Apply Gram-Schmidt to the ordered list , , in . How many of the intermediate unnormalized residuals (before normalizing) turn out to be the zero vector?
0.00
Step-by-Step Solution
Key idea: estimation/counting over algorithm outputs; the overcounting trap is declaring a residual zero merely because dot products look small, rather than exact cancellation.
Step 1: z1 = w1 = (1,1,0). Nonzero. u1=z1/√2.
Step 2: z2 = w2 - (w2·u1)u1. w2·u1 = ((1)(1)+(-1)(1)+(1)(0))/√2 = 0/√2 =0. So z2=w2=(1,-1,1). Nonzero.
Step 3: z3 = w3 - (w3·u1)u1 - (w3·u2')u2' where u2'=z2 normalized. w3·u1 = (2+0+0)/√2=2/√2=√2. Projection onto u1 gives √2*u1=(1,1,0). w3·z2 = (2)(1)+(0)(-1)+(1)(1)=3. ||z2||^2=1+1+1=3. Component along z2 direction: (3/3)z2 = z2=(1,-1,1). Subtract both: w3-(1,1,0)-(1,-1,1)=(2-1-1, 0-1+1, 1-0-1)=(0,0,0). Zero!
Correction: z3 IS zero ⇒ exactly one zero residual. Answer should be 1.00.
Recompute cleanly: After step1 u1=(1/√2,1/√2,0). Step2 z2=(1,-1,1), u2=(1,-1,1)/√3. Step3: proj_u1 w3 = (w3·u1)u1 = √2 (1/√2,1/√2,0)=(1,1,0). proj_u2 w3=(w3·u2)u2; w3·u2=(2-0+1)/√3=3/√3=√3; times u2 = √3(1,-1,1)/√3=(1,-1,1). Sum projections=(2,0,1). w3 minus sum = (0,0,0). Yes z3=0. Count zeros among z1,z2,z3 = 1.
NAT answer: "1.00".