Open navigation menu
Close suggestions
Search
Search
en
Change Language
Upload
Sign in
Sign in
Download free for days
0 ratings
0% found this document useful (0 votes)
47 views
6 pages
ch3 3
Uploaded by
checheche24
AI-enhanced title
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content,
claim it here
.
Available Formats
Download as PDF or read online on Scribd
Download
Save
Save ch3.3 For Later
0%
0% found this document useful, undefined
0%
, undefined
Embed
Share
Print
Report
0 ratings
0% found this document useful (0 votes)
47 views
6 pages
ch3 3
Uploaded by
checheche24
AI-enhanced title
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content,
claim it here
.
Available Formats
Download as PDF or read online on Scribd
Carousel Previous
Carousel Next
Download
Save
Save ch3.3 For Later
0%
0% found this document useful, undefined
0%
, undefined
Embed
Share
Print
Report
Download now
Download
You are on page 1
/ 6
Search
Fullscreen
190 Chapter 3 Orthogonality Problem Set 3.3 1. Find the best least-squares solution fto 3x = 10, 4x = 5. What error Eis minimized? Check that the error vector (10 — 3,5 — 4) is perpendicular to the column (3,4) 2. Suppose the values bj = 1 and bz =7 at times f; = 1 and fz = 2 are fitted by a line b = Dt through the origin. Solve D = 1 and 2D =7 by least squares, and sketch the best line.3.3 Projections and Least Squares 191 3. Solve Ax = b by least squares, and find p = ATif 10 1 A=|0 1], b=]1 1. 0 Verify that the error b— p is perpendicular to the columns of A. |. Write out E? = ||Ax—b||? and set to zero its derivatives with respect to u and y, i 4. Wi E? = ||Ax—b|j? and di th resp¢ dv, if 10 1 A=]0 1], -[ b=]3 v 11 4 Compare the resulting equations with ATA? = ATO, confirming that calculus as well cometry gives the normal equations. Find the solution ¥ and the projection p = AR. Why is p = b? 5. The following system has no solution: 1-1 4 c Ax=|1 0 = [51 = . [s|-|5} =» Ll 9 Sketch and solve a straight-line fit that leads to the minimization of the quadratic (C-D—4)? + (C—5)?+(C+D—9)?? What is the projection of b onto the column space of A? 6. Find the projection of b onto the column space of A: to A=]1 -1|, b= 24 aN Split b into p +-q, with p in the column space and q perpendicular to that space. Which of the four subspaces contains q? 7. Find the projection matrix P onto the space spanned by ay = (1,0,1) and az (1,1,-1), 8. If P is the projection matrix onto a k-dimensional subspace S of the whole space R”, what is the column space of P and what is its rank? 9. (a) If P= P™P, show that P is a projection matrix. (b) What subspace does the matrix P = 0 project onto? 10. If the vectors ay, az, and b are orthogonal, what are ATA and AT? What is the projection of b onto the plane of a; and ay?192 Chapter 3 Orthogonality 11. Suppose P is the projection matrix onto the subspace $ and @ is the projection onto the orthogonal complement S*. What are P+ Q and PQ? Show that P— Qs its own inverse. 12. If Vis the subspace spanned by (1, 1,0, 1) and (0,0, 1,0), find (a) a basis for the orthogonal complement V+ (b) the projection matrix P onto V. (©) the vector in V closest to the vector b = (0,1,0,—1) in V+. 13. Find the best straight-line fit (least squares) to the measurements b=4 at t=-2, b=3 at b= =0, b=0 at t= at Then find the projection of b = (4,3, 1,0) onto the column space of 1-2 1-1 10 12 14, The vectors a; = (1, 1,0) and az = (1, 1,1) span a plane in R?. Find the projection matrix P onto the plane, and find a nonzero vector b that is projected to zero 15. If P is the projection matrix onto a line in the x-y plane, draw a figure to describe the effect of the “reflection matrix” H =1—2P. Explain both geometrically and algebraically why H? =I. 16. Show that if uv has unit length, then the rank-1 matrix P = wu? is a projection matrix: Ithas properties (i) and (ii) in 3N. By choosing u = a/lal|, P becomes the projection onto the line through a, and Pb is the point p = Sa, Rank-1 projections correspond exactly to least-squares problems in one unknown, 17. What 2 by 2 matrix projects the x-y plane onto the —45° line x+y = 0? 18. We want to fit a plane y = C + Dt + Ez to the four points y=3 at f=1zr=1 y=6 a t=0,7=3 y=S at t=22=1 y=0 at £=0,2=0 (a) Find 4 equations in 3 unknowns to pass a plane through the points (if there is such a plane). (b) Find 3 equations in 3 unknowns for the best least-squares solution. 19. If Pe = A(ATA)~!AT is the projection onto the column space of A, what is the pro- jection Pe onto the row space? (It is not PZ!)20. 21. 22. 25. 26. 27. 28. 3.3 Projections and Least Squares 193 If P is the projection onto the column space of A, what is the projection onto the left nullspace? Suppose L; is the line through the origin in the direction of ay and Zz is the line through b in the direction of a. To find the closest points xa; and b +x2a2 on the two lines, write the two equations for the x1 and x2 that minimize ||x1a1 —x2a2 —_b| Solve for x if a1 = (1,1,0), a2 = (0,1,0), = (2,1,4) Find the best line C+ Dr to fit b = 4,2, —1,0,0 at times ¢ = ~2,~1,0,1,2 Show that the best least-squares fit to a set of measurements y1, ... ,ym by a horizontal line (a constant function y = C) is their average ca yn m Find the best straight-line fit to the following measurements, and sketch your solu- tion: y=2 at r=-1, = y=0 at 1=0, y=-3 at r= 1, y=-5 at 1=2 Suppose that instead of a straight line, we fit the data in Problem 24 by a parabola: y=C+Dt+ Et’. In the inconsistent system Ax = b that comes from the four mea- surements, what are the coefficient matrix A, the unknown vector x, and the data vector b? You need not compute £. A Middle-Aged man was stretched on a rack to lengths L = 5, 6, and 7 feet under applied forces of F = 1, 2, and 4 tons, Assuming Hooke’s law L = a+ BF, find his normal length a by least squares. Problems 27-31 introduce basic ideas of statistics—the foundation for least squares. (Recommended) This problem projects b = (b1,...,bm) onto the line through a = (1,...,1). We solve m equations ax = b in 1 unknown (by least squares). (a) Solve aa% = ab to show that is the mean (the average) of the b's, (b) Find e = b — af, the variance |le||?, and the standard deviation |le| (c) The horizontal line 5 = 3 is closest to b = (1,2,6), Check that p = (3,3,3) is perpendicular to ¢ and find the projection matrix P. First assumption behind least squares: Each measurement error has mean zero. Mul- tiply the 8 error vectors b — Ax = (£1,+1,£1) by (ATA)~!AT to show that the 8 vectors ¥—x also average to zero. The estimate Fis unbiased.194 29. 3. 32. 33. 34, 35. 36. 37. Chapter 3 Orthogonality Second assumption behind least squares: The m errors ¢; are independent with variance 67, so the average of (b —Ax)(b—Ax)" is o7J. Multiply on the left by (ATA)~1AT and on the right by A(A™A)~! to show that the average of (—x)(¥—x)? is o?(ATA)~1. This is the all-important covariance matrix for the error in ¥. . A doctor takes four readings of your heart rate, The best solution tox = by,... x= by is the average £ of b,,...,bs. The matrix A is a column of 1s. Problem 29 gives the expected error (f—x)? as 6?7(ATA)“! =__. By averaging, the variance drops from 0? to 07/4. If you know the average % of 9 numbers by, ...,b9, how can you quickly find the average jo with one more number bio? The idea of recursive least squares is to avoid adding 10 numbers. What coefficient of % correctly gives £19? Ko = bio t+ __%9 = Albi +--+ dio). Problems 32-37 use four points b = (0,8, 8,20) to bring out more ideas. With b = 0,8,8,20 att =0, 1,3,4, set up and solve the normal equations ATAT= ATD. For the best straight line as in Figure 3.9a, find its four heights p; and four errors e). What is the minimum value E? = e} +e} + 3 + 3? (Line C+ Dr does go through p's) With b = 0,8,8,20 at times ¢ = 0,1,3,4, write the four equations Ax = b (unsolvable). Change the measurements to p = 1,5, 13, 17 and find an exact solution to A = p. Check that e = b— p = (~1,3,—5,3) is perpendicular to both columns of A, What is the shortest distance ||el] from b to the column space of A? For the closest parabola b = C + Dt + E?* to the same four points, write the unsolv- able equations Ax = b in three unknowns x = (C,D,E). Set up the three normal equations ATA¥ = ATb (solution not required). You are now fitting a parabola to four points—what is happening in Figure 3.96? For the closest cubic b = C + Dt + Et? + FP to the same four points, write the four equations Ax = b. Solve them by elimination, This cubic now goes exactly through the points. What are p and e? The average of the four times is? = }(0+1+3-+4) =2. The average of the four b's isb = }(0+8+8+20) =9. (a) Verify that the best line goes through the center point (f,b) = (2,9) (b) Explain why C+ Df ATb, comes from the first equation in ATAT . What happens to the weighted average Sy = (3b, + w3b2)/(wy + 03) if the first weight w; approaches zero? The measurement by is totally unreliable39, 40. 41. 42, 34 Orthogonal Bases and Gram-Schmidt 195 From m independent measurements b1, ...,bm of your pulse rate, weighted by W1,..., Wns what is the weighted average that replaces equation (9)? It is the best estimate when the statistical variances are o? = 1/w?. If W = [29], find the W-inner product of x = (2,3) and y = (1,1), and the W-length of x. What line of vectors is W-perpendicular to y? Find the weighted least-squares solution Sy to Ax = b: 10 0 200 A=|11 b=lL w=|o10 12 1 oo1 Check that the projection Ai is still perpendicular (in the W-inner product!) to the error b — Ati. (a) Suppose you guess your professor's age, making errors e = —2, —1,5 with prob- abilities 3, 3,4. Check that the expected error E(e) is zero and find the variance E(e?) (b) If the professor guesses too (or tries to remember), making errors —1, 0, 1 with probabilities ¢,$,%, what weights w; and w» give the reliability of your guess and the professor's guess?
You might also like
Linear Algebra: Concepts and Methods Solutions To Problems
PDF
No ratings yet
Linear Algebra: Concepts and Methods Solutions To Problems
155 pages
EE263 Homework 3 Solutions
PDF
No ratings yet
EE263 Homework 3 Solutions
16 pages
HW1 Solution
PDF
No ratings yet
HW1 Solution
7 pages
3.3 3.4 Some Solutions
PDF
No ratings yet
3.3 3.4 Some Solutions
12 pages
Least Square Ion
PDF
No ratings yet
Least Square Ion
12 pages
MA 106: Linear Algebra: J. K. Verma Department of Mathematics Indian Institute of Technology Bombay
PDF
No ratings yet
MA 106: Linear Algebra: J. K. Verma Department of Mathematics Indian Institute of Technology Bombay
13 pages
HW 6 Solutions
PDF
No ratings yet
HW 6 Solutions
11 pages
Homework Sheet 5 Solutions
PDF
No ratings yet
Homework Sheet 5 Solutions
7 pages
Lec 10
PDF
No ratings yet
Lec 10
31 pages
Linear Least Squares
PDF
No ratings yet
Linear Least Squares
21 pages
Least Squares Aproximations
PDF
No ratings yet
Least Squares Aproximations
10 pages
Fa 2
PDF
No ratings yet
Fa 2
7 pages
L16 - 17 Linear Algebra - Least Square Approximation
PDF
No ratings yet
L16 - 17 Linear Algebra - Least Square Approximation
11 pages
Least Square by Nicholson-Linear Algebra-2018
PDF
No ratings yet
Least Square by Nicholson-Linear Algebra-2018
12 pages
a bình phương tối tiểu
PDF
No ratings yet
a bình phương tối tiểu
11 pages
MA 106: Linear Algebra: Prof. B.V. Limaye IIT Dharwad
PDF
No ratings yet
MA 106: Linear Algebra: Prof. B.V. Limaye IIT Dharwad
19 pages
Chapter 6: Application: The Least-Square Solution and The Least-Squares Error. 1 Best Approximation
PDF
No ratings yet
Chapter 6: Application: The Least-Square Solution and The Least-Squares Error. 1 Best Approximation
7 pages
Suggested Problems W11
PDF
No ratings yet
Suggested Problems W11
4 pages
HEA2 S
PDF
No ratings yet
HEA2 S
4 pages
KKKQ1223 Engineering Mathematics (Linear Algebra) : Best Approximation Least Squares Least Squares Fitting To Data
PDF
No ratings yet
KKKQ1223 Engineering Mathematics (Linear Algebra) : Best Approximation Least Squares Least Squares Fitting To Data
22 pages
Least Squares and Data Fitting
PDF
No ratings yet
Least Squares and Data Fitting
6 pages
3.1 Least-Squares Problems
PDF
No ratings yet
3.1 Least-Squares Problems
28 pages
Asgn2 LST
PDF
No ratings yet
Asgn2 LST
3 pages
Chap 03
PDF
No ratings yet
Chap 03
59 pages
Worksheet15 Solutions
PDF
No ratings yet
Worksheet15 Solutions
6 pages
Linear Least Squares Problems
PDF
No ratings yet
Linear Least Squares Problems
38 pages
Math1250 1Test2Sol
PDF
No ratings yet
Math1250 1Test2Sol
5 pages
Further Mathematical Methods (Linear Algebra) Solutions For Problem Sheet 8
PDF
No ratings yet
Further Mathematical Methods (Linear Algebra) Solutions For Problem Sheet 8
15 pages
MIT18 06SCF11 FinalRevsum
PDF
No ratings yet
MIT18 06SCF11 FinalRevsum
7 pages
Homework 9 Solutions: U + C v. If y Is Orthogonal To U U + C V)
PDF
No ratings yet
Homework 9 Solutions: U + C v. If y Is Orthogonal To U U + C V)
6 pages
Some Notes On Least Squares, QR-factorization, SVD and Fitting
PDF
No ratings yet
Some Notes On Least Squares, QR-factorization, SVD and Fitting
12 pages
MML Book (091 120)
PDF
No ratings yet
MML Book (091 120)
30 pages
TUT 2 Solutions
PDF
100% (1)
TUT 2 Solutions
22 pages
Least-Square Method
PDF
No ratings yet
Least-Square Method
32 pages
Lecture19 D1
PDF
No ratings yet
Lecture19 D1
14 pages
Sol51-Linear Algebra
PDF
No ratings yet
Sol51-Linear Algebra
3 pages
EE263s Homework 4
PDF
No ratings yet
EE263s Homework 4
11 pages
MIT18 06S10 Pset1 s10 Soln
PDF
No ratings yet
MIT18 06S10 Pset1 s10 Soln
9 pages
Lec Continuous Least Squares Annotated
PDF
No ratings yet
Lec Continuous Least Squares Annotated
31 pages
Linear Algebra 2
PDF
No ratings yet
Linear Algebra 2
16 pages
Homework 4 MATH2050
PDF
No ratings yet
Homework 4 MATH2050
7 pages
3.1 Projections & Reflections
PDF
No ratings yet
3.1 Projections & Reflections
4 pages
Lecture25 Ps
PDF
No ratings yet
Lecture25 Ps
10 pages
HW4 Solution
PDF
No ratings yet
HW4 Solution
10 pages
(MATH2111) (2017) (F) Final In5mue 14501
PDF
No ratings yet
(MATH2111) (2017) (F) Final In5mue 14501
12 pages
Student: Kyle Hoang Date: 04/15/22 Instructor: Larissa Williamson Course: MAS3114-Assignment: M30 (HW) Approximation &
PDF
No ratings yet
Student: Kyle Hoang Date: 04/15/22 Instructor: Larissa Williamson Course: MAS3114-Assignment: M30 (HW) Approximation &
26 pages
Unit-2 Linear Algebra - 2 Notes
PDF
No ratings yet
Unit-2 Linear Algebra - 2 Notes
14 pages
MATH 223, Linear Algebra Fall, 2007 Assignment 9, Due November 23, 2007
PDF
No ratings yet
MATH 223, Linear Algebra Fall, 2007 Assignment 9, Due November 23, 2007
2 pages
Wa0072
PDF
No ratings yet
Wa0072
12 pages
Lectures 17 To 20
PDF
No ratings yet
Lectures 17 To 20
76 pages
Lecture-13 - ESO208 - Aug 31 - 2022
PDF
No ratings yet
Lecture-13 - ESO208 - Aug 31 - 2022
28 pages
2018 2 Solutions
PDF
No ratings yet
2018 2 Solutions
9 pages
602 603
PDF
No ratings yet
602 603
5 pages
Section 6.4 (Extra Details)
PDF
No ratings yet
Section 6.4 (Extra Details)
10 pages
MAI391
PDF
No ratings yet
MAI391
52 pages