def plot_prediction_relevance1():
plt.figure(figsize=(7.2, 2))
plt.subplots_adjust(wspace=0.4)
for sigt in sg_sl:
ax = plt.subplot(1, 3, sigt + 1)
plt.title(titles1[sigt], fontsize=8)
if (sigt == 0): plt.text(-50, 450, titles[1], fontsize=14)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.xlabel("Predicted rate[Hz]", fontsize=8)
plt.ylabel("Measured rate[Hz]", fontsize=8)
R2s = np.zeros(len(sg_sl))
plt.plot(np.arange(0, 400, 1), np.arange(0, 400, 1), '-', color='k')
for ch in ch_sl:
obser, preds = ob_fr_data[sigt][ch][:,
1], np.array(simdata[ch][sigt])
outputd.append(np.array([obser, preds]).T)
#pvalue=round(stats.ranksums(np.array(simdata[ch][sigt]),
#ob_fr_data[sigt][ch][:,1])[1],3)
[R21, pval] = stats.pearsonr(obser, preds)
print('Pvalue=' + str(pval))
space = int(len(stimuli[sigt]) / 5)
R2s = np.zeros(5)
for st in range(5):
x = np.array(simdata[ch][sigt])[st * space:(st + 1) * space]
y = ob_fr_data[sigt][ch][:, 1][st * space:(st + 1) * space]
fit = alt.curve_fit(x, y)
R2s[st] = alt.R2(fit[2], y)
plt.plot(fit[0],
fit[1],
'--',
linewidth=1,
color=mysim.color_bf[ch][st])
plt.scatter(x,
y,
marker=markers[ch],
color=mysim.color_bf[ch][st],
s=10)
avgR2 = round(np.average(R2s), 2)
plt.scatter(0,
0,
marker=markers[ch],
color=mysim.colors[ch],
s=10,
label=Ttype_buf[ch] + ' ' + ' $\mathrm{R}^{2}$=' +
str(avgR2) + ', P=' + str(pval))
plt.legend(loc=1, fontsize=6, edgecolor='k')
if (ch == 0):
plt.xticks(fyticks[0], fontsize=6)
plt.yticks(fyticks[0], fontsize=6)
if (ch == 1):
plt.xticks(fyticks[1], fontsize=6)
plt.yticks(fyticks[1], fontsize=6)
if (ch == 2):
plt.xticks(fyticks[2], fontsize=6)
plt.yticks(fyticks[2], fontsize=6)
plt.legend(loc=1, fontsize=6, edgecolor='gray')
def plot_mean_fring_rate():
plt.figure(figsize=(8, 7))
for ch in ch_sl:
for sigt in sg_sl:
plt.subplot(3, 3, 3 * sigt + ch + 1)
if (sigt == 1) & (ch == 0):
plt.ylabel('Firing rate [spikes/s]', fontsize=12)
if (sigt == 2) & (ch == 1): plt.xlabel('Depth [um]', fontsize=12)
if (ch == 2):
plt.text(ylimts[sigt],
1 * fyticks[2][-1] / 3,
titles1[sigt],
rotation=90,
fontsize=10)
if (sigt == 0): plt.title(Ttype_buf[ch], fontsize=8)
if ((ch == 0) & (sigt == 0)):
plt.text(0.01, 55, titles[0], fontsize=14)
for st in range(5):
space = int(len(stimuli[sigt]) / 5)
x = np.array(simdata[ch][sigt])[st * space:(st + 1) * space]
y = ob_fr_data[sigt][ch][st * space:(st + 1) * space, 1]
fit = alt.curve_fit(x, y)
R2 = "{0:.3f}".format(alt.R2(fit[2], y))
plt.plot(1e6 * stimuli[3][sigt][st * space:(st + 1) * space],
np.array(simdata[ch][sigt])[st * space:(st + 1) *
space],
linewidth=0.5,
label=str(stimuli[sigt][st * space][1][0]) +
' $\mathrm{R}^{2}$=' + str(R2),
c=mysim.color_bf[ch][st],
marker=marker_buf[st],
markerfacecolor='none',
markersize=4)
plt.plot(ob_fr_data[sigt][ch][st * space:(st + 1) * space, 0],
ob_fr_data[sigt][ch][st * space:(st + 1) * space, 1],
'--',
linewidth=0.4,
c='gray',
marker=marker_buf[st],
markerfacecolor='none',
markersize=3)
[R21, pvalue] = stats.pearsonr(y, x)
print('Pvalue=' + str(pvalue))
plt.xscale('log')
plt.xticks(fontsize=6)
if (ch == 0): plt.yticks(fyticks[0], fontsize=6)
if (ch == 1): plt.yticks(fyticks[1], fontsize=6)
if (ch == 2): plt.yticks(fyticks[2], fontsize=6)
plt.legend(loc=2, prop={'family': 'simSun', 'size': 6})
def plot_prediction_relevance():
plt.figure(figsize=(7, 6))
for ch in ch_sl:
for sigt in sg_sl:
plt.subplot(3, 3, 3 * ch + sigt + 1)
if (sigt == 0) & (ch == 1):
plt.ylabel('Observed Firing rate [Hz]', fontsize=10)
if (sigt == 1) & (ch == 2):
plt.xlabel('Predicted Firing rate [Hz]', fontsize=10)
if (sigt == 2):
plt.text(fyticks[ch][-1] * 1.05,
fyticks[ch][-1] / 2,
Ttype_buf[ch],
fontsize=10)
if (ch == 0): plt.title(titles1[sigt], fontsize=8)
if ((ch == 0) & (sigt == 0)):
plt.text(0.01, 60, titles[0], fontsize=14)
for st in range(5):
space = int(len(stimuli[sigt]) / 5)
x = np.array(simdata[ch][sigt])[st * space:(st + 1) * space]
y = ob_fr_data[sigt][ch][st * space:(st + 1) * space, 1]
#pvalue=pvalue=round(stats.ranksums(x, y)[1],2)
[R21, pvalue] = stats.pearsonr(y, x)
print('Pvalue=' + str(pvalue))
fit = alt.curve_fit(x, y)
R2 = round(alt.R2(fit[2], y), 2)
plt.scatter(x,
y,
c=mysim.color_bf[ch][st],
marker=marker_buf[st],
s=10,
label=str(stimuli[sigt][st * space][1][0]) + ' ' +
' $\mathrm{R}^{2}$=' + str(R2) + ', P=' +
str(pvalue))
if (ch == 0):
plt.xticks(fyticks[0], fontsize=6)
plt.yticks(fyticks[0], fontsize=6)
if (ch == 1):
plt.xticks(fyticks[1], fontsize=6)
plt.yticks(fyticks[1], fontsize=6)
if (ch == 2):
plt.xticks(fyticks[2], fontsize=6)
plt.yticks(fyticks[2], fontsize=6)
plt.legend(fontsize=6, edgecolor='gray')
def plot_prediction_relevance():
suptitles = ['MIPS', 'MIDP']
plotlabels = ['SA1, ', 'RA1, ', 'PC, ']
ticks = [[0, 50, 100, 150], [0, 100, 200, 300, 400]]
obdata = [
np.load('data/ob_Frate_Tdots.npy'),
np.load('data/ob_MIPD_Tdots.npy')
]
simdata = [
np.load('data/sim_Frate_Tdots.npy'),
np.load('data/sim_MIPD_Tdots.npy')
]
for sigt in [0, 1]:
ax = plt.subplot(2, 2, sigt + 3)
if (sigt == 0): plt.text(-50, 160, '(d)', fontsize=14)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.title(suptitles[sigt], fontsize=8)
plt.xlabel('Predicted ' + suptitles[sigt], fontsize=8)
plt.ylabel('Observed ' + suptitles[sigt], fontsize=8)
R2s = np.zeros(3)
for ch in [0, 1, 2]:
x = simdata[sigt][ch][:]
y = obdata[sigt][ch][:]
#pvalue=round(stats.ranksums(x, y)[1],2)
#print(pvalue)
[R21, pval1] = stats.pearsonr(y, x)
print(Ttype_buf[ch] + ' Pvalue=' + str(pval1))
fit = alt.curve_fit(x, y)
R2 = "{0:.3f}".format(alt.R2(fit[2], y))
R2s[ch] = R2
plt.scatter(x,
y,
color='w',
edgecolors=mysim.colors[ch],
marker=mysim.markers[ch],
s=15,
label=plotlabels[ch] + ' $\mathrm{R}^{2}$=' +
str(R2)) #+', P='+str(pvalue)
outputd.append(np.array([y, x]).T)
plt.plot(fit[0], fit[1], '--', color=mysim.colors[ch])
plt.xticks(ticks[sigt], fontsize=6)
plt.yticks(ticks[sigt], fontsize=6)
plt.legend(loc=2, fontsize=6, edgecolor='gray')
def plot_prediction_relevance_sine_20hz():
ch_sl = [0, 1, 2]
sg_sl = [0]
plt.figure(figsize=(8.1, 2.3))
plt.subplots_adjust(wspace=0.4)
for ch in ch_sl:
ax = plt.subplot(1, 3, ch + 1)
plt.title(Ttype_buf[ch], fontsize=8)
if (ch == 0): plt.text(-4, 30, '(b)', fontsize=14)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.xlabel("Predicted rate [spikes/s]", fontsize=8)
plt.ylabel("Measured rate [spikes/s]", fontsize=8)
for sigt in sg_sl:
obser, preds = ob_fr_data[sigt][ch][:,
1], np.array(simdata[ch][sigt])
outputd.append(np.array([obser, preds]).T)
#pvalue=round(stats.ranksums(np.array(simdata[ch][sigt]),
# ob_fr_data[sigt][ch][:,1])[1],3)
[R21, pval] = stats.pearsonr(obser, preds)
print('Pvalue=' + str(pval))
space = int(len(stimuli[sigt]) / 5)
for st in range(1):
x = np.array(simdata[ch][sigt])[st * space:(st + 1) * space]
y = ob_fr_data[sigt][ch][:, 1][st * space:(st + 1) * space]
fit = alt.curve_fit(x, y)
R2 = "{0:.3f}".format(alt.R2(fit[2], y))
plt.plot(fit[0],
fit[1],
'--',
linewidth=1,
color=mysim.colors[ch])
plt.scatter(x,
y,
marker=markers[ch],
color=mysim.colors[ch],
s=10,
label=' $\mathrm{R}^{2}$=' + str(R2))
plt.xticks(fontsize=8)
plt.yticks(fontsize=8)
plt.legend(fontsize=8, edgecolor='gray')
def plot_prediction_relevance():
suptitles=['Touchsim','Our work']
plotlabels=['SA1','RA1','PC']
ticks=[[0,10,20,30],[0,10,20,30]]
bensimia_rfz=np.hstack([alt.read_data('data/txtdata/bensimia_RFSIZE.txt',[2,1]),np.loadtxt('data/txtdata/bensimia_RFSIZE.txt')])
observed_rfz=np.hstack([alt.read_data('data/txtdata/observed_rF_size.txt',[1,2]),np.loadtxt('data/txtdata/observed_rF_size.txt')])
prfz=np.load('data/rfsize_probe_indent.npy')
y1=[bensimia_rfz[bensimia_rfz[:,0]==1,2],bensimia_rfz[bensimia_rfz[:,0]==2,2]]
y2=[observed_rfz[observed_rfz[:,0]==1,2],observed_rfz[observed_rfz[:,0]==2,2]]
simdata=[y1,prfz]
obdata=[y2,y2]
for model in [0,1]:
ax=plt.subplot(2,2,model+3)
if(model==0): plt.text(33,37,'(c)',fontsize=14)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.title(suptitles[model],fontsize=8)
plt.xlabel('Predicted '+'RF size',fontsize=8)
plt.ylabel('Observed '+'RF size',fontsize=8)
R2s=np.zeros(3)
for ch in [0,1]:
x=simdata[model][ch][:]
y=obdata[model][ch][:]
#pvalue=round(stats.ranksums(x, y)[1],3)
#print(pvalue)
[R21,pval1]=stats.pearsonr(y, x)
print(suptitles[model]+' '+Ttype_buf[ch]+' Pvalue='+str(pval1))
fit=alt.curve_fit(x,y)
R2="{0:.3f}".format(alt.R2(fit[2],y))
R2s[ch]=R2
plt.scatter(x,y,marker=mysim.markers[ch],
color='w',edgecolors=mysim.colors[ch],s=20,label=plotlabels[ch]+', $\mathrm{R}^{2}$='+str(R2)) #+', P='+str(pvalue)
plt.plot(fit[0],fit[1],'--',color=mysim.colors[ch])
plt.xticks(ticks[model],fontsize=6)
plt.yticks(ticks[model],fontsize=6)
plt.legend(fontsize=6,edgecolor='gray')
def print_gratings_spiking_trians():
ftiproi = mysim.fingertiproi #*rslib.rtm(-np.pi/2)
ftiproi = np.vstack([ftiproi, ftiproi[0, :]])
#colr=['g','b','c']
plt.figure(figsize=(2, 6))
plt.subplots_adjust(hspace=0.35)
buf = np.load('data/forms_one_bars.npy')[2]
sres = np.load('data/one_bars_simulation_res.npy')
#obdata=np.loadtxt('data/txtdata/fr_gratings.txt')
obdata = np.loadtxt('Data/txtdata/one_bar_res.txt')[0:-4, :]
tbuf = np.zeros([len(sres[0]), int(simT / simdt)])
ax = plt.subplot(3, 1, 1)
plt.text(-5, 14, "(d)", fontsize=14) #
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.scatter(buf[:, 0],
buf[:, 1],
s=5,
c=1e3 * buf[:, 5],
marker='s',
cmap=plt.cm.Greys,
vmin=0,
vmax=8)
'''
tx,ty=np.linspace(0,width,100),height/2*np.ones(100)
plt.plot(tx,ty,'k--')
plt.plot(ftiproi[:,0]+width/2,ftiproi[:,1]+height/2,'y-',linewidth=1)
plt.fill_between(ftiproi[:,0]+width/2,ftiproi[:,1]+height/2,facecolor='y',alpha=0.5)
plt.annotate('', xy=(6*width/6,height/2), xytext=(2.5*width/4,height/2),
arrowprops=dict(color='c',headwidth = 5,width = 0.05,shrink=0.00))
'''
plt.xticks([0, 2, 4, 6, 8], fontsize=8)
plt.yticks([0, height / 2, height], fontsize=8)
plt.ylabel("y [mm]", fontsize=8)
plt.xlabel("x [mm]", fontsize=8)
#recorded sites
num = 10
#sel_points=np.vstack([0*np.ones(num),np.linspace(-height/2,height/2,num)]).T #ms
#sel_points=np.array([[0,0]]) #mm
sel_points = np.array([[0, 0]])
#ax.plot(sel_points[:,0]+width/2,sel_points[:,1]+height/2,c='k',linewidth=0,
#marker='o',markersize=5,markerfacecolor='w')
ax = plt.subplot(3, 1, 2)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
Aobdata = np.zeros([len(tbuf), 2])
xt = np.linspace(0, width, len(tbuf))
for i in range(len(tbuf)):
sel = np.where(
np.abs(xt[i] - obdata[:, 0]) == np.min(np.abs(xt[i] -
obdata[:, 0])))[0][0]
Aobdata[i, 0] = obdata[sel, 0]
Aobdata[i, 1] = obdata[sel, 1]
res = np.zeros(len(sres[0]))
for ch in range(1):
sel_entry = tsensors[ch].points_mapping_entrys(sel_points)
for sp in range(len(tbuf)):
#buf1[sp]=len(sres[ch][sp][1][sel_entry])/tsensors[ch].T
st1 = int(0.6 / simdt)
st2 = int(0.95 / simdt)
A = (sres[ch][sp][sel_entry, st1:st2] == 0.04)
A = np.sum(A)
#A=A[A>0]
res[sp] = np.average(A) / 0.35
#tbuf[sp,:]=sres[ch][sp][0][sel_entry,:] # sel Vf signal
#res[13:35]=15
#res[40:43]=15
#res[50:53]=15
#res=np.average(tbuf,1)
plt.plot(Aobdata[:, 0],
res,
mysim.colors[ch],
label='Simulated',
marker='o',
markerfacecolor='none',
markersize=4)
plt.plot(Aobdata[:, 0],
Aobdata[:, 1],
'gray',
label='Recorded',
marker='o',
markerfacecolor='none',
markersize=4)
plt.yticks([0, 50, 100, 150], fontsize=8)
plt.xticks([0, 2, 4, 6, 8], fontsize=8)
plt.legend(loc=1, prop={'family': 'simSun', 'size': 7})
plt.xlabel("x [mm]", fontsize=8)
plt.ylabel("Firing rate", fontsize=8)
plt.savefig('saved_figs/gratings_firing.png', bbox_inches='tight', dpi=300)
ax = plt.subplot(3, 1, 3)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
for ch in [0]:
x = res
y = Aobdata[:, 1]
#pvalue=round(stats.ranksums(x, y)[1],2)
[R21, pval] = stats.pearsonr(y, x)
print(pval)
fit = alt.curve_fit(x, y)
R2 = "{0:.3f}".format(alt.R2(fit[2], y))
plt.scatter(x,
y,
color='w',
edgecolors=mysim.colors[ch],
marker=mysim.markers[ch],
s=15,
label=' $\mathrm{R}^{2}$=' + str(R2)) #+', P='+str(pvalue)
outputd.append(np.array([y, x]).T)
plt.plot(fit[0], fit[1], '--', color=mysim.colors[ch])
[R21, pval] = stats.pearsonr(y, x)
print('Pvalue=' + str(pval))
plt.yticks([0, 20, 40, 60, 80], fontsize=8)
plt.xticks([0, 20, 40, 60, 80], fontsize=8)
plt.legend(loc=1, prop={'family': 'simSun', 'size': 8})
plt.xlabel("Observed Firing rate", fontsize=8)
plt.ylabel("Simulated Firing rate", fontsize=8)
def pop_paras_fitting(ch):
lowbounds = [[0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0]]
'''
upbounds=[[5000,2,5000,100,100],
[500,2,100,100,100,0.2],
[2000,2,1000,100,100,1]]
'''
upbounds = [[1000, 2, 1000, 100, 100], [5000, 2, 100, 100, 100, 1],
[5000, 2, 100, 100, 100, 1]]
#np.load('data/ob_vibro_firing_rate.npy')
dataset = rate_to_dm_dt[ch]
xdata = dataset[:, 0]
ydata = dataset[:, 1]
plt.scatter(xdata, ydata, color='gray',
marker='+') #, label='Original data')
x = np.arange(np.min(xdata), np.max(xdata), 1)
if (ch == 0):
popt, pcov = curve_fit(Pop_SA1_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = Pop_SA1_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(
x,
Pop_SA1_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label='Cs=%5.3f\n Cd=%5.3f\n Ce=%5.3f\n Rc=%5.3f\n 1/τ=%5.3f' %
tuple(popt) + '\n\n $\mathrm{R}^{2}$=' + str(R2))
pop_paras.append(popt)
paras_dict[0]['Cs'], paras_dict[0]['Cd'], paras_dict[0][
'Ce'], paras_dict[0]['Rc'], paras_dict[0]['Kf'] = popt[0], popt[
1], popt[2], popt[3], popt[4]
if (ch == 1):
popt, pcov = curve_fit(Pop_RA1_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = Pop_RA1_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(
x,
Pop_RA1_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label=
'Cs=%5.3f\n Cd=%5.3f\n Ce=%5.3f\n Rc=%5.3f\n 1/τ=%5.3f\n w=%5.3f'
% tuple(popt) + '\n\n $\mathrm{R}^{2}$=' + str(R2))
pop_paras.append(popt)
paras_dict[1]['Cs'], paras_dict[1]['Cd'], paras_dict[1][
'Ce'], paras_dict[1]['Rc'], paras_dict[1]['Kf'], paras_dict[1][
'w'] = popt[0], popt[1], popt[2], popt[3], popt[4], popt[5]
elif (ch == 2):
popt, pcov = curve_fit(Pop_PC_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = Pop_PC_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(
x,
Pop_PC_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label=
'Cs=%5.3f\n Cd=%5.3f\n Ce=%5.3f\n Rc=%5.3f\n 1/τ=%5.3f\n w=%5.3f' %
tuple(popt) + '\n\n $\mathrm{R}^{2}$=' + str(R2))
pop_paras.append(popt)
paras_dict[2]['Cs'], paras_dict[2]['Cd'], paras_dict[2][
'Ce'], paras_dict[2]['Rc'], paras_dict[2]['Kf'], paras_dict[2][
'w'] = popt[0], popt[1], popt[2], popt[3], popt[4], popt[5]
plt.xscale('log')
plt.xticks([10, 100, 1000], fontsize=8)
plt.xlabel('Depth [um]')
if (ch == 0): plt.ylabel('Firing rate [Spikes/s]')
plt.legend(ncol=1, fontsize=8)
np.save('data/pop_fitting_paras.npy', pop_paras)
def single_paras_fitting(ch):
lowbounds = [[0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [0, 0, 0, 0]]
'''
upbounds=[[20000,10,10,10,1000,100,1],
[500,10,10,1000,1],
[40000,10,1000,1]]
'''
upbounds = [[10000, 10, 10, 10, 1000, 100, 1], [10000, 10, 10, 1000, 1],
[1000000, 10, 1000, 1]]
xdata = Th_data_Set[ch][:, 0]
ydata = Th_data_Set[ch][:, 1]
plt.scatter(xdata, ydata, color='gray', marker='+')
x = np.arange(np.min(xdata), np.max(xdata), 0.1)
if (ch == 0):
popt, pcov = curve_fit(single_SA1_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = single_SA1_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(
x,
single_SA1_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label=
' KN=%5.3f\n Kb1=%5.3f\n Kb2=%5.3e\n Ku=%5.3e\n fb=%5.3f\n fl=%5.3f\n Q=%5.3f\n'
% tuple(popt) + '\n $\mathrm{R}^{2}$=' + str(R2))
single_paras.append(popt)
paras_dict_SA1 = {
'Kb1': popt[1],
'Kb2': popt[2],
'Ku': popt[3],
'fb': popt[4],
'fl': popt[5],
'Q': popt[6]
}
paras_dict.append(paras_dict_SA1)
if (ch == 1):
popt, pcov = curve_fit(single_RA1_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = single_RA1_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(
x,
single_RA1_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label=' KN=%5.3f\n Kb1=%5.3f\n Kb2=%5.3e\n fb=%5.3f\n Q=%5.3f\n' %
tuple(popt) + '\n $\mathrm{R}^{2}$=' + str(R2))
single_paras.append(popt)
paras_dict_RA1 = {
'Kb1': popt[1],
'Kb2': popt[2],
'fb': popt[3],
'Q': popt[4]
}
paras_dict.append(paras_dict_RA1)
elif (ch == 2):
popt, pcov = curve_fit(single_PC_func,
xdata,
ydata,
bounds=(lowbounds[ch], upbounds[ch]))
observed = ydata
predicted = single_PC_func(xdata, *popt)
R2 = "{0:.3f}".format(alt.R2(observed, predicted))
plt.plot(x,
single_PC_func(x, *popt),
mysim.colors[ch],
linewidth=1.5,
label=' KN=%5.3e\n Kb2=%5.3f\n fb=%5.3e\n Q=%5.3f\n' %
tuple(popt) + '\n $\mathrm{R}^{2}$=' + str(R2))
single_paras.append(popt)
paras_dict_PC = {'Kb2': popt[1], 'fb': popt[2], 'Q': popt[3]}
paras_dict.append(paras_dict_PC)
plt.xscale('log')
plt.yscale('log')
plt.xticks([1, 10, 100, 1000], fontsize=8)
plt.yticks([0.01, 0.1, 1, 10, 100, 1000], fontsize=8)
plt.xlabel('Frequency [Hz]')
if (ch == 0): plt.ylabel('Depth [um]')
plt.legend(loc=3, ncol=1, fontsize=8)
np.save('data/single_fitting_paras.npy', single_paras)
psycho_act_data[0][:, 1],
'-',
color='k',
linewidth=lw,
marker='o',
markerfacecolor='none',
markersize=6,
label='Psycho')
Iv1 = np.average(psycho_act_data[0][:, 1])
Iv0 = np.average(Pdata[1])
Kpv = Iv1 / Iv0
x = psycho_act_data[0][:, 1]
y = Pdata[1] * Kpv
fit = alt.linear_curve_fit(x, y)
R2v = "{0:.2f}".format(alt.R2(fit[2], y))
plt.plot(Pdata[0],
Pdata[1],
'-',
color='red',
marker='.',
linewidth=lw,
label='$\mathrm{K}_{pv}$=1')
plt.plot(Pdata[0],
Pdata[1] * Kpv,
'-',
color='b',
marker='.',
markersize=6,
linewidth=lw,
def plot_prediction_relevance():
plotlabels1=[str(PFs[0]),str(PFs[1]),str(PFs[2])]
plotlabels2=[str(curves[0]),
str(curves[1]),
str(curves[2]),
str(curves[3]),
str(curves[4]),
str(curves[5]),
str(curves[6]),
]
ticks=[[0,25,50,75,100],[0,10,20,30],[0,25,50,75,100]]
obdata=[np.loadtxt('data/txtdata/fring_diff_cav.txt')[:,1],
np.loadtxt('data/txtdata/fring_diff_cav_RA1.txt')[:,0],
np.loadtxt('data/txtdata/fring_diff_distances.txt')[:,1]]
simdata=[np.load('data/sim_fr_cav_SA1.npy'),
np.load('data/sim_fr_cav_RA1.npy'),
np.load('data/sim_fr_dis.npy')]
print('Fr reponse change as a function of curvature')
for ch in range(1):
if(ch==0):ax=plt.subplot(3,2,5)
else:ax=plt.subplot(3,2,6)
#if(ch==0): plt.text(-2,160,'(e)',fontsize=14)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.xlabel('Predicted firing rate (spikes/s)',fontsize=8)
plt.ylabel('Observed firing rate (spikes/s)',fontsize=8)
R2s=np.zeros(len(PFs))
for sigt in range(len(PFs)):
x=simdata[ch][sigt][:]
y=obdata[ch][sigt*len(rads):(sigt+1)*len(rads)]
#pvalue=round(stats.ranksums(x, y)[1],3)
#print(pvalue)
[R21,pval1]=stats.pearsonr(y, x)
print(Ttype_buf[ch]+' Pvalue='+str(pval1))
fit=alt.curve_fit(x,y)
R2="{0:.3f}".format(alt.R2(fit[2],y))
R2s[sigt]=R2
plt.scatter(x,y,marker=mysim.markers[sigt],
color='w',edgecolors=mysim.otc1[sigt],s=20,label=plotlabels1[sigt]+'N, $\mathrm{R}^{2}$='+str(R2)) #+' P='+str(pvalue)
plt.plot(fit[0],fit[1],'--',color=mysim.otc1[sigt])
plt.legend(loc=1,fontsize=6,edgecolor='k')
plt.xticks(ticks[ch],fontsize=6)
plt.yticks(ticks[ch],fontsize=6)
plt.legend(fontsize=6,edgecolor='gray')
print('Fr changing with distance under diff cavature')
for ch in [2]:
ax=plt.subplot(3,2,6)
ax.spines['top'].set_color('None')
ax.spines['right'].set_color('None')
plt.xlabel('Predicted firing rate (spikes/s)',fontsize=8)
plt.ylabel('Observed firing rate (spikes/s)',fontsize=8)
R2s=np.zeros(len(rads))
for sigt in range(1,len(rads)):
x=simdata[ch][sigt-1][:]
y=obdata[ch][sigt*len(distances):(sigt+1)*len(distances)]
[R21,pval1]=stats.pearsonr(y, x)
print(Ttype_buf[0]+' Pvalue='+str(pval1))
fit=alt.curve_fit(x,y)
R2="{0:.3f}".format(alt.R2(fit[2],y))
R2s[sigt]=R2
plt.scatter(x,y,marker=mysim.markers[sigt],
color='w',edgecolors=mysim.otc1[sigt],s=20,label=plotlabels2[sigt]+'$\mathrm{m}^{-1}$, $\mathrm{R}^{2}$='+str(R2)) #+' P='+str(pvalue)
plt.plot(fit[0],fit[1],'--',color=mysim.otc1[sigt])
plt.legend(loc=1,fontsize=6,edgecolor='k')
plt.xticks(ticks[2],fontsize=6)
plt.yticks(ticks[2],fontsize=6)
plt.legend(fontsize=6,edgecolor='gray')
