Topic 10: Differential Equations Flashcards
(27 cards)
What a first order separable differentiable equations, and how do you solve them (1,2)
-These are of the form dy/dx = f(x) g(y)
-You first rearrange these, to get 1/g(y) dy/dx = f(X)
-You then integrate as ∫1/g(y) (dy/dx) dx = ∫1/g(y) dy = ∫f(x) dx + c
How can we solve dy/dx = xy (5)
dx/dy = xy
-(1/y) dy/dx = x
-∫(1/y)dy = ∫x dx
-lny = (1/2)x2 + c
-y = e(1/2)x2 + c
-Let A = ec, so that y = Ae(1/2)x2
What are first order linear equations (1)
-dy/dx + a(x)y = b(x)
How can we rearrange x2dy/dx + 2xy = b(x) to help us solve the first order differential equation (5,1)
-x2dy/dx + 2xy = (d/dx)(x2y)
-∫ (d/dx)(x2y) dx= ∫ b(x) dx
=x2y = ∫ b(x) dx
-y = (1/x2) ∫ b(x) dx
-y = (1/x2) ∫ b(x) dx + c/x2
-Solve the integral to get the general solution, then substitute in the initial/boundary conditions to find c for the required specific solution
How can we let u(x) = e∫ a(x) dx to help find the integrating factor (6,1)
-Let u(x) = e∫ a(x) dx
-du/dx = a(x)u(x)
-Given dy/dx + a(x)y = b(x), we get:
-u(x)dy/dx + a(x)u(x)y = u(x)b(x)
-This is equal to (d/dx)(u(x)y) = u(x)b(x)
-Then we can integrate dx
-We can ignore the constant of integration when calculating u(x) = e∫ a(x) dx
How can we solve the first-order differential equation dy/dx - y = x (6)
-Find u(x) = e∫-1 dx = e-x
-So then e-x(dy/dx) - ye-x = xe-x
-(d/dx)(e-xy) = xe-x (use integration by parts for the RHS)
-e-xy = -xe-x - e-x + c = -e-x(1+x) + c
-y = -(1+x) + cex
-This is the general solution, to find the specific solution cconsider various solution curves
What is the integrating factor (2)
-The integrating factor is used with differential functions in the form dy/dt + a(t)y = b(t)
-The end goal is getting the LHS in form d/dt(uv) = u(dv/dt) + v(du/dt)
How do you calculate the integrating factor (1,1,1,6,3)
-Start with the equation in form dy/dt + a(t)y = b(t)
-Evaluate ∫a(t) dt (no arbitrary constant needed)
-Determine u(t) = e∫a(t) dt (this is the integrating factor)
-Multiply the differential equation through by u(t)
-dy/dt (u(t)) + (u(t))(a(t))y = (u(t))b(t)
-Since u(t) = e∫a(t) dt, du/dt = a(t) e∫a(t) dt = a(t) u(t)
-d/dt(u(t)y) = (dy/dt)(u(t)) + (du/dt)(y)
-d/dt(u(t)y) = (dy.dt)(u(t) + a(t)u(t)y
-d/dt(u(t)y) = u(t)b(t)
-We then integrate this
-u(t)y = ∫u(t)b(t) dx + c
-y = (1/u(t))(∫u(t)b(t) dx + c)
How can we solve the linear equation dx/dt - 2tx = t (4)
-The integrating factor is u(t) = e∫-2t dt = e-t2
-The differential equation then becomes (d/dt)(e-t2x) = te-t2)
-e-t2x = ∫te-t2 dt + c = (-1/2)e-t2 + c (integrate with respect to dt)
-x = Cet2 - 1/2 (this is the general solution)
What are second order differential equations (1)
-(d2y/dx2) = F(x, y, dy/dx) or x̄ = F(t, x, x’)
How do we directly integrate second order differential equations, and what information do we need about x (3,3)
-take d2x/dt2 = k
-dx/dt = kt + c
-x = (1/2)kt2 + Ct + D
To determine C & D, we need 2 pieces of information about x:
-2 points through which the solution curve passes (boundary value problem)
-x and (dx/dt) at the same point (initial value problem)
-x at one point, dx/dt at another (mixed problem)
How can we solve d2x/dt2 = dx/dt + t, given x(0) = 1 and ẋ(0) = 2 (7,1)
-Replace dx/dt with new variable y, to make the equation a first order one
-ẏ = y + t, y(0) = 2
-y = -(1+t) + Cet
-When t = 0, y = 2 so c = 3
-ẋ = -(1+t) + 3et
-x = -t -(1/2)t2 + 3et + D
-x(0) = 1, D = -2
-x = -(2+t+(1/2)t2) + 3et
What is the form of linear second order equations, and when are they homogeneous vs inhomogeneous (3,2)
-a(t) d2y/dt2 + b(t)dy/dt + c(t)y = f(t), where a(t), b(t), c(t) f(t) are continuous functions of t on some interval a(t) ≠ 0
-Ly = f(t)
-L = the linear differential operator, a(t) d2/dt2 + b(t)d/dt + c(t)
-If f(t) = 0, the equation is said to be homogeneous
-if f(t) ≠ 0, the equation is inhomogeneous
How can we combine solutions of linear homogeneous second-order differential equations (1,2)
-Let V be the set of all solutions, and let u1, u2 ∈ ℝ such that Lu1, Lu2 = 0
-u1 + u2 is a solution
-ku is a solution
What do we call V when solving linear homogeneous second-order differential equations (2,2)
-V is the vector space consisting of all the functions which are mapped to the 0 function by the linear map L
-We can combine the 2 properties and say Au1 + Bu2 is a solution for all choices of real arbitrary constants A and B
However, the general solution to the equations is in the form Au1 + Bu2 only when:
-u1 and u2 span the solution space of Ly
-u1 and u2 are linearly independent
How do we solve linear non-homogeneous second-order differential equations (3)
The non-homogeneous differential equation has the general solution:
-y = Au1(t) + Bu2(t) + v(t)
-Au1(t) + Bu2(t) is called the complementary solution, and is the general solution of the related homogeneous equation
-v(t) is any particular solution so that Lv = f(t), provided v ∉ V
How can we write the linear differential operator for linear non-homogeneous second-order differential equations (3)
-L(Au1 + Bu2 + v) = AL(u1) + BL(u2) + Lv
=0 + 0 + Lv
-f(t) = Lv
How can we find the auxillary or characteristic equation of a second-order linear differential equation with constant coefficients (1,6)
-a(d2y/dt2) + b(dy/dt) + cy = 0
-If we try solution y = emt, dy/dt = memt, d2y/dt2 = m2emt
-a(m2emt) + b(memt) + c(emt) = 0
-emt(am2 + bm + c) = 0
-Since emt ≠ 0, am2 + bm + c = 0
-This is an auxillary or characteristic equation
-This has solutions m1, m2 = (-b±√(b2 - 4ac))/2a
How can we find the complementary solution to a second-order linear equation with constant coefficients (3,4,3,5)
-a(d2y/dt2) + b(dy/dt) + cy = 0
-emt(am2 + bm + c) = 0 (let y = emt)
-This has solutions m1, m2 = (-b±√(b2 - 4ac))/2a
-Say m1 and m2 are real and distinct (b2 > 4ac)
-u1 = em1t, u2 = em2t
-These are linearly independent functions, so the complementary solution is a linear combination
-Aem1t + Bem2t
-Say m1 = m2 =m (b2 = 4ac)
-u1 = emt, u2 = temt
-Complementary solution: (A + Bt)(emt)
-Say m1 and m2 = α + iβ (b2 < 4ac)
-u1 = e(α + iβ)t = eαt(cosβt + i(sinβt))
-u2 = e(α - iβ)t = eαt(cosβt - i(sinβt))
-Since these are not real, we choose 2 real linearly independent solutions from the complex subspace spanned by u1 and u2 ((u1 and u2)/2)
-General solution: eαt(Acosβt + Bsinβt)
How can we solve (d2x/dt2) + 4(dx/dt) + 4x = 0 (3)
-The characteristic equation is m2 + 4m + 4 = 0
-Solving this gives us m = -2
-The general solution is thus x = (A + Bt)e-2t
How can we solve (d2y/dt2) -6 (dy/dt) + 13y = 0
-The characteristic equation is m2 - 6m + 13
-Using the quadratic equation, we get 3 ± √-1 = 3 ± 2i
-The general solution is thus y = e3t(A cos(2t) + B sin(2t))
What is the particular solution, and how do we find it (1,1)
-The particular solution is the bit to add to the complementary to get the general solution to the original non homogeneous equation
-We first use a trial and error solution, then change the coefficients
What should our trial solutions for particular solutions be depending on the original nonhomogeneous equations (1,7,1)
-Remember a(t) d2y/dt2 + b(t)dy/dt + c(t)y = f(t)
-If f(t) is a constant, trial C
-If f(t) = t, trial Ct + D
-If f(t) = t2, trial ct2 _ Dt + e
-If f(t) = tn, trial a polynomial of the same degree
-If f(t) = ekt, trial Cekt
-If f(t) = sin(at), trial C(cos(at)) + D(sin(at))
-If f(t) = cos(at), trial C(cos(at)) + D(sin(at))
-However, since the trial solution has to be linearly independent from the complementary, for f(t) = ekt be prepared to use Ctekt or Ct2ekt
How can we solve (d2x/dt2) +5 (dx/dt) + 6x = 4 (1,1,3,1)
-The auxillary solution is m2 + 5m + 6 = 0, meaning m = -2, -3
-The complementary solution is thus Ae-2t + Be-3t
The particular solution is
-f(t) = 4
-Let x = c, so dx/dt = d2x/dt2 = 0
-This gives us 6C = 4, so C = 2/3
-The general solution is thus 2/3 + Ae-2t + Be-3t