floating point - Data type mismatch in fortran -

-

i've written rudimentary algorithm in fortran 95 calculate gradient of function (an example of prescribed in code) using central differences augmented procedure known richardson extrapolation. 

function f(n,x) ! scalar multivariable function differentiated  integer :: n real(kind = kind(1d0)) :: x(n), f  f = x(1)**5.d0 + cos(x(2)) + log(x(3)) - sqrt(x(4))  end function f !=====! !=====! !=====!  program gradient !==============================================================================! ! calculates gradient of scalar function f @ x=0using finite        ! ! difference approximation, low order richardson extrapolation.         ! !==============================================================================!  parameter (n = 4, m = 25) real(kind = kind(1d0)) :: x(n), xhup(n), xhdown(n), d(m), r(m), dfdxi, h0, h, gradf(n)  h0 = 1.d0 x  = 3.d0  ! loop through each component of vector x , calculate appropriate ! derivative = 1,n ! reset step size h = h0      ! carry out m successive central difference approximations of derivative j = 1,m         xhup = x         xhdown = x         xhup(i) = xhup(i) + h         xhdown(i) = xhdown(i) - h         d(j) = ( f(n,xhup) - f(n,xhdown) ) / (2.d0*h)     h = h / 2.d0     end      r = 0.d0     k = 3,m      r(k) = ( 64.d0*d(k) - 20.d0*d(k-1) + d(k-2) ) / 45.d0         if ( abs(r(k) - r(k-1)) < 0.0001d0 )         dfdxi = r(k)             exit         end if     end      gradf(i) = dfdxi end  ! print out gradient write(*,*) " " write(*,*) " grad(f(x)) = " write(*,*) " " = 1,n     write(*,*) gradf(i) end  end program gradient

in single precision runs fine , gives me decent results. when try change double precision shown in code, error when trying compile claiming assignment statement

d(j) = ( f(n,xhup) - f(n,xhdown) ) / (2.d0*h)

is producing type mismatch real(4)/real(8). have tried several different declarations of double precision, appended every appropriate double precision constant in code d0, , same error every time. i'm little stumped how function f possibly producing single precision number.

the problem f not explicitely defined in main program, therefore implicitly assumed of single precision, type real(4) gfortran.

i agree comment of high performance mark, should use implicit none in fortran code, make sure object explicitely declared. way, have obtained more appropriate error message f not being explicitely defined.

also, consider 2 more things:

  • define function within module , import module in main program. practice define subroutines/functions within modules only, compiler can make checks on number , type of arguments, when invoke function.

  • you (again in module) introduce constant precicision , use everywhere, kind of real must specified. taking example below, changing line

    integer, parameter :: dp = kind(1.0d0)

    into

    integer, parameter :: dp = kind(1.0)

    you change real variables double single precision. note _dp suffix literal constants instead of d0 suffix, automatically adjust precision well.

    module accuracy   implicit none    integer, parameter :: dp = kind(1.0d0)  end module accuracy   module myfunc   use accuracy   implicit none  contains    function f(n,x)     integer :: n     real(dp) :: x(n), f     f = 0.5_dp * x(1)**5 + cos(x(2)) + log(x(3)) - sqrt(x(4))   end function f  end module myfunc   program gradient   use myfunc   implicit none    real(dp) :: x(n), xhup(n), xhdown(n), d(m), r(m), dfdxi, h0, h, gradf(n)   :  end program gradient

Comments

Post a Comment

Popular posts from this blog

java - Jmockit String final length method mocking Issue -

-
using jmockit 1.2 jar, trying mock string's length method getting unexpected invocation exception: failed: test java.lang.illegalstateexception: missing invocation mocked type @ point; please make sure such invocations appear after declaration of suitable mock field or parameter @ stringdemo.testa$1.<init>(testa.java:17) @ stringdemo.testa.test(testa.java:13) i using testng: @test public void test() throws exception { new expectations() { @mocked("length") string astring; { astring.length(); result = 2; } }; system.out.println(a.showa("test")); } } actual class a: public class { public static int showa(string str){ int a= str.length(); return a; } } this wrong way of recording expected results. shouldn't mock , record string's length() method, record showa() instead. here solution @suppresswarnings("unused...
Read more

What is the difference between data design and data model(ERD) -

-
i have google question , find answer in day. know, data design many many table have list column,and mark pk , fk. why er-diagram have find this? data design process of designing database. main output of data design detailed logical data model of database. some people data design includes of needed logical , physical design choices , physical storage parameters needed generate design in data definition language. technically, database analysis, database analysts (dbas) trained do. while person can both data design , database analysis, these 2 different tasks. data design models data. database analysis takes model , applies 1 or more database engines (relational, hierarchical, nosql). a logical data model 1 of main outputs of data design. data model represented entity relationship diagram, or er diagram.
Read more

ios - Can NSManagedObject conform to NSCoding -

-
i need transfer single object across device. right converting nsmanagedobject dictionary , archiving , sending nsdata. upon receiving unarchiving it. transfer nsmanagedobject archiving , unarchiving instead of creating intermediate data object. @interface test : nsmanagedobject<nscoding> @property (nonatomic, retain) nsstring * title; @end @implementation test @dynamic title; - (id)initwithcoder:(nscoder *)coder { self = [super init]; if (self) { self.title = [coder decodeobjectforkey:@"title"]; //<crash } return self; } - (void)encodewithcoder:(nscoder *)coder { [coder encodeobject:self.title forkey:@"title"]; } @end nsdata *archivedobjects = [nskeyedarchiver archiveddatawithrootobject:testobj]; nsdata *objectsdata = archivedobjects; if ([objectsdata length] > 0) { nsarray *objects = [nskeyedunarchiver unarchiveobjectwithdata:objectsdata]; } the problem above code is. crashes @ self.title in initwithcoder sa...
Read more