########################################################################
Here follows a list of the routines that
become available if the module "avh_olo" is USEd.

  subroutine olo_precision( ndecimals )
  subroutine olo_unit( iunit ,message )
  subroutine olo_scale( muscale )
  subroutine olo_onshell( thrs )
  subroutine olo_a0( rslt ,mm ,rmu )
  subroutine olo_an( rslt ,rank ,mm ,rmu )
  subroutine olo_b0( rslt ,pp ,m1,m2 ,rmu )
  subroutine olo_b11( b11,b00,b1,b0 ,pp,m1,m2 ,rmu )
  subroutine olo_bn( rslt ,rank ,pp,m1,m2 ,rmu )
  subroutine olo_c0( rslt ,p1,p2,p3 ,m1,m2,m3 ,rmu )
  subroutine olo_d0( rslt ,p1,p2,p3,p4,p12,p23 ,m1,m2,m3,m4 ,rmu )
  integer function olo_dp_precision()
  integer function olo_qp_precision()
  integer function olo_dd_precision()
  integer function olo_qd_precision()
  integer function olo_mp_precision()

They are described one by one below. The routines for the evaluation
of loop integrals are generic, ie they can be called with real and
with complex arguments, and with or without the last argument. If the
program is compiled with several levels of precision, the same routines
are called with different kinds and types of input. The output is
complex of the same kind/type as the input.

The extensions "_a0", "_b0" etc. may be omitted, you can also just
  call olo( rslt ,mm ,rmu )
  call olo( rslt ,rank ,mm ,rmu )
  call olo( rslt ,pp ,m1,m2 ,rmu )
  call olo( b11,b00,b1,b0 ,pp,m1,m2 ,rmu )
  call olo( rslt ,rank ,pp,m1,m2 ,rmu )
  call olo( rslt ,p1,p2,p3 ,m1,m2,m3 ,rmu )
  call olo( rslt ,p1,p2,p3,p4,p12,p23 ,m1,m2,m3,m4 ,rmu )

If the names of routines in "avh_olo" lead to name clashes, realize that
these can be cured in the USE statement, eg
  use avh_olo, my_olo=>olo
and then
  call my_olo( rslt ,pp ,m1,m2 ,rmu )
etc.

########################################################################
  subroutine olo_precision( ndecimals )
  integer ,intent(in) :: ndecimals

If OneLOop is operating at arbitrary precision, the number of decimals
can be set with this routine. It is the responsibility of the user that
this number corresponds to the actual precision of the variables.
The precision can be changed during runtime.

This has no effect on routine calls with Fortran intrinsic types, or
types with fixed precision.

########################################################################
  subroutine olo_unit( iunit ,message )
  integer     ,intent(in) :: iunit
  character(*),intent(in),optional :: message

Set the units to which messages are send. The argument "message"
specifies the type of message for which you want to set the unit.
At the moment, the options and their defaults are

  'error'--> 6, 'warning'--> 6, 'message'--> 6, 'printall'--> 0.

-If the unit for 'printall' is positive, then all input and output for
 all calls to the loop functions will be printed to that unit.
-If iunit is not positive, then nothing will be printed.
-If the routine is called without the second argument, then all units
 will be set to iunit, except the unit for 'printall', which will be
 set to 0.

########################################################################
  subroutine olo_scale( muscale )
  real(kind(1d0)) ,intent(in) :: muscale

The type/kind of the input is hard-wired strictly to real(kind(1d0)).

Set the default renormalization scale. The input has the units of mass,
not mass^2.
If this routine is not called, the scale is set to 1.

########################################################################
  subroutine olo_onshell( thrs )
  real(kind(1d0)) ,intent(in) :: thrs

The type/kind of the input is hard-wired strictly to real(kind(1d0)).

Set threshold to consider internal masses identical zero and external
squared momenta identical zero or equal to internal masses, if this
leads to an IR divergent case. For example, if  |p1-m1|<thrs, and p1=m1
consitutes an IR-divergent case, then the loop integral is considered to
be IR-divergent.  Here  thrs  is the input for this routine.  If this
routine is not called,  thrs  will essentially be considerd identically
zero, but warnings will be given when an IR-divergent case is
approached. If this routine is called with thrs=0d0, then these warnings
will not appear anymore.

########################################################################
  subroutine olo_a0( rslt ,mm ,rmu )
  subroutine olo_b0( rslt ,pp ,m1,m2 ,rmu )
  subroutine olo_c0( rslt ,p1,p2,p3 ,m1,m2,m3 ,rmu )
  subroutine olo_d0( rslt ,p1,p2,p3,p4,p12,p23 ,m1,m2,m3,m4 ,rmu )

The scalar one-loop functions. All have complex output "rslt(0:2)",
where

  rslt(0) is the eps^0   -coefficient
  rslt(1) is the eps^(-1)-coefficient
  rslt(2) is the eps^(-2)-coefficient

The real input "rmu" sets the renormalization scale for the
particular call. It does overrule, but not change the default scale.
If the routines are called without this argument, the default scale
is used.

The other input is real or complex, with the
following possibilities:
  - everything is real,
  - all momenta are real and all masses are complex,
  - everything is complex.

Remarks:
  - If the momenta are complex and any has a non-zero imaginary part,
    an error-message is issued and the imaginary part is considered to
    be zero.
  - If the masses are complex and any has a positive imaginary part,
    an error-message is issued and the imaginary part is considered to
    have the opposite sign.

########################################################################
  subroutine olo_b11( b11,b00,b1,b0 ,pp,m1,m2 ,rmu )

The 2-point Passarino-Veltman functions.
The complex output is  "b11(0:2),b00(0:2),b1(0:2),b0(0:2)"

where bX(0) is the eps^0   -coefficient
      bX(1) is the eps^(-1)-coefficient
      bX(2) is the eps^(-2)-coefficient

for X={11,00,1,0} defined such that

  B{mu,nu} = g{mu,nu}*b00 + p1{mu}*p1{nu}*b11
  B{mu} = p1{mu}*b1

For the input, the same holds as for the scalar function.

########################################################################
  subroutine olo_an( rslt ,rank ,mm ,rmu )

The 1-point Passarino-Veltman functions, up to rank 4.
The complex output is of shape rslt(0:2,0:2). 
For input rank=0, only rslt(:,0  ) is filled,
For input rank=1, only rslt(:,0  ) is filled,
For input rank=2, only rslt(:,0:1) is filled.
For input rank=3, only rslt(:,0:1) is filled.
  rslt(:,0) = A0
  rslt(:,1) = A00
  rslt(:,2) = A0000

where rslt(0,:) is the eps^0   -coefficient
      rslt(1,:) is the eps^(-1)-coefficient
      rslt(2,:) is the eps^(-2)-coefficient

For the input, the same holds as for the scalar function.

########################################################################
  subroutine olo_bn( rslt ,rank ,pp,m1,m2 ,rmu )

The 2-point Passarino-Veltman functions, up to rank 4.
The complex output is of shape rslt(0:2,0:8). 
For input rank=0, only rslt(:,0  ) is filled,
For input rank=1, only rslt(:,0:1) is filled,
For input rank=2, only rslt(:,0:3) is filled.
For input rank=3, only rslt(:,0:5) is filled.
  rslt(:,0) = B0    rslt(:,4) = B001
  rslt(:,1) = B1    rslt(:,5) = B111
  rslt(:,2) = B00   rslt(:,6) = B0000
  rslt(:,3) = B11   rslt(:,7) = B0011
                    rslt(:,8) = B1111

where rslt(0,:) is the eps^0   -coefficient
      rslt(1,:) is the eps^(-1)-coefficient
      rslt(2,:) is the eps^(-2)-coefficient

For the input, the same holds as for the scalar function.

########################################################################
  integer function olo_dp_precision()
  integer function olo_qp_precision()
  integer function olo_dd_precision()
  integer function olo_qd_precision()
  integer function olo_mp_precision()

These functions give the precisions at which the routines are operating.
They return the number of decimals.

########################################################################