It is easier to visualize plotting the curves of cos x and e x separately: > plot({cos(x),exp(x)},x = -5*Pi..Pi,y = );

Similar documents
Math 467 Homework Set 1: some solutions > with(detools): with(plots): Warning, the name changecoords has been redefined

Numerical Solution of ODE

1.4 Application Separable Equations and the Logistic Equation

Engineering Mathematics 2018

Gauss-Jordan Algorithm

Project #1 Math 285 Name:

MATH Differential Equations September 15, 2008 Project 1, Fall 2008 Due: September 24, 2008

Sam knows that his MP3 player has 40% of its battery life left and that the battery charges by an additional 12 percentage points every 15 minutes.

Chapter Six Chapter Six

Fill in the following table for the functions shown below.

Quantitative macro models feature an infinite number of periods A more realistic (?) view of time

ME 406 Assignment #1 Solutions

Implementing Ray Casting in Tetrahedral Meshes with Programmable Graphics Hardware (Technical Report)

Motor Control. 5. Control. Motor Control. Motor Control

4. Minimax and planning problems

Mass-Spring Systems and Resonance

Chapter 3 MEDIA ACCESS CONTROL

THERMAL PHYSICS COMPUTER LAB #3 : Stability of Dry Air and Brunt-Vaisala Oscillations

DEFINITION OF THE LAPLACE TRANSFORM

Motion along a Line. Describing Motion along a Line

NEWTON S SECOND LAW OF MOTION

CENG 477 Introduction to Computer Graphics. Modeling Transformations

A Principled Approach to. MILP Modeling. Columbia University, August Carnegie Mellon University. Workshop on MIP. John Hooker.

4.1 3D GEOMETRIC TRANSFORMATIONS

Matlab5 5.3 symbolisches Lösen von DGLn

the marginal product. Using the rule for differentiating a power function,

Shortest Path Algorithms. Lecture I: Shortest Path Algorithms. Example. Graphs and Matrices. Setting: Dr Kieran T. Herley.

Scattering at an Interface: Normal Incidence

EECS 487: Interactive Computer Graphics

Traditional Rendering (Ray Tracing and Radiosity)

AML710 CAD LECTURE 11 SPACE CURVES. Space Curves Intrinsic properties Synthetic curves

REDUCTIONS BBM ALGORITHMS DEPT. OF COMPUTER ENGINEERING ERKUT ERDEM. Bird s-eye view. May. 12, Reduction.

Announcements For The Logic of Boolean Connectives Truth Tables, Tautologies & Logical Truths. Outline. Introduction Truth Functions

RULES OF DIFFERENTIATION LESSON PLAN. C2 Topic Overview CALCULUS

PART 1 REFERENCE INFORMATION CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONITOR

FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS

Michiel Helder and Marielle C.T.A Geurts. Hoofdkantoor PTT Post / Dutch Postal Services Headquarters

Parametric equations 8A

A non-stationary uniform tension controlled interpolating 4-point scheme reproducing conics

Nonparametric CUSUM Charts for Process Variability

4 Error Control. 4.1 Issues with Reliable Protocols

Real Time Integral-Based Structural Health Monitoring

MOTION DETECTORS GRAPH MATCHING LAB PRE-LAB QUESTIONS

7.2 Puiseux Expansions

STEREO PLANE MATCHING TECHNIQUE

Y. Tsiatouhas. VLSI Systems and Computer Architecture Lab

FLOW VISUALIZATION USING MOVING TEXTURES * Nelson Max Lawrence Livermore National Laboratory Livermore, California

Definition and examples of time series

An Improved Square-Root Nyquist Shaping Filter

Evaluation and Improvement of Region-based Motion Segmentation

Learning in Games via Opponent Strategy Estimation and Policy Search

Assignment 2. Due Monday Feb. 12, 10:00pm.

Integro-differential splines and quadratic formulae

A Numerical Study on Impact Damage Assessment of PC Box Girder Bridge by Pounding Effect

Network management and QoS provisioning - QoS in Frame Relay. . packet switching with virtual circuit service (virtual circuits are bidirectional);

Data Structures and Algorithms. The material for this lecture is drawn, in part, from The Practice of Programming (Kernighan & Pike) Chapter 2

(Structural Time Series Models for Describing Trend in All India Sunflower Yield Using SAS

In Proceedings of CVPR '96. Structure and Motion of Curved 3D Objects from. using these methods [12].

Coded Caching with Multiple File Requests

! errors caused by signal attenuation, noise.!! receiver detects presence of errors:!

FINITE DIFFERENCE SOLUTIONS TO THE TIME DEPENDENT HEAT CONDUCTION EQUATIONS. T T q 1 T + = (1)

Piecewise Linear Models

The Beer Dock: Three and a Half Implementations of the Beer Distribution Game

Scheduling. Scheduling. EDA421/DIT171 - Parallel and Distributed Real-Time Systems, Chalmers/GU, 2011/2012 Lecture #4 Updated March 16, 2012

Hyelim Oh. School of Computing, National University of Singapore, 13 Computing Drive, Singapore SINGAPORE

Boyce - DiPrima 8.4, Multistep Methods

COSC 3213: Computer Networks I Chapter 6 Handout # 7

Ray Casting. Outline. Outline in Code

MOBILE COMPUTING. Wi-Fi 9/20/15. CSE 40814/60814 Fall Wi-Fi:

MOBILE COMPUTING 3/18/18. Wi-Fi IEEE. CSE 40814/60814 Spring 2018

Dimmer time switch AlphaLux³ D / 27

Image segmentation. Motivation. Objective. Definitions. A classification of segmentation techniques. Assumptions for thresholding

A Matching Algorithm for Content-Based Image Retrieval

Why not experiment with the system itself? Ways to study a system System. Application areas. Different kinds of systems

Flow graph/networks MAX FLOW APPLICATIONS. Flow constraints. Max flow problem 4/26/12

Precise Voronoi Cell Extraction of Free-form Rational Planar Closed Curves

The Laplace Transform

Reinforcement Learning by Policy Improvement. Making Use of Experiences of The Other Tasks. Hajime Kimura and Shigenobu Kobayashi

Less Pessimistic Worst-Case Delay Analysis for Packet-Switched Networks

An Adaptive Spatial Depth Filter for 3D Rendering IP

CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL

MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES

Chapter 4 Sequential Instructions

Probabilistic Detection and Tracking of Motion Discontinuities

Modelling urban travel time variability with the Burr regression technique

MARSS Reference Sheet

A Progressive-ILP Based Routing Algorithm for Cross-Referencing Biochips

AUTOMATIC 3D FACE REGISTRATION WITHOUT INITIALIZATION

Using CANopen Slave Driver

Optimal Crane Scheduling

Optics and Light. Presentation

TUTORING TEXTS IN MATHCAD

Design Alternatives for a Thin Lens Spatial Integrator Array

In fmri a Dual Echo Time EPI Pulse Sequence Can Induce Sources of Error in Dynamic Magnetic Field Maps

PROCESS AUTOMATION MANUAL TIMER RELAY KF**-DU-EX1.D ISO9001

A Formalization of Ray Casting Optimization Techniques

Test - Accredited Configuration Engineer (ACE) Exam - PAN-OS 6.0 Version

Lecture 18: Mix net Voting Systems

Computer representations of piecewise

A High Accuracy Volume Renderer for Unstructured Data

Transcription:

Mah 467 Homework Se : some soluions > wih(deools): wih(plos): Warning, he name changecoords has been redefined Problem :..7 Find he fixed poins, deermine heir sabiliy, for x( ) = cos x e x > plo(cos(x) - exp(x),x = -5*Pi..Pi,y=-..); y 4 8 6 4 x I is easier o visualize ploing he curves of cos x and e x separaely: > plo({cos(x),exp(x)},x = -5*Pi..Pi,y = -.4..);.5 y.5 4 8 6 4 x.5 The fixed poins are a he inersecions of hese wo curves. Le s find he firs few numerically, specifying he inervals for he numerical soluion, which we know from he behaviour of cos x: > x := fsolve(exp(x)-cos(x),x,-..); x := fsolve(exp(x)-cos(x),x,-pi..-);

x := fsolve(exp(x)-cos(x),x,-pi..-); x := fsolve(exp(x)-cos(x),x,-*pi..-pi); x3 := fsolve(exp(x)-cos(x),x,-3*pi..-*pi); x4 := fsolve(exp(x)-cos(x),x,-4*pi..-3*pi); x :=. x := -.969579 x := -4.79759 x3 := -7.8535938 x4 := -.995596 We observe ha he roos xk for k are close o he zeros of cos x, since e x is small: > evalf(seq(-(k-/)*pi,k=..4)); -4.738898, -7.85398635, -.9955749 From he sign of cos x, we can see ha he fixed poins xk are sable if k is odd, and unsable if k is even. We can deduce he qualiaive behaviour of he soluions from he fixed poins and heir sabiliy. Direc numerical inegraion using Maple gives: > eqp := diff(x(),) = exp(x()) - cos(x()); iniconds := [[x()=],[x()=],[x()=-],[x()=-],[x()=-4],[x()=x],[x() =-5],[x()=-7], [x()=-8]]: DEplo(eqp,x(),=..8,x=-9..,iniconds,linecolor=black,sepsi ze=.); > eqp := x( ) = e x( ) cos ( x( ) ) e x 4 6 8 x() 4 6 8 A closed-form analyical soluion is no available, since i would require he inegraion of e x cos x.

Problem :..8 We seek a dynamical sysem yielding he given flow. A possible answer is given by d x( ) = ( x + ) x( x ) d > eqp := diff(x(),) = (x()+)^ * x() * (x()-): iniconds := [[x()=-.5],[x()=-],[x()=-.8],[x()=],[x()=.], [x()=],[x()=.]]: DEplo(eqp,x(),=..5,x=-..3,iniconds,linecolor=black,seps ize=.); 3 x() 3 4 5 This is one of many possible answers; ohers are obained by muliplying he vecor field by g( x ), where g is posiive excep possibly a one or more of he fixed poins. Problem 3:..3 - he Skydiver > m := m : g := g : k := k : eqp3 := diff(v(),) = g - k*v()^/m; k v( ) eqp3 := v( ) = g m The command odeadvisor indicaes he soluion mehod for his firs-order ODE. > odeadvisor(eqp3); [_quadraure] To find he soluion, we separae variables, and inegrae; he inegraion of ( a v ) (-) (where gm a = ) is bes performed using parial fracions. Maple can also solve his equaion k analyically: dsolve(eqp3);

v( ) = anh g m k ( + _C) m k g m k mg In fac, his soluion is only valid if m, g and k are all posiive, and if < v( ). I seems ha k Maple auomaically made hese assumpions; in his case hey are jusified, bu his example shows ha in general Maple s analyical soluions are no always reliable: do he calculaions by hand, and show your working! (you can use Maple o check, if you like). Tha is, he soluion produced by Maple is only he general soluion for a < v( ) (he problem is o ake care wih absolue value signs...). We can find he paricular soluion saisfying he iniial condiion v( ) = : > solp3 := rhs(dsolve({eqp3,v()=})); g m k anh g m k m solp3 := k Now we can use Maple o find he asympoic behaviour: > limi(solp3,=infiniy); g m k anh g m k m lim k Evidenly, now (finally!) Maple is concerned abou he sign of he variables. Le s ry specifying ha all variables are posiive: > limi(solp3,=infiniy) assuming (g>,m>,k>); g m k k This gives he correc erminal velociy. We can wrie he formula for v( ) in erms of he erminal velociy V: vsol := simplify(subs(m=k*v^/g,solp3)) assuming (k>,v>); vsol := anh g V V This answer is much more easily obained by he graphical mehod. We plo dv d agains v (choosing some values of he variables): g := : m :=.: k := : plo(g - k*v*v/m,v=-.5...5); m := m : g := g : k := k :

5 v.5.5.5.5 5 Now le s use he numbers given: The average velociy is (in f/sec) Vavg := (34-)/6; evalf(vavg); 735 Vavg := 9 5.58669 ds The disance ravelled as a funcion of ime is s( ) saisfying = v and s( ) =. d s := in(vsol,) assuming (V>,g>); V ln anh g V ln anh g + V V s := g g This soluion doesn look oo correc: i is giving negaive argumens of he ln funcion (and is hus complex-valued). Le s ry o help Maple a bi, by performing he appropriae subsiuion by hand... s := In(vsol,) assuming (V>,g>); > s3 := value(suden[changevar](au=*g/v,s,au)); > s := subs(au=*g/v,s3); s := anh g d V V s3 := V ln ( cosh( τ) ) g V ln cosh g V s := g > g := 3.: := 6: s; dis := 34-;

> g := 3.: := 6: s; dis := 34-; solve(s=dis,v);.35596 V ln cosh 3735. V dis := 93-5.58669 In he command solve, Maple aemps an analyical soluion; in his case i ges i wrong (I m no sure why; bu he given value is he average velociy compued previously, which canno also be he erminal velociy). For a problem wih purely floaing-poin soluions, we should use fsolve (and look for he posiive soluion): Verm := fsolve(s=dis,v,v=..infiniy); Verm := 65.685485 From his value of he erminal velociy, we can compue he drag consan k. Noe ha he weigh (in pounds) is mg. weigh := 6.: kdrag := solve(sqr(weigh/k)=verm,k); kdrag :=.373537 > m := m : g := g : k := k : s := s : := : Problem 4:.3. - Auocaalysis The fixed poins are readily found o be and k_ a k_ (-) > fp4 := k_*a*x - km_*x^; solve(fp4,x); fp4 := k_ a x km_ x, k_ a km_ x = is unsable, he oher fixed poin is sable. We can do a quick graphical analysis, and plo some ypical soluions, if we assume values for he consans: a := : k_ := : km_ := : plo(fp4,x=-.5...5);. x.4...4.6.8..4..4.6 > DEplo(diff(x(),)=a*k_*x()-km_*x()^,x(),=..5,x=-...,[[x()=-.],[x()=.],[x()=.7],[x()=.7]],linecolor=black)

,[[x()=-.],[x()=.],[x()=.7],[x()=.7]],linecolor=black) ;.5 x().5 3 4 5 Problem 5:.3.3 - Tumour growh We plo he vecor field and some soluions for some values of a and b: > a :=.: b :=.75: plo(-a*n*ln(b*n),n=..);.5 N..4.6.8..4.6.8.5.5 Clearly he fixed poin N = is unsable, and N = is sable. b > DEplo(diff(N(),)=-a*N()*ln(b*N()),N(),=..5,N=-...,[[N ()=.],[N()=.],[N()=.8]],linecolor=black);

.5 N().5 3 4 5 Problem 6:.4.7 - Pichfork bifurcaion > f6 := (x,a) -> a*x - x^3; f6 := ( x, a) a x x 3 We plo he hree vecor fields nex o each oher, using he array funcion: p6a := plo(f6(x,-),x=-.5...5,y=-..,ickmarks=[,]): p6b := plo(f6(x,),x=-.5...5,y=-..,ickmarks=[,]): p6c := plo(f6(x,),x=-.5...5,y=-..,ickmarks=[,]): > plos6 := array(..,..3): plos6[,]:=p6a: plos6[,]:=p6b: plos6[,3]:=p6c: display(plos6); y y y x x x

For a, here is a unique fixed poin a x =, which is sable; for a < his is found by linear sabiliy analysis (since f () = a < ), while for a =, linear sabiliy analysis does no prove sabiliy (decay owards he origin is slower han exponenial - see he nex problem), bu a look a he plo of x 3 shows ha he origin is sable. If < a, here are hree fixed poins, a x =, x = a and x = a. Now f () = a is posiive, so he origin is unsable, while he oher wo fixed poins are sable, wih f = a. This is also apparen from he graphs. > Problem 7:.4.9 - Criical slowing down Rese variables: x := x : > eqp7 := diff(x(),) = - x()^3; eqp7 := x( ) = x( ) 3 Find he analyical soluion wih arbirary iniial condiion: xsa := dsolve({eqp7,x()=x},x()) assuming x>; xsb := dsolve({eqp7,x()=x},x()) assuming x<; xsz := dsolve({eqp7,x()=},x()); xsa := x( ) = + x xsb := x( ) = + x xsz := x( ) = > limi(xsa,=infiniy); limi(xsb,=infiniy); lim x( ) = lim x( ) = So he soluions approach zero for arbirary iniial condiions; however, he decay is proporional o, no exponenial. dx We plo he soluions of his equaion and of = x on he same graph: d lineq := diff(x(),) = -x(): linsoln := dsolve({lineq,x()=},x()); crisoln := dsolve({eqp7,x()=},x()); linsoln := x( ) = e ( )

crisoln := x( ) = + > plo([rhs(linsoln),rhs(crisoln)],=..); 8 6 4 4 6 8 dx Noe ha he soluion o = x 3 decays much more rapidly iniially, bu hen slows down once d x <. Problem 8:.5. - Reaching origin in finie ime The origin x = is a sable fixed poin for any real < c. We plo a few represenaive vecor fields: > plo(-x^(/),x=..,ickmarks=[,]); plo(-x^,x=..4,-4...5,ickmarks=[,]); plo(-x^,x=..,y=-4...5,ickmarks=[,]);

We know ha for c =, he decay owards he origin is exponenial, and x approaches asympoically. When < c, he decay is slower han exponenial, as we derived in Problem 7. So he only possibiliy for he soluion o decay o zero in finie ime is for c <. The ime aken from x = o x = is T = in(-/x^c,x=..); x ( c + ) T = lim x + c When < c, he limi diverges; when c =, T is also infinie ( T = lim ln x). When c <, he ime is finie: T = in(-/x^c,x=..) assuming c < ; Problem 9:.5. - Blow-up T = c dy We know ha soluions y( ) of = + y blow up in finie ime. Now for < x, he soluions x( ) d dx of = + x grow more rapidly han y( ), since x < x for < x. Thus he soluions x( ) mus d also blow up in finie ime. This is no ye a complee argumen, hough, since i is only valid for < x; bu since + x, we know ha soluions beginning a any iniial condiion x will reach x = a he laes a ime = x; and since we reach x = in finie ime, we can hen begin he comparison wih y( ). An alernaive argumen: suppose x( ) = x. The ime aken o diverge (reach x = ) is given by T = In(/(+x^),x=x..infiniy);

T = x d + x x If his is finie for all x, hen we have finie-ime blow-up. Bu we have T < In(/(+x^),x=-infiniy..infiniy): so In(/(+x^),x=-..) + *In(/(+x^),x=..infiniy): and inroducing appropriae comparisons, we find T < In(/,x=-..) + *In(/(+x^),x=..infiniy); T < dx + d - + x x Thus an esimae of he upper bound for he blow-up ime for any iniial condiion is > in(,x=-..) + *in(/(+x^),x=..infiniy); evalf(%); (clearly finie) π + 3.5779637 The acual upper bound is in(/(+x^),x=-infiniy..infiniy); evalf(%); 5 sin π.338478 We plo some numerical soluions: DEplo(diff(x(),)=+x()^(),x(),=..3,x=-3..5,[[x()=-.]],sepsize=.,linecolor=black); π 4 x().5.5.5 3