Professional Documents
Culture Documents
Brice Carnahan Ha Luther James o Wilkes Applied Numerical Methods Krieger Pub Co 1990pdf PDF
Brice Carnahan Ha Luther James o Wilkes Applied Numerical Methods Krieger Pub Co 1990pdf PDF
Brice Carnahan Ha Luther James o Wilkes Applied Numerical Methods Krieger Pub Co 1990pdf PDF
las, evaluate both of these integrals for x = 0.8, and compare
with the tabulated value, erf (0.5) = 0.520500.
220 Since (n+ 1) = ni, the Gauss-Laguerre three-point
formula could be used 10 spproximate n!:
* ese des $ at
‘What is the largest integer n for which this formula would be
exact?
221 Modify the function GAUSS of Example 2.3 x0 that
integrals of the form
[ree
are evaluated using a composite formula, equivalent torepeated
application of the m-point Gauss-Legendre quadrature over
nonoverlapping subintervals of {a,b} of length ( — a)/n.
2.22 Write a function, named LAGUER, that employs the
m-point Gauss-Laguerre quadrature formula (2.109) to
evaluate numerically an integral of the form
lever as,
where f(x) is an arbitrary function and aii finite, The function
should incorporate the necessary Laguerze polynomist roots
and the corresponding weight factors (see Table 2.4) for the
-mepoint quadrature where m may be any of 2, 3,4, 5, 6, 10, or
35. Let the argument fist be (A. M,F), where A and M have
‘obvious interpretations, and F is a function that evaluates
FG) for any x.
To test the routine, write a short main program and approp-
riate functions to compute:
() the gamma function
TY)
for «= 1.0, 1,2, 14, 1.6, 1.8, 2.0.
(6) the exponential integral,
for a= 0.5, 1.0, 1.5, 2.0. Ia each case the integral should be
evaluated using the 2, 3,4, 5, 6, 10, and 1$-point quadratures.
Compare the results with tabulated values.
223 Write and test a function, named HERMIT, that
implements the Gauss-Hermite quadrature of (2.115) for
n= 1,2,3,4,9, and 19 (see Table 2.6 and reference (3)), Let
the argument ist be (N, F), where N and F correspond to n and.
Fin 2.115), Use HERMIT to evaluate the integral
e
ne
and compare with the true value, #/[2 sin(x/4)).
224 in a study by Carnahan [17], the fraction fas of
certain fission neutrons having energies above a threshold
‘energy Er, (Mev) was found to be
mo
fn oes | sinb( /2B)e"* dE.
Evaluate fra within 40.001 for Ex =0.5, 2.9, 5.3, and
8. Mev.
‘2.25 The following relation is available (3) for P(x), the
derivative of the mh-degree Legendre polynomial:
(2? — Pia) = mePx) ~ nP ea).
Show that the family of polynomials Pi(x) is orthogonal on the
interval {~1,1] with respect to the weighting function (x? — 1),
‘Show also that a Gaussian quadrature of the form
[i feree= Sms
‘can be developed for the case in witich (wo base points are
reassigned (xo —1, x4=1, that the quadrature is exaet
when f(2) is a polynomial of degree 2n—1 or less, that
Hes Kays Fant Ate the 20708 of Ux), and that the weight
factors are
1p
wen Eyl, MPAA,
Le
wo BRET + PHO de,
1 ot &
GR TF
Note. It can be shown by further manipulation that
2 2
meri ne DIPLO
The above is known as Lobatto quadrature.
226 Show that a Gaussian quadrature,
[iL ferae= Swsoco,
for which x= —1 and the remaining base points are the »
roots of
1 PAD + Pass)
+) l+x
is exact when f(2)is a polynomial of degree 2a or less. Find the
weight factors ws, i= 0,1, ...1m, for n= 1,2,3.
227 A useful quadrature formula, attributed to Chebys
shev, is given by
. 2
[se 4s > Sot,Problems 135,
where the x, are the n+ 1 roots of an (n+ I)th-degree poly-
nomial C,,,(x). The first few of these polynomials are:
It can be shown [26] that the roots of these polynomials all lie
{in (~1,}) and are real, whereas some ofthe roots of C,(x), for
8 and m> 10 are complex. Note that this quadrature
has the attractive feature that all the weight factors are equal,
And the weighting function forthe integral is w(x) = 1.
(@) Show that the polynomials C,(x) are not orthogonal
with respect to the weighting function w(x) = I on the interval
Cun
(©) Find the error term for the quadrature.
() What is the degree of precision of the formula for n even
and n odd?
(@) Show that the two-point formula is identical to the two-
point Gauss-Legendre quadrature of (2.84),
(©) Modify the quadrature to allow est
of the form
ation of integrals
frees,
where @ and 6 are finite,
2.28 Using the roots of the polynomials C,(x) of Problem
2.27 and found by the program developed in Problem 3.32,
Write a function, named CHEB2, with argument list (A. 8, F,
'M, R) that implements the composite version of the quadrature
developed in part (e) of Problem 2.27. Here, A and B are the
lower and upper integration limits, M is the number of points
in the quadrature formula, and R (integer) is the number of
repeated applications of the quadrature. Thus the M-point
quadrature is to be applied to R nonoverlapping subintervals
‘of (AB], each of length (8 — A)/R, F is another function that
evaluates f(x) for any x.
Write main program and accompanying function F to
allow computation of the complete elliptic integral of the
second kind,
Ba) = [7 — asin? 2) a,
‘The main program should read values for ALPHA (a), M, and
R, call upon CHEB to return the estimated value of the
integral, print the results, and return to read another data set,
Since CHEB2 should be a general routine, F should have only
‘one argument (say X). ALPHA should be available to the func-
tion F through a COMMON (or equivalent) declaration,
For a variety of values for M and R, calculate values of
E(e), for «= 0.,0.1, 0.25, and 0.5. Compare your results with
the tabulated vatues [3]:
(0.00) — 1.570796327
E(0.10) = 1.530757637
E(0.25) = 1.467462209
E(0.50) = 1.350643881
2.29 A very light spring of length Z hss Young's motuhos
Eand cross-sectional moment of inertia J; itis rigidly clamped
at its lower end B and is initially vertical (Fig. P2.29). A
downward force P at the free end A causes the spring to bend
‘over. If 8s the angle of slope at any point ané 3s the distance
‘along the spring measured from A, then integration of the
exact governing equation El(dijds)=~Py, noting that
dylds = sin 6, leads to [21}:
——
I@PIETV C68 6 cos
[2 {2°01 = sinterapin* HY
where a is the value of 8 at A
a
and
Figure P229
Show from the above that the Euler load Ps for which the
spring just begins to rend is given by
wEL
aw
‘Let x, and ys denote the vertical distance of 4 above the datum
plane and the horizontal distance of B from A, respectively.
Compute the values of P/P, for which x,/L = 0.99, 0.95, 0.9,
0.5, and 0. What are the corresponding values of ys/L and a?
(Note that the above expression for Landaa related expres-
sion for x/L involve elliptic integrals.)
2.30 A semiinfinite medium (x > 0) has a thermal diffuse
ivity « and a zero initial temperature at time = 0, For 1>0,
the surface at x= 0 is maintained at a temperature 7, ~ T(t).
By using Duhamel’s theorem (see p. 62 of Carstaw and Jaeger136 Numerical Integration
122, for example), the subsequent temperature T(x) inside
the medium can be shown to be given by
ar 1,2 Gar
Te) =
(An alternative form of the integral can be obtained by intro-
ducing a new variable: x = x/(2Vale— )])
Let T, represent the periodic temperature in "F ata point on
the earth’s surface. For example, the mean monthly air tem-
peratures in Table P2.30 have been reported [23] at the loca-
tions indicated.
Table P2.30
Nagpur Cape Royds Yakutsk
January 669 26.1 ~45.0
February ne 208 “354
March a5 49 —na
Apsil 399 199) 140
May 924 55 0
June 864 at 580
July 804 9 67
‘August 803 157 50
September 80.2 =57 2.0
October 783 45 165
November 23 119 21
December 67 300 $12
Compute the likely mean monthly ground temperatures at
5, 10, 20, and 50 feet below the earth’s surface at each of the
above locations. Plot these computed temperatures ¢o show
their relation to the corresponding surface temperatures. In
each case, assume: (a) dry ground with «= 0.0926 5q ft/hr,
(b) the mean monthly ground and air temperatures at the
surface are approximately equal, and (¢) the pattern of air
temperature repeats itself indefinitely from one year to the
ext
231 Surpose that (m+ 1)(n-+1) functional values
109) are ‘ailable forall combinations of m-+1 levels of
AFRO oot and nt T fevels Of 3)f= 051, vgn
Define Laerangian interpolation coefficients %,,(8), 0,1,
man YsA)),j-=0, 1, sm, asin Problem 1.38. Let the
integral in she rectangular domain a 1 in the region of interest. Generally, witen x, is
close to a, the approximate relation
af = (F(x2) — Fall < plxz — al < p|x, — at.
Fax; — 9)
(3.24)
holds true; F(a) is called the asymptotic convergence
‘factor.
‘An example is furnished by the equation cos x — x =0
1
Further discussion of this and other iterative methods
can be found in Traub [14].
Example. Consider the equation
{O)= 2 3x +1=0,
oh
Figure 3.2 Roots of fx)
+136 Ward's Method
‘The function f(x) is illustrated in Fig. 3.2; by inspection,
2>a,> 1,1 >a >0,and ~1 >a, > —2. Consider first the
version
dui = 509 +0, 6.260)
Corresponding to (3.260), F(x)=x%, and [F'(x)| <1 if
[x] <1. Consequently, (3.26a), if used to define the iterative
process, can be expected to yield as, but not to yield ay or
as If, however, in (3.25) we use k= 4, so that
31,
xere5 ZPD, (6.266)
then F(x) becomes (3/2)x— (x? + 1/6 and F’(x)= 3/2 —
27/2, In particular, for 1<|x|You might also like