1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_var_f32.c
4 * Description: Variance of the elements of a floating-point vector
5 *
6 * $Date: 23 April 2021
7 * $Revision: V1.9.0
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
13 *
14 * SPDX-License-Identifier: Apache-2.0
15 *
16 * Licensed under the Apache License, Version 2.0 (the License); you may
17 * not use this file except in compliance with the License.
18 * You may obtain a copy of the License at
19 *
20 * www.apache.org/licenses/LICENSE-2.0
21 *
22 * Unless required by applicable law or agreed to in writing, software
23 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
24 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
25 * See the License for the specific language governing permissions and
26 * limitations under the License.
27 */
28
29 #include "dsp/statistics_functions.h"
30
31 /**
32 @ingroup groupStats
33 */
34
35 /**
36 @defgroup variance Variance
37
38 Calculates the variance of the elements in the input vector.
39 The underlying algorithm used is the direct method sometimes referred to as the two-pass method:
40
41 <pre>
42 Result = sum(element - meanOfElements)^2) / numElement - 1
43
44 meanOfElements = ( pSrc[0] * pSrc[0] + pSrc[1] * pSrc[1] + ... + pSrc[blockSize-1] ) / blockSize
45 </pre>
46
47 There are separate functions for floating point, Q31, and Q15 data types.
48 */
49
50 /**
51 @addtogroup variance
52 @{
53 */
54
55 /**
56 @brief Variance of the elements of a floating-point vector.
57 @param[in] pSrc points to the input vector
58 @param[in] blockSize number of samples in input vector
59 @param[out] pResult variance value returned here
60 @return none
61 */
62 #if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
63
64 #include "arm_helium_utils.h"
65
arm_var_f32(const float32_t * pSrc,uint32_t blockSize,float32_t * pResult)66 void arm_var_f32(
67 const float32_t * pSrc,
68 uint32_t blockSize,
69 float32_t * pResult)
70 {
71 uint32_t blkCnt; /* loop counters */
72 f32x4_t vecSrc;
73 f32x4_t sumVec = vdupq_n_f32(0.0f);
74 float32_t fMean;
75 float32_t sum = 0.0f; /* accumulator */
76 float32_t in; /* Temporary variable to store input value */
77
78 if (blockSize <= 1U) {
79 *pResult = 0;
80 return;
81 }
82
83 arm_mean_f32(pSrc, blockSize, &fMean);
84
85 /* Compute 4 outputs at a time */
86 blkCnt = blockSize >> 2U;
87 while (blkCnt > 0U)
88 {
89
90 vecSrc = vldrwq_f32(pSrc);
91 /*
92 * sum lanes
93 */
94 vecSrc = vsubq(vecSrc, fMean);
95 sumVec = vfmaq(sumVec, vecSrc, vecSrc);
96
97 blkCnt --;
98 pSrc += 4;
99 }
100
101 sum = vecAddAcrossF32Mve(sumVec);
102
103 /*
104 * tail
105 */
106 blkCnt = blockSize & 0x3;
107 while (blkCnt > 0U)
108 {
109 in = *pSrc++ - fMean;
110 sum += in * in;
111
112 /* Decrement loop counter */
113 blkCnt--;
114 }
115
116 /* Variance */
117 *pResult = sum / (float32_t) (blockSize - 1);
118 }
119 #else
120 #if defined(ARM_MATH_NEON_EXPERIMENTAL) && !defined(ARM_MATH_AUTOVECTORIZE)
arm_var_f32(const float32_t * pSrc,uint32_t blockSize,float32_t * pResult)121 void arm_var_f32(
122 const float32_t * pSrc,
123 uint32_t blockSize,
124 float32_t * pResult)
125 {
126 float32_t mean;
127
128 float32_t sum = 0.0f; /* accumulator */
129 float32_t in; /* Temporary variable to store input value */
130 uint32_t blkCnt; /* loop counter */
131
132 float32x4_t sumV = vdupq_n_f32(0.0f); /* Temporary result storage */
133 float32x2_t sumV2;
134 float32x4_t inV;
135 float32x4_t avg;
136
137 arm_mean_f32(pSrc,blockSize,&mean);
138 avg = vdupq_n_f32(mean);
139
140 blkCnt = blockSize >> 2U;
141
142 /* Compute 4 outputs at a time.
143 ** a second loop below computes the remaining 1 to 3 samples. */
144 while (blkCnt > 0U)
145 {
146 /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
147 /* Compute Power and then store the result in a temporary variable, sum. */
148 inV = vld1q_f32(pSrc);
149 inV = vsubq_f32(inV, avg);
150 sumV = vmlaq_f32(sumV, inV, inV);
151 pSrc += 4;
152
153 /* Decrement the loop counter */
154 blkCnt--;
155 }
156
157 sumV2 = vpadd_f32(vget_low_f32(sumV),vget_high_f32(sumV));
158 sum = vget_lane_f32(sumV2, 0) + vget_lane_f32(sumV2, 1);
159
160 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
161 ** No loop unrolling is used. */
162 blkCnt = blockSize % 0x4U;
163
164 while (blkCnt > 0U)
165 {
166 /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
167 /* compute power and then store the result in a temporary variable, sum. */
168 in = *pSrc++;
169 in = in - mean;
170 sum += in * in;
171
172 /* Decrement the loop counter */
173 blkCnt--;
174 }
175
176 /* Variance */
177 *pResult = sum / (float32_t)(blockSize - 1.0f);
178
179 }
180
181 #else
arm_var_f32(const float32_t * pSrc,uint32_t blockSize,float32_t * pResult)182 void arm_var_f32(
183 const float32_t * pSrc,
184 uint32_t blockSize,
185 float32_t * pResult)
186 {
187 uint32_t blkCnt; /* Loop counter */
188 float32_t sum = 0.0f; /* Temporary result storage */
189 float32_t fSum = 0.0f;
190 float32_t fMean, fValue;
191 const float32_t * pInput = pSrc;
192
193 if (blockSize <= 1U)
194 {
195 *pResult = 0;
196 return;
197 }
198
199 #if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
200
201 /* Loop unrolling: Compute 4 outputs at a time */
202 blkCnt = blockSize >> 2U;
203
204 while (blkCnt > 0U)
205 {
206 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
207
208 sum += *pInput++;
209 sum += *pInput++;
210 sum += *pInput++;
211 sum += *pInput++;
212
213
214 /* Decrement loop counter */
215 blkCnt--;
216 }
217
218 /* Loop unrolling: Compute remaining outputs */
219 blkCnt = blockSize % 0x4U;
220
221 #else
222
223 /* Initialize blkCnt with number of samples */
224 blkCnt = blockSize;
225
226 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
227
228 while (blkCnt > 0U)
229 {
230 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
231
232 sum += *pInput++;
233
234 /* Decrement loop counter */
235 blkCnt--;
236 }
237
238 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
239 fMean = sum / (float32_t) blockSize;
240
241 pInput = pSrc;
242
243 #if defined (ARM_MATH_LOOPUNROLL) && !defined(ARM_MATH_AUTOVECTORIZE)
244
245 /* Loop unrolling: Compute 4 outputs at a time */
246 blkCnt = blockSize >> 2U;
247
248 while (blkCnt > 0U)
249 {
250 fValue = *pInput++ - fMean;
251 fSum += fValue * fValue;
252
253 fValue = *pInput++ - fMean;
254 fSum += fValue * fValue;
255
256 fValue = *pInput++ - fMean;
257 fSum += fValue * fValue;
258
259 fValue = *pInput++ - fMean;
260 fSum += fValue * fValue;
261
262 /* Decrement loop counter */
263 blkCnt--;
264 }
265
266 /* Loop unrolling: Compute remaining outputs */
267 blkCnt = blockSize % 0x4U;
268
269 #else
270
271 /* Initialize blkCnt with number of samples */
272 blkCnt = blockSize;
273
274 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
275
276 while (blkCnt > 0U)
277 {
278 fValue = *pInput++ - fMean;
279 fSum += fValue * fValue;
280
281 /* Decrement loop counter */
282 blkCnt--;
283 }
284
285 /* Variance */
286 *pResult = fSum / (float32_t)(blockSize - 1.0f);
287 }
288 #endif /* #if defined(ARM_MATH_NEON) */
289 #endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
290
291 /**
292 @} end of variance group
293 */
294