def test_xz_KF_CL(kf_decoder, dim_red_dict, enc_fname, wts, input_type, extra_suff='', session_length=90.,task='co', dec_method=''):
kw_dict =dict(n_neurons=kf_decoder.n_units, encoder_name=enc_fname,
fa_dict=dim_red_dict, input_type=input_type, assist_level=(0., 0.), decoder=kf_decoder, wts=wts)
if task == 'co':
task2 = sim_fa_decoding.main_xz_CL(session_length, task_kwargs=kw_dict)
elif task == 'obs':
task2 = sim_fa_decoding.main_xz_CL_obstacles(session_length, task_kwargs=kw_dict)
try:
save_dict = dict(fit_qr=dim_red_dict['fit_qr'])
except:
save_dict = {}
encoder_index = enc_fname[-5]
pnm_test = sim_fa_decoding.save_stuff(task2, suffix=input_type+task+'_'+extra_suff+'_encoder_'+encoder_index+'_'+dec_method,
save_dict=save_dict)
print 'done with ', input_type, ', saved to: ', pnm_test
#run hdf metrics file:
get_sim_tuning.all_hdf_mets(pnm_test[:-4]+'.hdf', input_type)
return pnm_test
def all_exec(figs_only = False, filenames_only = False, vfb_files=False):
'''
Summary:
Input param: figs_only:
Input param: filenames_only:
Input param: vfb_files:
Output param:
'''
if vfb_files:
input_types = ['vfb']
else:
input_types = input_types_master
if filenames_only:
file_dict = {}
#Type of decoder fitting:
for enc_name in master_encoder_fname:
#type of priv noise
for punt in p_unt:
#type of priv to shar tuning:
for priv_frac in frac_of_tun_priv:
#Calculate weights for priv and shared tuning:
#Tuned
tun = 1 - (s_unt + punt)
#Private:
p_tun = priv_frac*tun
s_tun = tun - p_tun
#Weights array: priv_untuned, private_tuned, shared_untuned, shared_tuned
w = np.array([punt, p_tun , s_unt, s_tun])
w_adj = []
for wi in w:
w_adj.append(float(str(wi)))
#Make sure w is correct, and sum(w) = 1:
assert int(np.sum(np.array(w)*1000)) == 1000
#If actually running simulation, run visual feedback:
if (not figs_only) and (not filenames_only):
pnm_train = main_sim_gen.vfb(w_adj, encoder_fname=enc_name, task_name=master_task, session_length=vfb_session_length)
for dec_fit in decoder_fit_method:
for j in input_types:
print 'Input: ', j
if figs_only or filenames_only:
#try:
hdf_fname, error = less_ridiculous_fname_fcn(w_adj, j, enc_name, dec_fit, task=master_task)
# except:
# error = True
# print 'ERROR'
# print moose
if filenames_only:
if not error:
file_dict[punt, priv_frac, j, enc_name, dec_fit] = hdf_fname
elif figs_only:
if not error:
get_sim_tuning.all_hdf_mets(hdf_fname, j)
else:
os.environ['w0'] = str(w[0])
os.environ['w1'] = str(w[1])
os.environ['w2'] = str(w[2])
os.environ['w3'] = str(w[3])
os.environ['input_tp'] = j
os.environ['p_trn'] = pnm_train
os.environ['tsk'] = master_task
os.environ['dec_mth'] = dec_fit
os.environ['enc_nm'] = enc_name
os.environ['sesh_len'] = bmi_session_length
#os.environ['p_trn'] = os.path.expandvars('$FA_GROM_DATA/sims/figs_v3/enc031116_2029vfb60_15_15_10_.pkl')
subprocess.Popen(['python main_sim_gen.py $w0 $w1 $w2 $w3 $input_tp $p_trn $tsk $dec_mth $enc_nm $sesh_len'],shell=True)
#Wait ten secs before spawning the next one:
time.sleep(10.)
if filenames_only:
return file_dict