1 Mathematics for Physical Chemistry Coordinate Systems Coordinate systems are essential to understand how objects behave in space. A good choice of a coordinate system takes advantage of the symmetry of a system. A spherical system is better described with a spherical coordinate system and a rectangular system is better described with a rectangular system. The easiest coordinate system that we will consider is the number line, a onedimensional coordinate system. We will be using the number line a great deal in our early introduction to quantum mechanics. When we consider a particle in more than one dimension, we have choices in the coordinate system we use. As stated above, the coordinate system should match the symmetry of the object studied. Two-dimensional coordinate systems Cartesian coordinates Objects that are “boxy” are best described in Cartesian coordinates. y y (x,y) dx dy x x In two-dimensions, integrals yield an area. Calculating integrals requires the summation over an appropriate range of the area of boxes that are defined in relation to the symmetry of the coordinate system. These boxes are known as differential area elements. In the two-dimensional Cartesian coordinate system, the differential area element is dA = dx dy. The box has differential length of dx and differential height of dy. 2 Polar coordinates Objects that are approximately circular are best described with polar coordinates. The differential area element is r dr d, that is, a box with length dr and height is r d. y y (r,) r d r dr d x x Transformation Equations y x x r cos tan 1 y r sin r x 2 y2 Sometimes one want to describe an object in one coordinate system with another coordinate system. The equations that are used to make the conversion are referred to as transformation equations. Below are the transformation equations between the Cartesian coordinate system and the polar coordinate system. 3 Three-dimensional coordinate systems Cartesian coordinates z (x,y,z) x y A three-dimensional integral yields a volume. Thus the calculation involves adding 3-d boxes known as differential volume elements. In the three-dimensional Cartesian coordinate system, the differential volume element is dV = dx dy dz. The box has differential length of dx, differential height of dy and differential depth, dz. Differential volume element dV = dx dy dz z dy (x,y,z) dx dz x y 4 Spherical polar coordinates While Cartesian coordinates are relatively simple to use for some applications, when an object has spherical symmetry, like an atom, stubbornly insisting on using Cartesian coordinates to do calculations becomes very messy. Using spherical polar coordinates for an atom is much better choice for a coordinate system because calculations become much simpler. In the spherical polar coordinate system, r, the radial coordinate is the distance from the origin, the polar coordinate, , is the angle from a “preferred” direction (label as the “z” coordinate in the diagram) and the line connecting the origin with the point. The choice of the “preferred” is usually obvious (as in the direction of the internuclear axis connecting atoms in a diatomic molecule) or arbitrary (as in a hydrogen atom). The azimuthal coordinate, , is the angle between the projection of line connecting the origin and the point on the “x-y” plane (the plane perpendicular to the “preferred” direction) and a line defining = 0 (usually considered to the “x” axis). z coordinate is like latitude Range of is from 0 to (R,,) coordinate is like longitude Range of is from 0 to 2 R R sin y x The differential volume element is dV r 2 sin dr d d . r d z dr x d d y 5 The transformation equations between the spherical polar coordinate system and the Cartesian coordinate system are x r sin cos y r sin sin z r cos Cylindrical coordinates When considering diatomic molecules (among other possible systems like molecular spectroscopy in applied electric or magnetic fields), using the spherical polar coordinates is awkward. It becomes more convenient to think of the molecule with basic cylindrical symmetry rather than a spherical symmetry. Thus cylindrical coordinates can be used. is the radial coordinate, is the azimuthal coordinate and z is simply known as the height. z (,,z) y x Transformation Equations x cos y sin Differential volume element dV d d dz zz 6 Complex Numbers Introduction and Motivation Recall the following: Solution to x 2 1 0 is x 1 What is solution to x 2 1 0 x 2 1 0 x 2 1 x 1 What is 1 ? Define i 1 . Note: i i 1 1 1 i 2 1 What is solution to x 2 2x 5 0 ? Find solution from quadratic equation. x b b2 4ac 2 4 20 1 1 16 1 2i 2a 2 2 Complex number may have two parts z a bi a – real part b – imaginary part 7 Representations of complex numbers Argand plane While real numbers can be plotted on a number line, complex numbers must be plotted on a plane (called the Argand plane). Rectangular coordinates (a+ bi) b a Polar coordinates A complex number can also be represented in polar coordinates. (a+ bi) r sin r r cos a bi r cos i r sin r cos isin a r cos b r sin r a 2 b2 - modulus b tan 1 - phase angle a Without proof, the following identity will be asserted Euler’s identity: e i cos i sin Therefore a complex number in rectangular coordinate can always be expressed in polar coordinates and vice versa. a bi r ei 8 Operations with complex numbers We will need to add complex numbers together as well as multiplying them. Addition of complex numbers Adding complex numbers in the rectangular representation is simple. The sum is simply the sum of the real part and the sum of the imaginary part. (a + bi) + (c + di) = [(a + c) + (b + d)i] Addition in the polar representation is enormously difficult. In practice, numbers in the polar representation are converted to the rectangular representation, added, and then reconverted to polar representation. Multiplication of complex numbers Multiplying numbers in the rectangular representation is performed with the same technique as multiplying binomials; with the FOIL method (keeping in mind that i ∙ i = –1). (a + bi) × (c + di) = (a × c) + (a × di) + (bi × c) + (bi × di) = ac + adi +bci + bdi2 = ac – bd + (ad + bc)i Multiplying numbers in the polar representation is easier. The moduli are multiplied together and the phase angles are added. u = (r1, 1) v = (r2, 2) u ∙ v = (r1 ∙ r2, 1 + 2) 9 Double and Triple Integrals Double Integrals Recall integration of a single variable. x y x 2 x1 n x2 n f (x) dx lim f (x i ) x n x1 i 0 xi x1 i(x) Integration is a summation of rectangle x x2 x1 x areas to yield an area under a curve. What if the bottom border of an area is not the x-axis? f(x) y Area between the curves A x2 f (x) g(x) dx x1 Rather than use g(x) rectangles with uniform x, we could vary x. x1 y x x2 x f(x) Note two limiting processes are needed. One for the xcoordinate and one for the y-coordinate. y g(x) x1 x x2 x A x 2 f (x) x1 g(x) same as above dy dx x2 y x1 f (x) g(x) dx x2 f ( x1 10 Examples: 1 4 1 3x 2 3x dx dy 0 2 0 2 1 3 4 2 3 2 2 dy dy 18 dy 18y |10 18 2 2 0 0 1 4 2 Note that for multiple integrals, there exist a defined order to the integrations. Integrations are done “from the inside, out”. In the above example, dx is “inside” dy; therefore, the “x” integration is done first, then the “y” integration. Note that the limits of 2 to 4 are “inside” the limits of 0 to 1; therefore, the 2 to 4 limits are used in the “x” integration and the 0 to 1 limits are used in the “y” integration. 2 4 2 dx dy x 0 2y dy 4 2y dy 4y y 2 0 2 2 y 2 1 2 4 2y x dx dy 1 y 3 8 4 (0 0) 4 0 x2 2 y2 y 2 y4 y y5 y 2 dy dy 2 2 10 4 1 2 1 32 4 1 1 31 3 62 15 47 10 4 10 4 10 4 20 20 20 r sin dr d sin d 3 0 0 2 0 0 3 3 r dr cos 0 0 r4 4 3 0 81 81 81 81 cos cos 0 0 1 1 2 4 4 2 4 Triple Integrals Evaluation of triple integrals yields a volume instead of an area. Parallelpipeds (boxes) are summed rather than rectangles. Examples: 2 2x y x 1 x 0 2 2 r 0 0 0 2 2x 2 y2 dy dx y x dy dx xy 2 1 x 1 x 1 2 2 2 x2 2x x2 x 2x x x dx dx 2 2 2 1 1 x3 2 8 1 7 1 6 6 6 6 2 2x dz dy dx z yx 0 2 5 sin dr d d r dr 5 0 2 0 0 2x x dx sin d d r 2 32 128 2 2 2 0 cos 0 0 6 3 3 Note that in the integral above, the integrand can be separated into three independent functions and that the bounds of integration have no functional dependencies. Therefore, the triple integral is the same as the product of three single integrals. We’ll find this circumstance a great deal in quantum chemistry. 6 11 Vector Differential Operators Introduction In quantum chemistry, many of the quantities we will consider are vector quantities. (Review your knowledge of vectors, including dot products and cross products from physics.) That is, the quantities will have a magnitude and a direction. The electric force between charges is a vector quantity as is the electric field emanating from a collection of charges. The dipole moment of a molecule is a vector quantity. The flow of a liquid is a vector quantity. Other quantities are simpler in that they only have a magnitude. Such quantities are called scalar quantities. Examples of scalar quantities are charge, electric potential, temperature. We often want to know how scalar quantities and vector quantities change. Recall from calculus that a change in a quantity is a derivative. Thus to find the changes of quantities all we need to is take a derivative. However, because vector quantities have direction and scalar quantities can change independently in each of the three spatial dimensions; we have to be careful about what we mean by a derivative. We will examine three kinds of vector derivatives, the gradient, the divergence and the curl. All the vector derivatives involve a vector quantity called the del operator. The symbol for the del operator is an upside-down delta, . Del operator in Cartesian coordinates ˆ ˆ ˆ i j k x y z Del operator in spherical polar coordinates 1 1 eˆ r eˆ eˆ r r r sin 12 Gradient The gradient operator operates on a scalar quantity to find how the scalar quantity changes in three orthogonal directions. The del operator is applied to the scalar quantity to find its gradient. Thus the change of a scalar quantity is a vector quantity. Since the gradient is a vector quantity, it has a direction. **The direction where the gradient points is the direction of maximum change of the scalar quantity.** Example: Within a 2 meter cubic box, the temperature has been measured and found to be a function of x, y and z. T x, y, z x 2 y 2 z 273 K . The thermal energy will flow from the box in the opposite direction of the maximal change of the temperature. Thus q T a) Calculate the gradient of the temperature function. T ˆi ˆj kˆ x 2 y 2 z 273 y z x x 2 y 2 z 273 x 2x ˆi 2yz ˆj y 2 kˆ ˆi x 2 y 2 z 273 y ˆj x 2 y 2 z 273 z kˆ b) Calculate the value of the gradient at the point x = 1 m, y = 1 m and z = 2 m. 2 T 2x ˆi 2yz ˆj y 2 kˆ T 2 1 iˆ 2 1 2 ˆj 1 kˆ 2 iˆ 4 ˆj kˆ K m c) Find the angle between the direction of thermal energy flow at the point x = 1 m, y = 1 m and z = 2 m and the x = 1 plane of the box. The x = 1 plane is perpendicular to the unit vector in the x direction. Recall the definition of the dot product, A B A B cos . Let A ˆi and B 2 ˆi 4 ˆj kˆ A B A B cos ˆi 2 ˆi 4 ˆj kˆ ˆi 2 ˆi 4 ˆj kˆ cos 2 0 0 1 2 4 1 cos 2 2 1 2 2 cos 2 21 cos 2 2 cos 1 2.02 rad 115.88 21 21 13 Divergence F The divergence operator is one of two operators used to find the derivative of a vector-valued function, that is, a function that has a value and direction everywhere in space. The flux of a vector-valued function to or away from enclosed volume is found by taking the divergence. The divergence is found by taking the dot product of the del and a vector-valued function to yield a scalar quantity. (In coordinates other than Cartesian coordinates, the calculation is more sophisticated.) Consider the flow of gas through a section of pipe. The divergence of the flow tells about how the flow is changing. Consider three different possibilities. 1) If the divergence of the flow is zero, the gas flowing into the section equals the gas flowing out of the section. 2) If the divergence is positive then the flow of gas into the pipe is less than the flow of gas out of the pipe. Thus overall, the pressure of the gas is increasing. The section of pipe contains a source of gas. 3) If the divergence is negative then the flow of gas into the pipe is more than the flow of gas out of the pipe. Thus overall, the pressure of the gas is decreasing. The section of pipe contains a sink of gas. Example: The flow of a polymer solution through a flexible Tygon tube is profiled as the following vector-valued function, F x, y, z 214 ˆi 2y 0.098y3 ˆj 2z 0.098z3 kˆ mol in . The coordinate system origin is at the start of the tube and the middle. Is the fluid expanding or compressing? When will the divergence be zero? The flux of a vector-valued function to or away from a point is found by taking the divergence. F ˆi ˆj kˆ 214 ˆi 2y 0.098y3 ˆj 2z 0.098z 3 kˆ y z x 214 2y 0.098y x y 3 2z 0.098z 3 z 0 2 0.294y 2 2 0.294z 2 4 0.294y 2 0.294z 2 14 We don’t know if the fluid is expanding or compressing because we don’t the radius of the tubing, that is, we don’t know y and z. Let us find the radius of the tubing when the divergence equals zero. F 4 0.294y2 0.294z 2 0 4 0.294r 2 r 4 3.69 0.294 Thus when the tube is smaller than 3.69 in, the divergence is positive and the tube will expand to 3.69 in. If the tube is larger than 3.69 in, the divergence is negative and the tube will contract to 3.69 in. Curl F The curl operator is used to find the derivative of a vector-valued function perpendicular to its flow. **A nonzero curl means that the flow is rotating.** The curl of a vector-valued function is found by taking the cross product of the del operator and the vector-valued function. Example: Is the flow in the above problem rotating? F ˆi ˆj kˆ 214 ˆi 2y 0.098y3 ˆj 2z 0.098z 3 kˆ y z x 3 3 214 ˆ ˆ 2y 0.098y ˆ ˆ 2z 0.098z ˆ ˆ ii i j ik x x x 3 3 214 ˆ ˆ 2y 0.098y ˆ ˆ 2z 0.098z ˆ ˆ j i j j j k y y y 214 ˆ ˆ 2y 0.098y ki z z 3 kˆ ˆj 2z 0.098z3 z F 0 0 0 kˆ 0 ˆj 0 ˆj 0 ˆi 2 0.294z 0 0 kˆ 2 0.294y 2 0 0 ˆi 2 0 Thus, the flow is not rotating. kˆ kˆ 15 Laplacian 2 The divergence of a gradient of a scalar function is the Laplacian of the scalar function. **The Laplacian is a three-dimensional second derivative of a multi-variable scalar function.** The Laplacian is a scalar function itself. In quantum mechanics, the Laplacian is used to express the kinetic energy of a microscopic particle. Laplacian in Cartesian coordinates 2 2 2 2 x 2 y2 z 2 Laplacian in spherical polar coordinates 2 1 2 1 1 2 r sin r 2 sin r r r 2 sin r 2 sin 2 2 Table of vector derivative operators. Cartesian coordinates. Gradient ˆ ˆ ˆ i j k x y z Divergence V Vx Vy Vz x y z Curl V V V V V V V y z ˆi z x ˆj x y kˆ y x z y x z Laplacian 2 2 2 2 x 2 y2 z 2 16 Spherical coordinates Gradient Divergence V 1 1 r r r sin 1 2 1 1 r Vr V sin V 2 r r r sin r sin Curl V 1 r sin V rV eˆ r r Vr r sin V eˆ r sin r V Vr eˆ r sin r r 2 Laplacian 2 1 2 r r 2 r r Differential Equations 1 1 2 sin 2 r 2 sin 2 2 r sin 17 Introduction What are differential equations? Consider an analogy with algebraic equations. Algebraic equation: 2x 6 0 x 3 x 3 is the solution to the equation Differential equation: dy y0 dx What y = f(x) solves the equation? Try y = ex dy y e x dx ex ex 0 ex is not a solution to the differential equation. Try y = e-x y e x e x e x 0 e-x is a solution to the differential equation. Types of differential equations Partial – diff. eq. with partial derivatives (solutions are multi-variable functions) 2 2 2 4 x 2 y2 z 2 Ordinary – diff. eq. with ordinary derivatives (solutions are single variable functions) 2 dy m2 2 d y 1 x dx 2 2x dx l l 1 y 1 x 2 y 0 Linear – diff. eq. with the form ay by cy dy f x All terms in the equation have derivatives raised to the first or zeroth power. Nonlinear – diff. eq. that is not linear One or more terms in the equations have derivatives raised to the second or higher power. Order of a differential equation – highest derivative within equation 18 Examples in the classification of differential equations y cos x y k 2 y 0 y yy 4 0 first order, linear second order, linear third order, nonlinear y 7 y 8y 0 3 second order, nonlinear Separable Equations Separable differential equations can be rewritten as integrals. Consider the following differential equation. dy y f (x) dy f (x) dx dx y f (x) dx dy f (x) dx Thus, a simple first-order differential equation is the same thing as an indefinite integral. The technique of separable equations is to separate all that depends on each variable to different sides of the equation and integrate. Consider the following example involving the first-order kinetics of nuclear decay. Example: Radioactive decay is proportional to the number of nuclei. If N = N0 at time t =0, find the number of atoms at time t. dN N dt - decay constant First separate variables N and t. dN dt N Now integrate N t dN N N 0 dt 0 ln N ln N ln N 0 ln or N N0 t N t N0 N N 0 e t t 0 19 Example: The first-order kinetics of radioactive decay is well described with a firstorder differential equation that can be solved with separation of variables technique. For 14C, = 1.750 10-4 yr-1. If an animal bone is assumed to have 1.0 mol of 14C at death and the bone is measured to have 0.13 mol presently, how old is the bone? ln N t N0 t 1 N 1 0.13 mol ln ln 12000 yr 4 1 N 0 1.750 10 yr 1.0 mol Linear First-order Equations The general form for a linear first-order equation is y P x y Q x The solution for the a general linear first-order equation is y e I Q(x)e I dx c e I where I P(x)dx and c is a arbitrary constant that depends on the specific “boundary values” of the problem. 20 1 , find y x First manipulate equation as to find P and Q. 1 2 1 x 2 y 2xy y y 3 x x x Example: If x 2 y 2xy Px 2 x Qx 1 x3 I P x dx 2 dx 2 ln x x e I x 2 e I e2ln x eln x x 2 1 y x 2 3 x 2 dx c x 2 x 2 x 5dx c x 2 x 1 4 1 x2 x c x2 2 c x2 4 4x 2 Check solution 1 x 3 y 2x 3 2c x 2c x 4 2 x 3 1 1 1 x 2 y 2xy x2 2c x 2x 2 c x 2 x 4x x 2 x 1 1 1 2c x 3 2c x 3 x 2 2x 1 1 1 2c x 3 2c x 3 2x 2x x 1 1 0 x x Example: Find the solution for the differential equation: y y ex I dx x y e x e x e x dx ce x y e x e 2x dx ce x 1 1 y e x e 2x ce x e x ce x 2 2 Check solution 1 y e x ce x 2 y y e x 1 1 1 1 e x ce x e x ce x e x e x e x 2 2 2 2 21 Homogeneous first-order equations General linear homogeneous first-order equation: y P(x)y 0 Note: Q(x) = 0 P(x)dx Solution: y c e If P(x) is a constant then y c1e P x All first-order homogeneous diff. eq. will have an exponential solution. Technique using auxiliary equation d dy y Let D so that Dy dx dx Then y P(x)y 0 Dy Py 0 D P y 0 - auxiliary equation The “roots” of the auxiliary equation are the values of the exponents. D P 0 D P y c1e Px Homogeneous second-order equations General linear second-order equation: A(x)y B(x)y C(x)y D(x) For a homogeneous second-order equation, D(x) = 0. If A(x), B(x) and C(x) are constants, then the second-order equation can be rewritten as a product of first-order equations. The solution of each first-order differential equation is an exponential function; thus, the solution of the second-order equation is a sum of exponential functions. d dy d2 y For the following, let D so that Dy and D2 y 2 . dx dx dx Example: Solve y 5y 4y 0 y 5y 4y 0 D D 4 D 1 y 0 D 4 y 0 or D 1 y 0 2 5D 4 y 0 D2 y 5Dy 4y 0 (auxiliary equation) dy 4y 0 y c1e4x dx dy y 0 y c2e x D 1 y 0 dx The general solution is y c1e4x c2e x D 4 y 0 22 Thus the solution of differential equation depends on finding the exponents for the exponential functions. ***These exponents can be found by finding the roots of the auxiliary equation.*** D 2 5D 4 y 0 a 2 5a 4 0 a 4, 1 Worth noting is that the roots of the auxiliary equation can be complex so the solution of the differential equation can involve complex exponential functions. Digression: Application of Differential Equations to the Harmonic Oscillator. Pure Harmonic Oscillator Consider a mass connected to a spring. -x +x x0 Hooke’s Law: Force of a spring is proportional to its displacement from equilibrium F k x x 0 k – spring constant (force constant, Hooke’s Law constant) When the coordinate system is defined conveniently to let x0 = 0, then F k x For a harmonic oscillator: Force of acceleration equals the force from Hooke’s law F ma m d2x dt 2 d2x k x0 dt 2 m F kx m d2x kx dt 2 x = x(t) equation for a pure harmonic oscillator For our later convenience, let define k 2 m Equation for a pure harmonic oscillator becomes d2 x 2 x 0 2 dt 23 Let us now apply the auxiliary equation technique to find a solution to harmonic oscillator equation. (The solution relates the position of the mass to the time elapsed.) d D2 x 2 x 0 dt D2 2 D i D D 2 2 x 0 Thus the solution to the pure harmonic oscillator is x c1eit c2eit Recall Euler’s identity e it cos t i sin t Using Euler’s identity, the solution can be rearranged to x c3 cos t ic4 sin t ***Thus the solution for a pure harmonic oscillator is a sinusoidal wave.*** Damped Harmonic Oscillator A frictional force on the spring can be introduced as a term that is proportional to the velocity Fd d dx dt d – damping constant The differential equation from equates the spring force with the inertial force becomes: d2x dx d 2 x d dx k m 2 kx d x0 dt dt dt 2 m dt m d 2b For convenience, let us define m The auxiliary equation becomes: D 2 2bD 2 x 0 The roots of the auxiliary equation are D b b2 2 Consider three cases of damping Case I: b 2 2 Overdamped oscillator Solution: x c1e1t c2e2 t - mass returns to equilibrium quickly - no oscillation around equilibrium point Case II: b 2 2 Critically damped oscillator Solution: x c1 c2 t et - mass returns to equilibrium in infinite time - no oscillation around equilibrium point 24 Case III: b 2 2 Underdamped oscillator Solution: x et c1eit c 2e it - mass returns to equilibrium in infinite time - oscillation occurs around equilibrium point - amplitude of oscillations decreases over time - oscillations have decay envelope Underdamped harmonic oscillator with decay envelope 1 1 0.5 t cos( 8 t) e t e 0 t e 0.5 1 1 0 2 4 6 0 8 10 12 t 14 12.56 Comparison of Damped Oscillators y 4y 4 0 Critically damped: y 3y 4 0 Underdamped: y 5y 4 0 Overdamped: 1.5 2t e 1.5 ( 1 2 t) 1 e 4t e t 2 0 cos 7 t e 1.5t 1.5 1.5 0 0.5 1 0 Linear Algebra/Matrix Algebra 1.5 2 t 2.5 3 3.5 4 4 25 Definitions Rows, Columns and Elements a b c M d e f g h i row 1 row 2 row 3 column 3 column 2 column 1 Diagonal a e i For a single element of the matrix, Mrow, column For example, M11 = a, M13 = c, M21 = d, M32 = h Transpose Flip matrix along diagonal I. e., row 1 column 1 row 2 column 2 a d g b e h c f i etc … 1 4 2 1 8 3 T M 8 7 1 M 4 7 9 3 9 3 2 1 3 Determinant Definition for a 2 2 matrix a b a b M ad bc det M c d c d 1 4 Example: Find det M where M 2 5 det M 1 5 4 2 5 8 13 For a 3 3 matrix a b c e f d f d e d e f a b c h i g i g h g h i aei afh bdi bfg cdh ceg Shortcut for 3 3 matrices only! + + + - - a b c a b d e f g h i d e aei bfg cdh ceg afh bdi g h 26 1 5 2 Example: Find det W where W 7 3 4 2 1 5 1 5 2 1 5 7 3 4 7 3 1 3 5 5 4 2 2 7 1 2 3 2 1 4 1 5 7 2 1 5 2 1 15 40 14 12 4 175 144 Matrix Operations Addition To add two matrices together, matrices must be same size. Elements in the same row/column position are added together. 7 3 2 9 6 11 Example: Calculate P + Q where P 4 8 1 and Q 0 3 5 0 2 4 1 2 2 7 3 2 9 6 11 7 9 3 6 2 11 2 3 13 P Q 4 8 1 0 3 5 4 0 8 3 1 5 4 5 6 0 2 4 1 2 2 0 1 2 2 4 2 1 0 6 Multiplication by a scalar (constant) Multiply every element by the scalar Example: Calculate 4P 28 12 8 4P 16 32 4 0 8 16 Multiplication of matrices To multiply matrices the number of columns of the first matrix must equal the number of rows of the second matrix. The i,j element of the product matrix is formed by taking the “dot product” of the ith row of the first matrix with the jth column of the second matrix. 4 2 1 3 Example: If A and B , calculate AB 1 9 4 2 27 4 1 2 4 4 3 2 2 12 16 AB 11 9 4 1 3 9 2 35 15 Example: For A and B given in the problem above, calculate BA 1 2 3 9 1 29 1 4 3 1 BA 4 4 2 1 4 2 2 9 14 26 **NOTE: AB BA In other words, matrix multiplication is not commutative.** 1 0 2 1 1 0 Example: Calculate AB – BA when A 3 1 0 and B 0 2 1 0 5 1 3 1 0 11 0 0 2 3 11 0 2 2 1 1 0 0 1 2 0 7 1 0 AB 3 1 1 0 0 3 3 1 1 2 0 1 3 0 1 1 0 0 3 1 1 0 1 5 0 1 3 0 1 5 2 1 1 0 0 5 1 1 0 3 9 5 11 1 3 0 0 1 0 1 1 0 5 1 2 1 0 0 1 4 1 2 BA 0 1 2 3 1 0 0 0 2 1 1 5 0 2 2 0 11 6 3 1 3 1 1 3 0 0 3 0 1 1 0 5 3 2 1 0 0 1 0 1 6 7 1 0 4 1 2 3 0 2 AB BA 3 1 1 6 3 1 3 2 2 3 9 5 0 1 6 3 8 1 Inverse of a matrix The inverse of matrix, A-1, obey the following matrix multiplication properties AA-1 = A-1A = 1 1 0 0 0 1 0 0 1 0 0 1 0 0 1 0 1 0 0 1 0 0 1 0 0 1 0 0 0 0 1 Finding the inverse of a matrix in general is not too difficult but it is more than we want to consider here. The inverse for any 2 2 matrix is given below. a a 1 a 22 a12 A 11 12 A 1 det A a 21 a11 a 21 a 22 Systems of Equations 28 Matrix algebra can be used to solve for n unknowns in n equations. Of the many methods possible, we will briefly consider two. Gaussian Elimination Consider the following set of equations. 2u v w 1 4u v 2 4u 2v w 9 What are the values for u, v and w? Let put this system of equations in matrix form. 2 1 1 u 1 4 1 0 v 2 4 2 1 w 9 Gaussian elimination entails using the following manipulations on a system. These manipulations do not change the solution of u, v and w. 1) Multiply a row by a number. 2) Add two rows together. To use the technique efficiently, we create an “expanded matrix” composed of the coefficient matrix and the vector of constants. 2 1 1 u 1 2 1 1 1 4 1 0 v 2 4 1 0 2 4 2 1 w 9 4 2 1 9 We will use the above operations to change the expanded matrix into an upper triangular matrix, that is, the coefficients below the diagonal will be zero. a b c x 0 d e y 0 0 f z Then u, v and w can be found be simple substitution. To begin let us eliminate element 2,1 from the expanded matrix. We can do this by multiplying the first row by –2 and adding it to the second row. 29 2 1 1 1 4 1 0 2 4 2 1 9 4 2 2 0 1 2 4 2 1 2 2 2 4 2 2 2 4 4 1 0 2 4 4 1 2 0 2 2 2 4 2 1 9 4 2 1 9 2 4 9 Now let us eliminate element 3,1 by multiplying row 1 by –1 and adding to row 3. 2 2 2 4 2 2 2 4 2 2 2 4 1 2 4 0 1 2 4 0 1 2 4 0 4 2 1 9 4 2 1 9 4 4 2 2 1 2 9 2 4 2 2 2 0 1 2 4 0 4 3 11 Now let us eliminate element 3,2 by multiplying row 2 by 4 and adding to row 3. 2 2 2 2 4 2 2 2 4 2 2 4 8 16 0 1 2 4 0 4 8 16 0 4 0 4 3 11 0 4 3 11 0 4 4 3 8 11 16 2 4 2 2 0 4 8 16 0 0 5 5 This matrix can now be used to find our results. Let us use the matrix to recreate our system of equations. 2 4u 2v 2w 2 4 2 2 4 2 2 u 2 0 4 8 16 0 4 8 v 16 4v 8w 16 0 0 5 5 0 0 5 w 5 5w 5 We can now use each equation in turn to find u, v and w. 5w 5 w 1 4v 8w 16 4v 8 1 16 4v 8 v 2 4u 2v 2w 2 4u 2 2 2 1 2 4u 4 u 1 This technique is often very handy and much more efficient than raw substitution (which will work but is very tedious with algebra). Cramer’s rule 30 Another method for solving systems of equations is Cramer’s rule. This method is especially handy if you have a computer program that can quickly calculate the determinants of matrices. Each unknown is solving by taking the ratio of two determinants. The denominator of each ratio is always the determinant of the coefficient matrix. For our example above, 2 1 1 u 1 4 1 0 v 2 4 2 1 w 9 2 1 1 4 1 0 2 11 1 0 4 1 4 2 2 0 2 1 4 1 11 4 10 4 2 1 The numerator of each ratio is the coefficient matrix with a column replaced by the constants vector. To solve for u, the first column is replaced with the constants vector. 1 1 1 2 1 0 9 2 1 111 1 0 9 1 2 2 1 0 2 1 2 1 11 9 10 1 u 2 1 1 2 11 1 0 4 1 4 2 2 0 2 1 4 1 11 4 10 4 1 0 4 2 1 To solve for v, the second column is replaced with the constants vector. 2 1 1 4 2 0 4 9 1 2 2 1 1 0 4 1 4 9 2 0 9 1 4 1 1 2 4 20 v 2 2 1 1 2 11 1 0 4 1 4 2 2 0 2 1 4 1 11 4 10 4 1 0 4 2 1 To solve for w, the third column is replaced with the constants vector. 2 1 1 4 1 2 4 2 9 2 1 9 1 2 4 1 4 2 2 2 2 1 4 9 11 4 10 1 w 2 1 1 2 11 1 0 4 1 4 2 2 0 2 1 4 1 11 4 10 4 1 0 4 2 1 Eigenvalues and Eigenvectors 1 3 For the 2 2 matrix, , the eigenvalue/eigenvector problem is stated as 2 2 31 1 3 x x 2 2 y y x What are values of (eigenvalues) and the vectors (eigenvectors) that make the y above equation true? Restatement of the problem: Multiplying a square matrix by an arbitrary vector yields another vector. When the resultant vector is a multiple of the arbitrary vector, we say that particular vector is an eigenvector. The scalar multiple is called the eigenvalue. The eigenvalue/eigenvector problem can be restated as 1 3 x x 0 2 2 y y 1 3 x 1 0 x 1 3 x 0 x 2 2 y 0 1 y 2 2 y 0 y 1 3 0 x 1 3 x 0 2 y 2 2 0 y 2 This last equation is called the characteristic equation of the matrix. To solve for the eigenvalues and eigenvectors of a matrix. 1. Find the eigenvalues by calculating the determinant of the matrix in the characteristic equation. 3 x 1 0 2 y 2 1 3 1 2 3 2 0 2 2 2 3 2 6 2 3 4 0 2. Set the determinant equal to zero and solve for the roots of the equation. The roots of the equation are the eigenvalues. 2 3 4 0 1 4 1, 4 3. Each dimension of the square matrix yields an eigenvalue. A 2 2 has two eigenvalues. A 3 3 matrix has three eigenvalues, etc… Each eigenvalue corresponds to an eigenvector. The eigenvalues are used to find the eigenvectors. 32 To find the eigenvectors, a relationship must be found be x and y of the unknown eigenvector. a) This relationship is found by plugging an eigenvalue into the characteristic equation of the matrix. 3 x 1 1 3 x 1 0 0 2 2 1 y 2 y 2 2 3 x 2 0 2x 3y 0 y x 3 2 3 y b) Thus the eigenvector in its most general form is x v 2 x 3 All of the following eigenvectors are valid eigenvectors for the eigenvalue, -1. 3 1 3 6 4 v 2 2 4 1 3 2 The preferred eigenvector is called a normalized eigenvector. Normalization will be considered shortly. c) Find the other eigenvector associated with the other eigenvalue. 3 x 1 4 3 x 1 0 0 2 2 4 y 2 y 2 3 3 x 0 3x 3y 0 y x 2 2 y x This eigenvector for the eigenvalue, 4, in its most general form is v x 1 2 1 Example: The eigenvectors for the matrix Z 2 3 0 are 1 0 3 1 0 5 v1 2 , v 2 1 and v3 2 . Find the eigenvalues corresponding to each 2 1 1 eigenvector. 33 1 2 1 1 11 2 2 11 4 1 Z 2 3 0 2 2 1 3 2 0 1 8 4 2 1 0 3 1 11 0 2 3 1 4 1 Thus the eigenvalue is 1 4 1 2 1 0 1 0 2 1 1 2 0 0 Z 2 3 0 1 2 0 3 1 0 2 3 3 1 1 0 3 2 1 0 0 1 3 2 6 2 Thus the eigenvalue is 2 3 1 2 1 5 1 5 2 2 1 1 10 5 Z 2 3 0 2 2 5 3 2 0 1 4 2 2 1 0 3 1 1 5 0 2 3 1 2 1 Thus the eigenvalue is 3 2 Normalization of eigenvectors The square of a normalized eigenvector is one. In other words, the dot product between a normalized eigenvector and itself is one. n n 1 Eigenvectors are normalized by multiplying them by an appropriate constant. Nv Nv 1 N 2 v v 1 N 1 vv 34 x x Example: Normalize the following eigenvectors: v , v 2 . x x 3 x v v x x 2x 2 x 1 1 1 2 vv 2x 2x 1 1 x 2 Therefore the normalized eigenvector is n Nv 2x x 1 2 N x 2 13 v v x x 2 x2 N 3 x 9 3 1 1 3 vv 13 2 13x x 9 3 x 3 13 Therefore the normalized eigenvector is n Nv 2 13x x 2 3 13