ECEN 314: Signals and Systems Solutions to HW 3 Problem 1.27 (a) y(t) = x(t − 2) + x(2 − t) Let us check for linearity. x1 (t) → y1 (t) = x1 (t − 2) + x1 (2 − t) x2 (t) → y2 (t) = x2 (t − 2) + x2 (2 − t) ax1 (t) + bx2 (t) = x3 (t) → y3 (t) = x3 (t − 2) + x3 (2 − t) = ax1 (t − 2) + bx2 (t − 2) + ax1 (2 − t) + bx2 (2 − t) = a(x1 (t − 2) + x1 (2 − t)) + b(x2 (t − 2) + x2 (2 − t)) = ay1 (t) + by2 (t) Hence linear. Let us check for time-invariance. x1 (t) → y1 (t) = x1 (t − 2) + x1 (2 − t) x1 (t − to ) = x2 (t) → y2 (t) = x2 (t − 2) + x2 (2 − t) = x1 (t − to − 2) + x2 (2 − t − to ) 6= y1 (t − to ) Note that y1 (t − to ) = x1 (t − to − 2) + x1 (2 − t + to ). Hence time-variant. Suppose |x(t)| < B. Then y(t) < B + B = 2B (because |x(t − 2)| < B and |x(2 − t)| < B). Hence stable. Not memoryless as the present output at time t depends on t − 2. Non-Causal because y(-1)=x(-3)+x(3). So depends on future inputs. (b) y(t) = [cos(3t)]x(t) 1 Let us check for linearity. x1 (t) → y1 (t) = [cos(3t)]x1 (t) x2 (t) → y2 (t) = [cos(3t)]x2 (t) ax1 (t) + bx2 (t) = x3 (t) → y3 (t) = [cos(3t)]x3 (t) ¡ ¢ = [cos(3t)] ax1 (t) + bx2 (t) = ay1 (t) + by2 (t) Hence linear. Let us check for time-invariance. x1 (t) → y1 (t) = [cos(3t)]x1 (t) x1 (t − to ) = x2 (t) → y2 (t) = [cos(3t)]x2 (t) = [cos(3t)]x1 (t − to ) 6= y1 (t − to ) Note that y1 (t − to ) = [cos(3(t − to ))]x1 (t − to ). Hence time-variant. Stable as |y(t)| < ∞, when |x(t)| < B. Memoryless as the output at time t depends only on inputs at time t. Clearly causal. (c) y(t) = R2t x(τ )dτ −∞ Let us check for linearity. Z2t x1 (τ )dτ x1 (t) → y1 (t) = −∞ Z2t x2 (t) → y2 (t) = x2 (τ )dτ −∞ Z2t ax1 (t) + bx2 (t) = x3 (t) → y3 (t) = x3 (τ )dτ −∞ 2 Z2t = ¡ ¢ ax1 (τ ) + bx2 (τ ) dτ −∞ Z2t = a Z2t x1 (τ )dτ + b −∞ x2 (τ )dτ −∞ = ay1 (t) + by2 (t) Hence linear. Let us check for time-invariance. Z2t x1 (τ )dτ x1 (t) → y1 (t) = −∞ Z2t x1 (t − to ) = x2 (t) → y2 (t) = x2 (τ )dτ −∞ Z2t = x1 (τ − to )dτ −∞ 2t−t Z o = x1 (τ )dτ −∞ 6= y1 (t − to ) Note that y1 (t − to ) = 2(t−t R o) x1 (τ )dτ . Hence time-variant. −∞ Suppose x(t) = 1. Then y(1) = ∞. Hence unstable. Non-causal because, y(1) depends on the value of x(t) at t = 2, as y(1) = R2 x1 (τ )dτ . −∞ Clearly has memory by the above argument. ½ (d) y(t) = 0 t<0 . x(t) + x(t − 2) t ≥ 0 By using the same method as we used for the above parts, it is linear, causal and stable and 3 not memoryless. Now let us check for time-invariance. ½ 0 t<0 x1 (t) → y1 (t) = x1 (t) + x1 (t − 2) t ≥ 0 ½ 0 t<0 x1 (t − to ) = x2 (t) → y2 (t) = x2 (t) + x2 (t − 2) t ≥ 0 ½ 0 t<0 = x1 (t − to ) + x1 (t − to − 2) t ≥ 0 6= y1 (t − to ) This is because ½ y1 (t − to ) = 0 t < to x1 (t − to ) + x2 (t − to − 2) t ≥ to Hence time-variant. ½ (e) y(t) = 0 x(t) < 0 . x(t) + x(t − 2) x(t) ≥ 0 By using the same technique as was used for the previous problems, this is time-invariant, not memoryless, stable, causal. Let us check for linearity. Suppose let the input be x1 (t) = 1 for all t. Then the output y1 (t) corresponding to the input x1 (t) is y1 (t) = 2. ∀ t Let us now take the input x2 (t) = −x1 (t) = −1. If the system is linear, then we should get y2 (t) = −y1 (t) = −2, where y2 (t) is the output to the input x2 (t). Since x2 (t) < 0, the output y2 (t) = 0 6= −y1 (t). Hence not linear. (f) y(t) = x(t/3) This is linear and stable. It is not memoryless (for example, the output at time t = −3 depends on input at t = −1). It is non-causal as well. Let us see whether it is time-invariant. x1 (t) → y1 (t) = x1 (t/3) x1 (t − to ) = x2 (t) → y2 (t) = x2 (t/3) µ ¶ t = x1 − to 3 6= y1 (t − to ) This is because y1 (t − to ) = x1 4 ¡ t − to ¢ . 3 Hence time-variant. (g) y(t) = dx(t) dt This is linear, as well as time-invariant. This is not memoryless as y(t) depends on x(t − δt) in calculating dx(t) , since dx(t) = limδt→0 x(t)−x(t−δt) . dt dt δt Problem 1.28 (a) y[n] = x[−n] Let us check for linearity. x1 [n] → y1 [n] = x1 [−n] x2 [n] → y2 [n] = x2 [−n] ax1 [n] + bx2 [n] = x3 [n] → y3 [n] = x3 [−n] = ax1 [−n] + bx2 [−n] = ay1 [n] + by2 [n] Hence linear. Let us check for time-invariance. x1 [n] → y1 [n] = x1 [−n] x1 [n − no ] = x2 [n] → y2 [n] = x2 [−n] = x1 [−n − no ] 6= y1 [n − no ] Note that y1 [n − no ] = x1 [−n + no ]. Hence time-variant. This is stable as |y[n]| < ∞, if |x[n]| < B. Non-causal because, y[−1] depends on the value of x[1]. Clearly has memory by the above argument. (b) y[n] = x[n − 2] − 2x[n − 8] By using the same technique as used for the above problem, it is linear, time-invariant, causal, stable. This is not memoryless. 5 (c) y[n] = nx[n] x1 [n] → y1 [n] = nx1 [n] x2 [n] → y2 [n] = nx2 [n] ax1 [n] + bx2 [n] = x3 [n] → y3 [n] = nx3 [n] = n(ax1 [n] + bx2 [n]) = ay1 [n] + by2 [n] Hence linear. Let us check for time-invariance. x1 [n] → y1 [n] = nx1 [n] x1 [n − no ] = x2 [n] → y2 [n] = nx2 [n] = nx1 [n − no ] 6= y1 [n − no ] Note that y1 [n − no ] = (n − no )x1 [n − no ]. Hence time-variant. This is not stable because if x[n] = 1 for all n, then y[n] → ∞ as n → ∞. Memoryless because, y[n] depends only on x[n]. It is also causal. (d) y[n] = E{x[n − 1]}, where E is the even part. E{x[n − 1]} = x[n − 1] + x[1 − n] 2 This is linear, stable, not memoryless, non-causal. This is time-variant which can be seen by using exactly the same steps as we used for Problem 1.27 (a) with t replaced by n. (e) y[n] = x[n] n≥1 0 n=0 . x[n + 1] n ≤ −1 This is linear and stable, not memoryless and non-causal. Let us check for time-invariance. 6 x1 [n] → x1 [n − no ] = x2 [n] → = 6= This is because x1 [n] n≥1 0 n=0 y1 [n] = x1 [n + 1] n ≤ −1 n≥1 x2 [n] 0 n=0 y2 [n] = x2 [n + 1] n ≤ −1 n≥1 x1 [n − no ] 0 n=0 x1 [n + 1 − no ] n ≤ −1 y1 [n − no ] y1 [n − no ] = x1 [n − no ] n ≥ no + 1 0 n = no x1 [n + 1 − no ] n ≤ no − 1 Hence time-variant. x[n] n ≥ 1 0 n=0 . (f) y[n] = x[n] n ≤ −1 Following exactly the same steps, it is easy to see from Problem 1.28 (e) that it is linear and time-variant. It is causal, memoryless and stable. (g) y[n] = x[4n + 1] This system is linear as well as stable. Further it is non-causal and memoryless. Let us check for time-invariance. x1 [n] → y1 [n] = x1 [4n + 1] x1 [n − no ] = x2 [n] → y2 [n] = x2 [4n + 1] = x1 [4n + 1 − no ] 6= y1 [n − no ] This is because y1 [n − no ] = x[4(n − no ) + 1] = x1 [4n − 4no + 1]. Hence time-variant. Problem 1.31 7 (a) Note that x2 (t) = x1 (t)−x1 (t−2). Therefore using linearity we get y2 (t) = y1 (t)−y1 (t−2). See the figure below. (b) We see that x3 (t) = x1 (t + 1) + x1 (t). y1 (t + 1) + y1 (t). See the figure below. Therefore using linearity we get y3 (t) = y(t) y(t) 2 3 2 0 2 4 −1 t 1 2 t −2 Problem 1.36 (a) If x[n] is periodic ejωo (n+N )T =e (2π/To )N T = 2πk jωo nT , where ωo = 2π/To . This implies that ⇒ (T /To ) = k/N = rational number (b) If T /To = p/q, then x[n] = ej2πn(p/q) . Then the fundamental period is q/gcd(p, q) (from Problem 1.35), and therefore the fundamental frequency is 2π 2π p ωo T gcd(p, q) = gcd(p, q) = gcd(p, q) q p q p (c) From part (b) above, p/gcd(p, q) periods of x(t) are needed to form a single period of x[n]. 8