# Create graph
G, colors = build_regular_structure(width_basis=15,
basis_type="star",
nb_shapes=5,
shape=["house"],
start=0,
add_random_edges=0)
W = nx.adjacency_matrix(G)
W.eliminate_zeros()
taus = [0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1]
# Apply kernel at every node
signal = np.eye(W.shape[0])
heat_kernel = heat.Heat(W=W, taus=taus)
feats = heat_kernel.featurize(signal)
print('simple-example.py: saving feats to %s.npy' % args.outpath)
np.save(args.outpath, feats)
# --
# Cluster resulting features
# Normalize features
nfeats = feats - feats.mean(axis=0, keepdims=True)
nfeats /= (1e-10 + nfeats.std(axis=0, keepdims=True))
nfeats[np.isnan(nfeats)] = 0
# Reduce dimension
pca_feats = PCA(n_components=10).fit_transform(nfeats)
np.random.seed(args.seed)
taus = map(float, args.taus.split(','))
print("main.py: loading %s" % args.inpath, file=sys.stderr)
edgelist = np.array(pd.read_csv(args.inpath, sep='\t', header=None, dtype=int))
W = sparse.csr_matrix((np.arange(edgelist.shape[0]), (edgelist[:,0], edgelist[:,1])))
W = ((W + W.T) > 0).astype(int)
if not (np.sum(W, axis=0) > 0).all():
print("main.py: dropping isolated nodes", file=sys.stderr)
keep = np.asarray(W.sum(axis=0) > 0).squeeze()
W = W[keep]
W = W[:,keep]
print("main.py: running on graph w/ %d edges" % W.nnz, file=sys.stderr)
t = time()
hk = heat.Heat(W=W, taus=taus)
if args.n_queries < 0:
args.n_queries = hk.num_nodes
print("main.py: running %d queries" % args.n_queries, file=sys.stderr)
pfeats = par_graphwave(hk, n_queries=args.n_queries, n_chunks=args.n_chunks, n_jobs=args.n_jobs, verbose=10)
run_time = time() - t
print("main.py: took %f seconds -- saving %s.npy" % (run_time, args.outpath), file=sys.stderr)
np.save(args.outpath, pfeats)
# Program that solve the heat equation using finite differences import heat import test_function # Factors r = 0.3 dx = 0.1 dt = r*dx*dx x_end = 3. t_end = 6. b = [0., 0.] coef = 1. # Creating and solving the problem H = heat.Heat(dx, dt, x_end, t_end, coef, b, test_function.sinx) H.heat_solve() # Ploting the result using matplotlui H.heat_plot()
# from heat import Heat
# from plot import Plot
# read data
df = pd.read_csv(
'/Users/lukasgehrke/Documents/temp/chatham/crd_gaze_phys-LOW_work-HIGH_equip-Bike_all_good_s.csv'
)
# select a subject
s1 = df[df['pID'] == 40]
s1[['X', 'Y']]
# read background image for heatmap
img = mpimg.imread('/Users/lukasgehrke/Documents/temp/matb.png')
h = heat.Heat('workhigh', s1, None, img)
# specify resolution
x_edges = np.linspace(0, 1280, 100)
y_edges = np.linspace(0, 720, 100)
bins = [x_edges, y_edges]
h.heatmap = h.histogram2d(bins)
h.heatmap = h.zscore(h.heatmap)
h.heatmap = h.gaussian(h.heatmap, 1)
p = plot.Plot(h)
p.cm = p.make_cm_transparent(p.cm)
p.heat_xy()
# # create plot and add plot data