Unrecognized Properties

dylanruth

Hello, I've been trying to work my way through all of the tutorials and I'm running into a problem with the error message
" Unrecognized property 'numberofpartitions' for class 'qmu' " for step 3 of the tutorial Pine Island Glacier, Uncertainty Quantification (requires Dakota),

and similarly I get the error message " Unrecognized property 'lithosphere_thickness' for class 'giamme' " for step 2 of the tutorial Subglacial Channel Formation from a Single Moulin (SHAKTI).

I checked the 2020 version of the User Guide, and on page 141 it states that typing md.gia should display the relevant parameters, mantle_viscosity, lithosphere_thickness and cross_section_shape, but when I type md.gia in my workspace what comes up is
gia mme parameters:

modelid : 0 -- index into the multi-model ensemble, determines which field will be used.
Ngia : (0x0) -- geoid (mm/yr).
Ugia : (0x0) -- uplift (mm/yr).

I'm using MATLAB and I've installed ISSM by downloading the pre-compiled binaries and opening the compressed file on my macbook pro (2020 with macOS Catalina).

Any help would be appreciated!

justinquinn

Hi Dylan. The good news is you are not crazy. The bad news is that we have a pending merge from our development branch to our production branch (the one you have checked out) that addresses the issues with the tutorials. I am mot sure exactly when this merge will occur, but I can let you know here as soon as it is.

Additionally, I am copying here the updated runme.m for the UQ tutorial. Let me know if you run into any issues with it.

%PIG Uncertainty Quantification Tutorial
steps=[1];

if any(steps==1)
	disp('   Step 1: plot flux gates');

	md = loadmodel('../Pig/Models/PIG_Control_drag');

	texts=cell(1,13);
	textpositions=cell(1,13);

	for i=1:13,
		contour=expread(['./MassFluxes/MassFlux' num2str(i) '.exp']);
		textpositions{i}=[contour.x(end) contour.y(end)];
	end
	vel=md.results.StressbalanceSolution.Vel; vel(vel==0)=nan;
	plotmodel(md,'data',vel,'log',10,'expdisp',...
		{'MassFluxes/MassFlux1.exp','MassFluxes/MassFlux2.exp',...
		'MassFluxes/MassFlux3.exp','MassFluxes/MassFlux4.exp',...
		'MassFluxes/MassFlux5.exp','MassFluxes/MassFlux6.exp',...
		'MassFluxes/MassFlux7.exp','MassFluxes/MassFlux8.exp',...
		'MassFluxes/MassFlux9.exp','MassFluxes/MassFlux10.exp',...
		'MassFluxes/MassFlux11.exp','MassFluxes/MassFlux12.exp',...
		'MassFluxes/MassFlux13.exp'},...
		'expstyle',{'k-','k-','k-','k-','k-','k-','k-',...
		'k-','k-','k-','k-','k-','k-'},'linewidth',2,...
		'text',{'1','2','3','4','5','6','7',...
		'8','9','10','11','12','13'},...
		'textposition',textpositions);
end

if any(steps==2)
	disp('   Step 2: compute cross overs from CRESIS');

	md = loadmodel('../Pig/Models/PIG_Control_drag');

	%load cross overs: CRESIS McCord Antarctica, 2009 (courtesy of John Paden)
	load('../Data/CrossOvers2009.mat');

	%interpolate cross over errors over our mesh vertices
	DeltaHH=InterpFromMeshToMesh2d(index,x,y,dhh,md.mesh.x,md.mesh.y);

	%avoid NaN values
	pos=find(isnan(DeltaHH)); DeltaHH(pos)=0;

	%filter out unrealistic error ranges
	flags=ContourToNodes(md.mesh.x,md.mesh.y,'ErrorContour.exp',1);
	pos=find(~flags); DeltaHH(pos)=0;

	%avoid large unrealistic values
	pos=find(DeltaHH>1); DeltaHH(pos)=1;
	pos=find(DeltaHH<-1); DeltaHH(pos)=-1;

	%transform into absolute errors and setup a minimum error everywhere.
	DeltaHH=abs(DeltaHH);
	pos=find(DeltaHH==0); pos2=find(DeltaHH~=0);
	DeltaHH(pos)=min(DeltaHH(pos2));

	save Models/PIG.CrossOvers DeltaHH
end

if any(steps==3)
	disp('   Step 3: sampling analysis');

	%load model and cross over errors
	md = loadmodel('../Pig/Models/PIG_Control_drag');
	load -mat Models/PIG.CrossOvers

	%partition the mesh
	npart=50;
	[partition,md]=partitioner(md,'package','chaco','npart',npart,'weighting','on');
	partition=partition-1;

	%make DeltaHH into our 3 sigma deviation
	DeltaHH=DeltaHH/6; %2 (to transform DeltaHH into a radius) x 3 (for 3 sigma)
	DeltaHH_on_partition=AreaAverageOntoPartition(md,DeltaHH,partition);
	DeltaHH_on_grids=DeltaHH_on_partition(partition+1); %just to check in case

	md.qmu.variables.thickness=normal_uncertain('descriptor','scaled_Thickness',...
		'mean',ones(npart,1),...
		'stddev',DeltaHH_on_partition,...
		'partition',partition);

	%responses
	md.qmu.responses.MassFlux1=response_function('descriptor','indexed_MassFlux_1');
	md.qmu.responses.MassFlux2=response_function('descriptor','indexed_MassFlux_2'); %grounding line
	md.qmu.responses.MassFlux3=response_function('descriptor','indexed_MassFlux_3');
	md.qmu.responses.MassFlux4=response_function('descriptor','indexed_MassFlux_4');
	md.qmu.responses.MassFlux5=response_function('descriptor','indexed_MassFlux_5');
	md.qmu.responses.MassFlux6=response_function('descriptor','indexed_MassFlux_6');
	md.qmu.responses.MassFlux7=response_function('descriptor','indexed_MassFlux_7');
	md.qmu.responses.MassFlux8=response_function('descriptor','indexed_MassFlux_8');
	md.qmu.responses.MassFlux9=response_function('descriptor','indexed_MassFlux_9');
	md.qmu.responses.MassFlux10=response_function('descriptor','indexed_MassFlux_10');
	md.qmu.responses.MassFlux11=response_function('descriptor','indexed_MassFlux_11');
	md.qmu.responses.MassFlux12=response_function('descriptor','indexed_MassFlux_12');
	md.qmu.responses.MassFlux13=response_function('descriptor','indexed_MassFlux_13');

	%mass flux profiles
	md.qmu.mass_flux_profiles={...
		'MassFlux1.exp',...
		'MassFlux2.exp',...
		'MassFlux3.exp',...
		'MassFlux4.exp',...
		'MassFlux5.exp',...
		'MassFlux6.exp',...
		'MassFlux7.exp',...
		'MassFlux8.exp',...
		'MassFlux9.exp',...
		'MassFlux10.exp',...
		'MassFlux11.exp',...
		'MassFlux12.exp',...
		'MassFlux13.exp'...
	};
	md.qmu.mass_flux_profile_directory='./MassFluxes/';

	%sampling analysis
	md.qmu.method		=dakota_method('nond_samp');
	md.qmu.method(end)	=dmeth_params_set(...
		md.qmu.method(end),...
		'seed',1234,...
		'samples',30,...
		'sample_type','lhs'...
	); %random or lhs

	%a variety of parameters
	md.qmu.params.direct=true;
	md.qmu.params.analysis_driver='';
	md.qmu.params.analysis_components='';
	md.qmu.params.evaluation_concurrency=1;
	md.qmu.params.tabular_graphics_data=true;

	md.stressbalance.restol=10^-5; %tighten tolerances for UQ analyses

	%Turn off verbosity
	md.verbose=verbose(0);

	%Here, we choose to run with 4 processors, 3 for Dakota
	% while one serves as the master
	md.cluster=generic('name',oshostname,'np',4);

	%Dakota runs in parallel with a master/slave configuration.
	% At least 2 CPUs are needed to run the UQ
	md.qmu.params.evaluation_scheduling='master';
	md.qmu.params.processors_per_evaluation=md.cluster.np-1;

	%Turn dakota on
	md.qmu.isdakota=1; md.inversion.iscontrol=0;
	md=solve(md,'Stressbalance','overwrite','y');

	save ./Models/PIG.Sampling md;
end

if any(steps==4)
	disp('   Step 4: plot partition');

	%load model
	md = loadmodel('./Models/PIG.Sampling');

	plotmodel(md,'data','mesh','partition',partition,...
	'linewidth',2, 'axis#all','image','unit','km','colorbar','off',...
	'title','Partition Edges on ISSM mesh','grid','on');
end

if any(steps==5)
	disp('   Step 5: sensitivity analysis');

	%load model
	md = loadmodel('../Pig/Models/PIG_Control_drag');

	%partition the mesh
	npart=10;
	[partition,md]=partitioner(md,'package','chaco','weighting','on');
	partition=partition-1; %switch partition to c-indexing

	%all types of variables and responses: scaled_Thickness, indexed_MassFlux_i,MaxVel,nodal_DragCoefficient_i. scaled variables are expanded.

	%variables
	md.qmu.variables.drag_coefficient=normal_uncertain('descriptor','scaled_FrictionCoefficient',...
		'mean',ones(npart,1),...
		'stddev',0.05*ones(npart,1),...
		'partition',partition);
	md.qmu.variables.rheology_B=normal_uncertain('descriptor','scaled_MaterialsRheologyB',...
		'mean',ones(npart,1),...
		'stddev',0.05*ones(npart,1),...
		'partition',partition);
	md.qmu.variables.thickness=normal_uncertain('descriptor','scaled_Thickness',...
		'mean',ones(npart,1),...
		'stddev',0.05*ones(npart,1),...
		'partition',partition);

	%responses
	md.qmu.responses.MassFlux1=response_function('descriptor','indexed_MassFlux_1');
	md.qmu.responses.MassFlux2=response_function('descriptor','indexed_MassFlux_2');
	md.qmu.responses.MassFlux3=response_function('descriptor','indexed_MassFlux_3');
	md.qmu.responses.MassFlux4=response_function('descriptor','indexed_MassFlux_4');
	md.qmu.responses.MassFlux5=response_function('descriptor','indexed_MassFlux_5');
	md.qmu.responses.MassFlux6=response_function('descriptor','indexed_MassFlux_6');
	md.qmu.responses.MassFlux7=response_function('descriptor','indexed_MassFlux_7');
	md.qmu.responses.MassFlux8=response_function('descriptor','indexed_MassFlux_8');
	md.qmu.responses.MassFlux9=response_function('descriptor','indexed_MassFlux_9');
	md.qmu.responses.MassFlux10=response_function('descriptor','indexed_MassFlux_10');
	md.qmu.responses.MassFlux11=response_function('descriptor','indexed_MassFlux_11');
	md.qmu.responses.MassFlux12=response_function('descriptor','indexed_MassFlux_12');
	md.qmu.responses.MassFlux13=response_function('descriptor','indexed_MassFlux_13');

	%mass flux profiles
	md.qmu.mass_flux_profiles={...
		'MassFlux1.exp',...
		'MassFlux2.exp',...
		'MassFlux3.exp',...
		'MassFlux4.exp',...
		'MassFlux5.exp',...
		'MassFlux6.exp',...
		'MassFlux7.exp',...
		'MassFlux8.exp',...
		'MassFlux9.exp',...
		'MassFlux10.exp',...
		'MassFlux11.exp',...
		'MassFlux12.exp',...
		'MassFlux13.exp'...
	};
	md.qmu.mass_flux_profile_directory='./MassFluxes/';

	%method: local reliability
	md.qmu.method     =dakota_method('nond_l');
	md.qmu.method(end)=dmeth_params_set(md.qmu.method(end),...
	'output','quiet');

	%parameters
	md.qmu.params.direct=true;
	md.qmu.params.analysis_driver='';
	md.qmu.params.analysis_components='';
	md.qmu.params.evaluation_concurrency=1;
	md.qmu.params.tabular_graphics_data=false;

	md.stressbalance.restol=10^-5; %tighten for qmu analyses

	%Here, we choose to run with 2 processors, 1 for Dakota
	% while one serves as the master
	md.cluster=generic('name',oshostname,'np',2);

	%Dakota runs in parallel with a master/slave configuration.
	% At least 2 CPUs are needed to run the UQ
	md.qmu.params.evaluation_scheduling='master';
	md.qmu.params.processors_per_evaluation=md.cluster.np-1;

	%Clear results and turn dakota on
	md.results=[];
	md.qmu.isdakota=1; md.inversion.iscontrol=0;
	md.verbose=verbose('qmu',true);
	md=solve(md,'Stressbalance','overwrite','y');

	save ./Models/PIG.Sensitivity md;
end

if any(steps==6)
	disp('   Step 6: plot histogram');

	%load model
	md = loadmodel('./Models/PIG.Sampling');

	%which profile are we looking at?
	index=1;

	%retrieve results for the specific profile, mass flux in m^3 water equiv/s
	result=md.results.dakota.dresp_dat(npart+index);
	result.sample=result.sample/1e12*60*60*24*365;

	%plot histogram
	plot_hist_norm(result,'cdfleg','off','cdfplt','off','nrmplt','off',...
	'xlabelplt','M (Gt/yr)','ylabelplt','F','FontSize',8,'FaceColor',...
	'none','EdgeColor','red');
end

if any(steps==7)
	disp('   Step 7: plot sensitivity');

	%load model
	md = loadmodel('./Models/PIG.Sensitivity');
	%copy dakota results into model qmu
	md.qmu.results=md.results.dakota;

	%which profile are we looking at?
	index=1;

	%To plot sensitivities
	sa=md.results.dakota.dresp_out(index).sens(1:10); sa=sa(partition+1)/1e12*60*60*24*365;
	sb=md.results.dakota.dresp_out(index).sens(11:20); sb=sb(partition+1)/1e12*60*60*24*365;
	sh=md.results.dakota.dresp_out(index).sens(21:30); sh=sh(partition+1)/1e12*60*60*24*365;

	plotmodel(md,'data',sh,'data',sa,'data',sb,'expdisp#all',...
		['MassFluxes/MassFlux' num2str(index) '.exp'],...
		'expstyle#all','b-','linewidth#all',2,...
		'nlines',3,'ncols',1, 'axis#all','image',...
		'caxis#1',[0 150],'caxis#2',[-20 0],'caxis#3',[-25 0],...
		'colorbar#all','on','colorbarfontsize#all',10,...
		'colorbartitle#1','S_{H}', 'colorbartitle#2','S_{\alpha}',...
		'colorbartitle#3','S_{B}','unit#all','km','figure',1,...
		'title','Sensitivities: H, \alpha, B');

	%To plot importance factors
	ifa=importancefactors(md,'scaled_FrictionCoefficient',['indexed_MassFlux_' num2str(index)],partition);
	ifb=importancefactors(md,'scaled_MaterialsRheologyB',['indexed_MassFlux_' num2str(index)],partition);
	ifh=importancefactors(md,'scaled_Thickness',['indexed_MassFlux_' num2str(index)],partition);

	plotmodel(md,'data',ifh,'data',ifa,'data',ifb,'expdisp#all',...
		['MassFluxes/MassFlux' num2str(index) '.exp'],...
		'expstyle#all','b-','linewidth#all',2,'log#all',10,...
		'nlines',3,'ncols',1, 'axis#all','image','caxis#all',[1e-10 1],...
		'colorbar#all','on','colorbarfontsize#all',10,...
		'colorbartitle#1','If_{H}', 'colorbartitle#2','If_{\alpha}',...
		'colorbartitle#3','If_{B}','unit#all','km','figure',2,...
		'title','Importance Factors: H, \alpha, B');
end

dylanruth

Hi, it's good to know I wasn't just missing something really obvious, and thanks for getting back to me!

I got further with the updated runme.m you sent, but I'm running into a different error in step 3:

Error_ using IssmConfig_matlab

Error in ==> ../IssmConfig/IssmConfig.cpp:106
mexFunction error message: _DAKOTA_VERSION_ not found in config.h

Error in IssmConfig (line 14)
value = IssmConfig_matlab(string);

Error in qmu/checkconsistency (line 79)
			version=IssmConfig('_DAKOTA_VERSION_'); version=str2num(version(1:3));

Error in ismodelselfconsistent (line 31)
	md=checkconsistency(md.(field),md,solution,analyses);

Error in solve (line 94)
	ismodelselfconsistent(md),

Error in runme (line 149)
	_md=solve(md,'Stressbalance','overwrite','y');

It seems that I'm missing something called _DAKOTA_VERSION_ when running config.

justinquinn

Hi Dylan,
Just tested from my end and it works: I downloaded the binaries for MATLAB, copied over the runme.m I sent you, update paths to the datasets, then ran steps 1:7. Did you make sure to add bin/ and lib/ to your path in MATLAB?

yunsar

justinquinn I met the same problem.. is that solved ? or how can I use md.gia in macOS matlab as tutorial show? thanks!