<- previous    index    next ->
Extend PDE solution to complex coefficients and region L31z 
Extend PDE solution to complex coefficients and region

Using C++ standard template library #include 
and defining a type  cmplx   typedef complex cmplx;

Global substitute  cmplx  for  double  and fix a few varaibles
that must remain double. Fix many formats to print complex (real,imag)

Almost all mathematical algorithms that work with real values,
also work with complex values.

This is just the initial PDE of second order
in three dimensions, in "C", Java, Fortran, Ada


Source code for ".cpp" version, second derivative, two dimensions

nuderiv.hpp  compute derivatives
nuderiv.cpp  on non uniform grid
simeq.hpp  solve simultaneous equations
simeq.cpp  with complex coefficients
pde22nu_eq.cpp  complex PDE solver
pde22nu_eq_cpp.out  solution

The errors were close to the same PDE using just real values.

Commands to compile and run above
g++ -o pde22nu_eq pde22nu_eq.cpp simeq.cpp nuderiv.cpp -lm
pde22nu_eq > pde22nu_eq_cpp.out # writes pde22nu_eq_cpp.out     


Source code for java version

Complex.java  Complex type and methods
Csimeq.java  complex simultaneous equations
Cinvert.java  complex matrix inversion
Cnuderiv.java  complex non uniform derivative
pdeCnu22_eq.java  complex PDE solver
pdeCnu22_eq_java.out  solution

The errors were close to the same PDE using just real values.

Commands to compile and run above
javac -cp . Complex.java
javac -cp . Csimeq.java
javac -cp . Cinvert.java
javac -cp . Cnuderiv.java
javac -cp . pdeCnu22_eq.java
java  -cp . pdeCnu22_eq > pdeCnu22_eq_java.out # writes pdeCnu22_eq_java.out


Source code for fortran version

The  _eq uses C zero based subscripts, the _eq1 used Fortran 1 based subscripts

nuderivC.f90  Complex non uniform derivative
inverseC.f90  Complex matrix inversion
simeqC1.f90  Complex simultaneous equations
pdeCnu22_eq.f90  Complex PDE source code
pdeCnu22_eq_f90.out  solution
pdeCnu22_eq1.f90  Complex PDE source code
pdeCnu22_eq1_f90.out  solution

Commands to compile and run above
gfortran pdeCnu22_eq.f90 nuderivC.f90 inverseC.f90
a.out > pdeCnu22_eq_f90.out

gfortran pdeCnu22_eq1.f90 nuderivC.f90 inverseC.f90 simeqC1.f90
a.out > pdeCnu22_eq1_f90.out


For more information on Numerical Computation, follow link below:
CMSC 455 Numerical Computation lectures


    <- previous    index    next ->

Other links

Go to top