def fitDataToLeakyDielectricModel(saveDirName,c0NomIn,Vpp,threshFit,dataStruct):
VRMS=Vpp/np.sqrt(2)
tauNomIn=calcTimeConstantOxideOnly(c0NomIn,params)
combDriveDispl=runCombDriveModels.combDriveModels()
frequencyDim=dataStruct.frequencyDim
omegaDimIn=frequencyDim*2*np.pi
frequencyDimStack=dataStruct.frequencyDimStack
omegaDimInStack=frequencyDimStack*2*np.pi;
displacementData = dataStruct.displacementData
lambda_Ox=params.lambda_Ox/1e-9
lambda_Ox0=lambda_Ox
bigWeight=1e10;
scaleTau=bigWeight;
COx=(params.epsilon0*params.epsilonOx)/lambda_Ox;
CBulk = (params.epsilon0*params.epsilonR)/params.L;
alpha0=1#1e-2
sigma = np.sqrt(displacementData.ravel())#np.ones((np.shape(KCL100uMDat.ravel())))
sigma[displacementData.ravel()<np.sqrt(threshFit)]=bigWeight#np.sqrt(10)
RBulkIn = (tauNomIn/COx)
epsilon0 = params.epsilon0;
epsilonOx = params.epsilonOx
COx = (epsilon0*epsilonOx)/(lambda_Ox*1e-9);
tauOx = tauNomIn;
ROx0 = tauOx/COx;#RBulkIn;#tauOx/COx;
tauBulk0 = 0.01*tauOx#/1000;
CStern0 =( epsilon0*params.epsilonStern)/params.lambda_Stern;
lambdad0=np.sqrt((params.epsilonR*params.epsilon0*params.kB*params.T)/(2*(params.eCharge**2)*c0NomIn*1000.0*params.NA))
CEDL0 =( epsilon0*params.epsilonR)/lambdad0;
optVals, pcov = curve_fit(combDriveDispl.makeVikramModelRCDielFitAlpha(VRMS,combDriveParams,params), omegaDimInStack.ravel(), displacementData.ravel(),p0=(ROx0,lambda_Ox0,tauBulk0,alpha0),bounds=(0,[inf,inf,inf,inf]),sigma=sigma,absolute_sigma=False)
ROxOpt=optVals[0]
lambdaOxOpt=optVals[1]
tauBulkOptRC=optVals[2]
alphaOptRC=optVals[3]
funcDisplRC = combDriveDispl.makeVikramModelRCDielFitAlpha(VRMS,combDriveParams,params);
modelDisplacement = funcDisplRC(omegaDimIn,ROxOpt,lambdaOxOpt,tauBulkOptRC,alphaOptRC);
#mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData))/displacementData)**2,axis=0));
mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))#/np.mean(displacementData,axis=0);
mseDisplacementDataPerc=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))/np.mean(displacementData,axis=0);
print("Before Fitting","alpha",alpha0, "ROx",ROx0,"tauBulk",tauBulk0,"lambdaOx",lambda_Ox0)
print("Leaky Optval","alpha",alphaOptRC, "ROx",ROxOpt,"tauBulk",tauBulkOptRC,"lambdaOx",lambdaOxOpt)
print("model displacement",modelDisplacement)
print("mse",mseDisplacementData)
#print(np.mean(np.abs((modelDisplacement-displacementData))/displacementData,axis=0))
np.savez_compressed(saveDirName+'fitToLeakyDielectricModel',ROxOpt=ROxOpt,lambdaOxOpt=lambdaOxOpt,tauBulkOptRC=tauBulkOptRC,alphaOptRC=alphaOptRC,frequencyDim=frequencyDim,modelDisplacement=modelDisplacement,Vpp=Vpp,mseDisplacementData=mseDisplacementData)
return {'frequencyDim':frequencyDim, 'mseDisplacementData':mseDisplacementData,'mseDisplacementDataPerc':mseDisplacementDataPerc,'modelDisplacement':modelDisplacement,'displacementData':displacementData}
def fitDataToClassicModel(saveDirName,c0NomIn,Vpp,threshFit,dataStruct):
#curve_fit=scipy.optimize.curve_fit
VRMS=Vpp/np.sqrt(2)
tauNomIn=calcTimeConstantOxideOnly(c0NomIn,params)
combDriveDispl=runCombDriveModels.combDriveModels()
frequencyDim=dataStruct.frequencyDim
omegaDimIn=frequencyDim*2*np.pi
frequencyDimStack=dataStruct.frequencyDimStack
omegaDimInStack=frequencyDimStack*2*np.pi;
displacementData = dataStruct.displacementData
lambda_Ox=params.lambda_Ox/1e-9
lambda_Ox0=lambda_Ox
bigWeight=1e10;
scaleTau=bigWeight;
COx=(params.epsilon0*params.epsilonOx)/lambda_Ox;
CBulk = (params.epsilon0*params.epsilonR)/params.L;
alpha0=1#1e-2
sigma = np.sqrt(displacementData.ravel())#np.ones((np.shape(KCL100uMDat.ravel())))
sigma[displacementData.ravel()<np.sqrt(threshFit)]=bigWeight#np.sqrt(10)
RBulkIn = (tauNomIn/COx)
optVals, pcov = curve_fit(combDriveDispl.makeVikramModelFitAlpha(VRMS,params), omegaDimInStack.ravel(), displacementData.ravel(),p0=(tauNomIn,RBulkIn,alpha0),sigma=sigma,absolute_sigma=False)
funcDispl=combDriveDispl.makeVikramModelFitAlpha(VRMS,params)
tauOpt=optVals[0]
RBulkOpt=optVals[1]
alphaOpt=optVals[2]
modelDisplacement=funcDispl(omegaDimIn,tauOpt,RBulkOpt,alphaOpt)
#mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData))/displacementData)**2,axis=0));
mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))#/np.mean(displacementData,axis=0);
mseDisplacementDataPerc=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))/np.mean(displacementData,axis=0);
print("Before fitting","alpha",alpha0,"RBulk", RBulkIn,"tau",tauNomIn)
print("Classic","alpha",alphaOpt,"RBulk", RBulkOpt,"tau",tauOpt)
print("model displacement",modelDisplacement)
print("mse",mseDisplacementData)
#print(np.mean(np.abs((modelDisplacement-displacementData))/displacementData,axis=0))
np.savez_compressed(saveDirName+'fitToClassicCircuitModel',tauOpt=tauOpt,RBulkOpt=RBulkOpt,alphaOpt=alphaOpt,frequencyDim=frequencyDim,modelDisplacement=modelDisplacement,Vpp=Vpp,mseDisplacementData=mseDisplacementData);
return {'frequencyDim':frequencyDim, 'mseDisplacementData':mseDisplacementData, 'mseDisplacementDataPerc':mseDisplacementDataPerc,'modelDisplacement':modelDisplacement,'displacementData':displacementData}
def fitDataToLeakyDielectricSternModel(saveDirName,c0NomIn,Vpp,threshFit,dataStruct):
VRMS=Vpp/np.sqrt(2)
tauNomIn=calcTimeConstantOxideOnly(c0NomIn,params)
combDriveDispl=runCombDriveModels.combDriveModels()
frequencyDim=dataStruct.frequencyDim
omegaDimIn=frequencyDim*2*np.pi
frequencyDimStack=dataStruct.frequencyDimStack
omegaDimInStack=frequencyDimStack*2*np.pi;
displacementData = dataStruct.displacementData
lambda_Ox=params.lambda_Ox/1e-9
lambda_Ox0=lambda_Ox
bigWeight=1e10;
scaleTau=bigWeight;
COx=(params.epsilon0*params.epsilonOx)/lambda_Ox;
CBulk = (params.epsilon0*params.epsilonR)/params.L;
alpha0=1#1e-2
sigma = np.sqrt(displacementData.ravel())#np.ones((np.shape(KCL100uMDat.ravel())))
sigma[displacementData.ravel()<np.sqrt(threshFit)]=bigWeight#np.sqrt(10)
RBulkIn = (tauNomIn/COx)
epsilon0 = params.epsilon0;
epsilonOx = params.epsilonOx
COx = (epsilon0*epsilonOx)/(lambda_Ox*1e-9);
tauOx = tauNomIn#tauIn0100uMKCl;
ROx0 = tauOx/COx;
tauBulk0 = tauOx;
CStern0 =( epsilon0*params.epsilonStern)/params.lambda_Stern;
lambdad0=np.sqrt((params.epsilonR*params.epsilon0*params.kB*params.T)/(2*(params.eCharge**2)*c0NomIn*1000.0*params.NA))
CEDL0 =( epsilon0*params.epsilonR)/lambdad0;
optVals, pcov = curve_fit(combDriveDispl.makeVikramModelRCDielStrnFitAlpha(VRMS,combDriveParams,params), omegaDimInStack.ravel(), displacementData.ravel(),p0=(ROx0,lambda_Ox0,CStern0,tauBulk0,alpha0),bounds=(0,[inf,inf,0.1,inf,inf]),sigma=sigma,absolute_sigma=False)
ROxOptRCStern=optVals[0]
lambdaOxOptRCStern=optVals[1]
tauBulkOptRCStern=optVals[3]
alphaOptRCStern=optVals[4]
CSternOptRCStern=optVals[2]
funcDisplRCStern = combDriveDispl.makeVikramModelRCDielStrnFitAlpha(VRMS,combDriveParams,params);
modelDisplacement = funcDisplRCStern(omegaDimIn,ROxOptRCStern,lambdaOxOptRCStern,CSternOptRCStern,tauBulkOptRCStern,alphaOptRCStern);
#mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData))/displacementData)**2,axis=0));
mseDisplacementData=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))#/np.mean(displacementData,axis=0);
mseDisplacementDataPerc=np.sqrt(np.mean((((modelDisplacement-displacementData)))**2,axis=0))/np.mean(displacementData,axis=0);
print("Before Fitting","alpha",alpha0, "ROx",ROx0,"tauBulk",tauBulk0,"lambdaOx",lambda_Ox0,"CStern",CStern0)
print("Leaky Stern Optval","alpha",alphaOptRCStern, "ROx",ROxOptRCStern,"tauBulk",tauBulkOptRCStern,"lambdaOx",lambdaOxOptRCStern,"CStern",CSternOptRCStern)
print("model displacement",modelDisplacement)
print("mse",mseDisplacementData)
#print(np.mean(np.abs((modelDisplacement-displacementData))/displacementData,axis=0))
np.savez_compressed(saveDirName+'fitToLeakyDielectricModel',ROxOptRCStern=ROxOptRCStern,lambdaOxOptRCStern=lambdaOxOptRCStern,tauBulkOptRCStern=tauBulkOptRCStern,alphaOptRCStern=alphaOptRCStern,CSternOptRCStern=CSternOptRCStern,frequencyDim=frequencyDim,modelDisplacement=modelDisplacement,Vpp=Vpp,mseDisplacementData=mseDisplacementData)
return {'frequencyDim':frequencyDim, 'mseDisplacementData':mseDisplacementData,'mseDisplacementDataPerc':mseDisplacementDataPerc,'modelDisplacement':modelDisplacement,'displacementData':displacementData}
# targetDisplacement = np.array([0.2,0.4,0.6,0.8,1.0,1.2,1.4,1.6,1.8,2.0,2.2,2.4,2.6,2.8,3.0])
# splVikram100uMKClTargetDispl = interp1d(KCL100uMDispl,frequencyDim100uMKCl,kind='linear')
# splVikram500uMKClTargetDispl = interp1d(KCL500uMDispl,frequencyDim500uMKCl,kind='linear')
# splVikram1000uMKClTargetDispl = interp1d(KCL1000uMDispl,frequencyDim1000uMKCl,kind='linear')
## vikram100uMKClTargetDispl = splVikram100uMKClTargetDispl(targetDisplacement)
## vikram500uMKClTargetDispl = splVikram500uMKClTargetDispl(targetDisplacement)
## vikram1000uMKClTargetDispl = splVikram1000uMKClTargetDispl(targetDisplacement)
# funcKCLDisplVikramDisp = combDriveDispl.makeVikramModelFitAlphaDisp(VRMS);
# vikram100uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# vikram500uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# vikram1000uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# for k, item in enumerate(targetDisplacement):
# vikram100uMKClTargetDispl[k] = fmin(funcKCLDisplVikramDisp,2e6*2*np.pi,args=(tau100uMOpt,RBulkOpt100uM,alphaOpt100uM,targetDisplacement[k]))/(2*np.pi);
# vikram500uMKClTargetDispl[k] = fmin(funcKCLDisplVikramDisp,2e6*2*np.pi,args=(tau500uMOpt,RBulkOpt500uM,alphaOpt500uM,targetDisplacement[k]))/(2*np.pi);
# vikram1000uMKClTargetDispl[k] = fmin(funcKCLDisplVikramDisp,2e6*2*np.pi,args=(tau1000uMOpt,RBulkOpt1000uM,alphaOpt1000uM,targetDisplacement[k]))/(2*np.pi);
# if (vikram100uMKClTargetDispl[k]<0):
# vikram100uMKClTargetDispl[k]=-1*vikram100uMKClTargetDispl[k];
# if (vikram500uMKClTargetDispl[k]<0):
# vikram500uMKClTargetDispl[k]=-1*vikram500uMKClTargetDispl[k];
# if (vikram1000uMKClTargetDispl[k]<0):
# vikram1000uMKClTargetDispl[k]=-1*vikram1000uMKClTargetDispl[k];
# displ100uMKCl = np.reshape(spl100uMKCl(vikram100uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
# displ500uMKCl = np.reshape(spl500uMKCl(vikram500uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
# displ1000uMKCl = np.reshape(spl1000uMKCl(vikram1000uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
## meanTrackErrorVikram100uM = (np.abs(displ100uMKCl-targetDisplacement))
## meanTrackErrorVikram500uM = (np.abs(displ500uMKCl-targetDisplacement))
## meanTrackErrorVikram1000uM = (np.abs(displ1000uMKCl-targetDisplacement))
# meanTrackErrorVikram100uM = np.mean(np.abs(displ100uMKCl-targetDisplacement),axis=0)
# meanTrackErrorVikram500uM = np.mean(np.abs(displ500uMKCl-targetDisplacement),axis=0)
# meanTrackErrorVikram1000uM = np.mean(np.abs(displ1000uMKCl-targetDisplacement),axis=0)
#
# splRC100uMKClTargetDispl = interp1d(KCL100uMDisplRC,frequencyDim100uMKCl,kind='linear')
# splRC500uMKClTargetDispl = interp1d(KCL500uMDisplRC,frequencyDim500uMKCl,kind='linear')
# splRC1000uMKClTargetDispl = interp1d(KCL1000uMDisplRC,frequencyDim1000uMKCl,kind='linear')
## RC100uMKClTargetDispl = splRC100uMKClTargetDispl(targetDisplacement)
## RC500uMKClTargetDispl = splRC500uMKClTargetDispl(targetDisplacement)
## RC1000uMKClTargetDispl = splRC1000uMKClTargetDispl(targetDisplacement)
# funcKCLDisplRCDisp = combDriveDispl.makeVikramModelRCDielFitAlphaDisp(VRMS,combDriveParams,params);
# RC100uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# RC500uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# RC1000uMKClTargetDispl=np.zeros((len(targetDisplacement),1));
# for k, item in enumerate(targetDisplacement):
# RC100uMKClTargetDispl[k] = fmin(funcKCLDisplRCDisp,1e6*2*np.pi,args=(ROxOpt100uMRC,lambdaOxOpt100uMRC,tauBulkOpt100uMRC,alphaOpt100uMRC,targetDisplacement[k]))/(2*np.pi);
# RC500uMKClTargetDispl[k] = fmin(funcKCLDisplRCDisp,1e6*2*np.pi,args=(ROxOpt500uMRC,lambdaOxOpt500uMRC,tauBulkOpt500uMRC,alphaOpt500uMRC,targetDisplacement[k]))/(2*np.pi);
# RC1000uMKClTargetDispl[k] = fmin(funcKCLDisplRCDisp,1e6*2*np.pi,args=(ROxOpt1000uMRC,lambdaOxOpt1000uMRC,tauBulkOpt1000uMRC,alphaOpt1000uMRC,targetDisplacement[k]))/(2*np.pi);
# if (RC100uMKClTargetDispl[k]<0):
# RC100uMKClTargetDispl[k]=-1*RC100uMKClTargetDispl[k];
# if (RC500uMKClTargetDispl[k]<0):
# RC500uMKClTargetDispl[k]=-1*RC500uMKClTargetDispl[k];
# if (RC1000uMKClTargetDispl[k]<0):
# RC1000uMKClTargetDispl[k]=-1*RC1000uMKClTargetDispl[k];
# displ100uMKCl = np.reshape(spl100uMKCl(RC100uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
# displ500uMKCl = np.reshape(spl500uMKCl(RC500uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
# displ1000uMKCl = np.reshape(spl1000uMKCl(RC1000uMKClTargetDispl),(datEnd-datLen,len(targetDisplacement)));
## meanTrackErrorRC100uM = (np.abs(displ100uMKCl-targetDisplacement))
## meanTrackErrorRC500uM = (np.abs(displ500uMKCl-targetDisplacement)
## meanTrackErrorRC1000uM = (np.abs(displ1000uMKCl-targetDisplacement))
# meanTrackErrorRC100uM = np.mean(np.abs(displ100uMKCl-targetDisplacement),axis=0)
# meanTrackErrorRC500uM = np.mean(np.abs(displ500uMKCl-targetDisplacement),axis=0)
# meanTrackErrorRC1000uM = np.mean(np.abs(displ1000uMKCl-targetDisplacement),axis=0)
#
#
#
##
#
#
#
##
#
# mrkSZ=20
# lnSZ=6
# fntSZ=40
# mrkScale=1.3
# mplib.rcParams.update({'font.size':fntSZ})
# multiple_bars = plt.figure(figsize=(18,10))
# plt.semilogy(targetDisplacement, meanTrackErrorVikram100uM*100,'k-^',targetDisplacement, meanTrackErrorRC100uM*100,'b-*',markeredgewidth=0,markersize=mrkSZ,linewidth=lnSZ)
# plt.title('Mean Tracking Error at 100$\mu$$\it{M}$ KCl',fontsize=fntSZ)
# plt.xlabel('Target Displacement ($\mu$$\it{m}$)',fontsize=fntSZ)
# plt.ylabel('Tracking Error (%)',fontsize=fntSZ)
# plt.legend(['Classic Model','Leaky Model'],fontsize=fntSZ,markerscale=mrkScale,loc=0)
# plt.tight_layout()
# plt.grid(linewidth=3)
# plt.show()
# multiple_bars = plt.figure(figsize=(18,10))
# plt.semilogy(targetDisplacement, meanTrackErrorVikram500uM*100,'k-^',targetDisplacement, meanTrackErrorRC500uM*100,'b-*',markeredgewidth=0,markersize=mrkSZ,linewidth=lnSZ)
# plt.title('Mean Tracking Error at 500$\mu$$\it{M}$ KCl',fontsize=fntSZ)
# plt.xlabel('Target Displacement ($\mu$$\it{m}$)',fontsize=fntSZ)
# plt.ylabel('Tracking Error (%)',fontsize=fntSZ)
# plt.legend(['Classic Model','Leaky Model'],fontsize=fntSZ,markerscale=mrkScale,loc=0)
# plt.tight_layout()
# plt.grid(linewidth=3)
# plt.show()
# multiple_bars = plt.figure(figsize=(18,10))
# plt.semilogy(targetDisplacement, meanTrackErrorVikram1000uM*100,'k-^',targetDisplacement, meanTrackErrorRC1000uM*100,'b-*',markeredgewidth=0,markersize=mrkSZ,linewidth=lnSZ)
# plt.title('Mean Tracking Error at 1000$\mu$$\it{M}$ KCl',fontsize=fntSZ)
# plt.xlabel('Target Displacement ($\mu$$\it{m}$)',fontsize=fntSZ)
# plt.ylabel('Tracking Error (%)',fontsize=fntSZ)
# plt.legend(['Classic Model','Leaky Model'],fontsize=fntSZ,markerscale=mrkScale,loc=0)
# plt.tight_layout()
# plt.grid(linewidth=3)
# plt.show()