Laplace transforms These are a neat way of solving constant coefficient DE’s and also are used in process control. The Laplace transform is an operator L(⋅) that maps a function f (t ) , defined for t ≥ 0 , onto another function F (s ) ∞ L( f (t )) = F (s ) ≡ ∫ e − st f (t ) dt (1) 0 Equation (1) is well defined provided that both the following conditions hold: • The function f (t ) is piecewise continuous (or smoother) • The function f (t ) is of exponential order as t → ∞ Definition: The function f (t ) is of exponential order as t → ∞ if there exist constants: a, A, K , such that: f (t ) ≤ Ke at , t≥A (In simple terms, if f (t ) doesn’t grow any faster than exponential.) Laplace transform of a derivative? L( f ′(t )) = − f (0 ) + sL( f (t )) Frigaard 1 (2) MATH 256 What about if f (t ) has jumps in it? Suppose that f ′(t ) is piecewise continuous and of exponential order as t → ∞ and that f (t ) is of exponential order. What is L( f ′(t )) ? Let {t j }nj = 0 be the points of discontinuity of f ′(t ) , with t0 = 0 and tn = A . Formally, we have that: A L( f ′(t )) = lim ∫ e A→ ∞ − st 0 f ′(t ) dt = lim n tj − st e f ′(t ) dt ∑ ∫ A→∞ j =1t j −1 We integrate by parts on each interval: tj ⎞ n ⎛ n n tj tj st st − − st st − − j j − 1 L( f ′) = lim ∑ ⎜ e f t j −1 + s ∫ e f dt ⎟ = lim ∑ e f (t j ) − e f (t j −1 ) + s lim ∑ ∫ e − st f (t ) dt ⎟ A→ ∞ j =1 A→ ∞ j =1⎜ A → ∞ j =1 t j −1 t j −1 ⎝ ⎠ [ = − f (0 ) + lim e A→∞ [ ] − sA ] A f ( A) + s lim ∫ e − st f (t ) dt A→ ∞ 0 Therefore the expression (2) is valid even with jumps in f (t ) Frigaard 2 MATH 256 Higher derivatives? These can be handled iteratively: L( f ′′(t )) = − f ′(0) + sL( f ′(t )) = − f ′(0) − sf (0 ) + s 2 L( f (t )) L( f ′′′(t )) = − f ′′(0 ) + sL( f ′′(t )) = − f ′′(0) − sf ′(0) + s 2 L( f ′(t )) = − f ′′(0 ) − sf ′(0 ) − s 2 f (0) + s 3 L( f (t )) etc… Theorem 1: Suppose that f , f ′, f ′′.... f ( n −1) are continuous and that f ( n ) is piecewise continuous, on any interval [0, A]. Suppose also that each of f , f ′, f ′′.... f ( n −1) , f (n ) is of exponential order as t → ∞ . Then ( ) = s L( f (t )) − f L f (n ) n (n −1) (0) − sf (n − 2 ) (0).... − s n −1 n −1 f (0 ) = s L( f (t )) − ∑ s j f (n −1− j ) (0 ) n j =0 Linearity of the Laplace transform Note that the Laplace transform is a linear operator, i.e. suppose that f (t ), g (t ) have Laplace transforms F (s ), G (s ) . Then L(af (t ) + bg (t )) = aF (s ) + bG (s ) for any constants a and b . Why is this true? Frigaard 3 MATH 256 Solving linear IVPs with constant coefficients Example 1: Use Laplace transforms to solve the following IVP: y′ + ay = 0 y (0 ) = y0 Solution: Take the transform of the equation Therefore, the Laplace transform of the solutionY (s ) satisfies: Y (s ) = Recall that if f (t ) = e − at then F (s ) = y0 s+a 1 . We immediately recognise that the inverse is: s+a y (t ) = y0e − at which we know is the solution of the IVP, (from our first lecture). Frigaard 4 MATH 256 Points to note: 1. We reduce the differential equation to an algebraic equation. 2. After solving the algebraic equation, we are left with the problem of how to invert the Laplace transform to recover y (t ) . Example 2: Use Laplace transforms to solve the following IVP: y′′ − 9 y = 0 Frigaard y(0) = 1, y′(0) = −1 5 MATH 256 Inverting Laplace transforms • There is a 1-1 correspondence between f (t ) and F (s ) , i.e. we can meaningfully work in s-space and then invert to t-space. • General formula exists but is not covered in this course. • Tables exist for a wide range of common functions (see p. 319 in Boyce & DiPrima, 8th ed). • Use tables in combination with general results, to identify how to invert a particular formula Frigaard 6 MATH 256 Example 3: Use Laplace transforms to solve the following IVP: y′′ − 2 y′ − 2 y = 0 Frigaard y(0) = 1, y′(0) = 1 7 MATH 256 Frigaard 8 MATH 256 Properties of Laplace transforms • Frequently the problem is to invert Y (s ) , the solution to our IVP in s-space. • We have a limited number of "known" transforms of common functions and their inverses. • To expand our range of known functions & inverses, we apply a number of general results Assume that f (t ) , g (t ) , h(t ) have Laplace transforms F (s ) , G (s ) , H (s ) and a , b , c , are positive constants. • Linearity: L(af (t ) + bg (t )) = aF (s ) + bG (s ) Utility: solve IVP in s-space & the solution Y (s ) is of form: Y (s ) = aF (s ) + bG (s ) + cH (s ) + .... Then the solution in t-space, will be: y (t ) = af (t ) + bg (t ) + ch(t ) + .... • Stretching: L( f (at )) = 1 ⎛s⎞ F⎜ ⎟ a ⎝a⎠ Proof: Frigaard 9 MATH 256 • Differentiation I: (see theorem 1) L( f ′(t )) = − f (0 ) + sL( f (t )) , or applied iteratively: ( L f (n ) (t )) = s L( f (t )) − ∑ s j f (n −1− j ) (0) , n n −1 j =0 Utility: turning IVP's into algebraic equations • Differentiation II: (not the inverse of the above) ( L (− t ) n ) dn f (t ) = n F (s ) = F (n ) (s ). ds Proof: Utility: possibly we can recognise that the solution in s-space, Y (s ) , is a derivative with respect to s of a function that we know the transform of. Example: Suppose that Y (s ) = 6 / s 4 , find y (t ) Frigaard 10 MATH 256 • Shifting I: ( ) L e at f (t ) = F (s − a ) Proof: Utility: we can recognise that the solution in s-space, Y (s ) , is of form F (s − a ) , where we know the inverse transform of F (s ) . Example: Suppose that Y (s ) = Frigaard 5 , find y (t ) 2 (s − 2) + 25 11 MATH 256 • Shifting II: We introduce the unit step function uc (t ): ⎧0 0 ≤ t ≤ c uc (t ) = ⎨ ⎩1 c < t and note that the Laplace transform of uc (t ) is simply ∞ L(u c (t )) = ∫ u c (t )e 0 − st ∞ dt = ∫ e − st c e −cs dt = , s>0 s Now for a given function f (t ) , suppose we "shift" the function to the right by an amount c, defining f (t ) = 0 for t < 0 , i.e. 0≤t ≤c ⎧0 g (t ) = ⎨ ⎩ f (t − c ) c < t The function g (t ) is a delayed form of f (t ) . We can also write g (t ) = uc (t ) f (t − c ) . The second form of shifting theorem is: L(uc (t ) f (t − c )) = e − cs F (s ) Proof: Frigaard 12 MATH 256 Utility: we can recognise that the solution in s-space, Y (s ) , is of form F (s )e − cs , where we know the inverse transform of F (s ) . e− s + e − 2 s − e −3s − e − 4 s Example: If Y (s ) = , find y (t ) . s Frigaard 13 MATH 256 • Convolution theorem: If H (s ) = F (s )G (s ) then t t 0 0 h(t ) = ∫ f (τ )g (t − τ ) dτ = ∫ g (τ ) f (t − τ ) dτ Proof: (see text for reverse version). Notation: we denote the convolution of two functions by the symbol f ∗ g , i.e. t h(t ) = ( f ∗ g )(t ) = ∫ f (τ )g (t − τ ) dτ 0 Utility: Often we identify the solution in s-space, Y (s ) , as a product of form F (s )G (s ) , where we know the inverse transforms of F (s ) and G (s ) . This arises with inhomogeneous equations. Frigaard 14 MATH 256 Example: Consider the inhomogeneous 2nd order constant coefficient IVP: ay′′ + by′ + cy = g (t ) y (0 ) = y0 , y′(0 ) = y0′ Take the Laplace transform of this equation and use theorem 1: (as 2 ) + bs + c Y (s ) − (as + b ) y0 − ay0′ = G (s ) Therefore, Y (s ) = Φ(s ) + Ψ (s ) , where: (as + b ) y0 + ay0′ G (s ) as 2 + bs + c as 2 + bs + c We break Φ (s ) into partial fractions & invert: Φ(s ) → φ (t ) . Define the transfer function F (s ) by: Φ (s ) = , Ψ (s ) = 1 as 2 + bs + c We can find the inverse of F (s ) , say f (t ) , by breaking into partial fractions. The function f (t ) is called the impulse response of the system. Using the convolution theorem, Ψ (s ) = F (s )G (s ) inverts to: ψ (t ) = f ∗ g F (s ) = and the solution of the inhomogeneous IVP will be: y (t ) = φ (t ) + ψ (t ) Frigaard 15 MATH 256 Notes: 1. The function φ (t ) satisfies the homogeneous IVP: ay ′′ + by ′ + cy = 0 y (0 ) = y0 , y ′(0 ) = y0′ 2. The function ψ (t ) satisfies the inhomogeneous IVP: ay′′ + by′ + cy = g (t ) y (0 ) = 0, y′(0 ) = 0 Frigaard 16 MATH 256 Differential Equations with Piecewise Continuous Data Functions Consider the IVP: y′′ + p(t ) y′ + q(t ) y = g (t ) y (0 ) = y0 y′(0 ) = y0′ (3) Suppose that the data functions p(t ), q(t ) , g (t ) are piecewise continuous on the interval I = (0, β ), with points of discontinuity t = t j : j = 0,...., n , where t0 = 0 and t n = β . Questions: 1. Can we find a solution to (3) defined on I = (0, β )? 2. How smooth can a solution be? 3. What happens if p(t ), q(t ) , g (t ) are continuous on I , but are only piecewise differentiable? Frigaard 17 MATH 256 Example 1: Consider the IVP: 1 (u2 (t )(t − 2) − u4 (t )(t − 4)) − u8 (t ) 2 y (0 ) = 0 y′(0 ) = 0 y′′ + y = • Describe the qualitative behaviour of the solution for t ∈ (0,10) • Solve the IVP using Laplace transforms Frigaard 18 MATH 256 Example 2: Consider the IVP: 17 y′′ + 3 y′ + y = uπ (t ) − u2π (t ) 2 y (0 ) = 0 y′(0 ) = 0 • Describe the qualitative behaviour of the solution for t > 0 • Solve the IVP using Laplace transforms Frigaard 19 MATH 256 Example 3: Consider the IVP: y (iv ) − 16 y = sin t − u2π (t )sin (t − 2π ) y (0 ) = 0 y′(0 ) = 0 y′′(0 ) = 0 y′′′(0 ) = 0 Solve the IVP using Laplace transforms Frigaard 20 MATH 256 Differential Equations with Impulses Consider the function dτ (t ) : ⎧1 ⎪ dτ (t ) = ⎨ 2τ ⎪⎩ 0 Define the function I (τ ) by: I (τ ) = ∞ τ −∞ −τ t ≤τ (4) t >τ ∫ dτ (t ) dt = ∫ dτ (t ) dt = τ 1 ∫ 2τ dt = 1 −τ Now consider what happens to both I (τ ) and dτ (t ) as τ → 0 : • Straightforwardly, we get: lim I (τ ) = 1 τ →0 • For any fixed t ≠ 0 we have: lim dτ (t ) = 0 , but integral of dτ (t ) remains uniformly equal to 1. τ →0 This prompts the definition of the Dirac delta function δ (t ): δ (t ) = lim dτ (t ) (5) τ →0 which is not really well defined in a conventional sense. The delta function acts like an intense pulse, of unit strength. Frigaard 21 MATH 256 For differentiable functions f (t ) , the delta function δ (t ) has the following property: ∞ ∫ δ (t − t0 ) f (t ) dt = f (t0 ) −∞ Sketch proof: ∞ ∞ ∫ δ (t − t0 ) f (t ) dt = ∫ −∞ −∞ [lim d (t − t )]f (t ) dt τ →0 τ 0 ∞ = lim ∫ dτ (t − t0 ) f (t ) dt τ →0 −∞ ⎡ 1 t 0 +τ ⎤ = lim ⎢ ∫ f (t ) dt ⎥ τ → 0 2τ ⎢⎣ t 0 −τ ⎥⎦ = lim[ f (t0 ) + O(τ )] = f (t0 ) τ →0 Directly from the above, we can derive the Laplace transform of the delta function: ∞ L(δ (t − t 0 )) = ∫ δ (t − t 0 )e − st ∞ dt = ∫ δ (t − t 0 )e − st dt = e − st0 0 −∞ Now taking t0 → 0 L(δ (t − t0 )) = lim e − st 0 = 1 t0 →0 Frigaard 22 MATH 256 Example 1: Consider the IVP: 2 y′′ + y′ + 2 y = δ (t − 5) y (0) = 0 y′(0) = 0 Describe the qualitative behaviour of the solution for t > 0 . Solve the IVP using Laplace transforms Frigaard 23 MATH 256 Example 2: Consider the IVP: y′′ + 4 y = δ (t − π ) − δ (t − 2π ) y (0 ) = 0 y′(0 ) = 0 Describe the qualitative behaviour of the solution for t > 0 . Solve the IVP using Laplace transforms Frigaard 24 MATH 256