mirror of
https://gitlab.com/chrony/chrony.git
synced 2025-12-03 16:35:06 -05:00
The algorithm was designed for estimating quantiles in streams of integer values. When the estimate is equal to the input value, the step state variable does not change. This causes problems for the floating-point adaptation used for measurents of delay in chrony. One problem is numerical instability due to the strict comparison of the input value and the current estimate. Another problem is with signals that are so stable that the nanosecond resolution of the system functions becomes the limitation. There is a large difference in the value of the step state variable, which determines how quickly the estimate will adapt to a new distribution, between signals that are constant in the nanosecond resolution and signals that can move in two nanoseconds. Change the estimate update to never consider the input value equal to the current estimate and don't set the estimate exactly to the input value. Keep it off by a quarter of the minimum step to force jumping around the input value if it's constant and decreasing the step variable to negative values. Also fix the initial adjustment to step at least by the minimum step (the original algorithm is described with ceil(), not fabs()).
230 lines
5.7 KiB
C
230 lines
5.7 KiB
C
/*
|
|
chronyd/chronyc - Programs for keeping computer clocks accurate.
|
|
|
|
**********************************************************************
|
|
* Copyright (C) Miroslav Lichvar 2022
|
|
*
|
|
* This program is free software; you can redistribute it and/or modify
|
|
* it under the terms of version 2 of the GNU General Public License as
|
|
* published by the Free Software Foundation.
|
|
*
|
|
* This program is distributed in the hope that it will be useful, but
|
|
* WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
* General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU General Public License along
|
|
* with this program; if not, write to the Free Software Foundation, Inc.,
|
|
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
*
|
|
**********************************************************************
|
|
|
|
=======================================================================
|
|
|
|
Estimation of quantiles using the Frugal-2U streaming algorithm
|
|
(https://arxiv.org/pdf/1407.1121v1.pdf)
|
|
*/
|
|
|
|
#include "config.h"
|
|
|
|
#include "logging.h"
|
|
#include "memory.h"
|
|
#include "quantiles.h"
|
|
#include "regress.h"
|
|
#include "util.h"
|
|
|
|
/* Maximum number of repeated estimates for stabilisation */
|
|
#define MAX_REPEAT 64
|
|
|
|
struct Quantile {
|
|
double est;
|
|
double step;
|
|
int sign;
|
|
};
|
|
|
|
struct QNT_Instance_Record {
|
|
struct Quantile *quants;
|
|
int n_quants;
|
|
int repeat;
|
|
int q;
|
|
int min_k;
|
|
double min_step;
|
|
int n_set;
|
|
};
|
|
|
|
/* ================================================== */
|
|
|
|
QNT_Instance
|
|
QNT_CreateInstance(int min_k, int max_k, int q, int repeat, double min_step)
|
|
{
|
|
QNT_Instance inst;
|
|
long seed;
|
|
|
|
if (q < 2 || min_k > max_k || min_k < 1 || max_k >= q ||
|
|
repeat < 1 || repeat > MAX_REPEAT || min_step <= 0.0)
|
|
assert(0);
|
|
|
|
inst = MallocNew(struct QNT_Instance_Record);
|
|
inst->n_quants = (max_k - min_k + 1) * repeat;
|
|
inst->quants = MallocArray(struct Quantile, inst->n_quants);
|
|
inst->repeat = repeat;
|
|
inst->q = q;
|
|
inst->min_k = min_k;
|
|
inst->min_step = min_step;
|
|
|
|
QNT_Reset(inst);
|
|
|
|
/* Seed the random number generator, which will not be isolated from
|
|
other instances and other random() users */
|
|
UTI_GetRandomBytes(&seed, sizeof (seed));
|
|
srandom(seed);
|
|
|
|
return inst;
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
void
|
|
QNT_DestroyInstance(QNT_Instance inst)
|
|
{
|
|
Free(inst->quants);
|
|
Free(inst);
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
void
|
|
QNT_Reset(QNT_Instance inst)
|
|
{
|
|
int i;
|
|
|
|
inst->n_set = 0;
|
|
|
|
for (i = 0; i < inst->n_quants; i++) {
|
|
inst->quants[i].est = 0.0;
|
|
inst->quants[i].step = inst->min_step;
|
|
inst->quants[i].sign = 1;
|
|
}
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
static void
|
|
insert_initial_value(QNT_Instance inst, double value)
|
|
{
|
|
int i, j, r = inst->repeat;
|
|
|
|
if (inst->n_set * r >= inst->n_quants)
|
|
assert(0);
|
|
|
|
/* Keep the initial estimates repeated and ordered */
|
|
for (i = inst->n_set; i > 0 && inst->quants[(i - 1) * r].est > value; i--) {
|
|
for (j = 0; j < r; j++)
|
|
inst->quants[i * r + j].est = inst->quants[(i - 1) * r].est;
|
|
}
|
|
|
|
for (j = 0; j < r; j++)
|
|
inst->quants[i * r + j].est = value;
|
|
inst->n_set++;
|
|
|
|
/* Duplicate the largest value in unset quantiles */
|
|
for (i = inst->n_set * r; i < inst->n_quants; i++)
|
|
inst->quants[i].est = inst->quants[i - 1].est;
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
static void
|
|
update_estimate(struct Quantile *quantile, double value, double p, double rand,
|
|
double min_step)
|
|
{
|
|
if (value >= quantile->est) {
|
|
if (rand < (1.0 - p))
|
|
return;
|
|
quantile->step += quantile->sign > 0 ? min_step : -min_step;
|
|
quantile->est += quantile->step > min_step ? quantile->step : min_step;
|
|
if (quantile->est > value) {
|
|
quantile->step += value - quantile->est;
|
|
quantile->est = value + min_step / 4.0;
|
|
}
|
|
if (quantile->sign < 0 && quantile->step > min_step)
|
|
quantile->step = min_step;
|
|
quantile->sign = 1;
|
|
} else {
|
|
if (rand < p)
|
|
return;
|
|
quantile->step += quantile->sign < 0 ? min_step : -min_step;
|
|
quantile->est -= quantile->step > min_step ? quantile->step : min_step;
|
|
if (quantile->est < value) {
|
|
quantile->step += quantile->est - value;
|
|
quantile->est = value - min_step / 4.0;
|
|
}
|
|
if (quantile->sign > 0 && quantile->step > min_step)
|
|
quantile->step = min_step;
|
|
quantile->sign = -1;
|
|
}
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
void
|
|
QNT_Accumulate(QNT_Instance inst, double value)
|
|
{
|
|
double p, rand;
|
|
int i;
|
|
|
|
/* Initialise the estimates with first received values */
|
|
if (inst->n_set * inst->repeat < inst->n_quants) {
|
|
insert_initial_value(inst, value);
|
|
return;
|
|
}
|
|
|
|
for (i = 0; i < inst->n_quants; i++) {
|
|
p = (double)(i / inst->repeat + inst->min_k) / inst->q;
|
|
rand = (double)random() / ((1U << 31) - 1);
|
|
|
|
update_estimate(&inst->quants[i], value, p, rand, inst->min_step);
|
|
}
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
int
|
|
QNT_GetMinK(QNT_Instance inst)
|
|
{
|
|
return inst->min_k;
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
int
|
|
QNT_GetMaxK(QNT_Instance inst)
|
|
{
|
|
return inst->min_k + (inst->n_quants / inst->repeat) - 1;
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
double
|
|
QNT_GetMinStep(QNT_Instance inst)
|
|
{
|
|
return inst->min_step;
|
|
}
|
|
|
|
/* ================================================== */
|
|
|
|
double
|
|
QNT_GetQuantile(QNT_Instance inst, int k)
|
|
{
|
|
double estimates[MAX_REPEAT];
|
|
int i;
|
|
|
|
if (k < inst->min_k || (k - inst->min_k) * inst->repeat >= inst->n_quants)
|
|
assert(0);
|
|
|
|
for (i = 0; i < inst->repeat; i++)
|
|
estimates[i] = inst->quants[(k - inst->min_k) * inst->repeat + i].est;
|
|
|
|
return RGR_FindMedian(estimates, inst->repeat);
|
|
}
|