35 |
|
* |
36 |
|
* [1] Meineke, et al., J. Comp. Chem. 26, 252-271 (2005). |
37 |
|
* [2] Fennell & Gezelter, J. Chem. Phys. 124, 234104 (2006). |
38 |
< |
* [3] Sun, Lin & Gezelter, J. Chem. Phys. 128, 24107 (2008). |
39 |
< |
* [4] Vardeman & Gezelter, in progress (2009). |
38 |
> |
* [3] Sun, Lin & Gezelter, J. Chem. Phys. 128, 234107 (2008). |
39 |
> |
* [4] Kuang & Gezelter, J. Chem. Phys. 133, 164101 (2010). |
40 |
> |
* [5] Vardeman, Stocker & Gezelter, J. Chem. Theory Comput. 7, 834 (2011). |
41 |
|
*/ |
42 |
|
|
43 |
|
#include "math/CubicSpline.hpp" |
43 |
– |
#include "utils/simError.h" |
44 |
|
#include <cmath> |
45 |
+ |
#include <cassert> |
46 |
|
#include <cstdio> |
47 |
|
#include <algorithm> |
48 |
+ |
#include <numeric> |
49 |
|
|
50 |
|
using namespace OpenMD; |
51 |
|
using namespace std; |
52 |
|
|
53 |
|
CubicSpline::CubicSpline() : generated(false), isUniform(true) { |
54 |
< |
data_.clear(); |
54 |
> |
x_.clear(); |
55 |
> |
y_.clear(); |
56 |
|
} |
57 |
|
|
58 |
|
void CubicSpline::addPoint(const RealType xp, const RealType yp) { |
59 |
< |
data_.push_back(make_pair(xp, yp)); |
59 |
> |
x_.push_back(xp); |
60 |
> |
y_.push_back(yp); |
61 |
|
} |
62 |
|
|
63 |
|
void CubicSpline::addPoints(const vector<RealType>& xps, |
64 |
|
const vector<RealType>& yps) { |
65 |
|
|
66 |
< |
if (xps.size() != yps.size()) { |
67 |
< |
printf( painCave.errMsg, |
68 |
< |
"CubicSpline::addPoints was passed vectors of different length!\n"); |
69 |
< |
painCave.severity = OPENMD_ERROR; |
70 |
< |
painCave.isFatal = 1; |
67 |
< |
simError(); |
66 |
> |
assert(xps.size() == yps.size()); |
67 |
> |
|
68 |
> |
for (unsigned int i = 0; i < xps.size(); i++){ |
69 |
> |
x_.push_back(xps[i]); |
70 |
> |
y_.push_back(yps[i]); |
71 |
|
} |
69 |
– |
|
70 |
– |
for (int i = 0; i < xps.size(); i++) |
71 |
– |
data_.push_back(make_pair(xps[i], yps[i])); |
72 |
|
} |
73 |
|
|
74 |
|
void CubicSpline::generate() { |
76 |
|
// |
77 |
|
// class values constructed: |
78 |
|
// n = number of data_ points. |
79 |
< |
// x = vector of independent variable values |
80 |
< |
// y = vector of dependent variable values |
81 |
< |
// b = vector of S'(x[i]) values. |
82 |
< |
// c = vector of S"(x[i])/2 values. |
83 |
< |
// d = vector of S'''(x[i]+)/6 values (i < n). |
79 |
> |
// x_ = vector of independent variable values |
80 |
> |
// y_ = vector of dependent variable values |
81 |
> |
// b = vector of S'(x_[i]) values. |
82 |
> |
// c = vector of S"(x_[i])/2 values. |
83 |
> |
// d = vector of S'''(x_[i]+)/6 values (i < n). |
84 |
|
// Local variables: |
85 |
|
|
86 |
|
RealType fp1, fpn, h, p; |
87 |
|
|
88 |
|
// make sure the sizes match |
89 |
|
|
90 |
< |
n = data_.size(); |
90 |
> |
n = x_.size(); |
91 |
|
b.resize(n); |
92 |
|
c.resize(n); |
93 |
|
d.resize(n); |
97 |
|
bool sorted = true; |
98 |
|
|
99 |
|
for (int i = 1; i < n; i++) { |
100 |
< |
if ( (data_[i].first - data_[i-1].first ) <= 0.0 ) sorted = false; |
100 |
> |
if ( (x_[i] - x_[i-1] ) <= 0.0 ) sorted = false; |
101 |
|
} |
102 |
|
|
103 |
|
// sort if necessary |
104 |
|
|
105 |
< |
if (!sorted) sort(data_.begin(), data_.end()); |
105 |
> |
if (!sorted) { |
106 |
> |
vector<int> p = sort_permutation(x_); |
107 |
> |
x_ = apply_permutation(x_, p); |
108 |
> |
y_ = apply_permutation(y_, p); |
109 |
> |
} |
110 |
|
|
111 |
+ |
|
112 |
|
// Calculate coefficients for the tridiagonal system: store |
113 |
|
// sub-diagonal in B, diagonal in D, difference quotient in C. |
114 |
|
|
115 |
< |
b[0] = data_[1].first - data_[0].first; |
116 |
< |
c[0] = (data_[1].second - data_[0].second) / b[0]; |
115 |
> |
b[0] = x_[1] - x_[0]; |
116 |
> |
c[0] = (y_[1] - y_[0]) / b[0]; |
117 |
|
|
118 |
|
if (n == 2) { |
119 |
|
|
120 |
|
// Assume the derivatives at both endpoints are zero. Another |
121 |
|
// assumption could be made to have a linear interpolant between |
122 |
|
// the two points. In that case, the b coefficients below would be |
123 |
< |
// (data_[1].second - data_[0].second) / (data_[1].first - data_[0].first) |
123 |
> |
// (y_[1] - y_[0]) / (x_[1] - x_[0]) |
124 |
|
// and the c and d coefficients would both be zero. |
125 |
|
b[0] = 0.0; |
126 |
< |
c[0] = -3.0 * pow((data_[1].second - data_[0].second) / |
127 |
< |
(data_[1].first-data_[0].first), 2); |
123 |
< |
d[0] = -2.0 * pow((data_[1].second - data_[0].second) / |
124 |
< |
(data_[1].first-data_[0].first), 3); |
126 |
> |
c[0] = -3.0 * pow((y_[1] - y_[0]) / (x_[1] - x_[0]), 2); |
127 |
> |
d[0] = -2.0 * pow((y_[1] - y_[0]) / (x_[1] - x_[0]), 3); |
128 |
|
b[1] = b[0]; |
129 |
|
c[1] = 0.0; |
130 |
|
d[1] = 0.0; |
131 |
< |
dx = 1.0 / (data_[1].first - data_[0].first); |
131 |
> |
dx = 1.0 / (x_[1] - x_[0]); |
132 |
|
isUniform = true; |
133 |
|
generated = true; |
134 |
|
return; |
137 |
|
d[0] = 2.0 * b[0]; |
138 |
|
|
139 |
|
for (int i = 1; i < n-1; i++) { |
140 |
< |
b[i] = data_[i+1].first - data_[i].first; |
140 |
> |
b[i] = x_[i+1] - x_[i]; |
141 |
|
if ( fabs( b[i] - b[0] ) / b[0] > 1.0e-5) isUniform = false; |
142 |
< |
c[i] = (data_[i+1].second - data_[i].second) / b[i]; |
142 |
> |
c[i] = (y_[i+1] - y_[i]) / b[i]; |
143 |
|
d[i] = 2.0 * (b[i] + b[i-1]); |
144 |
|
} |
145 |
|
|
151 |
|
fp1 = c[0] - b[0]*(c[1] - c[0])/(b[0] + b[1]); |
152 |
|
if (n > 3) fp1 = fp1 + b[0]*((b[0] + b[1]) * (c[2] - c[1]) / |
153 |
|
(b[1] + b[2]) - |
154 |
< |
c[1] + c[0]) / (data_[3].first - data_[0].first); |
154 |
> |
c[1] + c[0]) / (x_[3] - x_[0]); |
155 |
|
|
156 |
|
fpn = c[n-2] + b[n-2]*(c[n-2] - c[n-3])/(b[n-3] + b[n-2]); |
157 |
< |
|
157 |
> |
|
158 |
|
if (n > 3) fpn = fpn + b[n-2] * |
159 |
< |
(c[n-2] - c[n-3] - (b[n-3] + b[n-2]) * |
160 |
< |
(c[n-3] - c[n-4])/(b[n-3] + b[n-4]))/(data_[n-1].first - data_[n-4].first); |
159 |
> |
(c[n-2] - c[n-3] - (b[n-3] + b[n-2]) * |
160 |
> |
(c[n-3] - c[n-4])/(b[n-3] + b[n-4])) / |
161 |
> |
(x_[n-1] - x_[n-4]); |
162 |
|
|
159 |
– |
|
163 |
|
// Calculate the right hand side and store it in C. |
164 |
|
|
165 |
|
c[n-1] = 3.0 * (fpn - c[n-2]); |
183 |
|
// Calculate the coefficients defining the spline. |
184 |
|
|
185 |
|
for (int i = 0; i < n-1; i++) { |
186 |
< |
h = data_[i+1].first - data_[i].first; |
186 |
> |
h = x_[i+1] - x_[i]; |
187 |
|
d[i] = (c[i+1] - c[i]) / (3.0 * h); |
188 |
< |
b[i] = (data_[i+1].second - data_[i].second)/h - h * (c[i] + h * d[i]); |
188 |
> |
b[i] = (y_[i+1] - y_[i])/h - h * (c[i] + h * d[i]); |
189 |
|
} |
190 |
|
|
191 |
|
b[n-1] = b[n-2] + h * (2.0 * c[n-2] + h * 3.0 * d[n-2]); |
192 |
|
|
193 |
< |
if (isUniform) dx = 1.0 / (data_[1].first - data_[0].first); |
193 |
> |
if (isUniform) dx = 1.0 / (x_[1] - x_[0]); |
194 |
|
|
195 |
|
generated = true; |
196 |
|
return; |
197 |
|
} |
198 |
|
|
199 |
< |
RealType CubicSpline::getValueAt(RealType t) { |
199 |
> |
RealType CubicSpline::getValueAt(const RealType& t) { |
200 |
|
// Evaluate the spline at t using coefficients |
201 |
|
// |
202 |
|
// Input parameters |
205 |
|
// value of spline at t. |
206 |
|
|
207 |
|
if (!generated) generate(); |
205 |
– |
RealType dt; |
208 |
|
|
209 |
< |
if ( t < data_[0].first || t > data_[n-1].first ) { |
210 |
< |
sprintf( painCave.errMsg, |
209 |
< |
"CubicSpline::getValueAt was passed a value outside the range of the spline!\n"); |
210 |
< |
painCave.severity = OPENMD_ERROR; |
211 |
< |
painCave.isFatal = 1; |
212 |
< |
simError(); |
213 |
< |
} |
209 |
> |
assert(t >= x_.front()); |
210 |
> |
assert(t <= x_.back()); |
211 |
|
|
212 |
|
// Find the interval ( x[j], x[j+1] ) that contains or is nearest |
213 |
|
// to t. |
214 |
|
|
218 |
– |
int j; |
219 |
– |
|
215 |
|
if (isUniform) { |
216 |
|
|
217 |
< |
j = max(0, min(n-1, int((t - data_[0].first) * dx))); |
217 |
> |
j = max(0, min(n-1, int((t - x_[0]) * dx))); |
218 |
|
|
219 |
|
} else { |
220 |
|
|
221 |
|
j = n-1; |
222 |
|
|
223 |
|
for (int i = 0; i < n; i++) { |
224 |
< |
if ( t < data_[i].first ) { |
224 |
> |
if ( t < x_[i] ) { |
225 |
|
j = i-1; |
226 |
|
break; |
227 |
|
} |
230 |
|
|
231 |
|
// Evaluate the cubic polynomial. |
232 |
|
|
233 |
< |
dt = t - data_[j].first; |
234 |
< |
return data_[j].second + dt*(b[j] + dt*(c[j] + dt*d[j])); |
233 |
> |
dt = t - x_[j]; |
234 |
> |
return y_[j] + dt*(b[j] + dt*(c[j] + dt*d[j])); |
235 |
> |
} |
236 |
> |
|
237 |
> |
|
238 |
> |
void CubicSpline::getValueAt(const RealType& t, RealType& v) { |
239 |
> |
// Evaluate the spline at t using coefficients |
240 |
> |
// |
241 |
> |
// Input parameters |
242 |
> |
// t = point where spline is to be evaluated. |
243 |
> |
// Output: |
244 |
> |
// value of spline at t. |
245 |
|
|
246 |
+ |
if (!generated) generate(); |
247 |
+ |
|
248 |
+ |
assert(t >= x_.front()); |
249 |
+ |
assert(t <= x_.back()); |
250 |
+ |
|
251 |
+ |
// Find the interval ( x[j], x[j+1] ) that contains or is nearest |
252 |
+ |
// to t. |
253 |
+ |
|
254 |
+ |
if (isUniform) { |
255 |
+ |
|
256 |
+ |
j = max(0, min(n-1, int((t - x_[0]) * dx))); |
257 |
+ |
|
258 |
+ |
} else { |
259 |
+ |
|
260 |
+ |
j = n-1; |
261 |
+ |
|
262 |
+ |
for (int i = 0; i < n; i++) { |
263 |
+ |
if ( t < x_[i] ) { |
264 |
+ |
j = i-1; |
265 |
+ |
break; |
266 |
+ |
} |
267 |
+ |
} |
268 |
+ |
} |
269 |
+ |
|
270 |
+ |
// Evaluate the cubic polynomial. |
271 |
+ |
|
272 |
+ |
dt = t - x_[j]; |
273 |
+ |
v = y_[j] + dt*(b[j] + dt*(c[j] + dt*d[j])); |
274 |
|
} |
275 |
|
|
276 |
+ |
pair<RealType, RealType> CubicSpline::getLimits(){ |
277 |
+ |
if (!generated) generate(); |
278 |
+ |
return make_pair( x_.front(), x_.back() ); |
279 |
+ |
} |
280 |
|
|
281 |
< |
pair<RealType, RealType> CubicSpline::getValueAndDerivativeAt(RealType t) { |
281 |
> |
void CubicSpline::getValueAndDerivativeAt(const RealType& t, RealType& v, |
282 |
> |
RealType &dv) { |
283 |
|
// Evaluate the spline and first derivative at t using coefficients |
284 |
|
// |
285 |
|
// Input parameters |
286 |
|
// t = point where spline is to be evaluated. |
249 |
– |
// Output: |
250 |
– |
// pair containing value of spline at t and first derivative at t |
287 |
|
|
288 |
|
if (!generated) generate(); |
253 |
– |
RealType dt; |
289 |
|
|
290 |
< |
if ( t < data_.front().first || t > data_.back().first ) { |
291 |
< |
sprintf( painCave.errMsg, |
257 |
< |
"CubicSpline::getValueAndDerivativeAt was passed a value outside the range of the spline!\n"); |
258 |
< |
painCave.severity = OPENMD_ERROR; |
259 |
< |
painCave.isFatal = 1; |
260 |
< |
simError(); |
261 |
< |
} |
290 |
> |
assert(t >= x_.front()); |
291 |
> |
assert(t <= x_.back()); |
292 |
|
|
293 |
|
// Find the interval ( x[j], x[j+1] ) that contains or is nearest |
294 |
|
// to t. |
295 |
|
|
266 |
– |
int j; |
267 |
– |
|
296 |
|
if (isUniform) { |
297 |
|
|
298 |
< |
j = max(0, min(n-1, int((t - data_[0].first) * dx))); |
298 |
> |
j = max(0, min(n-1, int((t - x_[0]) * dx))); |
299 |
|
|
300 |
|
} else { |
301 |
|
|
302 |
|
j = n-1; |
303 |
|
|
304 |
|
for (int i = 0; i < n; i++) { |
305 |
< |
if ( t < data_[i].first ) { |
305 |
> |
if ( t < x_[i] ) { |
306 |
|
j = i-1; |
307 |
|
break; |
308 |
|
} |
311 |
|
|
312 |
|
// Evaluate the cubic polynomial. |
313 |
|
|
314 |
< |
dt = t - data_[j].first; |
314 |
> |
dt = t - x_[j]; |
315 |
|
|
316 |
< |
RealType yval = data_[j].second + dt*(b[j] + dt*(c[j] + dt*d[j])); |
317 |
< |
RealType dydx = b[j] + dt*(2.0 * c[j] + 3.0 * dt * d[j]); |
316 |
> |
v = y_[j] + dt*(b[j] + dt*(c[j] + dt*d[j])); |
317 |
> |
dv = b[j] + dt*(2.0 * c[j] + 3.0 * dt * d[j]); |
318 |
> |
} |
319 |
> |
|
320 |
> |
std::vector<int> CubicSpline::sort_permutation(std::vector<RealType>& v) { |
321 |
> |
std::vector<int> p(v.size()); |
322 |
> |
|
323 |
> |
// 6 lines to replace std::iota(p.begin(), p.end(), 0); |
324 |
> |
int value = 0; |
325 |
> |
std::vector<int>::iterator i; |
326 |
> |
for (i = p.begin(); i != p.end(); ++i) { |
327 |
> |
(*i) = value; |
328 |
> |
++value; |
329 |
> |
} |
330 |
|
|
331 |
< |
return make_pair(yval, dydx); |
331 |
> |
std::sort(p.begin(), p.end(), OpenMD::Comparator(v) ); |
332 |
> |
return p; |
333 |
|
} |
334 |
+ |
|
335 |
+ |
std::vector<RealType> CubicSpline::apply_permutation(std::vector<RealType> const& v, |
336 |
+ |
std::vector<int> const& p) { |
337 |
+ |
int n = p.size(); |
338 |
+ |
std::vector<double> sorted_vec(n); |
339 |
+ |
for (int i = 0; i < n; ++i) { |
340 |
+ |
sorted_vec[i] = v[p[i]]; |
341 |
+ |
} |
342 |
+ |
return sorted_vec; |
343 |
+ |
} |