Next: , Previous:   [Contents][Index]

19 Integration


19.1 Introduction to Integration

Maxima has several routines for handling integration. The integrate function makes use of most of them. There is also the antid package, which handles an unspecified function (and its derivatives, of course). For numerical uses, there is a set of adaptive integrators from QUADPACK, named quad_qag, quad_qags, etc., which are described under the heading QUADPACK. Hypergeometric functions are being worked on, see specint for details. Generally speaking, Maxima only handles integrals which are integrable in terms of the "elementary functions" (rational functions, trigonometrics, logs, exponentials, radicals, etc.) and a few extensions (error function, dilogarithm). It does not handle integrals in terms of unknown functions such as g(x) and h(x).


19.2 Functions and Variables for Integration

Function: changevar (expr, f(x,y), y, x)

Makes the change of variable given by f(x,y) = 0 in all integrals occurring in expr with integration with respect to x. The new variable is y.

The change of variable can also be written f(x) = g(y).

(%i1) assume(a > 0)$
(%i2) 'integrate (%e**sqrt(a*y), y, 0, 4);
                      4
                     /
                     [    sqrt(a) sqrt(y)
(%o2)                I  %e                dy
                     ]
                     /
                      0
(%i3) changevar (%, y-z^2/a, z, y);
                      0
                     /
                     [                abs(z)
                   2 I            z %e       dz
                     ]
                     /
                      - 2 sqrt(a)
(%o3)            - ----------------------------
                                a

An expression containing a noun form, such as the instances of 'integrate above, may be evaluated by ev with the nouns flag. For example, the expression returned by changevar above may be evaluated by ev (%o3, nouns).

changevar may also be used to make changes in the indices of a sum or product. However, it must be realized that when a change is made in a sum or product, this change must be a shift, i.e., i = j+ ..., not a higher degree function. E.g.,

(%i4) sum (a[i]*x^(i-2), i, 0, inf);
                         inf
                         ====
                         \         i - 2
(%o4)                     >    a  x
                         /      i
                         ====
                         i = 0
(%i5) changevar (%, i-2-n, n, i);
                        inf
                        ====
                        \               n
(%o5)                    >      a      x
                        /        n + 2
                        ====
                        n = - 2
Categories: Integral calculus ·
Function: dblint (f, r, s, a, b)

A double-integral routine which was written in top-level Maxima and then translated and compiled to machine code. Use load ("dblint") to access this package. It uses the Simpson’s rule method in both the x and y directions to calculate

/b /s(x)
|  |
|  |    f(x,y) dy dx
|  |
/a /r(x)

The function f must be a translated or compiled function of two variables, and r and s must each be a translated or compiled function of one variable, while a and b must be floating point numbers. The routine has two global variables which determine the number of divisions of the x and y intervals: dblint_x and dblint_y, both of which are initially 10, and can be changed independently to other integer values (there are 2*dblint_x+1 points computed in the x direction, and 2*dblint_y+1 in the y direction). The routine subdivides the X axis and then for each value of X it first computes r(x) and s(x); then the Y axis between r(x) and s(x) is subdivided and the integral along the Y axis is performed using Simpson’s rule; then the integral along the X axis is done using Simpson’s rule with the function values being the Y-integrals. This procedure may be numerically unstable for a great variety of reasons, but is reasonably fast: avoid using it on highly oscillatory functions and functions with singularities (poles or branch points in the region). The Y integrals depend on how far apart r(x) and s(x) are, so if the distance s(x) - r(x) varies rapidly with X, there may be substantial errors arising from truncation with different step-sizes in the various Y integrals. One can increase dblint_x and dblint_y in an effort to improve the coverage of the region, at the expense of computation time. The function values are not saved, so if the function is very time-consuming, you will have to wait for re-computation if you change anything (sorry). It is required that the functions f, r, and s be either translated or compiled prior to calling dblint. This will result in orders of magnitude speed improvement over interpreted code in many cases!

demo ("dblint") executes a demonstration of dblint applied to an example problem.

Categories: Integral calculus ·
Function: defint (expr, x, a, b)

Attempts to compute a definite integral. defint is called by integrate when limits of integration are specified, i.e., when integrate is called as integrate (expr, x, a, b). Thus from the user’s point of view, it is sufficient to call integrate.

defint returns a symbolic expression, either the computed integral or the noun form of the integral. See quad_qag and related functions for numerical approximation of definite integrals.

Categories: Integral calculus ·
Option variable: erfflag

Default value: true

When erfflag is false, prevents risch from introducing the erf function in the answer if there were none in the integrand to begin with.

Categories: Integral calculus ·
Function: ilt (expr, s, t)

Computes the inverse Laplace transform of expr with respect to s and parameter t. expr must be a ratio of polynomials whose denominator has only linear and quadratic factors; there is an extension of ilt, called pwilt (Piece-Wise Inverse Laplace Transform) that handles several other cases where ilt fails.

By using the functions laplace and ilt together with the solve or linsolve functions the user can solve a single differential or convolution integral equation or a set of them.

(%i1) 'integrate (sinh(a*x)*f(t-x), x, 0, t) + b*f(t) = t**2;
              t
             /
             [                                    2
(%o1)        I  f(t - x) sinh(a x) dx + b f(t) = t
             ]
             /
              0
(%i2) laplace (%, t, s);
                               a laplace(f(t), t, s)   2
(%o2)  b laplace(f(t), t, s) + --------------------- = --
                                       2    2           3
                                      s  - a           s
(%i3) linsolve ([%], ['laplace(f(t), t, s)]);
                                        2      2
                                     2 s  - 2 a
(%o3)     [laplace(f(t), t, s) = --------------------]
                                    5         2     3
                                 b s  + (a - a  b) s
(%i4) ilt (rhs (first (%)), s, t);
Is  a b (a b - 1)  positive, negative, or zero?

pos;
               sqrt(a b (a b - 1)) t
        2 cosh(---------------------)       2
                         b               a t
(%o4) - ----------------------------- + -------
              3  2      2               a b - 1
             a  b  - 2 a  b + a

                                                       2
                                             + ------------------
                                                3  2      2
                                               a  b  - 2 a  b + a
Categories: Laplace transform ·
Option variable: intanalysis

Default value: true

When true, definite integration tries to find poles in the integrand in the interval of integration. If there are, then the integral is evaluated appropriately as a principal value integral. If intanalysis is false, this check is not performed and integration is done assuming there are no poles.

See also ldefint.

Examples:

Maxima can solve the following integrals, when intanalysis is set to false:

(%i1) integrate(1/(sqrt(x)+1),x,0,1);
                                1
                               /
                               [       1
(%o1)                          I  ----------- dx
                               ]  sqrt(x) + 1
                               /
                                0

(%i2) integrate(1/(sqrt(x)+1),x,0,1),intanalysis:false;
(%o2)                            2 - 2 log(2)

(%i3) integrate(cos(a)/sqrt((tan(a))^2 +1),a,-%pi/2,%pi/2);
The number 1 isn't in the domain of atanh
 -- an error. To debug this try: debugmode(true);

(%i4) intanalysis:false$
(%i5) integrate(cos(a)/sqrt((tan(a))^2+1),a,-%pi/2,%pi/2);
                                      %pi
(%o5)                                 ---
                                       2
Categories: Integral calculus ·
Function: integrate
    integrate (expr, x)
    integrate (expr, x, a, b)

Attempts to symbolically compute the integral of expr with respect to x. integrate (expr, x) is an indefinite integral, while integrate (expr, x, a, b) is a definite integral, with limits of integration a and b. The limits should not contain x, although integrate does not enforce this restriction. a need not be less than b. If b is equal to a, integrate returns zero.

See quad_qag and related functions for numerical approximation of definite integrals. See residue for computation of residues (complex integration). See antid for an alternative means of computing indefinite integrals.

The integral (an expression free of integrate) is returned if integrate succeeds. Otherwise the return value is the noun form of the integral (the quoted operator 'integrate) or an expression containing one or more noun forms. The noun form of integrate is displayed with an integral sign.

In some circumstances it is useful to construct a noun form by hand, by quoting integrate with a single quote, e.g., 'integrate (expr, x). For example, the integral may depend on some parameters which are not yet computed. The noun may be applied to its arguments by ev (i, nouns) where i is the noun form of interest.

integrate handles definite integrals separately from indefinite, and employs a range of heuristics to handle each case. Special cases of definite integrals include limits of integration equal to zero or infinity (inf or minf), trigonometric functions with limits of integration equal to zero and %pi or 2 %pi, rational functions, integrals related to the definitions of the beta and psi functions, and some logarithmic and trigonometric integrals. Processing rational functions may include computation of residues. If an applicable special case is not found, an attempt will be made to compute the indefinite integral and evaluate it at the limits of integration. This may include taking a limit as a limit of integration goes to infinity or negative infinity; see also ldefint.

Special cases of indefinite integrals include trigonometric functions, exponential and logarithmic functions, and rational functions. integrate may also make use of a short table of elementary integrals.

integrate may carry out a change of variable if the integrand has the form f(g(x)) * diff(g(x), x). integrate attempts to find a subexpression g(x) such that the derivative of g(x) divides the integrand. This search may make use of derivatives defined by the gradef function. See also changevar and antid.

If none of the preceding heuristics find the indefinite integral, the Risch algorithm is executed. The flag risch may be set as an evflag, in a call to ev or on the command line, e.g., ev (integrate (expr, x), risch) or integrate (expr, x), risch. If risch is present, integrate calls the risch function without attempting heuristics first. See also risch.

integrate works only with functional relations represented explicitly with the f(x) notation. integrate does not respect implicit dependencies established by the depends function.

integrate may need to know some property of a parameter in the integrand. integrate will first consult the assume database, and, if the variable of interest is not there, integrate will ask the user. Depending on the question, suitable responses are yes; or no;, or pos;, zero;, or neg;.

integrate is not, by default, declared to be linear. See declare and linear.

integrate attempts integration by parts only in a few special cases.

Examples:

  • Elementary indefinite and definite integrals.
    (%i1) integrate (sin(x)^3, x);
                               3
                            cos (x)
    (%o1)                   ------- - cos(x)
                               3
    
    (%i2) integrate (x/ sqrt (b^2 - x^2), x);
                                     2    2
    (%o2)                    - sqrt(b  - x )
    
    (%i3) integrate (cos(x)^2 * exp(x), x, 0, %pi);
                                   %pi
                               3 %e      3
    (%o3)                      ------- - -
                                  5      5
    
    (%i4) integrate (x^2 * exp(-x^2), x, minf, inf);
                                sqrt(%pi)
    (%o4)                       ---------
                                    2
    
  • Use of assume and interactive query.
    (%i1) assume (a > 1)$
    
    (%i2) integrate (x**a/(x+1)**(5/2), x, 0, inf);
        2 a + 2
    Is  -------  an integer?
           5
    
    no;
    Is  2 a - 3  positive, negative, or zero?
    
    neg;
                                       3
    (%o2)                  beta(a + 1, - - a)
                                       2
    
  • Change of variable. There are two changes of variable in this example: one using a derivative established by gradef, and one using the derivation diff(r(x)) of an unspecified function r(x).
    (%i3) gradef (q(x), sin(x**2));
    (%o3)                         q(x)
    
    (%i4) diff (log (q (r (x))), x);
                          d               2
                         (-- (r(x))) sin(r (x))
                          dx
    (%o4)                ----------------------
                                q(r(x))
    
    (%i5) integrate (%, x);
    (%o5)                     log(q(r(x)))
    
  • Return value contains the 'integrate noun form. In this example, Maxima can extract one factor of the denominator of a rational function, but cannot factor the remainder or otherwise find its integral. grind shows the noun form 'integrate in the result. See also integrate_use_rootsof for more on integrals of rational functions.
    (%i1) expand ((x-4) * (x^3+2*x+1));
                        4      3      2
    (%o1)              x  - 4 x  + 2 x  - 7 x - 4
    
    (%i2) integrate (1/%, x);
                                  /  2
                                  [ x  + 4 x + 18
                                  I ------------- dx
                                  ]  3
                     log(x - 4)   / x  + 2 x + 1
    (%o2)            ---------- - ------------------
                         73               73
    
    (%i3) grind (%);
    log(x-4)/73-('integrate((x^2+4*x+18)/(x^3+2*x+1),x))/73$
    
  • Defining a function in terms of an integral. The body of a function is not evaluated when the function is defined. Thus the body of f_1 in this example contains the noun form of integrate. The quote-quote operator '' causes the integral to be evaluated, and the result becomes the body of f_2.
    (%i1) f_1 (a) := integrate (x^3, x, 1, a);
                                         3
    (%o1)           f_1(a) := integrate(x , x, 1, a)
    
    (%i2) ev (f_1 (7), nouns);
    (%o2)                          600
    
    (%i3) /* Note parentheses around integrate(...) here */
          f_2 (a) := ''(integrate (x^3, x, 1, a));
                                       4
                                      a    1
    (%o3)                   f_2(a) := -- - -
                                      4    4
    
    (%i4) f_2 (7);
    (%o4)                          600
    
Categories: Integral calculus ·
System variable: integration_constant

Default value: %c

When a constant of integration is introduced by indefinite integration of an equation, the name of the constant is constructed by concatenating integration_constant and integration_constant_counter.

integration_constant may be assigned any symbol.

Examples:

(%i1) integrate (x^2 = 1, x);
                           3
                          x
(%o1)                     -- = x + %c1
                          3
(%i2) integration_constant : 'k;
(%o2)                           k
(%i3) integrate (x^2 = 1, x);
                            3
                           x
(%o3)                      -- = x + k2
                           3
Categories: Integral calculus ·
System variable: integration_constant_counter

Default value: 0

When a constant of integration is introduced by indefinite integration of an equation, the name of the constant is constructed by concatenating integration_constant and integration_constant_counter.

integration_constant_counter is incremented before constructing the next integration constant.

Examples:

(%i1) integrate (x^2 = 1, x);
                           3
                          x
(%o1)                     -- = x + %c1
                          3
(%i2) integrate (x^2 = 1, x);
                           3
                          x
(%o2)                     -- = x + %c2
                          3
(%i3) integrate (x^2 = 1, x);
                           3
                          x
(%o3)                     -- = x + %c3
                          3
(%i4) reset (integration_constant_counter);
(%o4)            [integration_constant_counter]
(%i5) integrate (x^2 = 1, x);
                           3
                          x
(%o5)                     -- = x + %c1
                          3
Categories: Integral calculus ·
Option variable: integrate_use_rootsof

Default value: false

When integrate_use_rootsof is true and the denominator of a rational function cannot be factored, integrate returns the integral in a form which is a sum over the roots (not yet known) of the denominator.

For example, with integrate_use_rootsof set to false, integrate returns an unsolved integral of a rational function in noun form:

(%i1) integrate_use_rootsof: false$
(%i2) integrate (1/(1+x+x^5), x);
        /  2
        [ x  - 4 x + 5
        I ------------ dx                            2 x + 1
        ]  3    2                2            5 atan(-------)
        / x  - x  + 1       log(x  + x + 1)          sqrt(3)
(%o2)   ----------------- - --------------- + ---------------
                7                 14             7 sqrt(3)

Now we set the flag to be true and the unsolved part of the integral will be expressed as a summation over the roots of the denominator of the rational function:

(%i3) integrate_use_rootsof: true$
(%i4) integrate (1/(1+x+x^5), x);
      ====        2
      \       (%r4  - 4 %r4 + 5) log(x - %r4)
       >      -------------------------------
      /                    2
      ====            3 %r4  - 2 %r4
                        3      2
      %r4 in rootsof(%r4  - %r4  + 1, %r4)
(%o4) ----------------------------------------------------------
               7

                                                      2 x + 1
                                  2            5 atan(-------)
                             log(x  + x + 1)          sqrt(3)
                           - --------------- + ---------------
                                   14             7 sqrt(3)

Alternatively the user may compute the roots of the denominator separately, and then express the integrand in terms of these roots, e.g., 1/((x - a)*(x - b)*(x - c)) or 1/((x^2 - (a+b)*x + a*b)*(x - c)) if the denominator is a cubic polynomial. Sometimes this will help Maxima obtain a more useful result.

Categories: Integral calculus ·
Function: laplace (expr, t, s)

Attempts to compute the Laplace transform of expr with respect to the variable t and transform parameter s. The Laplace transform of the function f(t) is the one-sided transform defined by $$ F(s) = \int_0^{\infty} f(t) e^{-st} dt $$

where F(s) is the transform of f(t), represented by expr.

laplace recognizes in expr the functions delta, exp, log, sin, cos, sinh, cosh, and erf, as well as derivative, integrate, sum, and ilt. If laplace fails to find a transform the function specint is called. specint can find the laplace transform for expressions with special functions like the bessel functions bessel_j, bessel_i, … and can handle the unit_step function. See also specint.

If specint cannot find a solution too, a noun laplace is returned.

expr may also be a linear, constant coefficient differential equation in which case atvalue of the dependent variable is used. The required atvalue may be supplied either before or after the transform is computed. Since the initial conditions must be specified at zero, if one has boundary conditions imposed elsewhere he can impose these on the general solution and eliminate the constants by solving the general solution for them and substituting their values back.

laplace recognizes convolution integrals of the form $$ \int_0^t f(x) g(t-x) dx $$

Other kinds of convolutions are not recognized.

Functional relations must be explicitly represented in expr; implicit relations, established by depends, are not recognized. That is, if f depends on x and y, f (x, y) must appear in expr.

See also ilt, the inverse Laplace transform.

Examples:

(%i1) laplace (exp (2*t + a) * sin(t) * t, t, s);
                            a
                          %e  (2 s - 4)
(%o1)                    ---------------
                           2           2
                         (s  - 4 s + 5)
(%i2) laplace ('diff (f (x), x), x, s);
(%o2)             s laplace(f(x), x, s) - f(0)
(%i3) diff (diff (delta (t), t), t);
                          2
                         d
(%o3)                    --- (delta(t))
                           2
                         dt
(%i4) laplace (%, t, s);
                            !
               d            !         2
(%o4)        - -- (delta(t))!      + s  - delta(0) s
               dt           !
                            !t = 0
(%i5) assume(a>0)$
(%i6) laplace(gamma_incomplete(a,t),t,s),gamma_expand:true;
                                              - a - 1
                         gamma(a)   gamma(a) s
(%o6)                    -------- - -----------------
                            s            1     a
                                        (- + 1)
                                         s
(%i7) factor(laplace(gamma_incomplete(1/2,t),t,s));
                                              s + 1
                      sqrt(%pi) (sqrt(s) sqrt(-----) - 1)
                                                s
(%o7)                 -----------------------------------
                                3/2      s + 1
                               s    sqrt(-----)
                                           s
(%i8) assume(exp(%pi*s)>1)$
(%i9) laplace(sum((-1)^n*unit_step(t-n*%pi)*sin(t),n,0,inf),t,s),
         simpsum;
                         %i                         %i
              ------------------------ - ------------------------
                              - %pi s                    - %pi s
              (s + %i) (1 - %e       )   (s - %i) (1 - %e       )
(%o9)         ---------------------------------------------------
                                       2
(%i9) factor(%);
                                      %pi s
                                    %e
(%o9)                   -------------------------------
                                             %pi s
                        (s - %i) (s + %i) (%e      - 1)

Function: ldefint (expr, x, a, b)

Attempts to compute the definite integral of expr by using limit to evaluate the indefinite integral of expr with respect to x at the upper limit b and at the lower limit a. If it fails to compute the definite integral, ldefint returns an expression containing limits as noun forms.

ldefint is not called from integrate, so executing ldefint (expr, x, a, b) may yield a different result than integrate (expr, x, a, b). ldefint always uses the same method to evaluate the definite integral, while integrate may employ various heuristics and may recognize some special cases.

Categories: Integral calculus ·
Function: pwilt (expr, s, t)

Computes the inverse Laplace transform of expr with respect to s and parameter t. Unlike ilt, pwilt is able to return piece-wise and periodic functions and can also handle some cases with polynomials of degree greater than 3 in the denominator.

Two examples where ilt fails:

(%i1) pwilt (exp(-s)*s/(s^3-2*s-s+2), s, t);
                                       t - 1       - 2 (t - 1)
                             (t - 1) %e        2 %e
(%o1)         hstep(t - 1) (--------------- - ---------------)
                                    3                 9
                                    
(%i2) pwilt ((s^2+2)/(s^2-1), s, t);
                                         t       - t
                                     3 %e    3 %e
(%o2)                    delta(t) + ----- - -------
                                       2        2
Categories: Laplace transform ·
Function: potential (givengradient)

The calculation makes use of the global variable potentialzeroloc[0] which must be nonlist or of the form

[indeterminatej=expressionj, indeterminatek=expressionk, ...]

the former being equivalent to the nonlist expression for all right-hand sides in the latter. The indicated right-hand sides are used as the lower limit of integration. The success of the integrations may depend upon their values and order. potentialzeroloc is initially set to 0.

Option variable: prefer_d

Default value: false

When prefer_d is true, specint will prefer to express solutions using parabolic_cylinder_d rather than hypergeometric functions.

In the example below, the solution contains parabolic_cylinder_d when prefer_d is true.

(%i1) assume(s>0);
(%o1)                               [s > 0]
(%i2) factor(specint(ex:%e^-(t^2/8)*exp(-s*t),t));
                                         2
                                      2 s
(%o2)           - sqrt(2) sqrt(%pi) %e     (erf(sqrt(2) s) - 1)
(%i3) specint(ex,t),prefer_d=true;
                                                          2
                                                         s
                                                         --
                                                 s       8
                    parabolic_cylinder_d(- 1, -------) %e
                                              sqrt(2)
(%o3)               ---------------------------------------
                                    sqrt(2)

Function: residue (expr, z, z_0)

Computes the residue in the complex plane of the expression expr when the variable z assumes the value z_0. The residue is the coefficient of (z - z_0)^(-1) in the Laurent series for expr.

(%i1) residue (s/(s**2+a**2), s, a*%i);
                                1
(%o1)                           -
                                2
(%i2) residue (sin(a*x)/x**4, x, 0);
                                 3
                                a
(%o2)                         - --
                                6
Function: risch (expr, x)

Integrates expr with respect to x using the transcendental case of the Risch algorithm. (The algebraic case of the Risch algorithm has not been implemented.) This currently handles the cases of nested exponentials and logarithms which the main part of integrate can’t do. integrate will automatically apply risch if given these cases.

erfflag, if false, prevents risch from introducing the erf function in the answer if there were none in the integrand to begin with.

(%i1) risch (x^2*erf(x), x);
                                                        2
             3                      2                - x
        %pi x  erf(x) + (sqrt(%pi) x  + sqrt(%pi)) %e
(%o1)   -------------------------------------------------
                              3 %pi
(%i2) diff(%, x), ratsimp;
                             2
(%o2)                       x  erf(x)
Categories: Integral calculus ·
Function: specint (exp(- s*t) * expr, t)

Compute the Laplace transform of expr with respect to the variable t. The integrand expr may contain special functions. The parameter s maybe be named something else; it is determined automatically, as the examples below show where p is used in some places.

The following special functions are handled by specint: incomplete gamma function, error functions (but not the error function erfi, it is easy to transform erfi e.g. to the error function erf), exponential integrals, bessel functions (including products of bessel functions), hankel functions, hermite and the laguerre polynomials.

Furthermore, specint can handle the hypergeometric function %f[p,q]([],[],z), the Whittaker function of the first kind %m[u,k](z) and of the second kind %w[u,k](z).

The result may be in terms of special functions and can include unsimplified hypergeometric functions. If variable prefer_d is true then the parabolic_cylinder_d function may be used in the result in preference to hypergeometric functions.

When laplace fails to find a Laplace transform, specint is called. Because laplace knows more general rules for Laplace transforms, it is preferable to use laplace and not specint.

demo("hypgeo") displays several examples of Laplace transforms computed by specint.

Examples:

(%i1) assume (p > 0, a > 0)$
(%i2) specint (t^(1/2) * exp(-a*t/4) * exp(-p*t), t);
                           sqrt(%pi)
(%o2)                     ------------
                                 a 3/2
                          2 (p + -)
                                 4
(%i3) specint (t^(1/2) * bessel_j(1, 2 * a^(1/2) * t^(1/2))
              * exp(-p*t), t);
                                   - a/p
                         sqrt(a) %e
(%o3)                    ---------------
                                2
                               p

Examples for exponential integrals:

(%i4) assume(s>0,a>0,s-a>0)$
(%i5) ratsimp(specint(%e^(a*t)
                      *(log(a)+expintegral_e1(a*t))*%e^(-s*t),t));
                             log(s)
(%o5)                        ------
                             s - a
(%i6) logarc:true$

(%i7) gamma_expand:true$

radcan(specint((cos(t)*expintegral_si(t)
                     -sin(t)*expintegral_ci(t))*%e^(-s*t),t));
                             log(s)
(%o8)                        ------
                              2
                             s  + 1
ratsimp(specint((2*t*log(a)+2/a*sin(a*t)
                      -2*t*expintegral_ci(a*t))*%e^(-s*t),t));
                               2    2
                          log(s  + a )
(%o9)                     ------------
                                2
                               s

Results when using the expansion of gamma_incomplete and when changing the representation to expintegral_e1:

(%i10) assume(s>0)$
(%i11) specint(1/sqrt(%pi*t)*unit_step(t-k)*%e^(-s*t),t);
                                            1
                            gamma_incomplete(-, k s)
                                            2
(%o11)                      ------------------------
                               sqrt(%pi) sqrt(s)

(%i12) gamma_expand:true$
(%i13) specint(1/sqrt(%pi*t)*unit_step(t-k)*%e^(-s*t),t);
                              erfc(sqrt(k) sqrt(s))
(%o13)                        ---------------------
                                     sqrt(s)

(%i14) expintrep:expintegral_e1$
(%i15) ratsimp(specint(1/(t+a)^2*%e^(-s*t),t));
                              a s
                        a s %e    expintegral_e1(a s) - 1
(%o15)                - ---------------------------------
                                        a
Categories: Laplace transform ·
Function: tldefint (expr, x, a, b)

Equivalent to ldefint with tlimswitch set to true.

Categories: Integral calculus ·

19.3 Introduction to QUADPACK

QUADPACK is a collection of functions for the numerical computation of one-dimensional definite integrals. It originated from a joint project of R. Piessens 1, E. de Doncker 2, C. Ueberhuber 3, and D. Kahaner 4.

The QUADPACK library included in Maxima is an automatic translation (via the program f2cl) of the Fortran source code of QUADPACK as it appears in the SLATEC Common Mathematical Library, Version 4.1 5. The SLATEC library is dated July 1993, but the QUADPACK functions were written some years before. There is another version of QUADPACK at Netlib 6; it is not clear how that version differs from the SLATEC version.

The QUADPACK functions included in Maxima are all automatic, in the sense that these functions attempt to compute a result to a specified accuracy, requiring an unspecified number of function evaluations. Maxima’s Lisp translation of QUADPACK also includes some non-automatic functions, but they are not exposed at the Maxima level.

Further information about QUADPACK can be found in the QUADPACK book 7.

19.3.1 Overview

quad_qag

Integration of a general function over a finite interval. quad_qag implements a simple globally adaptive integrator using the strategy of Aind (Piessens, 1973). The caller may choose among 6 pairs of Gauss-Kronrod quadrature formulae for the rule evaluation component. The high-degree rules are suitable for strongly oscillating integrands.

quad_qags

Integration of a general function over a finite interval. quad_qags implements globally adaptive interval subdivision with extrapolation (de Doncker, 1978) by the Epsilon algorithm (Wynn, 1956).

quad_qagi

Integration of a general function over an infinite or semi-infinite interval. The interval is mapped onto a finite interval and then the same strategy as in quad_qags is applied.

quad_qawo

Integration of \(\cos(\omega x) f(x)\) or \(\sin(\omega x) f(x)\) over a finite interval, where \(\omega\) is a constant. The rule evaluation component is based on the modified Clenshaw-Curtis technique. quad_qawo applies adaptive subdivision with extrapolation, similar to quad_qags.

quad_qawf

Calculates a Fourier cosine or Fourier sine transform on a semi-infinite interval. The same approach as in quad_qawo is applied on successive finite intervals, and convergence acceleration by means of the Epsilon algorithm (Wynn, 1956) is applied to the series of the integral contributions.

quad_qaws

Integration of \(w(x)f(x)\) over a finite interval [a, b], where w is a function of the form \((x-a)^\alpha (b-x)^\beta v(x)\) and v(x) is 1 or \(\log(x-a)\) or \(\log(b-x)\) or \(\log(x-a)\log(b-x),\) and \(\alpha > -1\) and \(\beta > -1.\)

A globally adaptive subdivision strategy is applied, with modified Clenshaw-Curtis integration on the subintervals which contain a or b.

quad_qawc

Computes the Cauchy principal value of f(x)/(x - c) over a finite interval (a, b) and specified c. The strategy is globally adaptive, and modified Clenshaw-Curtis integration is used on the subranges which contain the point x = c.

quad_qagp

Basically the same as quad_qags but points of singularity or discontinuity of the integrand must be supplied. This makes it easier for the integrator to produce a good solution.


19.4 Functions and Variables for QUADPACK

Function: quad_qag
    quad_qag (f(x), x, a, b, key, [epsrel, epsabs, limit])
    quad_qag (f, x, a, b, key, [epsrel, epsabs, limit])

Integration of a general function over a finite interval. quad_qag implements a simple globally adaptive integrator using the strategy of Aind (Piessens, 1973). The caller may choose among 6 pairs of Gauss-Kronrod quadrature formulae for the rule evaluation component. The high-degree rules are suitable for strongly oscillating integrands.

quad_qag computes the integral

$$ \int_a^b f(x)\, dx $$

The function to be integrated is f(x), with dependent variable x, and the function is to be integrated between the limits a and b. key is the integrator to be used and should be an integer between 1 and 6, inclusive. The value of key selects the order of the Gauss-Kronrod integration rule. High-order rules are suitable for strongly oscillating integrands.

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The numerical integration is done adaptively by subdividing the integration region into sub-intervals until the desired accuracy is achieved.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit is the maximum number of subintervals to use. Default is 200.

quad_qag returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

if no problems were encountered;

1

if too many sub-intervals were done;

2

if excessive roundoff error is detected;

3

if extremely bad integrand behavior occurs;

6

if the input is invalid.

Examples:

(%i1) quad_qag (x^(1/2)*log(1/x), x, 0, 1, 3, 'epsrel=5d-8);
(%o1)    [.4444444444492108, 3.1700968502883E-9, 961, 0]
(%i2) integrate (x^(1/2)*log(1/x), x, 0, 1);
                                4
(%o2)                           -
                                9
Function: quad_qags
    quad_qags (f(x), x, a, b, [epsrel, epsabs, limit])
    quad_qags (f, x, a, b, [epsrel, epsabs, limit])

Integration of a general function over a finite interval. quad_qags implements globally adaptive interval subdivision with extrapolation (de Doncker, 1978) by the Epsilon algorithm (Wynn, 1956).

quad_qags computes the integral

$$ \int_a^b f(x)\, dx $$

The function to be integrated is f(x), with dependent variable x, and the function is to be integrated between the limits a and b.

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit is the maximum number of subintervals to use. Default is 200.

quad_qags returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

4

failed to converge

5

integral is probably divergent or slowly convergent

6

if the input is invalid.

Examples:

(%i1) quad_qags (x^(1/2)*log(1/x), x, 0, 1, 'epsrel=1d-10);
(%o1)   [.4444444444444448, 1.11022302462516E-15, 315, 0]

Note that quad_qags is more accurate and efficient than quad_qag for this integrand.

Function: quad_qagi
    quad_qagi (f(x), x, a, b, [epsrel, epsabs, limit])
    quad_qagi (f, x, a, b, [epsrel, epsabs, limit])

Integration of a general function over an infinite or semi-infinite interval. The interval is mapped onto a finite interval and then the same strategy as in quad_qags is applied.

quad_qagi evaluates one of the following integrals

$$ \int_a^\infty f(x) \, dx $$ $$ \int_\infty^a f(x) \, dx $$ $$ \int_{-\infty}^\infty f(x) \, dx $$

using the Quadpack QAGI routine. The function to be integrated is f(x), with dependent variable x, and the function is to be integrated over an infinite range.

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

One of the limits of integration must be infinity. If not, then quad_qagi will just return the noun form.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit is the maximum number of subintervals to use. Default is 200.

quad_qagi returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

4

failed to converge

5

integral is probably divergent or slowly convergent

6

if the input is invalid.

Examples:

(%i1) quad_qagi (x^2*exp(-4*x), x, 0, inf, 'epsrel=1d-8);
(%o1)        [0.03125, 2.95916102995002E-11, 105, 0]
(%i2) integrate (x^2*exp(-4*x), x, 0, inf);
                               1
(%o2)                          --
                               32
Function: quad_qawc
    quad_qawc (f(x), x, c, a, b, [epsrel, epsabs, limit])
    quad_qawc (f, x, c, a, b, [epsrel, epsabs, limit])

Computes the Cauchy principal value of f(x)/(x - c) over a finite interval. The strategy is globally adaptive, and modified Clenshaw-Curtis integration is used on the subranges which contain the point x = c.

quad_qawc computes the Cauchy principal value of

$$ \int_{a}^{b}{{{f\left(x\right)}\over{x-c}}\>dx} $$

using the Quadpack QAWC routine. The function to be integrated is f(x)/(x-c), with dependent variable x, and the function is to be integrated over the interval a to b.

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit is the maximum number of subintervals to use. Default is 200.

quad_qawc returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

6

if the input is invalid.

Examples:

(%i1) quad_qawc (2^(-5)*((x-1)^2+4^(-5))^(-1), x, 2, 0, 5,
                 'epsrel=1d-7);
(%o1)    [- 3.130120337415925, 1.306830140249558E-8, 495, 0]
(%i2) integrate (2^(-alpha)*(((x-1)^2 + 4^(-alpha))*(x-2))^(-1),
      x, 0, 5);
Principal Value
                       alpha
        alpha       9 4                 9
       4      log(------------- + -------------)
                      alpha           alpha
                  64 4      + 4   64 4      + 4
(%o2) (-----------------------------------------
                        alpha
                     2 4      + 2

       3 alpha                       3 alpha
       -------                       -------
          2            alpha/2          2          alpha/2
    2 4        atan(4 4       )   2 4        atan(4       )   alpha
  - --------------------------- - -------------------------)/2
              alpha                        alpha
           2 4      + 2                 2 4      + 2
(%i3) ev (%, alpha=5, numer);
(%o3)                    - 3.130120337415917
Function: quad_qawf
    quad_qawf (f(x), x, a, omega, trig, [epsabs, limit, maxp1, limlst])
    quad_qawf (f, x, a, omega, trig, [epsabs, limit, maxp1, limlst])

Calculates a Fourier cosine or Fourier sine transform on a semi-infinite interval using the Quadpack QAWF function. The same approach as in quad_qawo is applied on successive finite intervals, and convergence acceleration by means of the Epsilon algorithm (Wynn, 1956) is applied to the series of the integral contributions.

quad_qawf computes the integral

$$ \int_a^\infty f(x) \, w(x) \, dx $$

The weight function w is selected by trig:

cos
\(w(x) = \cos\omega x\)
sin
\(w(x) = \sin\omega x\)

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsabs

Desired absolute error of approximation. Default is 1d-10.

limit

Size of internal work array. (limit - limlst)/2 is the maximum number of subintervals to use. Default is 200.

maxp1

Maximum number of Chebyshev moments. Must be greater than 0. Default is 100.

limlst

Upper bound on the number of cycles. Must be greater than or equal to 3. Default is 10.

quad_qawf returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

6

if the input is invalid.

Examples:

(%i1) quad_qawf (exp(-x^2), x, 0, 1, 'cos, 'epsabs=1d-9);
(%o1)   [.6901942235215714, 2.84846300257552E-11, 215, 0]
(%i2) integrate (exp(-x^2)*cos(x), x, 0, inf);
                          - 1/4
                        %e      sqrt(%pi)
(%o2)                   -----------------
                                2
(%i3) ev (%, numer);
(%o3)                   .6901942235215714
Function: quad_qawo
    quad_qawo (f(x), x, a, b, omega, trig, [epsrel, epsabs, limit, maxp1, limlst])
    quad_qawo (f, x, a, b, omega, trig, [epsrel, epsabs, limit, maxp1, limlst])

Integration of \(\cos(\omega x) f(x)\) or \(\sin(\omega x)\) over a finite interval, where \(\omega\) is a constant. The rule evaluation component is based on the modified Clenshaw-Curtis technique. quad_qawo applies adaptive subdivision with extrapolation, similar to quad_qags.

quad_qawo computes the integral using the Quadpack QAWO routine:

$$ \int_a^b f(x) \, w(x) \, dx $$

The weight function w is selected by trig:

cos
\(w(x) = \cos\omega x\)
sin
\(w(x) = \sin\omega x\)

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit/2 is the maximum number of subintervals to use. Default is 200.

maxp1

Maximum number of Chebyshev moments. Must be greater than 0. Default is 100.

limlst

Upper bound on the number of cycles. Must be greater than or equal to 3. Default is 10.

quad_qawo returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

6

if the input is invalid.

Examples:

(%i1) quad_qawo (x^(-1/2)*exp(-2^(-2)*x), x, 1d-8, 20*2^2, 1, cos);
(%o1)     [1.376043389877692, 4.72710759424899E-11, 765, 0]
(%i2) rectform (integrate (x^(-1/2)*exp(-2^(-alpha)*x) * cos(x),
      x, 0, inf));
                   alpha/2 - 1/2            2 alpha
        sqrt(%pi) 2              sqrt(sqrt(2        + 1) + 1)
(%o2)   -----------------------------------------------------
                               2 alpha
                         sqrt(2        + 1)
(%i3) ev (%, alpha=2, numer);
(%o3)                     1.376043390090716
Function: quad_qaws
    quad_qaws (f(x), x, a, b, alpha, beta, wfun, [epsrel, epsabs, limit])
    quad_qaws (f, x, a, b, alpha, beta, wfun, [epsrel, epsabs, limit])

Integration of w(x) f(x) over a finite interval, where w(x) is a certain algebraic or logarithmic function. A globally adaptive subdivision strategy is applied, with modified Clenshaw-Curtis integration on the subintervals which contain the endpoints of the interval of integration.

quad_qaws computes the integral using the Quadpack QAWS routine:

$$ \int_a^b f(x) \, w(x) \, dx $$

The weight function w is selected by wfun:

1
\(w(x) = (x - a)^\alpha (b - x)^\beta\)
2
\(w(x) = (x - a)^\alpha (b - x)^\beta \log(x - a)\)
3
\(w(x) = (x - a)^\alpha (b - x)^\beta \log(b - x)\)
4
\(w(x) = (x - a)^\alpha (b - x)^\beta \log(x - a) \log(b - x)\)

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limitis the maximum number of subintervals to use. Default is 200.

quad_qaws returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

6

if the input is invalid.

Examples:

(%i1) quad_qaws (1/(x+1+2^(-4)), x, -1, 1, -0.5, -0.5, 1,
                 'epsabs=1d-9);
(%o1)     [8.750097361672832, 1.24321522715422E-10, 170, 0]
(%i2) integrate ((1-x*x)^(-1/2)/(x+1+2^(-alpha)), x, -1, 1);
       alpha
Is  4 2      - 1  positive, negative, or zero?

pos;
                          alpha         alpha
                   2 %pi 2      sqrt(2 2      + 1)
(%o2)              -------------------------------
                               alpha
                            4 2      + 2
(%i3) ev (%, alpha=4, numer);
(%o3)                     8.750097361672829
Function: quad_qagp
    quad_qagp (f(x), x, a, b, points, [epsrel, epsabs, limit])
    quad_qagp (f, x, a, b, points, [epsrel, epsabs, limit])

Integration of a general function over a finite interval. quad_qagp implements globally adaptive interval subdivision with extrapolation (de Doncker, 1978) by the Epsilon algorithm (Wynn, 1956).

quad_qagp computes the integral

$$ \int_a^b f(x) \, dx $$

The function to be integrated is f(x), with dependent variable x, and the function is to be integrated between the limits a and b.

The integrand may be specified as the name of a Maxima or Lisp function or operator, a Maxima lambda expression, or a general Maxima expression.

To help the integrator, the user must supply a list of points where the integrand is singular or discontinuous.

The keyword arguments are optional and may be specified in any order. They all take the form key=val. The keyword arguments are:

epsrel

Desired relative error of approximation. Default is 1d-8.

epsabs

Desired absolute error of approximation. Default is 0.

limit

Size of internal work array. limit is the maximum number of subintervals to use. Default is 200.

quad_qagp returns a list of four elements:

  • an approximation to the integral,
  • the estimated absolute error of the approximation,
  • the number integrand evaluations,
  • an error code.

The error code (fourth element of the return value) can have the values:

0

no problems were encountered;

1

too many sub-intervals were done;

2

excessive roundoff error is detected;

3

extremely bad integrand behavior occurs;

4

failed to converge

5

integral is probably divergent or slowly convergent

6

if the input is invalid.

Examples:

(%i1) quad_qagp(x^3*log(abs((x^2-1)*(x^2-2))),x,0,3,[1,sqrt(2)]);
(%o1)   [52.74074838347143, 2.6247632689546663e-7, 1029, 0]
(%i2) quad_qags(x^3*log(abs((x^2-1)*(x^2-2))), x, 0, 3);
(%o2)   [52.74074847951494, 4.088443219529836e-7, 1869, 0]

The integrand has singularities at 1 and sqrt(2) so we supply these points to quad_qagp. We also note that quad_qagp is more accurate and more efficient that quad_qags.

Function: quad_control (parameter, [value])

Control error handling for quadpack. The parameter should be one of the following symbols:

current_error

The current error number

control

Controls if messages are printed or not. If it is set to zero or less, messages are suppressed.

max_message

The maximum number of times any message is to be printed.

If value is not given, then the current value of the parameter is returned. If value is given, the value of parameter is set to the given value.


Footnotes

(1)

Applied Mathematics and Programming Division, K.U. Leuven

(2)

Applied Mathematics and Programming Division, K.U. Leuven

(3)

Institut für Mathematik, T.U. Wien

(4)

National Bureau of Standards, Washington, D.C., U.S.A

(5)

https://www.netlib.org/slatec

(6)

https://www.netlib.org/quadpack

(7)

R. Piessens, E. de Doncker-Kapenga, C.W. Uberhuber, and D.K. Kahaner. QUADPACK: A Subroutine Package for Automatic Integration. Berlin: Springer-Verlag, 1983, ISBN 0387125531.


Next: , Previous:   [Contents][Index]