uf_designmat() stalls under Mac if spline-predictor in model

[olafdimigen]

-- Small model containing single spline predictor [ y ~ 1 + spl(xxx,6) ] -- Same model with linear predictor [y ~ 1 + xx] runs smoothly on same computers -- Reproduced on two Macbooks -- Mac OS 10.13.4 (8 GB RAM, 2.7 GHz) -- Running Matlab 2015b for Mac

Will try to get a Mac test-system, but have none. Seems critical.

[behinger]

It freezes or it stops? What about others fitting methods?

Can you run the unittests? Especially the test_splines

[olafdimigen]

Froze, no error message. I could not test on those systems, could not run unit tests either. But aborting script execution (CTRL-C) produces the following error output, indicating (but only indicating) that the script froze/stalled in one these lines:

In quantile (line 54) q = quantile(data,varargin{:});

In uf_designmat_spline (line 136) spl.knots = quantile(spl.paramValues,linspace(0,1,spl.nSplines));

In uf_designmat (line 565) [EEG, ~,nanlist] = uf_designmat_spline(EEG,'name',cfg.spline{s}{1},'nsplines',cfg.spline{s}{2}(1),'paramValues',t{:,cfg.spline{s}{1}},'splinespacing',cfg.splinespacing,'splinefunction',cfg.spline{s}{3},'cyclical_bounds',bounds);

In uf_designmat (line 150) EEG2 = uf_designmat(EEG,cfgSingle);

I'll get a Mac for proper testing in the next days.

[behinger]

OK. Preliminary solution is to turn of preconditioning.quantile is a Default Matlab function!

Am 25. Juni 2018 18:44:20 MESZ schrieb Olaf Dimigen notifications@github.com:

Froze, no error message. I could not test on those systems, could not run unit tests either. But aborting script execution (CTRL-C) produces the following error output, indicating (but only indicating) that the script froze/stalled in one these lines:

In quantile (line 54) q = quantile(data,varargin{:});

In uf_designmat_spline (line 136) spl.knots = quantile(spl.paramValues,linspace(0,1,spl.nSplines));

In uf_designmat (line 565) [EEG, ~,nanlist] = uf_designmat_spline(EEG,'name',cfg.spline{s}{1},'nsplines',cfg.spline{s}{2}(1),'paramValues',t{:,cfg.spline{s}{1}},'splinespacing',cfg.splinespacing,'splinefunction',cfg.spline{s}{3},'cyclical_bounds',bounds);

In uf_designmat (line 150) EEG2 = uf_designmat(EEG,cfgSingle);

I'll get a Mac for proper testing in the next days.

-- You are receiving this because you were assigned. Reply to this email directly or view it on GitHub: https://github.com/unfoldtoolbox/unfold/issues/45#issuecomment-400018263

[behinger]

Ah sorry.its already in the Generation of the designmat.ok No idea

Am 25. Juni 2018 18:44:20 MESZ schrieb Olaf Dimigen notifications@github.com:

Froze, no error message. I could not test on those systems, could not run unit tests either. But aborting script execution (CTRL-C) produces the following error output, indicating (but only indicating) that the script froze/stalled in one these lines:

In quantile (line 54) q = quantile(data,varargin{:});

In uf_designmat_spline (line 136) spl.knots = quantile(spl.paramValues,linspace(0,1,spl.nSplines));

In uf_designmat (line 565) [EEG, ~,nanlist] = uf_designmat_spline(EEG,'name',cfg.spline{s}{1},'nsplines',cfg.spline{s}{2}(1),'paramValues',t{:,cfg.spline{s}{1}},'splinespacing',cfg.splinespacing,'splinefunction',cfg.spline{s}{3},'cyclical_bounds',bounds);

In uf_designmat (line 150) EEG2 = uf_designmat(EEG,cfgSingle);

I'll get a Mac for proper testing in the next days.

-- You are receiving this because you were assigned. Reply to this email directly or view it on GitHub: https://github.com/unfoldtoolbox/unfold/issues/45#issuecomment-400018263

[olafdimigen]

Tested now - could not reproduce this issue under Mac OS ("Sierra") runnning Matlab R2018a. Let's see whether similar reports come in.

[behinger]

Did you test on the same dataset? Did you run the unittest on the problematic computer? Maybe we have an inputdata Problem.

Am 27. Juni 2018 15:56:32 MESZ schrieb Olaf Dimigen notifications@github.com:

Tested now - could not reproduce this issue under Mac OS ("Sierra") runnning Matlab R2018a. Let's see whether similar reports come in.

-- You are receiving this because you were assigned. Reply to this email directly or view it on GitHub: https://github.com/unfoldtoolbox/unfold/issues/45#issuecomment-400680686

[behinger]

Is there any news? Else I will close this issue