]
    Nodes_C = [n.ID for n in model.reactionNodes]
    Nodes_R = Nodes_I + Nodes_C
    Nodes_L = [n.ID for n in model.Nodes if n.ID not in Nodes_R]

    # --- Count DOFs - NOTE: we count node by node
    nDOF___  = sum([len(n.data['DOFs_c'])                                 for n in model.Nodes])
    # Interface DOFs
    nDOFI__  = sum([len(n.data['DOFs_c'])              for n in model.interfaceNodes])
    nDOFI_B = sum([sum(np.array(n.data['IBC'])==idDOF_Leader)   for n in model.interfaceNodes])
    nDOFI_F  = sum([sum(np.array(n.data['IBC'])==idDOF_Fixed )   for n in model.interfaceNodes])
    if nDOFI__!=nDOFI_B+nDOFI_F: raise Exception('Wrong distribution of interface DOFs')
    # DOFs of reaction nodes
    nDOFC__ = sum([len(n.data['DOFs_c'])              for n in model.reactionNodes]) 
    nDOFC_B = sum([sum(np.array(n.data['RBC'])==idDOF_Leader)   for n in model.reactionNodes])
    nDOFC_F = sum([sum(np.array(n.data['RBC'])==idDOF_Fixed)    for n in model.reactionNodes])
    nDOFC_L = sum([sum(np.array(n.data['RBC'])==idDOF_Internal) for n in model.reactionNodes])
    if nDOFC__!=nDOFC_B+nDOFC_F+nDOFC_L: raise Exception('Wrong distribution of reaction DOFs')
    # DOFs of reaction + interface nodes
    nDOFR__ = nDOFI__ + nDOFC__ # Total number, used to be called "nDOFR"
    # DOFs of internal nodes
    nDOFL_L  = sum([len(n.data['DOFs_c']) for n in model.internalNodes])
    if nDOFL_L!=nDOF___-nDOFR__: raise Exception('Wrong distribution of internal DOF')
    # Total number of DOFs in each category:
    nDOF__B = nDOFC_B + nDOFI_B
    nDOF__F = nDOFC_F + nDOFI_F          
    nDOF__L = nDOFC_L           + nDOFL_L 

    # --- Distibutes the I, L, C nodal DOFs into  B, F, L sub-categories 
    # NOTE: order is importatn for compatibility with SubDyn
    IDI__ = []
    IDI_B = []
    IDI_F = []
    for n in model.interfaceNodes:
        IDI__ += n.data['DOFs_c'] # NOTE: respects order
        IDI_B += [dof for i,dof in enumerate(n.data['DOFs_c']) if n.data['IBC'][i]==idDOF_Leader]
        IDI_F += [dof for i,dof in enumerate(n.data['DOFs_c']) if n.data['IBC'][i]==idDOF_Fixed ]
    IDI__ = IDI_B+IDI_F
    IDC__ = []
    IDC_B = []
    IDC_L = []
    IDC_F = []
    for n in model.reactionNodes:
        IDC__ += n.data['DOFs_c'] # NOTE: respects order
        IDC_B += [dof for i,dof in enumerate(n.data['DOFs_c']) if n.data['RBC'][i]==idDOF_Leader  ]
        IDC_L += [dof for i,dof in enumerate(n.data['DOFs_c']) if n.data['RBC'][i]==idDOF_Internal]
        IDC_F += [dof for i,dof in enumerate(n.data['DOFs_c']) if n.data['RBC'][i]==idDOF_Fixed   ]
    IDR__=IDC__+IDI__
    IDL_L = []
    for n in model.internalNodes:
        IDL_L += n.data['DOFs_c']

    # Storing variables similar to SubDyn
    SD_Vars={}
    SD_Vars['nDOF___']=nDOF___;
    SD_Vars['nDOFI__']=nDOFI__; SD_Vars['nDOFI_B']=nDOFI_B; SD_Vars['nDOFI_F']=nDOFI_F;
    SD_Vars['nDOFC__']=nDOFC__; SD_Vars['nDOFC_B']=nDOFC_B; SD_Vars['nDOFC_F']=nDOFC_F; SD_Vars['nDOFC_L']=nDOFC_L;
    SD_Vars['nDOFR__']=nDOFR__; SD_Vars['nDOFL_L']=nDOFL_L;
    SD_Vars['nDOF__B']=nDOF__B; SD_Vars['nDOF__F']=nDOF__F; SD_Vars['nDOF__L']=nDOF__L;
    SD_Vars['IDC__']=IDC__;
    SD_Vars['IDC_B']=IDC_B;
    SD_Vars['IDC_F']=IDC_F;
    SD_Vars['IDC_L']=IDC_L;
    SD_Vars['IDI__']=IDI__;
    SD_Vars['IDR__']=IDR__;
    SD_Vars['IDI_B']=IDI_B;
    SD_Vars['IDI_F']=IDI_F;
    SD_Vars['IDL_L']=IDL_L;
    SD_Vars['ID__B']=model.DOFc_Leader
    SD_Vars['ID__F']=model.DOFc_Fixed
    SD_Vars['ID__L']=model.DOFc_Follower
    return SD_Vars

def subdyntoYAMLSum(model, filename, more=False):
    """ 
    Write a YAML summary file, similar to SubDyn
    """
    # --- Helper functions
    def nodeID(nodeID):
        if hasattr(nodeID,'__len__'):
            return [model.Nodes.index(model.getNode(n))+1 for n in nodeID]
        else:
            return model.Nodes.index(model.getNode(nodeID))+1

    def elemID(elemID):
        #e=model.getElement(elemID)
        for ie,e in enumerate(model.Elements):
            if e.ID==elemID:
                return ie+1
    def elemType(elemType):
        from welib.FEM.fem_elements import idMemberBeam, idMemberCable, idMemberRigid
        return {'SubDynBeam3d':idMemberBeam, 'SubDynFrame3d':idMemberBeam, 'Beam':idMemberBeam, 'Frame3d':idMemberBeam,
                'SubDynTimoshenko3d':idMemberBeam,
                'SubDynCable3d':idMemberCable, 'Cable':idMemberCable,
                'Rigid':idMemberRigid,
                'SubDynRigid3d':idMemberRigid}[elemType]

    def propID(propID, propset):
        prop = model.NodePropertySets[propset]
        for ip, p in enumerate(prop):
            if p.ID == propID:
                return ip+1

    SD_Vars = subdynPartitionVars(model)

    # --- Helper functions
    s=''
    s += '#____________________________________________________________________________________________________\n'
    s += '# RIGID BODY EQUIVALENT DATA\n'
    s += '#____________________________________________________________________________________________________\n'
    s0 = 'Mass: {:15.6e} # Total Mass\n'.format(model.M_O[0,0])
    s += s0.replace('e+','E+').replace('e-','E-')
    s0 = 'CM_point: [{:15.6e},{:15.6e},{:15.6e},] # Center of mass coordinates (Xcm,Ycm,Zcm)\n'.format(model.center_of_mass[0],model.center_of_mass[1],model.center_of_mass[2])
    s += s0.replace('e+','E+').replace('e-','E-')
    s0 = 'TP_point: [{:15.6e},{:15.6e},{:15.6e},] # Transition piece reference point\n'.format(model.refPoint[0],model.refPoint[1],model.refPoint[2])
    s += s0.replace('e+','E+').replace('e-','E-')
    s += yaml_array('MRB',  model.M_O,  comment = 'Rigid Body Equivalent Mass Matrix w.r.t. (0,0,0).')
    s += yaml_array('M_P' , model.M_ref,comment = 'Rigid Body Equivalent Mass Matrix w.r.t. TP Ref point')
    s += yaml_array('M_G' , model.M_G,  comment = 'Rigid Body Equivalent Mass Matrix w.r.t. CM (Xcm,Ycm,Zcm).')
    s += '#____________________________________________________________________________________________________\n'
    s += '# GUYAN MATRICES at the TP reference point\n'
    s += '#____________________________________________________________________________________________________\n'
    s += yaml_array('KBBt' , model.KBBt,  comment = '')
    s += yaml_array('MBBt' , model.MBBt,  comment = '')
    s += yaml_array('CBBt' , model.CBBt,  comment = '(user Guyan Damping + potential joint damping from CB-reduction)')
    s += '#____________________________________________________________________________________________________\n'
    s += '# SYSTEM FREQUENCIES\n'
    s += '#____________________________________________________________________________________________________\n'
    s += '#Eigenfrequencies [Hz] for full system, with reaction constraints (+ Soil K/M + SoilDyn K0) \n'
    s += yaml_array('Full_frequencies', model.freq)
    s += '#Frequencies of Guyan modes [Hz]\n'
    s += yaml_array('GY_frequencies', model.f_G)
    s += '#Frequencies of Craig-Bampton modes [Hz]\n'
    s += yaml_array('CB_frequencies', model.f_CB)
    s += '#____________________________________________________________________________________________________\n'
    s += '# Internal FEM representation\n'
    s += '#____________________________________________________________________________________________________\n'
    s += 'nNodes_I: {:7d} # Number of Nodes: "interface" (I)\n'.format(len(model.interfaceNodes))
    s += 'nNodes_C: {:7d} # Number of Nodes: "reactions" (C)\n'.format(len(model.reactionNodes))
    s += 'nNodes_L: {:7d} # Number of Nodes: "internal"  (L)\n'.format(len(model.internalNodes))
    s += 'nNodes  : {:7d} # Number of Nodes: total   (I+C+L)\n'.format(len(model.Nodes))
    if more:
        s += 'nDOFI__ : {:7d} # Number of DOFs: "interface"          (I__)\n'.format(len(SD_Vars['IDI__']))
        s += 'nDOFI_B : {:7d} # Number of DOFs: "interface" retained (I_B)\n'.format(len(SD_Vars['IDI_B']))
        s += 'nDOFI_F : {:7d} # Number of DOFs: "interface" fixed    (I_F)\n'.format(len(SD_Vars['IDI_F']))
        s += 'nDOFC__ : {:7d} # Number of DOFs: "reactions"          (C__)\n'.format(len(SD_Vars['IDC__']))
        s += 'nDOFC_B : {:7d} # Number of DOFs: "reactions" retained (C_B)\n'.format(len(SD_Vars['IDC_B']))
        s += 'nDOFC_L : {:7d} # Number of DOFs: "reactions" internal (C_L)\n'.format(len(SD_Vars['IDC_L']))
        s += 'nDOFC_F : {:7d} # Number of DOFs: "reactions" fixed    (C_F)\n'.format(len(SD_Vars['IDC_F']))
        s += 'nDOFR__ : {:7d} # Number of DOFs: "intf+react"         (__R)\n'.format(len(SD_Vars['IDR__']))
        s += 'nDOFL_L : {:7d} # Number of DOFs: "internal"  internal (L_L)\n'.format(len(SD_Vars['IDL_L']))
    s += 'nDOF__B : {:7d} # Number of DOFs:             retained (__B)\n'.format(SD_Vars['nDOF__B'])
    s += 'nDOF__L : {:7d} # Number of DOFs:             internal (__L)\n'.format(SD_Vars['nDOF__L'])
    s += 'nDOF__F : {:7d} # Number of DOFs:             fixed    (__F)\n'.format(SD_Vars['nDOF__F'])
    s += 'nDOF_red: {:7d} # Number of DOFs: total\n'                     .format(SD_Vars['nDOF___'])
    s += yaml_array('Nodes_I', nodeID([n.ID for n in model.interfaceNodes]), Fmt='{:7d}', comment='"interface" nodes"');
    s += yaml_array('Nodes_C', nodeID([n.ID for n in model.reactionNodes ]), Fmt='{:7d}', comment='"reaction" nodes"');
    s += yaml_array('Nodes_L', nodeID([n.ID for n in model.internalNodes ]), Fmt='{:7d}', comment='"internal" nodes"');
    if more:
        s += yaml_array('DOF_I__', np.array(SD_Vars['IDI__'])+1,   Fmt='{:7d}', comment = '"interface"           DOFs"')
        s += yaml_array('DOF_I_B', np.array(SD_Vars['IDI_B'])+1,   Fmt='{:7d}', comment = '"interface" retained  DOFs')
        s += yaml_array('DOF_I_F', np.array(SD_Vars['IDI_F'])+1,   Fmt='{:7d}', comment = '"interface" fixed     DOFs')
        s += yaml_array('DOF_C__', np.array(SD_Vars['IDC__'])+1,   Fmt='{:7d}', comment = '"reaction"            DOFs"')
        s += yaml_array('DOF_C_B', np.array(SD_Vars['IDC_B'])+1,   Fmt='{:7d}', comment = '"reaction"  retained  DOFs')
        s += yaml_array('DOF_C_L', np.array(SD_Vars['IDC_L'])+1,   Fmt='{:7d}', comment = '"reaction"  internal  DOFs')
        s += yaml_array('DOF_C_F', np.array(SD_Vars['IDC_F'])+1,   Fmt='{:7d}', comment = '"reaction"  fixed     DOFs')
        s += yaml_array('DOF_L_L', np.array(SD_Vars['IDL_L'])+1,   Fmt='{:7d}', comment = '"internal"  internal  DOFs')
        s += yaml_array('DOF_R_' , np.array(SD_Vars['IDR__'])+1,   Fmt='{:7d}', comment = '"interface&reaction"  DOFs')
    s += yaml_array('DOF___B', np.array(model.DOFc_Leader  )+1, Fmt='{:7d}',  comment='all         retained  DOFs');
    s += yaml_array('DOF___F', np.array(model.DOFc_Fixed   )+1, Fmt='{:7d}',  comment='all         fixed     DOFs');
    s += yaml_array('DOF___L', np.array(model.DOFc_Follower)+1, Fmt='{:7d}',  comment='all         internal  DOFs');
    s += '\n'
    s += '#Index map from DOF to nodes\n'
    s += '#     Node No.,  DOF/Node,   NodalDOF\n'
    s += 'DOF2Nodes: # {} x 3 (nDOFRed x 3, for each constrained DOF, col1: node index, col2: number of DOF, col3: DOF starting from 1)\n'.format(model.nDOFc)
    DOFc2Nodes = model.DOFc2Nodes
    for l in DOFc2Nodes:
        s +='  - [{:7d},{:7d},{:7d}] # {}\n'.format(l[1]+1, l[2], l[3], l[0]+1 )
    s += '#     Node_[#]          X_[m]           Y_[m]           Z_[m]       JType_[-]       JDirX_[-]       JDirY_[-]       JDirZ_[-]  JStff_[Nm/rad]\n'
    s += 'Nodes: # {} x 9\n'.format(len(model.Nodes))
    for n in model.Nodes:
        s += '  - [{:7d}.,{:15.3f},{:15.3f},{:15.3f},{:14d}.,   0.000000E+00,   0.000000E+00,   0.000000E+00,   0.000000E+00]\n'.format(nodeID(n.ID), n.x, n.y, n.z, int(n.data['Type']) )
    s += '#    Elem_[#]    Node_1   Node_2   Prop_1   Prop_2     Type     Length_[m]      Area_[m^2]  Dens._[kg/m^3]        E_[N/m2]        G_[N/m2]       shear_[-]       Ixx_[m^4]       Iyy_[m^4]       Jzz_[m^4]          T0_[N]\n'
    s += 'Elements: # {} x 16\n'.format(len(model.Elements))
    for e in model.Elements:
        I = e.inertias
        s0='  - [{:7d}.,{:7d}.,{:7d}.,{:7d}.,{:7d}.,{:7d}.,{:15.3f},{:15.3f},{:15.3f},{:15.6e},{:15.6e},{:15.6e},{:15.6e},{:15.6e},{:15.6e},{:15.6e}]\n'.format(
            elemID(e.ID), nodeID(e.nodeIDs[0]), nodeID(e.nodeIDs[1]), propID(e.propIDs[0], e.propset), propID(e.propIDs[1], e.propset), elemType(e.data['Type']), 
            e.length, e.area, e.rho, e.E, e.G, e.kappa, I[0], I[1], I[2], e.T0)
        s += s0.replace('e+','E+').replace('e-','E-')
    s += '#____________________________________________________________________________________________________\n'
    s += '#User inputs\n'
    s += '\n'
    s += '#Number of properties (NProps):{:6d}\n'.format(len(model.NodePropertySets['Beam']))
    s += '#Prop No         YoungE         ShearG        MatDens          XsecD          XsecT\n'
    for ip,p in enumerate(model.NodePropertySets['Beam']):
        s0='#{:8d}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}\n'.format(p.ID, p['E'],p['G'],p['rho'],p['D'],p['t'])
        s +=  s0.replace('e+','E+').replace('e-','E-')
    s +='\n'
    s += '#No. of Reaction DOFs:{:6d}\n'.format(len(SD_Vars['IDC__']) )
    s += '#React. DOF_ID    BC\n'
    s += '\n'.join(['#{:10d}{:10s}'.format(idof+1,'     Fixed' ) for idof in SD_Vars['IDC_F']])
    s += '\n'.join(['#{:10d}{:10s}'.format(idof+1,'     Free'  ) for idof in SD_Vars['IDC_L']])
    s += '\n'.join(['#{:10d}{:10s}'.format(idof+1,'     Leader') for idof in SD_Vars['IDC_B']])
    s += '\n\n'
    s += '#No. of Interface DOFs:{:6d}\n'.format(len(SD_Vars['IDI__']))
    s += '#Interf. DOF_ID    BC\n'
    s += '\n'.join(['#{:10d}{:10s}'.format(idof+1,'    Fixed' ) for idof in SD_Vars['IDI_F']])
    s += '\n'.join(['#{:10d}{:10s}'.format(idof+1,'    Leader') for idof in SD_Vars['IDI_B']])
    s += '\n\n'
    CM = []
    from welib.yams.utils import identifyRigidBodyMM
    for n in model.Nodes:
        if 'addedMassMatrix' in n.data:
            mass, J_G, ref2COG = identifyRigidBodyMM(n.data['addedMassMatrix'])
            CM.append( (n.ID, mass, J_G, ref2COG) )
    s += '#Number of concentrated masses (NCMass):{:6d}\n'.format(len(CM))
    s += '#JointCMas           Mass            JXX            JYY            JZZ            JXY            JXZ            JYZ           MCGX           MCGY           MCGZ\n'
    for cm in CM:
        s0 = '# {:9.0f}.{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}\n'.format( nodeID(cm[0]),  cm[1], cm[2][0,0], cm[2][1,1], cm[2][2,2], cm[2][0,1], cm[2][0,2], cm[2][1,2],cm[3][0],cm[3][1],cm[3][2] )
        s += s0.replace('e+','E+').replace('e-','E-')
    s += '\n'
    #s += '#Number of members    18\n'
    #s += '#Number of nodes per member:     2\n'
    #s += '#Member I Joint1_ID Joint2_ID    Prop_I    Prop_J           Mass         Length     Node IDs...\n'
    #s += '#       77        61        60        11        11   1.045888E+04   2.700000E+00       19    18\n'
    #s += '#____________________________________________________________________________________________________\n'
    #s += '#Direction Cosine Matrices for all Members: GLOBAL-2-LOCAL. No. of 3x3 matrices=    18\n'
    #s += '#Member I        DC(1,1)        DC(1,2)        DC(1,3)        DC(2,1)        DC(2,2)        DC(2,3)        DC(3,1)        DC(3,2)        DC(3,3)\n'
    #s += '#       77  1.000E+00  0.000E+00  0.000E+00  0.000E+00 -1.000E+00  0.000E+00  0.000E+00  0.000E+00 -1.000E+00\n'
    s += '#____________________________________________________________________________________________________\n'
    s += '#FEM Eigenvectors ({} x {}) [m or rad], full system with reaction constraints (+ Soil K/M + SoilDyn K0)\n'.format(*model.Q.shape)
    s += yaml_array('Full_Modes', model.Q)
    s += '#____________________________________________________________________________________________________\n'
    s += '#CB Matrices (PhiM,PhiR) (reaction constraints applied)\n'
    s += yaml_array('PhiM', model.Phi_CB[:,:model.nModesCB] ,comment='(CB modes)')
    s += yaml_array('PhiR', model.Phi_G,  comment='(Guyan modes)')
    s += '\n'
    if more:
        s += '#____________________________________________________________________________________________________\n'
        s += '# ADDITIONAL DEBUGGING INFORMATION\n'
        s += '#____________________________________________________________________________________________________\n'
        s +=  ''
        e = model.Elements[0]
        rho=e.rho
        A = e.area
        L = e.length
        t= rho*A*L
        s0 = '{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}{:15.6e}\n'.format(model.gravity,e.area, e.length, e.inertias[0], e.inertias[1], e.inertias[2], e.kappa, e.E, e.G, e.rho, t)
        s0 = s0.replace('e+','E+').replace('e-','E-')
        s += s0
        s += yaml_array('KeLocal' +str(), model.Elements[0].Ke(local=True))

        for ie,e in enumerate(model.Elements):
            s += yaml_array('DC' +str(ie+1), e.DCM.transpose())
            s += yaml_array('Ke' +str(ie+1), e.Ke())
            s += yaml_array('Me' +str(ie+1), e.Me())
            s += yaml_array('FGe'+str(ie+1), e.Fe_g(model.gravity))
            s += yaml_array('FCe'+str(ie+1), e.Fe_o())

            s += yaml_array('KeLocal' +str(ie+1), e.Ke(local=True))
            s += yaml_array('MeLocal' +str(ie+1), e.Me(local=True))
            s += yaml_array('FGeLocal'+str(ie+1), e.Fe_g(model.gravity, local=True))
            s += yaml_array('FCeLocal'+str(ie+1), e.Fe_o(local=True))

        s += '#____________________________________________________________________________________________________\n'
        e = model.Elements[0]
        s += yaml_array('Ke', e.Ke(local=True), comment='First element stiffness matrix'); # TODO not in local
        s += yaml_array('Me', e.Me(local=True), comment='First element mass matrix');
        s += yaml_array('FGe', e.Fe_g(model.gravity,local=True), comment='First element gravity vector');
        s += yaml_array('FCe', e.Fe_o(local=True), comment='First element cable pretension');
        s += '#____________________________________________________________________________________________________\n'
        s += '#FULL FEM K and M matrices. TOTAL FEM TDOFs:    {}\n'.format(model.nDOF); # NOTE: wrong in SubDyn, should be nDOFc
        s += yaml_array('K', model.KK, comment='Stiffness matrix');
        s += yaml_array('M', model.MM, comment='Mass matrix');
        s += '#____________________________________________________________________________________________________\n'
        s += '#Gravity and cable loads applied at each node of the system (before DOF elimination with T matrix)\n'
        s += yaml_array('FG', model.FF_init, comment=' ');
        s += '#____________________________________________________________________________________________________\n'
        s += '#Additional CB Matrices (MBB,MBM,KBB) (constraint applied)\n'
        s += yaml_array('MBB'    , model.MBB, comment='');
        s += yaml_array('MBM'    , model.MBM[:,:model.nModesCB], comment='');
        s += yaml_array('CMMdiag', model.CMM, comment='(2 Zeta OmegaM)');
        s += yaml_array('KBB'    , model.KBB, comment='');
        s += yaml_array('KMM'    , np.diag(model.KMM), comment='(diagonal components, OmegaL^2)');
        s += yaml_array('KMMdiag', np.diag(model.KMM)[:model.nModesCB], comment='(diagonal components, OmegaL^2)');
        s += yaml_array('PhiL'   , model.Phi_CB, comment='');
        s += 'PhiLOm2-1: # 18 x 18 \n'
        s += 'KLL^-1: # 18 x 18 \n'
    s += '#____________________________________________________________________________________________________\n'
    s += yaml_array('T_red', model.T_c, Fmt = '{:9.2e}', comment='(Constraint elimination matrix)');
    s += 'AA: # 16 x 16 (State matrix dXdx)\n'
    s += 'BB: # 16 x 48 (State matrix dXdu)\n'
    s += 'CC: # 6 x 16 (State matrix dYdx)\n'
    s += 'DD: # 6 x 48 (State matrix dYdu)\n'
    s += '#____________________________________________________________________________________________________\n'
    s += yaml_array('TI', model.T_refPoint,  Fmt = '{:9.2e}',comment='(TP refpoint Transformation Matrix TI)');
    if filename is not None:
        with open(filename, 'w') as f:
            f.write(s)