def exercise2(N_lst=(100, 250, 500), tests=10, confidence=0.95):
print 'Exercise 2:'
with open('../tex/dat/ex2-table-N%d-Q%d-C%d.tex' %
(N_lst[-1], tests , confidence*100),'w') as fp:
fp.write(
'\\begin{tabular}{|l|l|l|l|l|l|l|} \n'
'\\hline \n'
'N & tests & C-level & $\alpha_{N,max}$ & $\alpha_{N,max}+CI$ & '
'$\alpha_{N,max}-CI$ & $P_{N,max}$\\\\ \n'
' \\hline \\hline \n'
)
for N in N_lst:
print N
[load_mean, load_mean_lb, load_mean_ub, pmax_mean,
load_std] = load_max(
N=N, flip_ratio=0.1, pcut=0,
tests=tests, confidence=confidence)
fp.write('%d & %d & %0.1f $ \\%% $& %0.4f & %0.4f & %0.4f & %0.2f \\\\ \n' % \
(N, tests, confidence*100, \
load_mean, load_mean_lb, load_mean_ub, \
pmax_mean))
fp.write(
'\\hline \n'
'\\end{tabular} \n'
)
fp.closed
print 'The maximal load of the Hopfield Network' \
'with N=%d is %.5f with a 0.95 confidence intervall ' \
'of [%.5f,%.5f].\nThe capacity of the network is %.5f' \
% (N, load_mean,load_mean_lb ,load_mean_ub,pmax_mean)
def exercise4(N=200, c=0.1, numint=11, tests=10, confidence=0.95):
print('Exercise 4:')
i = 0
load_mean = zeros(numint,float)
load_std = zeros(numint,float)
for e in np.linspace(0,1,numint):
[load_mean[i], load_mean_lb, load_mean_ub, pmax_mean,
load_std[i]] = load_max(
N=N, flip_ratio=c, pcut=0, tests=tests,
confidence=confidence, excitatory=e)
i += 1
alpha_null = 0.1278
fig = figure()
ax1 = fig.add_subplot(111)
lp = errorbar(np.linspace(0,1,num=numint),load_mean/alpha_null,load_std/alpha_null)
ax1.set_ylim(0,1)
ax1.yaxis.grid(True, linestyle='-', which='major',
color=(0.2,0.2,0.2), alpha=0.5)
ax1.set_axisbelow(True)
ax1.set_title('Mean maximal load over different E (N=%d, Q=%d)' %
(N, tests))
ax1.set_ylabel('mean maximal load')
ax1.set_xlabel('E')
savefig('../tex/dat/ex4-mean_max_load-N%d-Q%d-C%d.png' %
(N, tests , confidence*100))
close()
def exercise3(N = 200, c= 0.1, confidence=0.95, numint = 11, tests=10):
print 'Exercise 3:'
i = 0
load_mean = zeros(numint,float)
load_std = zeros(numint,float)
with open('../tex/dat/ex3-table-N%d-Q%d-C%d.tex' %
(N, tests , confidence*100),'w') as fp:
fp.write(
'\\begin{tabular}{|l|l|l|l|l|l|l|l|} \n'
'\\hline \n'
'$P_{cut}$ & N & tests & C-level & maximal load & lower bound & '
'upper bound & $P_{N,max}$\\\\ \n'
' \\hline \\hline \n'
)
for pcut in np.linspace(0,1,num=numint):
[load_mean[i], load_mean_lb, load_mean_ub, pmax_mean,
load_std[i]] = load_max(N=N, flip_ratio=c, pcut=pcut,
tests=tests, confidence=confidence)
fp.write('%0.1f & %d & %d & %0.1f $\\%%$ '
' & %0.2f & %0.2f & %0.2f & %0.2f \\\\ \n' %
(pcut, N, tests, confidence*100,
load_mean[i], load_mean_lb, load_mean_ub, pmax_mean))
i += 1
fp.write('\hline'
'\end{tabular}'
)
fp.closed
alpha_null = 0.128
fig = figure()
ax1 = fig.add_subplot(111)
lp = errorbar(np.linspace(0,1,num=numint),load_mean/alpha_null,load_std/alpha_null)
axhline(y=0.5,linewidth=1, color='r')
ax1.set_ylim(0,1)
ax1.yaxis.grid(True, linestyle='-', which='major',
color=(0.2,0.2,0.2), alpha=0.5)
ax1.set_axisbelow(True)
ax1.set_title('Mean maximal load over different P_cut (N=%d, Q=%d)' %
(N, tests))
ax1.set_ylabel('mean maximal load (alpha_max/alpha_null)')
ax1.set_xlabel('P_cut')
savefig('../tex/dat/ex3-mean_max_load-N%d-Q%d-C%d.png' %
(N, tests , confidence*100))
close()