Exercises Notebook
Converted from
exercises.ipynbfor web reading.
Spectral Graph Theory - Exercises
This notebook contains 10 progressive exercises for 04-Spectral-Graph-Theory. Each exercise has a learner workspace followed by a complete reference solution. The examples emphasize graph math used in retrieval, GNNs, spectral methods, and network analysis.
Code cell 2
import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
try:
import seaborn as sns
sns.set_theme(style="whitegrid", palette="colorblind")
HAS_SNS = True
except ImportError:
plt.style.use("seaborn-v0_8-whitegrid")
HAS_SNS = False
mpl.rcParams.update({
"figure.figsize": (10, 6),
"figure.dpi": 120,
"font.size": 13,
"axes.titlesize": 15,
"axes.labelsize": 13,
"xtick.labelsize": 11,
"ytick.labelsize": 11,
"legend.fontsize": 11,
"legend.framealpha": 0.85,
"lines.linewidth": 2.0,
"axes.spines.top": False,
"axes.spines.right": False,
"savefig.bbox": "tight",
"savefig.dpi": 150,
})
np.random.seed(42)
print("Plot setup complete.")
Code cell 3
import numpy as np
import numpy.linalg as la
from collections import deque, defaultdict
import heapq
np.set_printoptions(precision=8, suppress=True)
np.random.seed(42)
def header(title):
print("\n" + "=" * len(title))
print(title)
print("=" * len(title))
def check_true(name, cond):
ok=bool(cond)
print(f"{'PASS' if ok else 'FAIL'} - {name}")
return ok
def check_close(name, got, expected, tol=1e-8):
ok=np.allclose(got, expected, atol=tol, rtol=tol)
print(f"{'PASS' if ok else 'FAIL'} - {name}")
if not ok:
print(' got =', got)
print(' expected=', expected)
return ok
def adj_from_edges(n, edges, directed=False, weighted=False):
A=np.zeros((n,n), dtype=float)
for e in edges:
if weighted:
u,v,w=e
else:
u,v=e; w=1.0
A[u,v]=w
if not directed: A[v,u]=w
return A
def components(A):
n=A.shape[0]; seen=set(); comps=[]
for s in range(n):
if s in seen: continue
q=deque([s]); seen.add(s); comp=[]
while q:
u=q.popleft(); comp.append(u)
for v in np.flatnonzero(A[u]):
if int(v) not in seen:
seen.add(int(v)); q.append(int(v))
comps.append(comp)
return comps
def laplacian(A):
return np.diag(A.sum(axis=1))-A
def softmax(x):
x=np.asarray(x,float); e=np.exp(x-x.max()); return e/e.sum()
print("Chapter 11 helper setup complete.")
Exercise 1: Graph Laplacian
Compute and verify row sums are zero.
Code cell 5
# Your Solution
# Exercise 1 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 1.")
Code cell 6
# Solution
# Exercise 1 - Graph Laplacian
header("Exercise 1: Laplacian")
A=adj_from_edges(4,[(0,1),(1,2),(2,3)])
L=laplacian(A)
check_close("row sums", L.sum(axis=1), np.zeros(4))
check_true("PSD", np.min(la.eigvalsh(L))>-1e-10)
Exercise 2: Multiplicity of Zero
Use Laplacian zero eigenvalues to count components.
Code cell 8
# Your Solution
# Exercise 2 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 2.")
Code cell 9
# Solution
# Exercise 2 - Multiplicity of Zero
header("Exercise 2: zero multiplicity")
A=adj_from_edges(5,[(0,1),(1,2),(3,4)])
vals=la.eigvalsh(laplacian(A)); zeros=np.sum(vals<1e-10)
print(vals)
check_true("two components", zeros==2)
Exercise 3: Fiedler Vector
Compute the second eigenvector for a path graph.
Code cell 11
# Your Solution
# Exercise 3 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 3.")
Code cell 12
# Solution
# Exercise 3 - Fiedler Vector
header("Exercise 3: Fiedler")
A=adj_from_edges(5,[(0,1),(1,2),(2,3),(3,4)])
vals,V=la.eigh(laplacian(A)); f=V[:,1]
print("lambda2",vals[1],"fiedler",f)
check_true("lambda2 positive for connected graph", vals[1]>0)
Exercise 4: Normalized Laplacian
Compute .
Code cell 14
# Your Solution
# Exercise 4 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 4.")
Code cell 15
# Solution
# Exercise 4 - Normalized Laplacian
header("Exercise 4: normalized Laplacian")
A=adj_from_edges(3,[(0,1),(1,2)])
d=A.sum(axis=1); Lnorm=np.eye(3)-np.diag(1/np.sqrt(d))@A@np.diag(1/np.sqrt(d))
check_close("symmetric", Lnorm, Lnorm.T)
check_true("eigenvalues nonnegative", np.min(la.eigvalsh(Lnorm))>-1e-10)
Exercise 5: Graph Smoothness
Verify for undirected edges.
Code cell 17
# Your Solution
# Exercise 5 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 5.")
Code cell 18
# Solution
# Exercise 5 - Graph Smoothness
header("Exercise 5: smoothness")
edges=[(0,1),(1,2),(2,3)]; A=adj_from_edges(4,edges); L=laplacian(A); x=np.array([1.,2.,4.,7.])
energy=x@L@x; edge_sum=sum((x[i]-x[j])**2 for i,j in edges)
check_close("Dirichlet energy", energy, edge_sum)
Exercise 6: Spectral Clustering Toy
Partition a graph by the sign of the Fiedler vector.
Code cell 20
# Your Solution
# Exercise 6 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 6.")
Code cell 21
# Solution
# Exercise 6 - Spectral Clustering Toy
header("Exercise 6: spectral clustering")
A=adj_from_edges(6,[(0,1),(1,2),(3,4),(4,5),(2,3)])
vals,V=la.eigh(laplacian(A)); labels=V[:,1]>0
print(labels)
check_true("two nonempty sides", labels.any() and (~labels).any())
Exercise 7: Random Walk Matrix
Build and check row stochasticity.
Code cell 23
# Your Solution
# Exercise 7 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 7.")
Code cell 24
# Solution
# Exercise 7 - Random Walk Matrix
header("Exercise 7: random walk")
A=adj_from_edges(4,[(0,1),(1,2),(2,3),(3,0)])
P=A/A.sum(axis=1,keepdims=True)
check_close("rows sum one", P.sum(axis=1), np.ones(4))
Exercise 8: Graph Fourier Transform
Project a signal onto Laplacian eigenvectors and reconstruct.
Code cell 26
# Your Solution
# Exercise 8 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 8.")
Code cell 27
# Solution
# Exercise 8 - Graph Fourier Transform
header("Exercise 8: graph Fourier")
A=adj_from_edges(4,[(0,1),(1,2),(2,3)])
vals,U=la.eigh(laplacian(A)); x=np.array([1.,0.,2.,1.])
hat=U.T@x; rec=U@hat
check_close("reconstruction", rec, x)
Exercise 9: Heat Diffusion
Apply through eigendecomposition.
Code cell 29
# Your Solution
# Exercise 9 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 9.")
Code cell 30
# Solution
# Exercise 9 - Heat Diffusion
header("Exercise 9: heat diffusion")
A=adj_from_edges(4,[(0,1),(1,2),(2,3)]); L=laplacian(A); vals,U=la.eigh(L); x=np.array([1.,0.,0.,0.])
y=U@np.diag(np.exp(-0.5*vals))@U.T@x
print(y)
check_close("mass conserved", y.sum(), x.sum())
Exercise 10: Spectral Gap
Compare path and complete graph algebraic connectivity.
Code cell 32
# Your Solution
# Exercise 10 - learner workspace
# Write your solution here, then run the reference solution below to compare.
print("Learner workspace ready for Exercise 10.")
Code cell 33
# Solution
# Exercise 10 - Spectral Gap
header("Exercise 10: spectral gap")
P=adj_from_edges(5,[(0,1),(1,2),(2,3),(3,4)]); K=np.ones((5,5))-np.eye(5)
gapP=la.eigvalsh(laplacian(P))[1]; gapK=la.eigvalsh(laplacian(K))[1]
print(gapP,gapK)
check_true("complete graph better connected", gapK>gapP)