1 | /* |
---|
2 | Copyright (c) 2003-2006 Gino van den Bergen / Erwin Coumans http://continuousphysics.com/Bullet/ |
---|
3 | |
---|
4 | This software is provided 'as-is', without any express or implied warranty. |
---|
5 | In no event will the authors be held liable for any damages arising from the use of this software. |
---|
6 | Permission is granted to anyone to use this software for any purpose, |
---|
7 | including commercial applications, and to alter it and redistribute it freely, |
---|
8 | subject to the following restrictions: |
---|
9 | |
---|
10 | 1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required. |
---|
11 | 2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software. |
---|
12 | 3. This notice may not be removed or altered from any source distribution. |
---|
13 | */ |
---|
14 | |
---|
15 | |
---|
16 | #ifndef BT_MATRIX3x3_H |
---|
17 | #define BT_MATRIX3x3_H |
---|
18 | |
---|
19 | #include "btVector3.h" |
---|
20 | #include "btQuaternion.h" |
---|
21 | |
---|
22 | #ifdef BT_USE_DOUBLE_PRECISION |
---|
23 | #define btMatrix3x3Data btMatrix3x3DoubleData |
---|
24 | #else |
---|
25 | #define btMatrix3x3Data btMatrix3x3FloatData |
---|
26 | #endif //BT_USE_DOUBLE_PRECISION |
---|
27 | |
---|
28 | |
---|
29 | /**@brief The btMatrix3x3 class implements a 3x3 rotation matrix, to perform linear algebra in combination with btQuaternion, btTransform and btVector3. |
---|
30 | * Make sure to only include a pure orthogonal matrix without scaling. */ |
---|
31 | class btMatrix3x3 { |
---|
32 | |
---|
33 | ///Data storage for the matrix, each vector is a row of the matrix |
---|
34 | btVector3 m_el[3]; |
---|
35 | |
---|
36 | public: |
---|
37 | /** @brief No initializaion constructor */ |
---|
38 | btMatrix3x3 () {} |
---|
39 | |
---|
40 | // explicit btMatrix3x3(const btScalar *m) { setFromOpenGLSubMatrix(m); } |
---|
41 | |
---|
42 | /**@brief Constructor from Quaternion */ |
---|
43 | explicit btMatrix3x3(const btQuaternion& q) { setRotation(q); } |
---|
44 | /* |
---|
45 | template <typename btScalar> |
---|
46 | Matrix3x3(const btScalar& yaw, const btScalar& pitch, const btScalar& roll) |
---|
47 | { |
---|
48 | setEulerYPR(yaw, pitch, roll); |
---|
49 | } |
---|
50 | */ |
---|
51 | /** @brief Constructor with row major formatting */ |
---|
52 | btMatrix3x3(const btScalar& xx, const btScalar& xy, const btScalar& xz, |
---|
53 | const btScalar& yx, const btScalar& yy, const btScalar& yz, |
---|
54 | const btScalar& zx, const btScalar& zy, const btScalar& zz) |
---|
55 | { |
---|
56 | setValue(xx, xy, xz, |
---|
57 | yx, yy, yz, |
---|
58 | zx, zy, zz); |
---|
59 | } |
---|
60 | /** @brief Copy constructor */ |
---|
61 | SIMD_FORCE_INLINE btMatrix3x3 (const btMatrix3x3& other) |
---|
62 | { |
---|
63 | m_el[0] = other.m_el[0]; |
---|
64 | m_el[1] = other.m_el[1]; |
---|
65 | m_el[2] = other.m_el[2]; |
---|
66 | } |
---|
67 | /** @brief Assignment Operator */ |
---|
68 | SIMD_FORCE_INLINE btMatrix3x3& operator=(const btMatrix3x3& other) |
---|
69 | { |
---|
70 | m_el[0] = other.m_el[0]; |
---|
71 | m_el[1] = other.m_el[1]; |
---|
72 | m_el[2] = other.m_el[2]; |
---|
73 | return *this; |
---|
74 | } |
---|
75 | |
---|
76 | /** @brief Get a column of the matrix as a vector |
---|
77 | * @param i Column number 0 indexed */ |
---|
78 | SIMD_FORCE_INLINE btVector3 getColumn(int i) const |
---|
79 | { |
---|
80 | return btVector3(m_el[0][i],m_el[1][i],m_el[2][i]); |
---|
81 | } |
---|
82 | |
---|
83 | |
---|
84 | /** @brief Get a row of the matrix as a vector |
---|
85 | * @param i Row number 0 indexed */ |
---|
86 | SIMD_FORCE_INLINE const btVector3& getRow(int i) const |
---|
87 | { |
---|
88 | btFullAssert(0 <= i && i < 3); |
---|
89 | return m_el[i]; |
---|
90 | } |
---|
91 | |
---|
92 | /** @brief Get a mutable reference to a row of the matrix as a vector |
---|
93 | * @param i Row number 0 indexed */ |
---|
94 | SIMD_FORCE_INLINE btVector3& operator[](int i) |
---|
95 | { |
---|
96 | btFullAssert(0 <= i && i < 3); |
---|
97 | return m_el[i]; |
---|
98 | } |
---|
99 | |
---|
100 | /** @brief Get a const reference to a row of the matrix as a vector |
---|
101 | * @param i Row number 0 indexed */ |
---|
102 | SIMD_FORCE_INLINE const btVector3& operator[](int i) const |
---|
103 | { |
---|
104 | btFullAssert(0 <= i && i < 3); |
---|
105 | return m_el[i]; |
---|
106 | } |
---|
107 | |
---|
108 | /** @brief Multiply by the target matrix on the right |
---|
109 | * @param m Rotation matrix to be applied |
---|
110 | * Equivilant to this = this * m */ |
---|
111 | btMatrix3x3& operator*=(const btMatrix3x3& m); |
---|
112 | |
---|
113 | /** @brief Set from a carray of btScalars |
---|
114 | * @param m A pointer to the beginning of an array of 9 btScalars */ |
---|
115 | void setFromOpenGLSubMatrix(const btScalar *m) |
---|
116 | { |
---|
117 | m_el[0].setValue(m[0],m[4],m[8]); |
---|
118 | m_el[1].setValue(m[1],m[5],m[9]); |
---|
119 | m_el[2].setValue(m[2],m[6],m[10]); |
---|
120 | |
---|
121 | } |
---|
122 | /** @brief Set the values of the matrix explicitly (row major) |
---|
123 | * @param xx Top left |
---|
124 | * @param xy Top Middle |
---|
125 | * @param xz Top Right |
---|
126 | * @param yx Middle Left |
---|
127 | * @param yy Middle Middle |
---|
128 | * @param yz Middle Right |
---|
129 | * @param zx Bottom Left |
---|
130 | * @param zy Bottom Middle |
---|
131 | * @param zz Bottom Right*/ |
---|
132 | void setValue(const btScalar& xx, const btScalar& xy, const btScalar& xz, |
---|
133 | const btScalar& yx, const btScalar& yy, const btScalar& yz, |
---|
134 | const btScalar& zx, const btScalar& zy, const btScalar& zz) |
---|
135 | { |
---|
136 | m_el[0].setValue(xx,xy,xz); |
---|
137 | m_el[1].setValue(yx,yy,yz); |
---|
138 | m_el[2].setValue(zx,zy,zz); |
---|
139 | } |
---|
140 | |
---|
141 | /** @brief Set the matrix from a quaternion |
---|
142 | * @param q The Quaternion to match */ |
---|
143 | void setRotation(const btQuaternion& q) |
---|
144 | { |
---|
145 | btScalar d = q.length2(); |
---|
146 | btFullAssert(d != btScalar(0.0)); |
---|
147 | btScalar s = btScalar(2.0) / d; |
---|
148 | btScalar xs = q.x() * s, ys = q.y() * s, zs = q.z() * s; |
---|
149 | btScalar wx = q.w() * xs, wy = q.w() * ys, wz = q.w() * zs; |
---|
150 | btScalar xx = q.x() * xs, xy = q.x() * ys, xz = q.x() * zs; |
---|
151 | btScalar yy = q.y() * ys, yz = q.y() * zs, zz = q.z() * zs; |
---|
152 | setValue(btScalar(1.0) - (yy + zz), xy - wz, xz + wy, |
---|
153 | xy + wz, btScalar(1.0) - (xx + zz), yz - wx, |
---|
154 | xz - wy, yz + wx, btScalar(1.0) - (xx + yy)); |
---|
155 | } |
---|
156 | |
---|
157 | |
---|
158 | /** @brief Set the matrix from euler angles using YPR around YXZ respectively |
---|
159 | * @param yaw Yaw about Y axis |
---|
160 | * @param pitch Pitch about X axis |
---|
161 | * @param roll Roll about Z axis |
---|
162 | */ |
---|
163 | void setEulerYPR(const btScalar& yaw, const btScalar& pitch, const btScalar& roll) |
---|
164 | { |
---|
165 | setEulerZYX(roll, pitch, yaw); |
---|
166 | } |
---|
167 | |
---|
168 | /** @brief Set the matrix from euler angles YPR around ZYX axes |
---|
169 | * @param eulerX Roll about X axis |
---|
170 | * @param eulerY Pitch around Y axis |
---|
171 | * @param eulerZ Yaw aboud Z axis |
---|
172 | * |
---|
173 | * These angles are used to produce a rotation matrix. The euler |
---|
174 | * angles are applied in ZYX order. I.e a vector is first rotated |
---|
175 | * about X then Y and then Z |
---|
176 | **/ |
---|
177 | void setEulerZYX(btScalar eulerX,btScalar eulerY,btScalar eulerZ) { |
---|
178 | ///@todo proposed to reverse this since it's labeled zyx but takes arguments xyz and it will match all other parts of the code |
---|
179 | btScalar ci ( btCos(eulerX)); |
---|
180 | btScalar cj ( btCos(eulerY)); |
---|
181 | btScalar ch ( btCos(eulerZ)); |
---|
182 | btScalar si ( btSin(eulerX)); |
---|
183 | btScalar sj ( btSin(eulerY)); |
---|
184 | btScalar sh ( btSin(eulerZ)); |
---|
185 | btScalar cc = ci * ch; |
---|
186 | btScalar cs = ci * sh; |
---|
187 | btScalar sc = si * ch; |
---|
188 | btScalar ss = si * sh; |
---|
189 | |
---|
190 | setValue(cj * ch, sj * sc - cs, sj * cc + ss, |
---|
191 | cj * sh, sj * ss + cc, sj * cs - sc, |
---|
192 | -sj, cj * si, cj * ci); |
---|
193 | } |
---|
194 | |
---|
195 | /**@brief Set the matrix to the identity */ |
---|
196 | void setIdentity() |
---|
197 | { |
---|
198 | setValue(btScalar(1.0), btScalar(0.0), btScalar(0.0), |
---|
199 | btScalar(0.0), btScalar(1.0), btScalar(0.0), |
---|
200 | btScalar(0.0), btScalar(0.0), btScalar(1.0)); |
---|
201 | } |
---|
202 | |
---|
203 | static const btMatrix3x3& getIdentity() |
---|
204 | { |
---|
205 | static const btMatrix3x3 identityMatrix(btScalar(1.0), btScalar(0.0), btScalar(0.0), |
---|
206 | btScalar(0.0), btScalar(1.0), btScalar(0.0), |
---|
207 | btScalar(0.0), btScalar(0.0), btScalar(1.0)); |
---|
208 | return identityMatrix; |
---|
209 | } |
---|
210 | |
---|
211 | /**@brief Fill the values of the matrix into a 9 element array |
---|
212 | * @param m The array to be filled */ |
---|
213 | void getOpenGLSubMatrix(btScalar *m) const |
---|
214 | { |
---|
215 | m[0] = btScalar(m_el[0].x()); |
---|
216 | m[1] = btScalar(m_el[1].x()); |
---|
217 | m[2] = btScalar(m_el[2].x()); |
---|
218 | m[3] = btScalar(0.0); |
---|
219 | m[4] = btScalar(m_el[0].y()); |
---|
220 | m[5] = btScalar(m_el[1].y()); |
---|
221 | m[6] = btScalar(m_el[2].y()); |
---|
222 | m[7] = btScalar(0.0); |
---|
223 | m[8] = btScalar(m_el[0].z()); |
---|
224 | m[9] = btScalar(m_el[1].z()); |
---|
225 | m[10] = btScalar(m_el[2].z()); |
---|
226 | m[11] = btScalar(0.0); |
---|
227 | } |
---|
228 | |
---|
229 | /**@brief Get the matrix represented as a quaternion |
---|
230 | * @param q The quaternion which will be set */ |
---|
231 | void getRotation(btQuaternion& q) const |
---|
232 | { |
---|
233 | btScalar trace = m_el[0].x() + m_el[1].y() + m_el[2].z(); |
---|
234 | btScalar temp[4]; |
---|
235 | |
---|
236 | if (trace > btScalar(0.0)) |
---|
237 | { |
---|
238 | btScalar s = btSqrt(trace + btScalar(1.0)); |
---|
239 | temp[3]=(s * btScalar(0.5)); |
---|
240 | s = btScalar(0.5) / s; |
---|
241 | |
---|
242 | temp[0]=((m_el[2].y() - m_el[1].z()) * s); |
---|
243 | temp[1]=((m_el[0].z() - m_el[2].x()) * s); |
---|
244 | temp[2]=((m_el[1].x() - m_el[0].y()) * s); |
---|
245 | } |
---|
246 | else |
---|
247 | { |
---|
248 | int i = m_el[0].x() < m_el[1].y() ? |
---|
249 | (m_el[1].y() < m_el[2].z() ? 2 : 1) : |
---|
250 | (m_el[0].x() < m_el[2].z() ? 2 : 0); |
---|
251 | int j = (i + 1) % 3; |
---|
252 | int k = (i + 2) % 3; |
---|
253 | |
---|
254 | btScalar s = btSqrt(m_el[i][i] - m_el[j][j] - m_el[k][k] + btScalar(1.0)); |
---|
255 | temp[i] = s * btScalar(0.5); |
---|
256 | s = btScalar(0.5) / s; |
---|
257 | |
---|
258 | temp[3] = (m_el[k][j] - m_el[j][k]) * s; |
---|
259 | temp[j] = (m_el[j][i] + m_el[i][j]) * s; |
---|
260 | temp[k] = (m_el[k][i] + m_el[i][k]) * s; |
---|
261 | } |
---|
262 | q.setValue(temp[0],temp[1],temp[2],temp[3]); |
---|
263 | } |
---|
264 | |
---|
265 | /**@brief Get the matrix represented as euler angles around YXZ, roundtrip with setEulerYPR |
---|
266 | * @param yaw Yaw around Y axis |
---|
267 | * @param pitch Pitch around X axis |
---|
268 | * @param roll around Z axis */ |
---|
269 | void getEulerYPR(btScalar& yaw, btScalar& pitch, btScalar& roll) const |
---|
270 | { |
---|
271 | |
---|
272 | // first use the normal calculus |
---|
273 | yaw = btScalar(btAtan2(m_el[1].x(), m_el[0].x())); |
---|
274 | pitch = btScalar(btAsin(-m_el[2].x())); |
---|
275 | roll = btScalar(btAtan2(m_el[2].y(), m_el[2].z())); |
---|
276 | |
---|
277 | // on pitch = +/-HalfPI |
---|
278 | if (btFabs(pitch)==SIMD_HALF_PI) |
---|
279 | { |
---|
280 | if (yaw>0) |
---|
281 | yaw-=SIMD_PI; |
---|
282 | else |
---|
283 | yaw+=SIMD_PI; |
---|
284 | |
---|
285 | if (roll>0) |
---|
286 | roll-=SIMD_PI; |
---|
287 | else |
---|
288 | roll+=SIMD_PI; |
---|
289 | } |
---|
290 | }; |
---|
291 | |
---|
292 | |
---|
293 | /**@brief Get the matrix represented as euler angles around ZYX |
---|
294 | * @param yaw Yaw around X axis |
---|
295 | * @param pitch Pitch around Y axis |
---|
296 | * @param roll around X axis |
---|
297 | * @param solution_number Which solution of two possible solutions ( 1 or 2) are possible values*/ |
---|
298 | void getEulerZYX(btScalar& yaw, btScalar& pitch, btScalar& roll, unsigned int solution_number = 1) const |
---|
299 | { |
---|
300 | struct Euler |
---|
301 | { |
---|
302 | btScalar yaw; |
---|
303 | btScalar pitch; |
---|
304 | btScalar roll; |
---|
305 | }; |
---|
306 | |
---|
307 | Euler euler_out; |
---|
308 | Euler euler_out2; //second solution |
---|
309 | //get the pointer to the raw data |
---|
310 | |
---|
311 | // Check that pitch is not at a singularity |
---|
312 | if (btFabs(m_el[2].x()) >= 1) |
---|
313 | { |
---|
314 | euler_out.yaw = 0; |
---|
315 | euler_out2.yaw = 0; |
---|
316 | |
---|
317 | // From difference of angles formula |
---|
318 | btScalar delta = btAtan2(m_el[0].x(),m_el[0].z()); |
---|
319 | if (m_el[2].x() > 0) //gimbal locked up |
---|
320 | { |
---|
321 | euler_out.pitch = SIMD_PI / btScalar(2.0); |
---|
322 | euler_out2.pitch = SIMD_PI / btScalar(2.0); |
---|
323 | euler_out.roll = euler_out.pitch + delta; |
---|
324 | euler_out2.roll = euler_out.pitch + delta; |
---|
325 | } |
---|
326 | else // gimbal locked down |
---|
327 | { |
---|
328 | euler_out.pitch = -SIMD_PI / btScalar(2.0); |
---|
329 | euler_out2.pitch = -SIMD_PI / btScalar(2.0); |
---|
330 | euler_out.roll = -euler_out.pitch + delta; |
---|
331 | euler_out2.roll = -euler_out.pitch + delta; |
---|
332 | } |
---|
333 | } |
---|
334 | else |
---|
335 | { |
---|
336 | euler_out.pitch = - btAsin(m_el[2].x()); |
---|
337 | euler_out2.pitch = SIMD_PI - euler_out.pitch; |
---|
338 | |
---|
339 | euler_out.roll = btAtan2(m_el[2].y()/btCos(euler_out.pitch), |
---|
340 | m_el[2].z()/btCos(euler_out.pitch)); |
---|
341 | euler_out2.roll = btAtan2(m_el[2].y()/btCos(euler_out2.pitch), |
---|
342 | m_el[2].z()/btCos(euler_out2.pitch)); |
---|
343 | |
---|
344 | euler_out.yaw = btAtan2(m_el[1].x()/btCos(euler_out.pitch), |
---|
345 | m_el[0].x()/btCos(euler_out.pitch)); |
---|
346 | euler_out2.yaw = btAtan2(m_el[1].x()/btCos(euler_out2.pitch), |
---|
347 | m_el[0].x()/btCos(euler_out2.pitch)); |
---|
348 | } |
---|
349 | |
---|
350 | if (solution_number == 1) |
---|
351 | { |
---|
352 | yaw = euler_out.yaw; |
---|
353 | pitch = euler_out.pitch; |
---|
354 | roll = euler_out.roll; |
---|
355 | } |
---|
356 | else |
---|
357 | { |
---|
358 | yaw = euler_out2.yaw; |
---|
359 | pitch = euler_out2.pitch; |
---|
360 | roll = euler_out2.roll; |
---|
361 | } |
---|
362 | } |
---|
363 | |
---|
364 | /**@brief Create a scaled copy of the matrix |
---|
365 | * @param s Scaling vector The elements of the vector will scale each column */ |
---|
366 | |
---|
367 | btMatrix3x3 scaled(const btVector3& s) const |
---|
368 | { |
---|
369 | return btMatrix3x3(m_el[0].x() * s.x(), m_el[0].y() * s.y(), m_el[0].z() * s.z(), |
---|
370 | m_el[1].x() * s.x(), m_el[1].y() * s.y(), m_el[1].z() * s.z(), |
---|
371 | m_el[2].x() * s.x(), m_el[2].y() * s.y(), m_el[2].z() * s.z()); |
---|
372 | } |
---|
373 | |
---|
374 | /**@brief Return the determinant of the matrix */ |
---|
375 | btScalar determinant() const; |
---|
376 | /**@brief Return the adjoint of the matrix */ |
---|
377 | btMatrix3x3 adjoint() const; |
---|
378 | /**@brief Return the matrix with all values non negative */ |
---|
379 | btMatrix3x3 absolute() const; |
---|
380 | /**@brief Return the transpose of the matrix */ |
---|
381 | btMatrix3x3 transpose() const; |
---|
382 | /**@brief Return the inverse of the matrix */ |
---|
383 | btMatrix3x3 inverse() const; |
---|
384 | |
---|
385 | btMatrix3x3 transposeTimes(const btMatrix3x3& m) const; |
---|
386 | btMatrix3x3 timesTranspose(const btMatrix3x3& m) const; |
---|
387 | |
---|
388 | SIMD_FORCE_INLINE btScalar tdotx(const btVector3& v) const |
---|
389 | { |
---|
390 | return m_el[0].x() * v.x() + m_el[1].x() * v.y() + m_el[2].x() * v.z(); |
---|
391 | } |
---|
392 | SIMD_FORCE_INLINE btScalar tdoty(const btVector3& v) const |
---|
393 | { |
---|
394 | return m_el[0].y() * v.x() + m_el[1].y() * v.y() + m_el[2].y() * v.z(); |
---|
395 | } |
---|
396 | SIMD_FORCE_INLINE btScalar tdotz(const btVector3& v) const |
---|
397 | { |
---|
398 | return m_el[0].z() * v.x() + m_el[1].z() * v.y() + m_el[2].z() * v.z(); |
---|
399 | } |
---|
400 | |
---|
401 | |
---|
402 | /**@brief diagonalizes this matrix by the Jacobi method. |
---|
403 | * @param rot stores the rotation from the coordinate system in which the matrix is diagonal to the original |
---|
404 | * coordinate system, i.e., old_this = rot * new_this * rot^T. |
---|
405 | * @param threshold See iteration |
---|
406 | * @param iteration The iteration stops when all off-diagonal elements are less than the threshold multiplied |
---|
407 | * by the sum of the absolute values of the diagonal, or when maxSteps have been executed. |
---|
408 | * |
---|
409 | * Note that this matrix is assumed to be symmetric. |
---|
410 | */ |
---|
411 | void diagonalize(btMatrix3x3& rot, btScalar threshold, int maxSteps) |
---|
412 | { |
---|
413 | rot.setIdentity(); |
---|
414 | for (int step = maxSteps; step > 0; step--) |
---|
415 | { |
---|
416 | // find off-diagonal element [p][q] with largest magnitude |
---|
417 | int p = 0; |
---|
418 | int q = 1; |
---|
419 | int r = 2; |
---|
420 | btScalar max = btFabs(m_el[0][1]); |
---|
421 | btScalar v = btFabs(m_el[0][2]); |
---|
422 | if (v > max) |
---|
423 | { |
---|
424 | q = 2; |
---|
425 | r = 1; |
---|
426 | max = v; |
---|
427 | } |
---|
428 | v = btFabs(m_el[1][2]); |
---|
429 | if (v > max) |
---|
430 | { |
---|
431 | p = 1; |
---|
432 | q = 2; |
---|
433 | r = 0; |
---|
434 | max = v; |
---|
435 | } |
---|
436 | |
---|
437 | btScalar t = threshold * (btFabs(m_el[0][0]) + btFabs(m_el[1][1]) + btFabs(m_el[2][2])); |
---|
438 | if (max <= t) |
---|
439 | { |
---|
440 | if (max <= SIMD_EPSILON * t) |
---|
441 | { |
---|
442 | return; |
---|
443 | } |
---|
444 | step = 1; |
---|
445 | } |
---|
446 | |
---|
447 | // compute Jacobi rotation J which leads to a zero for element [p][q] |
---|
448 | btScalar mpq = m_el[p][q]; |
---|
449 | btScalar theta = (m_el[q][q] - m_el[p][p]) / (2 * mpq); |
---|
450 | btScalar theta2 = theta * theta; |
---|
451 | btScalar cos; |
---|
452 | btScalar sin; |
---|
453 | if (theta2 * theta2 < btScalar(10 / SIMD_EPSILON)) |
---|
454 | { |
---|
455 | t = (theta >= 0) ? 1 / (theta + btSqrt(1 + theta2)) |
---|
456 | : 1 / (theta - btSqrt(1 + theta2)); |
---|
457 | cos = 1 / btSqrt(1 + t * t); |
---|
458 | sin = cos * t; |
---|
459 | } |
---|
460 | else |
---|
461 | { |
---|
462 | // approximation for large theta-value, i.e., a nearly diagonal matrix |
---|
463 | t = 1 / (theta * (2 + btScalar(0.5) / theta2)); |
---|
464 | cos = 1 - btScalar(0.5) * t * t; |
---|
465 | sin = cos * t; |
---|
466 | } |
---|
467 | |
---|
468 | // apply rotation to matrix (this = J^T * this * J) |
---|
469 | m_el[p][q] = m_el[q][p] = 0; |
---|
470 | m_el[p][p] -= t * mpq; |
---|
471 | m_el[q][q] += t * mpq; |
---|
472 | btScalar mrp = m_el[r][p]; |
---|
473 | btScalar mrq = m_el[r][q]; |
---|
474 | m_el[r][p] = m_el[p][r] = cos * mrp - sin * mrq; |
---|
475 | m_el[r][q] = m_el[q][r] = cos * mrq + sin * mrp; |
---|
476 | |
---|
477 | // apply rotation to rot (rot = rot * J) |
---|
478 | for (int i = 0; i < 3; i++) |
---|
479 | { |
---|
480 | btVector3& row = rot[i]; |
---|
481 | mrp = row[p]; |
---|
482 | mrq = row[q]; |
---|
483 | row[p] = cos * mrp - sin * mrq; |
---|
484 | row[q] = cos * mrq + sin * mrp; |
---|
485 | } |
---|
486 | } |
---|
487 | } |
---|
488 | |
---|
489 | |
---|
490 | |
---|
491 | |
---|
492 | /**@brief Calculate the matrix cofactor |
---|
493 | * @param r1 The first row to use for calculating the cofactor |
---|
494 | * @param c1 The first column to use for calculating the cofactor |
---|
495 | * @param r1 The second row to use for calculating the cofactor |
---|
496 | * @param c1 The second column to use for calculating the cofactor |
---|
497 | * See http://en.wikipedia.org/wiki/Cofactor_(linear_algebra) for more details |
---|
498 | */ |
---|
499 | btScalar cofac(int r1, int c1, int r2, int c2) const |
---|
500 | { |
---|
501 | return m_el[r1][c1] * m_el[r2][c2] - m_el[r1][c2] * m_el[r2][c1]; |
---|
502 | } |
---|
503 | |
---|
504 | void serialize(struct btMatrix3x3Data& dataOut) const; |
---|
505 | |
---|
506 | void serializeFloat(struct btMatrix3x3FloatData& dataOut) const; |
---|
507 | |
---|
508 | void deSerialize(const struct btMatrix3x3Data& dataIn); |
---|
509 | |
---|
510 | void deSerializeFloat(const struct btMatrix3x3FloatData& dataIn); |
---|
511 | |
---|
512 | void deSerializeDouble(const struct btMatrix3x3DoubleData& dataIn); |
---|
513 | |
---|
514 | }; |
---|
515 | |
---|
516 | |
---|
517 | SIMD_FORCE_INLINE btMatrix3x3& |
---|
518 | btMatrix3x3::operator*=(const btMatrix3x3& m) |
---|
519 | { |
---|
520 | setValue(m.tdotx(m_el[0]), m.tdoty(m_el[0]), m.tdotz(m_el[0]), |
---|
521 | m.tdotx(m_el[1]), m.tdoty(m_el[1]), m.tdotz(m_el[1]), |
---|
522 | m.tdotx(m_el[2]), m.tdoty(m_el[2]), m.tdotz(m_el[2])); |
---|
523 | return *this; |
---|
524 | } |
---|
525 | |
---|
526 | SIMD_FORCE_INLINE btScalar |
---|
527 | btMatrix3x3::determinant() const |
---|
528 | { |
---|
529 | return btTriple((*this)[0], (*this)[1], (*this)[2]); |
---|
530 | } |
---|
531 | |
---|
532 | |
---|
533 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
534 | btMatrix3x3::absolute() const |
---|
535 | { |
---|
536 | return btMatrix3x3( |
---|
537 | btFabs(m_el[0].x()), btFabs(m_el[0].y()), btFabs(m_el[0].z()), |
---|
538 | btFabs(m_el[1].x()), btFabs(m_el[1].y()), btFabs(m_el[1].z()), |
---|
539 | btFabs(m_el[2].x()), btFabs(m_el[2].y()), btFabs(m_el[2].z())); |
---|
540 | } |
---|
541 | |
---|
542 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
543 | btMatrix3x3::transpose() const |
---|
544 | { |
---|
545 | return btMatrix3x3(m_el[0].x(), m_el[1].x(), m_el[2].x(), |
---|
546 | m_el[0].y(), m_el[1].y(), m_el[2].y(), |
---|
547 | m_el[0].z(), m_el[1].z(), m_el[2].z()); |
---|
548 | } |
---|
549 | |
---|
550 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
551 | btMatrix3x3::adjoint() const |
---|
552 | { |
---|
553 | return btMatrix3x3(cofac(1, 1, 2, 2), cofac(0, 2, 2, 1), cofac(0, 1, 1, 2), |
---|
554 | cofac(1, 2, 2, 0), cofac(0, 0, 2, 2), cofac(0, 2, 1, 0), |
---|
555 | cofac(1, 0, 2, 1), cofac(0, 1, 2, 0), cofac(0, 0, 1, 1)); |
---|
556 | } |
---|
557 | |
---|
558 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
559 | btMatrix3x3::inverse() const |
---|
560 | { |
---|
561 | btVector3 co(cofac(1, 1, 2, 2), cofac(1, 2, 2, 0), cofac(1, 0, 2, 1)); |
---|
562 | btScalar det = (*this)[0].dot(co); |
---|
563 | btFullAssert(det != btScalar(0.0)); |
---|
564 | btScalar s = btScalar(1.0) / det; |
---|
565 | return btMatrix3x3(co.x() * s, cofac(0, 2, 2, 1) * s, cofac(0, 1, 1, 2) * s, |
---|
566 | co.y() * s, cofac(0, 0, 2, 2) * s, cofac(0, 2, 1, 0) * s, |
---|
567 | co.z() * s, cofac(0, 1, 2, 0) * s, cofac(0, 0, 1, 1) * s); |
---|
568 | } |
---|
569 | |
---|
570 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
571 | btMatrix3x3::transposeTimes(const btMatrix3x3& m) const |
---|
572 | { |
---|
573 | return btMatrix3x3( |
---|
574 | m_el[0].x() * m[0].x() + m_el[1].x() * m[1].x() + m_el[2].x() * m[2].x(), |
---|
575 | m_el[0].x() * m[0].y() + m_el[1].x() * m[1].y() + m_el[2].x() * m[2].y(), |
---|
576 | m_el[0].x() * m[0].z() + m_el[1].x() * m[1].z() + m_el[2].x() * m[2].z(), |
---|
577 | m_el[0].y() * m[0].x() + m_el[1].y() * m[1].x() + m_el[2].y() * m[2].x(), |
---|
578 | m_el[0].y() * m[0].y() + m_el[1].y() * m[1].y() + m_el[2].y() * m[2].y(), |
---|
579 | m_el[0].y() * m[0].z() + m_el[1].y() * m[1].z() + m_el[2].y() * m[2].z(), |
---|
580 | m_el[0].z() * m[0].x() + m_el[1].z() * m[1].x() + m_el[2].z() * m[2].x(), |
---|
581 | m_el[0].z() * m[0].y() + m_el[1].z() * m[1].y() + m_el[2].z() * m[2].y(), |
---|
582 | m_el[0].z() * m[0].z() + m_el[1].z() * m[1].z() + m_el[2].z() * m[2].z()); |
---|
583 | } |
---|
584 | |
---|
585 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
586 | btMatrix3x3::timesTranspose(const btMatrix3x3& m) const |
---|
587 | { |
---|
588 | return btMatrix3x3( |
---|
589 | m_el[0].dot(m[0]), m_el[0].dot(m[1]), m_el[0].dot(m[2]), |
---|
590 | m_el[1].dot(m[0]), m_el[1].dot(m[1]), m_el[1].dot(m[2]), |
---|
591 | m_el[2].dot(m[0]), m_el[2].dot(m[1]), m_el[2].dot(m[2])); |
---|
592 | |
---|
593 | } |
---|
594 | |
---|
595 | SIMD_FORCE_INLINE btVector3 |
---|
596 | operator*(const btMatrix3x3& m, const btVector3& v) |
---|
597 | { |
---|
598 | return btVector3(m[0].dot(v), m[1].dot(v), m[2].dot(v)); |
---|
599 | } |
---|
600 | |
---|
601 | |
---|
602 | SIMD_FORCE_INLINE btVector3 |
---|
603 | operator*(const btVector3& v, const btMatrix3x3& m) |
---|
604 | { |
---|
605 | return btVector3(m.tdotx(v), m.tdoty(v), m.tdotz(v)); |
---|
606 | } |
---|
607 | |
---|
608 | SIMD_FORCE_INLINE btMatrix3x3 |
---|
609 | operator*(const btMatrix3x3& m1, const btMatrix3x3& m2) |
---|
610 | { |
---|
611 | return btMatrix3x3( |
---|
612 | m2.tdotx( m1[0]), m2.tdoty( m1[0]), m2.tdotz( m1[0]), |
---|
613 | m2.tdotx( m1[1]), m2.tdoty( m1[1]), m2.tdotz( m1[1]), |
---|
614 | m2.tdotx( m1[2]), m2.tdoty( m1[2]), m2.tdotz( m1[2])); |
---|
615 | } |
---|
616 | |
---|
617 | /* |
---|
618 | SIMD_FORCE_INLINE btMatrix3x3 btMultTransposeLeft(const btMatrix3x3& m1, const btMatrix3x3& m2) { |
---|
619 | return btMatrix3x3( |
---|
620 | m1[0][0] * m2[0][0] + m1[1][0] * m2[1][0] + m1[2][0] * m2[2][0], |
---|
621 | m1[0][0] * m2[0][1] + m1[1][0] * m2[1][1] + m1[2][0] * m2[2][1], |
---|
622 | m1[0][0] * m2[0][2] + m1[1][0] * m2[1][2] + m1[2][0] * m2[2][2], |
---|
623 | m1[0][1] * m2[0][0] + m1[1][1] * m2[1][0] + m1[2][1] * m2[2][0], |
---|
624 | m1[0][1] * m2[0][1] + m1[1][1] * m2[1][1] + m1[2][1] * m2[2][1], |
---|
625 | m1[0][1] * m2[0][2] + m1[1][1] * m2[1][2] + m1[2][1] * m2[2][2], |
---|
626 | m1[0][2] * m2[0][0] + m1[1][2] * m2[1][0] + m1[2][2] * m2[2][0], |
---|
627 | m1[0][2] * m2[0][1] + m1[1][2] * m2[1][1] + m1[2][2] * m2[2][1], |
---|
628 | m1[0][2] * m2[0][2] + m1[1][2] * m2[1][2] + m1[2][2] * m2[2][2]); |
---|
629 | } |
---|
630 | */ |
---|
631 | |
---|
632 | /**@brief Equality operator between two matrices |
---|
633 | * It will test all elements are equal. */ |
---|
634 | SIMD_FORCE_INLINE bool operator==(const btMatrix3x3& m1, const btMatrix3x3& m2) |
---|
635 | { |
---|
636 | return ( m1[0][0] == m2[0][0] && m1[1][0] == m2[1][0] && m1[2][0] == m2[2][0] && |
---|
637 | m1[0][1] == m2[0][1] && m1[1][1] == m2[1][1] && m1[2][1] == m2[2][1] && |
---|
638 | m1[0][2] == m2[0][2] && m1[1][2] == m2[1][2] && m1[2][2] == m2[2][2] ); |
---|
639 | } |
---|
640 | |
---|
641 | ///for serialization |
---|
642 | struct btMatrix3x3FloatData |
---|
643 | { |
---|
644 | btVector3FloatData m_el[3]; |
---|
645 | }; |
---|
646 | |
---|
647 | ///for serialization |
---|
648 | struct btMatrix3x3DoubleData |
---|
649 | { |
---|
650 | btVector3DoubleData m_el[3]; |
---|
651 | }; |
---|
652 | |
---|
653 | |
---|
654 | |
---|
655 | |
---|
656 | SIMD_FORCE_INLINE void btMatrix3x3::serialize(struct btMatrix3x3Data& dataOut) const |
---|
657 | { |
---|
658 | for (int i=0;i<3;i++) |
---|
659 | m_el[i].serialize(dataOut.m_el[i]); |
---|
660 | } |
---|
661 | |
---|
662 | SIMD_FORCE_INLINE void btMatrix3x3::serializeFloat(struct btMatrix3x3FloatData& dataOut) const |
---|
663 | { |
---|
664 | for (int i=0;i<3;i++) |
---|
665 | m_el[i].serializeFloat(dataOut.m_el[i]); |
---|
666 | } |
---|
667 | |
---|
668 | |
---|
669 | SIMD_FORCE_INLINE void btMatrix3x3::deSerialize(const struct btMatrix3x3Data& dataIn) |
---|
670 | { |
---|
671 | for (int i=0;i<3;i++) |
---|
672 | m_el[i].deSerialize(dataIn.m_el[i]); |
---|
673 | } |
---|
674 | |
---|
675 | SIMD_FORCE_INLINE void btMatrix3x3::deSerializeFloat(const struct btMatrix3x3FloatData& dataIn) |
---|
676 | { |
---|
677 | for (int i=0;i<3;i++) |
---|
678 | m_el[i].deSerializeFloat(dataIn.m_el[i]); |
---|
679 | } |
---|
680 | |
---|
681 | SIMD_FORCE_INLINE void btMatrix3x3::deSerializeDouble(const struct btMatrix3x3DoubleData& dataIn) |
---|
682 | { |
---|
683 | for (int i=0;i<3;i++) |
---|
684 | m_el[i].deSerializeDouble(dataIn.m_el[i]); |
---|
685 | } |
---|
686 | |
---|
687 | #endif //BT_MATRIX3x3_H |
---|
688 | |
---|