<chapter name="Four-Vectors">
<h2>Four-Vectors</h2>
The <code>Vec4</code> class gives a simple implementation of four-vectors.
The member function names are based on the assumption that these
represent four-momentum vectors. Thus one can get or set
<ei>p_x, p_y, p_z</ei> and <ei>e</ei>, but not <ei>x, y, z</ei>
or <ei>t</ei>. This is only a matter of naming, however; a
<code>Vec4</code> can equally well be used to store a space-time
four-vector.
<p/>
The <code>Particle</code> object contains a <code>Vec4 p</code> that
stores the particle four-momentum, and another <code>Vec4 vProd</code>
for the production vertex. For the latter the input/output method
names are adapted to the space-time character rather than the normal
energy-momentum one. Thus a user would not normally access the
<code>Vec4</code> classes directly, but only via the methods of the
<code>Particle</code> class,
see <aloc href="ParticleProperties">Particle Properties</aloc>.
<p/>
Nevertheless you are free to use the PYTHIA four-vectors, e.g. as
part of some simple analysis code based directly on the PYTHIA output,
say to define the four-vector sum of a set of particles. But note that
this class was never set up to allow complete generality, only to
provide the operations that are of use inside PYTHIA. There is no
separate class for three-vectors, since such can easily be represented
by four-vectors where the fourth component is not used.
<p/>
Four-vectors have the expected functionality: they can be created,
copied, added, multiplied, rotated, boosted, and manipulated in other
ways. Operator overloading is implemented where reasonable. Properties
can be read out, not only the components themselves but also for derived
quantities such as absolute momentum and direction angles.
<h3>Constructors and basic operators</h3>
A few methods are available to create or copy a four-vector:
<method name="Vec4::Vec4(double x = 0., double y = 0., double z = 0.,
double t = 0.)">
creates a four-vector, by default with all components set to 0.
</method>
<method name="Vec4::Vec4(const Vec4& v)">
creates a four-vector copy of the input four-vector.
</method>
<method name="Vec4& Vec4::operator=(const Vec4& v)">
copies the input four-vector.
</method>
<method name="Vec4& Vec4::operator=(double value)">
gives a four-vector with all components set to <ei>value</ei>.
</method>
<h3>Member methods for input</h3>
The values stored in a four-vector can be modified in a few different
ways:
<method name="void Vec4::reset()">
sets all components to 0.
</method>
<method name="void Vec4::p(double pxIn, double pyIn, double pzIn,
double eIn)">
sets all components to their input values.
</method>
<method name="void Vec4::p(Vec4 pIn)">
sets all components equal to those of the input four-vector.
</method>
<method name="void Vec4::px(double pxIn)">
</method>
<methodmore name="void Vec4::py(double pyIn)">
</methodmore>
<methodmore name="void Vec4::pz(double pzIn)">
</methodmore>
<methodmore name="void Vec4::e(double eIn)">
sets the respective component to the input value.
</methodmore>
<h3>Member methods for output</h3>
A number of methods provides output of basic or derived quantities:
<method name="double Vec4::px()">
</method>
<methodmore name="double Vec4::py()">
</methodmore>
<methodmore name="double Vec4::pz()">
</methodmore>
<methodmore name="double Vec4::e()">
gets the respective component.
</methodmore>
<method name="double Vec4::mCalc()">
</method>
<methodmore name="double Vec4::m2Calc()">
the (squared) mass, calculated from the four-vectors.
If <ei>m^2 < 0</ei> the mass is given with a minus sign,
<ei>-sqrt(-m^2)</ei>. Note the possible loss of precision
in the calculation of <ei>E^2 - p^2</ei>; for particles the
correct mass is stored separately to avoid such problems.
</methodmore>
<method name="double Vec4::pT()">
</method>
<methodmore name="double Vec4::pT2()">
the (squared) transverse momentum.
</methodmore>
<method name="double Vec4::pAbs()">
</method>
<methodmore name="double Vec4::pAbs2()">
the (squared) absolute momentum.
</methodmore>
<method name="double Vec4::eT()">
</method>
<methodmore name="double Vec4::eT2()">
the (squared) transverse energy,
<ei>eT = e * sin(theta) = e * pT / pAbs</ei>.
</methodmore>
<method name="double Vec4::theta()">
the polar angle, in the range 0 through
<ei>pi</ei>.
</method>
<method name="double Vec4::phi()">
the azimuthal angle, in the range <ei>-pi</ei> through <ei>pi</ei>.
</method>
<method name="double Vec4::thetaXZ()">
the angle in the <ei>xz</ei> plane, in the range <ei>-pi</ei> through
<ei>pi</ei>, with 0 along the <ei>+z</ei> axis.
</method>
<method name="double Vec4::pPos()">
</method>
<methodmore name="double Vec4::pNeg()">
the combinations <ei>E+-p_z</ei>.</methodmore>
<method name="double Vec4::rap()">
</method>
<methodmore name="double Vec4::eta()">
true rapidity <ei>y</ei> and pseudorapidity <ei>eta</ei>.
</methodmore>
<h3>Friend methods for output</h3>
There are also some <code>friend</code> methods that take one, two
or three four-vectors as argument. Several of them only use the
three-vector part of the four-vector.
<method name="friend ostream& operator<<(ostream&, const Vec4& v)">
writes out the values of the four components of a <code>Vec4</code> and,
within brackets, a fifth component being the invariant length of the
four-vector, as provided by <code>mCalc()</code> above, and it all
ended with a newline.
</method>
<method name="friend double m(const Vec4& v1, const Vec4& v2)">
</method>
<methodmore name="friend double m2(const Vec4& v1, const Vec4& v2)">
the (squared) invariant mass.
</methodmore>
<method name="friend double dot3(const Vec4& v1, const Vec4& v2)">
the three-product.
</method>
<method name="friend double cross3(const Vec4& v1, const Vec4& v2)">
the cross-product.
</method>
<method name="friend double theta(const Vec4& v1, const Vec4& v2)">
</method>
<methodmore name="friend double costheta(const Vec4& v1, const Vec4& v2)">
the (cosine) of the opening angle between the vectors,
in the range 0 through <ei>pi</ei>.
</methodmore>
<method name="friend double phi(const Vec4& v1, const Vec4& v2)">
</method>
<methodmore name="friend double cosphi(const Vec4& v1, const Vec4& v2)">
the (cosine) of the azimuthal angle between the vectors around the
<ei>z</ei> axis, in the range 0 through <ei>pi</ei>.
</methodmore>
<method name="friend double phi(const Vec4& v1, const Vec4& v2,
const Vec4& v3)">
</method>
<methodmore name="friend double cosphi(const Vec4& v1, const Vec4& v2,
const Vec4& v3)">
the (cosine) of the azimuthal angle between the first two vectors
around the direction of the third, in the range 0 through <ei>pi</ei>.
</methodmore>
<method name="friend double RRapPhi(const Vec4& v1, const Vec4& v2)">
</method>
<methodmore name="friend double REtaPhi(const Vec4& v1, const Vec4& v2)">
the <ei>R</ei> distance measure, in <ei>(y, phi)</ei> or
<ei>(eta, phi)</ei> cylindrical coordinates, i.e.
<ei>R^2 = (y_1 - y_2)^2 + (phi_1 - phi_2)^2</ei> and equivalent.
</methodmore>
<h3>Operations with four-vectors</h3>
Of course one should be able to add, subtract and scale four-vectors,
and more:
<method name="Vec4 Vec4::operator-()">
return a vector with flipped sign for all components, while leaving
the original vector unchanged.
</method>
<method name="Vec4& Vec4::operator+=(const Vec4& v)">
add a four-vector to an existing one.
</method>
<method name="Vec4& Vec4::operator-=(const Vec4& v)">
subtract a four-vector from an existing one.
</method>
<method name="Vec4& Vec4::operator*=(double f)">
multiply all four-vector components by a real number.
</method>
<method name="Vec4& Vec4::operator/=(double f)">
divide all four-vector components by a real number.
</method>
<method name="friend Vec4 operator+(const Vec4& v1, const Vec4& v2)">
add two four-vectors.
</method>
<method name="friend Vec4 operator-(const Vec4& v1, const Vec4& v2)">
subtract two four-vectors.
</method>
<method name="friend Vec4 operator*(double f, const Vec4& v)">
</method>
<methodmore name="friend Vec4 operator*(const Vec4& v, double f)">
multiply a four-vector by a real number.
</methodmore>
<method name="friend Vec4 operator/(const Vec4& v, double f)">
divide a four-vector by a real number.
</method>
<method name="friend double operator*(const Vec4& v1, const Vec4 v2)">
four-vector product.
</method>
<p/>
There are also a few related operations that are normal member methods:
<method name="void Vec4::rescale3(double f)">
multiply the three-vector components by <ei>f</ei>, but keep the
fourth component unchanged.
</method>
<method name="void Vec4::rescale4(double f)">
multiply all four-vector components by <ei>f</ei>.
</method>
<method name="void Vec4::flip3()">
flip the sign of the three-vector components, but keep the
fourth component unchanged.
</method>
<method name="void Vec4::flip4()">
flip the sign of all four-vector components.
</method>
<h3>Rotations and boosts</h3>
A common task is to rotate or boost four-vectors. In case only one
four-vector is affected the operation may be performed directly on it.
However, in case many particles are affected, the helper class
<code>RotBstMatrix</code> can be used to speed up operations.
<method name="void Vec4::rot(double theta, double phi)">
rotate the three-momentum with the polar angle <ei>theta</ei>
and the azimuthal angle <ei>phi</ei>.
</method>
<method name="void Vec4::rotaxis(double phi, double nx, double ny,
double nz)">
rotate the three-momentum with the azimuthal angle <ei>phi</ei>
around the direction defined by the <ei>(n_x, n_y, n_z)</ei>
three-vector.
</method>
<method name="void Vec4::rotaxis(double phi, Vec4& n)">
rotate the three-momentum with the azimuthal angle <ei>phi</ei>
around the direction defined by the three-vector part of <ei>n</ei>.
</method>
<method name="void Vec4::bst(double betaX, double betaY, double betaZ)">
boost the four-momentum by <ei>beta = (beta_x, beta_y, beta_z)</ei>.
</method>
<method name="void Vec4::bst(double betaX, double betaY, double betaZ,
double gamma)">
boost the four-momentum by <ei>beta = (beta_x, beta_y, beta_z)</ei>,
where the <ei>gamma = 1/sqrt(1 - beta^2)</ei> is also input to allow
better precision when <ei>beta</ei> is close to unity.
</method>
<method name="void Vec4::bst(const Vec4& p)">
boost the four-momentum by <ei>beta = (p_x/E, p_y/E, p_z/E)</ei>.
</method>
<method name="void Vec4::bst(const Vec4& p, double m)">
boost the four-momentum by <ei>beta = (p_x/E, p_y/E, p_z/E)</ei>,
where the <ei>gamma = E/m</ei> is also calculated from input to allow
better precision when <ei>beta</ei> is close to unity.
</method>
<method name="void Vec4::bstback(const Vec4& p)">
boost the four-momentum by <ei>beta = (-p_x/E, -p_y/E, -p_z/E)</ei>.
</method>
<method name="void Vec4::bstback(const Vec4& p, double m)">
boost the four-momentum by <ei>beta = (-p_x/E, -p_y/E, -p_z/E)</ei>,
where the <ei>gamma = E/m</ei> is also calculated from input to allow
better precision when <ei>beta</ei> is close to unity.
</method>
<method name="void Vec4::rotbst(const RotBstMatrix& M)">
perform a combined rotation and boost; see below for a description
of the <code>RotBstMatrix</code>.
</method>
<p/>
For a longer sequence of rotations and boosts, and where several
<code>Vec4</code> are to be rotated and boosted in the same way,
a more efficient approach is to define a <code>RotBstMatrix</code>,
which forms a separate auxiliary class. You can build up this
4-by-4 matrix by successive calls to the methods of the class,
such that the matrix encodes the full sequence of operations.
The order in which you do these calls must agree with the imagined
order in which the rotations/boosts should be applied to a
four-momentum, since in general the operations do not commute.
<method name="RotBstMatrix::RotBstMatrix()">
creates a diagonal unit matrix, i.e. one that leaves a four-vector
unchanged.
</method>
<method name="RotBstMatrix::RotBstMatrix(const RotBstMatrix& Min)">
creates a copy of the input matrix.
</method>
<method
name="RotBstMatrix& RotBstMatrix::operator=(const RotBstMatrix4& Min)">
copies the input matrix.
</method>
<method name="void RotBstMatrix::rot(double theta = 0., double phi = 0.)">
rotate by this polar and azimuthal angle.
</method>
<method name="void RotBstMatrix::rot(const Vec4& p)">
rotate so that a vector originally along the <ei>+z</ei> axis becomes
parallel with <ei>p</ei>. More specifically, rotate by <ei>-phi</ei>,
<ei>theta</ei> and <ei>phi</ei>, with angles defined by <ei>p</ei>.
</method>
<method name="void RotBstMatrix::bst(double betaX = 0., double betaY = 0.,
double betaZ = 0.)">
boost by this <ei>beta</ei> vector.
</method>
<method name="void RotBstMatrix::bst(const Vec4&)">
</method>
<methodmore name="void RotBstMatrix::bstback(const Vec4&)">
boost with a <ei>beta = p/E</ei> or <ei>beta = -p/E</ei>, respectively.
</methodmore>
<method name="void RotBstMatrix::bst(const Vec4& p1, const Vec4& p2)">
boost so that <ei>p_1</ei> is transformed to <ei>p_2</ei>. It is assumed
that the two vectors obey <ei>p_1^2 = p_2^2</ei>.
</method>
<method name="void RotBstMatrix::toCMframe(const Vec4& p1, const Vec4& p2)">
boost and rotate to the rest frame of <ei>p_1</ei> and <ei>p_2</ei>,
with <ei>p_1</ei> along the <ei>+z</ei> axis.
</method>
<method name="void RotBstMatrix::fromCMframe(const Vec4& p1, const Vec4& p2)">
rotate and boost from the rest frame of <ei>p_1</ei> and <ei>p_2</ei>,
with <ei>p_1</ei> along the <ei>+z</ei> axis, to the actual frame of
<ei>p_1</ei> and <ei>p_2</ei>, i.e. the inverse of the above.
</method>
<method name="void RotBstMatrix::rotbst(const RotBstMatrix& Min);">
combine the current matrix with another one.
</method>
<method name="void RotBstMatrix::invert()">
invert the matrix, which corresponds to an opposite sequence and sign
of rotations and boosts.
</method>
<method name="void RotBstMatrix::reset()">
reset to no rotation/boost; i.e. the default at creation.
</method>
<method name="double RotBstMatrix::deviation()">
crude estimate how much a matrix deviates from the unit matrix:
the sum of the absolute values of all non-diagonal matrix elements
plus the sum of the absolute deviation of the diagonal matrix
elements from unity.
</method>
<method name="friend ostream& operator<<(ostream&,
const RotBstMatrix& M)">
writes out the values of the sixteen components of a
<code>RotBstMatrix</code>, on four consecutive lines and
ended with a newline.
</method>
</chapter>
<!-- Copyright (C) 2014 Torbjorn Sjostrand -->
Marcin Chrzaszcz