Develop Reference

Linear interpolation between two numbers with steps

I've a little trouble finding out how to linearly interpolate between two numbers with a defined number of intermediate steps.
Let's say I want to interpolate between 4 and 22 with 8 intermediate steps like so : Example
It's easy to figure out that it's x+2 here. But what if the starting value was 5 and the final value 532 with 12 intermediate steps? (In my special case I would need starting and ending value with 16 steps in between)

If you have two fence posts and you put k fence posts between them, you create k + 1 spaces. For instance:
| |
post1 post2
adding one posts creates two spaces
| | |
post1 post2
If you want those k + 1 spaces to be equal you can divide the total distance by k + 1 to get the distance between adjacent posts.
d = 22 - 4 = 18
k = 8
e = d / (k + 1) = 18 / 9 = 2
In your other case example, the answer is
d = 532 - 5 = 527
k = 12
e = d / (k + 1) = 527 / 13 ~ 40.5

I hesitate to produce two separate answers, but I feel this methodology is sufficiently unique from the other one. There's a useful function which may be exactly what you need which is appropriately called Mathf.Lerp().
var start = 5;
var end = 532;
var steps = 13;
for (int i = 0; i <= steps; i++) {
// The type conversion is necessary because both i and steps are integers
var value = Mathf.Lerp(start, end, i / (float)steps);
Debug.Log(value);
}

For actually doing the linear interpolation, use Mathf.MoveTowards().
For figuring out your maximum delta (i.e. the amount you want it to move each step), take the difference, and then divide it by the number of desired steps.
var start = 4;
var end = 22;
var distance = end - start;
var steps = 9; // Your example technically has 9 steps, not 8
var delta = distance / steps;
Note that this conveniently assumes your distance is a clean multiple of steps. If you don't know this is the case and it's important that you never exceed that number of steps, you may want to explicitly check for it. Here's a crude example for an integer. Floating point methods may be more complicated:
if (distance % delta > 0) { delta += 1; }

How to interpolate through 3 points/numbers with a defined number of samples? (in c#)

So for example we have 1, 5, and 10 and we want to interpolate between these with 12 points, we should get:
1.0000
1.7273
2.4545
3.1818
3.9091
4.6364
5.4545
6.3636
7.2727
8.1818
9.0909
10.0000
say we have 5, 10, and 4 and again 12 points, we should get:
5.0000
5.9091
6.8182
7.7273
8.6364
9.5455
9.4545
8.3636
7.2727
6.1818
5.0909
4.0000

This is a generalized solution that works by these principles:
Performs linear interpolation
It calculates a "floating point index" into the input array
This index is used to select 1 (if the fractional parts is very close to 0) or 2 numbers from the input array
The integer part of this index is the base input array index
The fractional part says how far towards the next array element we should move
This should work with whatever size input arrays and output collections you would need.
public IEnumerable<double> Interpolate(double[] inputs, int count)
{
double maxCountForIndexCalculation = count - 1;
for (int index = 0; index < count; index++)
{
double floatingIndex = (index / maxCountForIndexCalculation) * (inputs.Length - 1);
int baseIndex = (int)floatingIndex;
double fraction = floatingIndex - baseIndex;
if (Math.Abs(fraction) < 1e-5)
yield return inputs[baseIndex];
else
{
double delta = inputs[baseIndex + 1] - inputs[baseIndex];
yield return inputs[baseIndex] + fraction * delta;
}
}
}
It produces the two collections of outputs you showed in your question but beyond that, I have not tested it. Little error checking is performed so you should add the necessary bits.

The problem is an interpolation of two straight lines with different slopes given the end points and the intersection.
Interpolation is defined as following : In the mathematical field of numerical analysis, interpolation is a method of constructing new data points within the range of a discrete set of known data points.
I'm tired of people giving negative points for solutions to hard problems. This is not a simply problem, but a problem that require "thinking out of the box". lets looks at the solution for following input : 1 12 34
I picked these numbers because the results are all integers
The step size L (Lower) = distance of elements from 1 to 12 = 2
The step size H (Higher) = distance of elements from 12 to 34 = 4
So the answer is : 1 3 5 7 9 11 [12] 14 18 22 26 30 34
Notice the distance between the 6th point 11 and center is 1 (half of L)
Notice the distance between the center point 12 and the 7th point is 2 (half of H)
Finally notice the distance between the 6th and 7th points is 3.
My results are scaled exactly the same as the OPs first example.
It is hard to see the sequence with the fractional inputs the OP posted. If you look at the OP first example and calculate the step distance of the first 6 points you get 0.72. The last 6 points the distance is 0.91. Then calculate the distance from the 6th point to the center is .36 (half 0.72). Then center to 7th point 0.45 (half 0.91). Excuse me for rounding the numbers a little bit.
It is a sequence problem just like the in junior high school where you learned arithmetic and geometric sequences. Then as a bonus question you got the sequence 23, 28, 33, 42,51,59,68,77,86 which turns out to be the train stations on the NYC 3rd Ave subway system. Solving problems like this you need to think "Outside the Box" which comes from the tests IBM gives to Job Applicants. These are the people who can solve the Nine Point Problem : http://www.brainstorming.co.uk/puzzles/ninedotsnj.html
I did the results when the number of points is EVEN which in you case is 12. You will need to complete the code if the number of points is ODD.
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
namespace ConsoleApplication1
{
class Program
{
const int NUMBER_POINTS = 12;
static void Main(string[] args)
{
List<List<float>> tests = new List<List<float>>() {
new List<float>() { 1,5, 10},
new List<float>() { 5,10, 4}
};
foreach (List<float> test in tests)
{
List<float> output = new List<float>();
float midPoint = test[1];
if(NUMBER_POINTS % 2 == 0)
{
//even number of points
//add lower numbers
float lowerDelta = (test[1] - test[0])/((NUMBER_POINTS / 2) - .5F);
for (int i = 0; i < NUMBER_POINTS / 2; i++)
{
output.Add(test[0] + (i * lowerDelta));
}
float upperDelta = (test[2] - test[1]) / ((NUMBER_POINTS / 2) - .5F); ;
for (int i = 0; i < NUMBER_POINTS / 2; i++)
{
output.Add(test[1] + (i * upperDelta) + (upperDelta / 2F));
}
}
else
{
}
Console.WriteLine("Numbers = {0}", string.Join(" ", output.Select(x => x.ToString())));
}
Console.ReadLine();
}
}
}

Fast Exp calculation: possible to improve accuracy without losing too much performance?

I am trying out the fast Exp(x) function that previously was described in this answer to an SO question on improving calculation speed in C#:
public static double Exp(double x)
{
var tmp = (long)(1512775 * x + 1072632447);
return BitConverter.Int64BitsToDouble(tmp << 32);
}
The expression is using some IEEE floating point "tricks" and is primarily intended for use in neural sets. The function is approximately 5 times faster than the regular Math.Exp(x) function.
Unfortunately, the numeric accuracy is only -4% -- +2% relative to the regular Math.Exp(x) function, ideally I would like to have accuracy within at least the sub-percent range.
I have plotted the quotient between the approximate and the regular Exp functions, and as can be seen in the graph the relative difference appears to be repeated with practically constant frequency.
Is it possible to take advantage of this regularity to improve the accuracy of the "fast exp" function further without substantially reducing the calculation speed, or would the computational overhead of an accuracy improvement outweigh the computational gain of the original expression?
(As a side note, I have also tried one of the alternative approaches proposed in the same SO question, but this approach does not seem to be computationally efficient in C#, at least not for the general case.)
UPDATE MAY 14
Upon request from #Adriano, I have now performed a very simple benchmark. I have performed 10 million computations using each of the alternative exp functions for floating point values in the range [-100, 100]. Since the range of values I am interested in spans from -20 to 0 I have also explicitly listed the function value at x = -5. Here are the results:
Math.Exp: 62.525 ms, exp(-5) = 0.00673794699908547
Empty function: 13.769 ms
ExpNeural: 14.867 ms, exp(-5) = 0.00675211846828461
ExpSeries8: 15.121 ms, exp(-5) = 0.00641270968867667
ExpSeries16: 32.046 ms, exp(-5) = 0.00673666189488182
exp1: 15.062 ms, exp(-5) = -12.3333325982094
exp2: 15.090 ms, exp(-5) = 13.708332516253
exp3: 16.251 ms, exp(-5) = -12.3333325982094
exp4: 17.924 ms, exp(-5) = 728.368055056781
exp5: 20.972 ms, exp(-5) = -6.13293614238501
exp6: 24.212 ms, exp(-5) = 3.55518353166184
exp7: 29.092 ms, exp(-5) = -1.8271053775984
exp7 +/-: 38.482 ms, exp(-5) = 0.00695945286970704
ExpNeural is equivalent to the Exp function specified in the beginning of this text. ExpSeries8 is the formulation that I originally claimed was not very efficient on .NET; when implementing it exactly like Neil it was actually very fast. ExpSeries16 is the analogous formula but with 16 multiplications instead of 8. exp1 through exp7 are the different functions from Adriano's answer below. The final variant of exp7 is a variant where the sign of x is checked; if negative the function returns 1/exp(-x) instead.
Unfortunately, neither of the expN functions listed by Adriano are sufficient in the broader negative value range I am considering. The series expansion approach by Neil Coffey seems to be more suitable in "my" value range, although it is too rapidly diverging with larger negative x, especially when using "only" 8 multiplications.

Taylor series approximations (such as the expX() functions in Adriano's answer) are most accurate near zero and can have huge errors at -20 or even -5. If the input has a known range, such as -20 to 0 like the original question, you can use a small look up table and one additional multiply to greatly improve accuracy.
The trick is to recognize that exp() can be separated into integer and fractional parts. For example:
exp(-2.345) = exp(-2.0) * exp(-0.345)
The fractional part will always be between -1 and 1, so a Taylor series approximation will be pretty accurate. The integer part has only 21 possible values for exp(-20) to exp(0), so these can be stored in a small look up table.

Try following alternatives (exp1 is faster, exp7 is more precise).
Code
public static double exp1(double x) {
return (6+x*(6+x*(3+x)))*0.16666666f;
}
public static double exp2(double x) {
return (24+x*(24+x*(12+x*(4+x))))*0.041666666f;
}
public static double exp3(double x) {
return (120+x*(120+x*(60+x*(20+x*(5+x)))))*0.0083333333f;
}
public static double exp4(double x) {
return 720+x*(720+x*(360+x*(120+x*(30+x*(6+x))))))*0.0013888888f;
}
public static double exp5(double x) {
return (5040+x*(5040+x*(2520+x*(840+x*(210+x*(42+x*(7+x)))))))*0.00019841269f;
}
public static double exp6(double x) {
return (40320+x*(40320+x*(20160+x*(6720+x*(1680+x*(336+x*(56+x*(8+x))))))))*2.4801587301e-5;
}
public static double exp7(double x) {
return (362880+x*(362880+x*(181440+x*(60480+x*(15120+x*(3024+x*(504+x*(72+x*(9+x)))))))))*2.75573192e-6;
}
Precision
Function Error in [-1...1] Error in [3.14...3.14]
exp1 0.05 1.8% 8.8742 38.40%
exp2 0.01 0.36% 4.8237 20.80%
exp3 0.0016152 0.59% 2.28 9.80%
exp4 0.0002263 0.0083% 0.9488 4.10%
exp5 0.0000279 0.001% 0.3516 1.50%
exp6 0.0000031 0.00011% 0.1172 0.50%
exp7 0.0000003 0.000011% 0.0355 0.15%
Credits
These implementations of exp() have been calculated by "scoofy" using Taylor series from a tanh() implementation of "fuzzpilz" (whoever they are, I just had these references on my code).

In case anyone wants to replicate the relative error function shown in the question, here's a way using Matlab (the "fast" exponent is not very fast in Matlab, but it is accurate):
t = 1072632447+[0:ceil(1512775*pi)];
x = (t - 1072632447)/1512775;
ex = exp(x);
t = uint64(t);
import java.lang.Double;
et = arrayfun( #(n) java.lang.Double.longBitsToDouble(bitshift(n,32)), t );
plot(x, et./ex);
Now, the period of the error exactly coincides with when the binary value of tmp overflows from the mantissa into the exponent. Let's break our data into bins by discarding the bits that become the exponent (making it periodic), and keeping only the high eight remaining bits (to make our lookup table a reasonable size):
index = bitshift(bitand(t,uint64(2^20-2^12)),-12) + 1;
Now we calculate the mean required adjustment:
relerrfix = ex./et;
adjust = NaN(1,256);
for i=1:256; adjust(i) = mean(relerrfix(index == i)); end;
et2 = et .* adjust(index);
The relative error is decreased to +/- .0006. Of course, other tables sizes are possible as well (for example, a 6-bit table with 64 entries gives +/- .0025) and the error is almost linear in table size. Linear interpolation between table entries would improve the error yet further, but at the expense of performance. Since we've already met the accuracy goal, let's avoid any further performance hits.
At this point it's some trivial editor skills to take the values computed by MatLab and create a lookup table in C#. For each computation, we add a bitmask, table lookup, and double-precision multiply.
static double FastExp(double x)
{
var tmp = (long)(1512775 * x + 1072632447);
int index = (int)(tmp >> 12) & 0xFF;
return BitConverter.Int64BitsToDouble(tmp << 32) * ExpAdjustment[index];
}
The speedup is very similar to the original code -- for my computer, this is about 30% faster compiled as x86 and about 3x as fast for x64. With mono on ideone, it's a substantial net loss (but so is the original).
Complete source code and testcase: http://ideone.com/UwNgx
using System;
using System.Diagnostics;
namespace fastexponent
{
class Program
{
static double[] ExpAdjustment = new double[256] {
1.040389835,
1.039159306,
1.037945888,
1.036749401,
1.035569671,
1.034406528,
1.033259801,
1.032129324,
1.031014933,
1.029916467,
1.028833767,
1.027766676,
1.02671504,
1.025678708,
1.02465753,
1.023651359,
1.022660049,
1.021683458,
1.020721446,
1.019773873,
1.018840604,
1.017921503,
1.017016438,
1.016125279,
1.015247897,
1.014384165,
1.013533958,
1.012697153,
1.011873629,
1.011063266,
1.010265947,
1.009481555,
1.008709975,
1.007951096,
1.007204805,
1.006470993,
1.005749552,
1.005040376,
1.004343358,
1.003658397,
1.002985389,
1.002324233,
1.001674831,
1.001037085,
1.000410897,
0.999796173,
0.999192819,
0.998600742,
0.998019851,
0.997450055,
0.996891266,
0.996343396,
0.995806358,
0.995280068,
0.99476444,
0.994259393,
0.993764844,
0.993280711,
0.992806917,
0.992343381,
0.991890026,
0.991446776,
0.991013555,
0.990590289,
0.990176903,
0.989773325,
0.989379484,
0.988995309,
0.988620729,
0.988255677,
0.987900083,
0.987553882,
0.987217006,
0.98688939,
0.98657097,
0.986261682,
0.985961463,
0.985670251,
0.985387985,
0.985114604,
0.984850048,
0.984594259,
0.984347178,
0.984108748,
0.983878911,
0.983657613,
0.983444797,
0.983240409,
0.983044394,
0.982856701,
0.982677276,
0.982506066,
0.982343022,
0.982188091,
0.982041225,
0.981902373,
0.981771487,
0.981648519,
0.981533421,
0.981426146,
0.981326648,
0.98123488,
0.981150798,
0.981074356,
0.981005511,
0.980944219,
0.980890437,
0.980844122,
0.980805232,
0.980773726,
0.980749562,
0.9807327,
0.9807231,
0.980720722,
0.980725528,
0.980737478,
0.980756534,
0.98078266,
0.980815817,
0.980855968,
0.980903079,
0.980955475,
0.981017942,
0.981085714,
0.981160303,
0.981241675,
0.981329796,
0.981424634,
0.981526154,
0.981634325,
0.981749114,
0.981870489,
0.981998419,
0.982132873,
0.98227382,
0.982421229,
0.982575072,
0.982735318,
0.982901937,
0.983074902,
0.983254183,
0.983439752,
0.983631582,
0.983829644,
0.984033912,
0.984244358,
0.984460956,
0.984683681,
0.984912505,
0.985147403,
0.985388349,
0.98563532,
0.98588829,
0.986147234,
0.986412128,
0.986682949,
0.986959673,
0.987242277,
0.987530737,
0.987825031,
0.988125136,
0.98843103,
0.988742691,
0.989060098,
0.989383229,
0.989712063,
0.990046579,
0.990386756,
0.990732574,
0.991084012,
0.991441052,
0.991803672,
0.992171854,
0.992545578,
0.992924825,
0.993309578,
0.993699816,
0.994095522,
0.994496677,
0.994903265,
0.995315266,
0.995732665,
0.996155442,
0.996583582,
0.997017068,
0.997455883,
0.99790001,
0.998349434,
0.998804138,
0.999264107,
0.999729325,
1.000199776,
1.000675446,
1.001156319,
1.001642381,
1.002133617,
1.002630011,
1.003131551,
1.003638222,
1.00415001,
1.004666901,
1.005188881,
1.005715938,
1.006248058,
1.006785227,
1.007327434,
1.007874665,
1.008426907,
1.008984149,
1.009546377,
1.010113581,
1.010685747,
1.011262865,
1.011844922,
1.012431907,
1.013023808,
1.013620615,
1.014222317,
1.014828902,
1.01544036,
1.016056681,
1.016677853,
1.017303866,
1.017934711,
1.018570378,
1.019210855,
1.019856135,
1.020506206,
1.02116106,
1.021820687,
1.022485078,
1.023154224,
1.023828116,
1.024506745,
1.025190103,
1.02587818,
1.026570969,
1.027268461,
1.027970647,
1.02867752,
1.029389072,
1.030114973,
1.030826088,
1.03155163,
1.032281819,
1.03301665,
1.033756114,
1.034500204,
1.035248913,
1.036002235,
1.036760162,
1.037522688,
1.038289806,
1.039061509,
1.039837792,
1.040618648
};
static double FastExp(double x)
{
var tmp = (long)(1512775 * x + 1072632447);
int index = (int)(tmp >> 12) & 0xFF;
return BitConverter.Int64BitsToDouble(tmp << 32) * ExpAdjustment[index];
}
static void Main(string[] args)
{
double[] x = new double[1000000];
double[] ex = new double[x.Length];
double[] fx = new double[x.Length];
Random r = new Random();
for (int i = 0; i < x.Length; ++i)
x[i] = r.NextDouble() * 40;
Stopwatch sw = new Stopwatch();
sw.Start();
for (int j = 0; j < x.Length; ++j)
ex[j] = Math.Exp(x[j]);
sw.Stop();
double builtin = sw.Elapsed.TotalMilliseconds;
sw.Reset();
sw.Start();
for (int k = 0; k < x.Length; ++k)
fx[k] = FastExp(x[k]);
sw.Stop();
double custom = sw.Elapsed.TotalMilliseconds;
double min = 1, max = 1;
for (int m = 0; m < x.Length; ++m) {
double ratio = fx[m] / ex[m];
if (min > ratio) min = ratio;
if (max < ratio) max = ratio;
}
Console.WriteLine("minimum ratio = " + min.ToString() + ", maximum ratio = " + max.ToString() + ", speedup = " + (builtin / custom).ToString());
}
}
}

The following code should address the accuracy requirements, as for inputs in [-87,88] the results have relative error <= 1.73e-3. I do not know C#, so this is C code, but I assume conversion should be fairly straightforward.
I assume that since the accuracy requirement is low, the use of single-precision computation is fine. A classic algorithm is being used in which the computation of exp() is mapped to computation of exp2(). After argument conversion via multiplication by log2(e), exponentation by the fractional part is handled using a minimax polynomial of degree 2, while exponentation by the integral part of the argument is performed by direct manipulation of the exponent part of the IEEE-754 single-precision number.
The volatile union facilitates re-interpretation of a bit pattern as either an integer or a single-precision floating-point number, needed for the exponent manipulation. It looks like C# offers decidated re-interpretation functions for this, which is much cleaner.
The two potential performance problems are the floor() function and float->int conversion. Traditionally both were slow on x86 due to the need to handle dynamic processor state. But SSE (in particular SSE 4.1) provides instructions that allow these operations to be fast. I do not know whether C# can make use of those instructions.
/* max. rel. error <= 1.73e-3 on [-87,88] */
float fast_exp (float x)
{
volatile union {
float f;
unsigned int i;
} cvt;
/* exp(x) = 2^i * 2^f; i = floor (log2(e) * x), 0 <= f <= 1 */
float t = x * 1.442695041f;
float fi = floorf (t);
float f = t - fi;
int i = (int)fi;
cvt.f = (0.3371894346f * f + 0.657636276f) * f + 1.00172476f; /* compute 2^f */
cvt.i += (i << 23); /* scale by 2^i */
return cvt.f;
}

I have studied the paper by Nicol Schraudolph where the original C implementation of the above function was defined in more detail now. It does seem that it is probably not possible to substantially approve the accuracy of the exp computation without severely impacting the performance. On the other hand, the approximation is valid also for large magnitudes of x, up to +/- 700, which is of course advantageous.
The function implementation above is tuned to obtain minimum root mean square error. Schraudolph describes how the additive term in the tmp expression can be altered to achieve alternative approximation properties.
"exp" >= exp for all x 1072693248 - (-1) = 1072693249
"exp" <= exp for all x - 90253 = 1072602995
"exp" symmetric around exp - 45799 = 1072647449
Mimimum possible mean deviation - 68243 = 1072625005
Minimum possible root-mean-square deviation - 60801 = 1072632447
He also points out that at a "microscopic" level the approximate "exp" function exhibits stair-case behavior since 32 bits are discarded in the conversion from long to double. This means that the function is piece-wise constant on a very small scale, but the function is at least never decreasing with increasing x.

I have developed for my purposes the following function that calculates quickly and accurately the natural exponent with single precision. The function works in the entire range of float values. The code is written under Visual Studio (x86).
_declspec(naked) float _vectorcall fexp(float x)
{
static const float ct[8] = // Constants table
{
1.44269502f, // lb(e)
1.92596299E-8f, // Correction to the value lb(e)
-9.21120925E-4f, // 16*b2
0.115524396f, // 4*b1
2.88539004f, // b0
4.29496730E9f, // 2^32
0.5f, // 0.5
2.32830644E-10f // 2^-32
};
_asm
{
mov ecx,offset ct // ecx contains the address of constants tables
vmulss xmm1,xmm0,[ecx] // xmm1 = x*lb(e)
vcvtss2si eax,xmm1 // eax = round(x*lb(e)) = k
vcvtsi2ss xmm1,xmm1,eax // xmm1 = k
dec eax // of=1, if eax=80000000h, i.e. overflow
jno exp_cont // Jump to form 2^k, if k is normal
vucomiss xmm0,xmm0 // Compare xmm0 with itself to identify NaN
jp exp_end // Complete with NaN result, if x=NaN
vmovd eax,xmm0 // eax contains the bits of x
test eax,eax // sf=1, if x<0; of=0 always
exp_break: // Get 0 for x<0 or Inf for x>0
vxorps xmm0,xmm0,xmm0 // xmm0 = 0
jle exp_end // Ready, if x<<0
vrcpss xmm0,xmm0,xmm0 // xmm0 = Inf at case x>>0
exp_end: // The result at xmm0 is ready
ret // Return
exp_cont: // Continue if |x| is not too big
vfmsub231ss xmm1,xmm0,[ecx] // xmm1 = x*lb(e)-k = t/2 in the range from -0,5 to 0,5
cdq // edx=-1, if x<0, othervice edx=0
vfmadd132ss xmm0,xmm1,[ecx+4] // xmm0 = t/2 (corrected value)
and edx,8 // edx=8, if x<0, othervice edx=0
vmulss xmm2,xmm0,xmm0 // xmm2 = t^2/4 - the argument of polynomial
vmovss xmm1,[ecx+8] // Initialize the sum with highest coefficient 16*b2
lea eax,[eax+8*edx+97] // The exponent of 2^(k-31), if x>0, othervice 2^(k+33)
vfmadd213ss xmm1,xmm2,[ecx+12] // xmm1 = 4*b1+4*b2*t^2
test eax,eax // Test the sign of x
jle exp_break // Result is 0 if the exponent is too small
vfmsub213ss xmm1,xmm2,xmm0 // xmm1 = -t/2+b1*t^2+b2*t^4
cmp eax,254 // Check that the exponent is not too large
ja exp_break // Jump to set Inf if overflow
vaddss xmm1,xmm1,[ecx+16] // xmm1 = b0-t/2+b1*t^2+b2*t^4 = f(t)-t/2
shl eax,23 // eax contains the bits of 2^(k-31) or 2^(k+33)
vdivss xmm0,xmm0,xmm1 // xmm0 = t/(2*f(t)-t)
vmovd xmm2,eax // xmm2 = 2^(k-31), if x>0; otherwice 2^(k+33)
vaddss xmm0,xmm0,[ecx+24] // xmm0 = t/(2*f(t)-t)+0,5
vmulss xmm0,xmm0,xmm2 // xmm0 = e^x with shifted exponent of +-32
vmulss xmm0,xmm0,[ecx+edx+20] // xmm0 = e^x with corrected exponent
ret // Return
}
}

Magnitude to Decibel always returns NaN in C#

So my question has changed from returning Infinity to returning NaN. If your FFT is always returning Infinity this may help (http://gerrybeauregard.wordpress.com/2011/04/01/an-fft-in-c/#comment-196). So I think it is returning NaN because C# is trying to get the Square Root of a negative number HOWEVER, the number should not be negative based on the code below, because I am squaring both numbers before getting the square root (which should make them positive). The number returned is negative, however. I have tried it the inefficient of using multiple variables to get re * re and im * im and the two results added together, but the results are negative as well. Math.Abs was no good either. I have contacted the creator of the FFT Class (see my link above) and am waiting for his next reply. I took some of the code below from an AS3 version of this I did before. If I get the answer from the class creator before I get one here then I will post that. Any insight is most helpful and thank you to everyone who has helped me so far in this. I am an AS3 programmer coming to C# (because it's much more capable), so it's possible I missed something simple in my newbness. I am using Unity.
private const uint LOGN = 11; // Log2 FFT Length
private const uint N = 1 << (int)LOGN; // FFT Length
private const uint BUF_LEN = N; // Audio buffer length
public FFT2 fft; // FFT Object
private double[] tempIm = new double[N]; // Temporary Imaginary Number array
private double[] m_mag = new double[N/2]; // Magnitude array
private double[] m_win = new double[N]; // Hanning Window
private int fftCount = 0; // How many times the FFT has been performed
private double SCALE = (double)20/System.Math.Log(10); // used to convert magnitude from FFT to usable dB
private double MIN_VALUE = (double)System.Double.MinValue;
...
// Hanning analysis window
for (int i = 0; i < N; i++) // for i < 2048
m_win[i] = (4.0/N) * 0.5*(1-Mathf.Cos(2*Mathf.PI*i/N)); // Hanning Vector [1] = 1 / 4595889085.750801
…
// Perform FFT
fft.run(tempRe, tempIm);
fftCount++;
// Convert from Decibel to Magnitude
for (int i = 0; i < N/2; i++) {
double re = tempRe[i]; // get the Real FFT Number at position i
double im = tempIm[i]; // get the Imaginary FFT Number at position i
m_mag[i] = Math.Sqrt(re * re + im * im); // Convert magnitude to decibels
m_mag[i] = SCALE * Math.Log(m_mag[i] + MIN_VALUE);
if (fftCount == 50 && i == 400) print ("dB # 399: " + m_mag[399]);
}
The prior code prints:
dB # 400: NaN
-5.56725062513722E+33
Thank you!!

You defined MIN_VALUE to be System.Double.MinValue, which is the smallest possible double precision number. Adding anything other than System.Double.MaxValue or positive infinity will give a negative result. Taking the logarithm of a negative number returns NaN.
I'm guessing you want MIN_VALUE to be the smallest positive number. In C#, you use System.Double.Epsilon for that. (I know... it shouldn't be called that...)
Any other tiny value, like 1e-100, will work as well.

for (int i = 0; i < tempRe.Length; i++) tempRe[i] = m_win[i];
with the multiplication it will always be zero, unless that is what you wanted.

I found the answer! So it turns out that unlike C++, AS3, and some other languages, C# has negative minimum double value.
System.Double.MIN_VALUE = -1.7976931348623157E+308;
Getting the Square root of negatives causes the NaN error. For more info see (http://www.codeproject.com/KB/cs/numprogrammingcs.aspx). I am having trouble getting an accurate dB reading now but I research and see what I can find.

how to always round up to the next integer [duplicate]

This question already has answers here:
How can I ensure that a division of integers is always rounded up?
(10 answers)
Closed 6 years ago.
i am trying to find total pages in building a pager on a website (so i want the result to be an integer. i get a list of records and i want to split into 10 per page (the page count)
when i do this:
list.Count() / 10
or
list.Count() / (decimal)10
and the list.Count() =12, i get a result of 1.
How would I code it so i get 2 in this case (the remainder should always add 1)

Math.Ceiling((double)list.Count() / 10);

(list.Count() + 9) / 10
Everything else here is either overkill or simply wrong (except for bestsss' answer, which is awesome). We do not want the overhead of a function call (Math.Truncate(), Math.Ceiling(), etc.) when simple math is enough.
OP's question generalizes (pigeonhole principle) to:
How many boxes do I need to store x objects if only y objects fit into each box?
The solution:
derives from the realization that the last box might be partially empty, and
is (x + y - 1) ÷ y using integer division.
You'll recall from 3rd grade math that integer division is what we're doing when we say 5 ÷ 2 = 2.
Floating-point division is when we say 5 ÷ 2 = 2.5, but we don't want that here.
Many programming languages support integer division. In languages derived from C, you get it automatically when you divide int types (short, int, long, etc.). The remainder/fractional part of any division operation is simply dropped, thus:
5 / 2 == 2
Replacing our original question with x = 5 and y = 2 we have:
How many boxes do I need to store 5 objects if only 2 objects fit into each box?
The answer should now be obvious: 3 boxes -- the first two boxes hold two objects each and the last box holds one.
(x + y - 1) ÷ y =
(5 + 2 - 1) ÷ 2 =
6 ÷ 2 =
3
So for the original question, x = list.Count(), y = 10, which gives the solution using no additional function calls:
(list.Count() + 9) / 10

A proper benchmark or how the number may lie
Following the argument about Math.ceil(value/10d) and (value+9)/10 I ended up coding a proper non-dead code, non-interpret mode benchmark.
I've been telling that writing micro benchmark is not an easy task. The code below illustrates this:
00:21:40.109 starting up....
00:21:40.140 doubleCeil: 19444599
00:21:40.140 integerCeil: 19444599
00:21:40.140 warming up...
00:21:44.375 warmup doubleCeil: 194445990000
00:21:44.625 warmup integerCeil: 194445990000
00:22:27.437 exec doubleCeil: 1944459900000, elapsed: 42.806s
00:22:29.796 exec integerCeil: 1944459900000, elapsed: 2.363s
The benchmark is in Java since I know well how Hotspot optimizes and ensures it's a fair result. With such results, no statistics, noise or anything can taint it.
Integer ceil is insanely much faster.
The code
package t1;
import java.math.BigDecimal;
import java.util.Random;
public class Div {
static int[] vals;
static long doubleCeil(){
int[] v= vals;
long sum = 0;
for (int i=0;i<v.length;i++){
int value = v[i];
sum+=Math.ceil(value/10d);
}
return sum;
}
static long integerCeil(){
int[] v= vals;
long sum = 0;
for (int i=0;i<v.length;i++){
int value = v[i];
sum+=(value+9)/10;
}
return sum;
}
public static void main(String[] args) {
vals = new int[7000];
Random r= new Random(77);
for (int i = 0; i < vals.length; i++) {
vals[i] = r.nextInt(55555);
}
log("starting up....");
log("doubleCeil: %d", doubleCeil());
log("integerCeil: %d", integerCeil());
log("warming up...");
final int warmupCount = (int) 1e4;
log("warmup doubleCeil: %d", execDoubleCeil(warmupCount));
log("warmup integerCeil: %d", execIntegerCeil(warmupCount));
final int execCount = (int) 1e5;
{
long time = System.nanoTime();
long s = execDoubleCeil(execCount);
long elapsed = System.nanoTime() - time;
log("exec doubleCeil: %d, elapsed: %.3fs", s, BigDecimal.valueOf(elapsed, 9));
}
{
long time = System.nanoTime();
long s = execIntegerCeil(execCount);
long elapsed = System.nanoTime() - time;
log("exec integerCeil: %d, elapsed: %.3fs", s, BigDecimal.valueOf(elapsed, 9));
}
}
static long execDoubleCeil(int count){
long sum = 0;
for(int i=0;i<count;i++){
sum+=doubleCeil();
}
return sum;
}
static long execIntegerCeil(int count){
long sum = 0;
for(int i=0;i<count;i++){
sum+=integerCeil();
}
return sum;
}
static void log(String msg, Object... params){
String s = params.length>0?String.format(msg, params):msg;
System.out.printf("%tH:%<tM:%<tS.%<tL %s%n", new Long(System.currentTimeMillis()), s);
}
}

This will also work:
c = (count - 1) / 10 + 1;

I think the easiest way is to divide two integers and increase by one :
int r = list.Count() / 10;
r += (list.Count() % 10 == 0 ? 0 : 1);
No need of libraries or functions.
edited with the right code.

You can use Math.Ceiling
http://msdn.microsoft.com/en-us/library/system.math.ceiling%28v=VS.100%29.aspx

Xform to double (and back) for a simple ceil?
list.Count()/10 + (list.Count()%10 >0?1:0) - this bad, div + mod
edit 1st:
on a 2n thought that's probably faster (depends on the optimization): div * mul (mul is faster than div and mod)
int c=list.Count()/10;
if (c*10<list.Count()) c++;
edit2 scarpe all. forgot the most natural (adding 9 ensures rounding up for integers)
(list.Count()+9)/10

Check by using mod - if there is a remainder, simply increment the value by one.