<div dir="ltr">To simply trim out all +/-150 uV, for sample, from all channel all dataponts, use trimOutlier.<div><a href="http://sccn.ucsd.edu/wiki/Plugin_list_process">http://sccn.ucsd.edu/wiki/Plugin_list_process</a><br>
</div><div><br></div><div>Makoto</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">2014-04-09 13:10 GMT-07:00 Jason Palmer <span dir="ltr"><<a href="mailto:japalmer29@gmail.com" target="_blank">japalmer29@gmail.com</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Dear Tatu,<br>
<br>
Apparently that data is causing a "stack overflow", which can occur with<br>
large datasets using the Windows version (which forces a maximum stack<br>
size). It may be just small enough with the block of data removed. However,<br>
since it also contains large artifacts, this probably contributes to the<br>
need for addition stack resources. The Mac and Linux versions should not<br>
have this issue.<br>
<br>
I typically remove large artifacts before running ICA by simply detecting<br>
timepoints where the potential magnitude is greater than say 150uV, and use<br>
pop_select.m to remove them. Or if the boundary events are always at the<br>
beginning/end, or known locations, you can use pop_select to remove them as<br>
well. Or you could search for boundary events and remove continuous section<br>
around the boundaries.<br>
<br>
FYI, I usually remove bad channels with simple heuristics, do average<br>
reference, and high-pass filter above 1 Hz as the remaining pre-processing.<br>
<br>
Best,<br>
Jason<br>
<div class="HOEnZb"><div class="h5"><br>
-----Original Message-----<br>
From: <a href="mailto:eeglablist-bounces@sccn.ucsd.edu">eeglablist-bounces@sccn.ucsd.edu</a><br>
[mailto:<a href="mailto:eeglablist-bounces@sccn.ucsd.edu">eeglablist-bounces@sccn.ucsd.edu</a>] On Behalf Of Tatu Huovilainen<br>
Sent: Wednesday, April 09, 2014 1:47 AM<br>
To: <a href="mailto:eeglablist@sccn.ucsd.edu">eeglablist@sccn.ucsd.edu</a><br>
Subject: [Eeglablist] AMICA on semicontinuous data<br>
<br>
Hello EEGlab community,<br>
<br>
I'm still somewhat new to EEGlab, but the features and their documentation<br>
have made it really easy to hop on. Still, I've run into a problem I don't<br>
have the knowhow to solve and the documentation on it is pretty scarce.<br>
<br>
I'm using AMICA for cleaning the data, which works unbelievably well by the<br>
way, but I've learned that if there's even one boundary event in the data<br>
the program crashes. This is problematic as some of the blocks I'm trying to<br>
run ICA on have some ~10 second windows with huge artefacts that have to be<br>
removed. Below is an output of the AMICA command. It's from a block that<br>
without a removed part would run through perfectly well.<br>
<br>
Has anyone encountered this before, and is there maybe some simple<br>
workaround I don't see?<br>
<br>
Any advice would be appreciated. Thank you in advance!<br>
<br>
-Tatu<br>
<br>
(Using Matlab R2012a & EEGlab 13.1.1<br>
[ EEG.icaweights, EEG.icasphere, mods ] = runamica12(<br>
EEG.data(chanInd('EEG'),:) ); %%% The chanInd('EEG') just returns a list of<br>
channels of the type.<br>
The system cannot find the path specified.<br>
The system cannot find the path specified.<br>
The system cannot find the path specified.<br>
No recognized parallel environment found. Run qconf -spl to get a list of<br>
available environments and use keyword use_pe.<br>
Running locally with maximum of 4 threads.<br>
Writing data file: D:\Tatu\AMICATMP\tmpdata61471.fdt<br>
1 processor name = XXXXXX<br>
1 host_num = 1376903278<br>
This is MPI process 1 of 1 ; I am process 1<br>
of<br>
1 on node: XXXXXX %Post removed, probably no need..<br>
1 : node root process 1 of 1<br>
Processing arguments ...<br>
num_files = 1<br>
FILES:<br>
D:\Tatu\AMICATMP\tmpdata61471.fdt<br>
num_dir_files = 1<br>
initial matrix block_size = 256<br>
do_opt_block = 0<br>
number of models = 1<br>
number of density mixture components = 3<br>
pdf type = 0<br>
max_iter = 2000<br>
num_samples = 1<br>
data_dim = 134<br>
field_dim = 382975<br>
do_history = 0<br>
histstep = 10<br>
share_comps = 0<br>
share_start = 100<br>
comp_thresh = 0.990000000000000<br>
share_int = 100<br>
initial lrate = 5.000000000000000E-002<br>
minimum lrate = 1.000000000000000E-008<br>
lrate factor = 0.500000000000000<br>
initial rholrate = 5.000000000000000E-002<br>
rho0 = 1.50000000000000<br>
min rho = 1.00000000000000<br>
max rho = 2.00000000000000<br>
rho lrate factor = 0.500000000000000<br>
kurt_start = 3<br>
num kurt = 5<br>
kurt interval = 1<br>
do_newton = 1<br>
newt_start = 50<br>
newt_ramp = 10<br>
initial newton lrate = 1.00000000000000<br>
do_reject = 0<br>
num reject = 3<br>
reject sigma = 3.00000000000000<br>
reject start = 2<br>
reject interval = 3<br>
max_thrds = 2<br>
write step = 10<br>
write_nd = 0<br>
write_LLt = 1<br>
dec window = 1<br>
max_decs = 3<br>
fix_init = 0<br>
update_A = 1<br>
update_c = 1<br>
update_gm = 1<br>
update_alpha = 1<br>
update_mu = 1<br>
update_beta = 1<br>
invsigmax = 100.000000000000<br>
invsigmin = 0.000000000000000E+000<br>
do_rho = 1<br>
load_rej = 0<br>
load_c = 0<br>
load_gm = 0<br>
load_alpha = 0<br>
load_mu = 0<br>
load_beta = 0<br>
load_rho = 0<br>
load_comp_list = 0<br>
do_mean = 1<br>
do_sphere = 1<br>
doPCA = 1<br>
pcakeep = 134<br>
pcadb = 30.0000000000000<br>
byte_size = 4<br>
doscaling = 1<br>
scalestep = 1<br>
A subdirectory or file D:\Tatu\AMICATMP\amicaouttmp\ already exists.<br>
output directory = D:\Tatu\AMICATMP\amicaouttmp\<br>
1 : setting num_thrds to 2 ...<br>
1 : using 2 threads.<br>
1 : node_thrds = 2<br>
bytes in real = 1<br>
1 : REAL nbyte = 1<br>
getting segment list ...<br>
blocks in sample = 382975<br>
total blocks = 382975<br>
node blocks = 382975<br>
node 1 start: file 1 sample 1 index<br>
1<br>
node 1 stop : file 1 sample 1 index<br>
382975<br>
1 : data = 19.4008998870850 8.51673507690430<br>
getting the mean ...<br>
mean = -1.38158794941323 9.108994773376549E-003<br>
-9.745015379274158E-003<br>
subtracting the mean ...<br>
getting the sphering matrix ...<br>
cnt = 382975<br>
doing eig nx = 134 lwork = 179560<br>
minimum eigenvalues = 1.32521304789175 1.60365130875427<br>
1.83976537562973<br>
maximum eigenvalues = 287079.512877785 8402.99938147310<br>
4669.66292334350<br>
num eigs kept = 134<br>
numeigs = 134<br>
sphering the data ...<br>
1 Allocating variables ...<br>
1 : Initializing variables ...<br>
1 : block size = 256<br>
1 : entering the main loop ...<br>
forrtl: severe (170): Program Exception - stack overflow<br>
Image PC Routine Line Source<br>
amica12win64.exe 000000013FCEE047 Unknown Unknown Unknown<br>
amica12win64.exe 000000013FC55847 Unknown Unknown Unknown<br>
libiomp5md.dll 000000001007BC0C Unknown Unknown Unknown<br>
libiomp5md.dll 000000001005C428 Unknown Unknown Unknown<br>
libiomp5md.dll 0000000010053BC8 Unknown Unknown Unknown<br>
libiomp5md.dll 000000001003591C Unknown Unknown Unknown<br>
amica12win64.exe 000000013FC41A35 Unknown Unknown Unknown<br>
amica12win64.exe 000000013FC1447C Unknown Unknown Unknown<br>
amica12win64.exe 000000014008F42C Unknown Unknown Unknown<br>
amica12win64.exe 000000013FCEE56F Unknown Unknown Unknown<br>
kernel32.dll 0000000076CA652D Unknown Unknown Unknown<br>
ntdll.dll 0000000076DDC541 Unknown Unknown Unknown<br>
No gm present, setting num_models to 1<br>
No W present, exiting<br>
Reference to non-existent field 'W'.<br>
<br>
Error in runamica12 (line 851)<br>
weights = mods.W(:,:,1);<br>
<br>
--------------------------------<br>
Tatu Huovilainen<br>
Research Assistant<br>
CBRU - Helsinki University<br>
<br>
_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to<br>
<a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.edu</a><br>
<br>
_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu">eeglablist-request@sccn.ucsd.edu</a><br>
</div></div></blockquote></div><br><br clear="all"><div><br></div>-- <br><div dir="ltr">Makoto Miyakoshi<br>Swartz Center for Computational Neuroscience<br>Institute for Neural Computation, University of California San Diego<br>
</div>
</div>