OPAL (Object Oriented Parallel Accelerator Library) 2024.2
OPAL
VerticalFFAMagnet.cpp
Go to the documentation of this file.
1//
2// Class VerticalFFAMagnet
3// Defines the abstract interface for a vertical FFA magnet
4// with vertical scaling fringe fields.
5//
6// Copyright (c) 2019 - 2023, Chris Rogers, STFC Rutherford Appleton Laboratory, Didcot, UK
7// All rights reserved
8//
9// This file is part of OPAL.
10//
11// OPAL is free software: you can redistribute it and/or modify
12// it under the terms of the GNU General Public License as published by
13// the Free Software Foundation, either version 3 of the License, or
14// (at your option) any later version.
15//
16// You should have received a copy of the GNU General Public License
17// along with OPAL. If not, see <https://www.gnu.org/licenses/>.
18//
20
23
24#include <cmath>
25
27 : Component(name), straightGeometry_m(1.) {
28}
29
31 Component(right),
32 straightGeometry_m(right.straightGeometry_m),
33 dummy(right.dummy),
34 maxOrder_m(right.maxOrder_m),
35 k_m(right.k_m),
36 Bz_m(right.Bz_m),
37 zNegExtent_m(right.zNegExtent_m),
38 zPosExtent_m(right.zPosExtent_m),
39 halfWidth_m(right.halfWidth_m),
40 bbLength_m(right.bbLength_m),
41 endField_m(right.endField_m->clone()),
42 dfCoefficients_m(right.dfCoefficients_m)
43{
45}
46
47
49}
50
52 VerticalFFAMagnet* magnet = new VerticalFFAMagnet(*this);
53 magnet->initialise();
54 return magnet;
55}
56
58 return dummy;
59}
60
62 return dummy;
63}
64
68}
69
71 double& /*startField*/, double& /*endField*/) {
72 RefPartBunch_m = bunch;
73 initialise();
74}
75
77 RefPartBunch_m = nullptr;
78}
79
81 return straightGeometry_m;
82}
83
85 return straightGeometry_m;
86}
87
89 visitor.visitVerticalFFAMagnet(*this);
90}
91
93 if (std::abs(R[0]) > halfWidth_m ||
94 R[2] < 0. || R[2] > bbLength_m ||
95 R[1] < -zNegExtent_m || R[1] > zPosExtent_m) {
96 return true;
97 }
98 std::vector<double> fringeDerivatives(maxOrder_m+2, 0.);
99 double zRel = R[2]-bbLength_m/2.; // z relative to centre of magnet
100 for (size_t i = 0; i < fringeDerivatives.size(); ++i) {
101 fringeDerivatives[i] = endField_m->function(zRel, i); // d^i_phi f
102 }
103
104 std::vector<double> x_n(maxOrder_m+1); // x^n
105 x_n[0] = 1.; // x^0
106 for (size_t i = 1; i < x_n.size(); ++i) {
107 x_n[i] = x_n[i-1] * R[0];
108 }
109
110 // note that the last element is always 0, because dfCoefficients_m is
111 // of size maxOrder_m+1. This leads to better Maxwellianness in testing.
112 std::vector<double> f_n(maxOrder_m + 2, 0.);
113 std::vector<double> dz_f_n(maxOrder_m + 1, 0.);
114 for (size_t n = 0; n < dfCoefficients_m.size(); ++n) {
115 const std::vector<double>& coefficients = dfCoefficients_m[n];
116 for (size_t i = 0; i < coefficients.size(); ++i) {
117 f_n[n] += coefficients[i] * fringeDerivatives[i];
118 dz_f_n[n] += coefficients[i] * fringeDerivatives[i+1];
119 }
120 }
121 double bref = Bz_m * std::exp(k_m * R[1]);
122 B[0] = 0.;
123 B[1] = 0.;
124 B[2] = 0.;
125 for (size_t n = 0; n < x_n.size(); ++n) {
126 B[0] += bref * f_n[n+1] * (n+1) / k_m * x_n[n];
127 B[1] += bref * f_n[n] * x_n[n];
128 B[2] += bref * dz_f_n[n] / k_m * x_n[n];
129 }
130 return false;
131}
132
134 dfCoefficients_m = std::vector< std::vector<double> >(maxOrder_m+1);
135 dfCoefficients_m[0] = std::vector<double>(1, 1.);
136 if (maxOrder_m > 0) {
137 dfCoefficients_m[1] = std::vector<double>();
138 }
139 // n indexes like the polynomial order of the midplane expansion
140 // e.g. Bz = exp(mz) f_n y^n
141 // where y is distance from the midplane and z is height
142 for (size_t n = 2; n < dfCoefficients_m.size(); n+=2) {
143 const std::vector<double>& oldCoefficients = dfCoefficients_m[n-2];
144 std::vector<double> coefficients(oldCoefficients.size() + 2, 0);
145 // j indexes the derivative of f_0
146 for (size_t j = 0; j < oldCoefficients.size(); ++j) {
147 coefficients[j] += -1./(n)/(n-1) * k_m * k_m * oldCoefficients[j];
148 coefficients[j+2] += -1./(n)/(n-1) * oldCoefficients[j];
149 }
150 dfCoefficients_m[n] = coefficients;
151 }
152}
153
155 endField_m.reset(endField);
156 endField_m->setMaximumDerivative(maxOrder_m);
157}
158
159void VerticalFFAMagnet::setMaxOrder(size_t maxOrder) {
160 if (endField_m.get()) {
161 endField_m->setMaximumDerivative(maxOrder);
162 }
163 maxOrder_m = maxOrder;
164}
Tps< T > exp(const Tps< T > &x)
Exponential.
Definition: TpsMath.h:165
PETE_TUTree< FnAbs, typename T::PETE_Expr_t > abs(const PETE_Expr< T > &l)
const std::string name
virtual void visitVerticalFFAMagnet(const VerticalFFAMagnet &)=0
Apply the algorithm to a vertical FFA magnet.
Interface for a single beam element.
Definition: Component.h:50
PartBunchBase< double, 3 > * RefPartBunch_m
Definition: Component.h:191
Bending magnet with an exponential dependence on field in the vertical plane.
void finalise()
Finalise the VerticalFFAMagnet - sets bunch to nullptr.
ElementBase * clone() const
Inheritable copy constructor.
void accept(BeamlineVisitor &visitor) const
Accept a beamline visitor.
bool getFieldValue(const Vector_t &R, Vector_t &B) const
Calculate the field at some arbitrary position in cartesian coordinates.
void initialise()
Initialise the VerticalFFAMagnet.
std::unique_ptr< endfieldmodel::EndFieldModel > endField_m
BGeometryBase & getGeometry()
Return the cell geometry.
BMultipoleField dummy
void setMaxOrder(size_t maxOrder)
Set the maximum power of x used in the off-midplane expansion;.
std::vector< std::vector< double > > dfCoefficients_m
StraightGeometry straightGeometry_m
EMField & getField()
Return a dummy (0.) field value (what is this for?)
VerticalFFAMagnet(const std::string &name)
Construct a new VerticalFFAMagnet.
~VerticalFFAMagnet()
Destructor - deletes the field.
void initialise(PartBunchBase< double, 3 > *bunch, double &startField, double &endField)
Initialise the VerticalFFAMagnet.
void setEndField(endfieldmodel::EndFieldModel *endField)
Set the fringe field.
Abstract base class for accelerator geometry classes.
Definition: Geometry.h:43
virtual void setElementLength(double length)
Set design length.
Abstract base class for electromagnetic fields.
Definition: EMField.h:188