Related
I have spent a lot of time to learn about implementing/visualizing dynamic programming problems using iteration but I find it very hard to understand, I can implement the same using recursion with memoization but it is slow when compared to iteration.
Can someone explain the same by a example of a hard problem or by using some basic concepts. Like the matrix chain multiplication, longest palindromic sub sequence and others. I can understand the recursion process and then memoize the overlapping sub problems for efficiency but I can't understand how to do the same using iteration.
Thanks!
Dynamic programming is all about solving the sub-problems in order to solve the bigger one. The difference between the recursive approach and the iterative approach is that the former is top-down, and the latter is bottom-up. In other words, using recursion, you start from the big problem you are trying to solve and chop it down to a bit smaller sub-problems, on which you repeat the process until you reach the sub-problem so small you can solve. This has an advantage that you only have to solve the sub-problems that are absolutely needed and using memoization to remember the results as you go. The bottom-up approach first solves all the sub-problems, using tabulation to remember the results. If we are not doing extra work of solving the sub-problems that are not needed, this is a better approach.
For a simpler example, let's look at the Fibonacci sequence. Say we'd like to compute F(101). When doing it recursively, we will start with our big problem - F(101). For that, we notice that we need to compute F(99) and F(100). Then, for F(99) we need F(97) and F(98). We continue until we reach the smallest solvable sub-problem, which is F(1), and memoize the results. When doing it iteratively, we start from the smallest sub-problem, F(1) and continue all the way up, keeping the results in a table (so essentially it's just a simple for loop from 1 to 101 in this case).
Let's take a look at the matrix chain multiplication problem, which you requested. We'll start with a naive recursive implementation, then recursive DP, and finally iterative DP. It's going to be implemented in a C/C++ soup, but you should be able to follow along even if you are not very familiar with them.
/* Solve the problem recursively (naive)
p - matrix dimensions
n - size of p
i..j - state (sub-problem): range of parenthesis */
int solve_rn(int p[], int n, int i, int j) {
// A matrix multiplied by itself needs no operations
if (i == j) return 0;
// A minimal solution for this sub-problem, we
// initialize it with the maximal possible value
int min = std::numeric_limits<int>::max();
// Recursively solve all the sub-problems
for (int k = i; k < j; ++k) {
int tmp = solve_rn(p, n, i, k) + solve_rn(p, n, k + 1, j) + p[i - 1] * p[k] * p[j];
if (tmp < min) min = tmp;
}
// Return solution for this sub-problem
return min;
}
To compute the result, we starts with the big problem:
solve_rn(p, n, 1, n - 1)
The key of DP is to remember all the solutions to the sub-problems instead of forgetting them, so we don't need to recompute them. It's trivial to make a few adjustments to the above code in order to achieve that:
/* Solve the problem recursively (DP)
p - matrix dimensions
n - size of p
i..j - state (sub-problem): range of parenthesis */
int solve_r(int p[], int n, int i, int j) {
/* We need to remember the results for state i..j.
This can be done in a matrix, which we call dp,
such that dp[i][j] is the best solution for the
state i..j. We initialize everything to 0 first.
static keyword here is just a C/C++ thing for keeping
the matrix between function calls, you can also either
make it global or pass it as a parameter each time.
MAXN is here too because the array size when doing it like
this has to be a constant in C/C++. I set it to 100 here.
But you can do it some other way if you don't like it. */
static int dp[MAXN][MAXN] = {{0}};
/* A matrix multiplied by itself has 0 operations, so we
can just return 0. Also, if we already computed the result
for this state, just return that. */
if (i == j) return 0;
else if (dp[i][j] != 0) return dp[i][j];
// A minimal solution for this sub-problem, we
// initialize it with the maximal possible value
dp[i][j] = std::numeric_limits<int>::max();
// Recursively solve all the sub-problems
for (int k = i; k < j; ++k) {
int tmp = solve_r(p, n, i, k) + solve_r(p, n, k + 1, j) + p[i - 1] * p[k] * p[j];
if (tmp < dp[i][j]) dp[i][j] = tmp;
}
// Return solution for this sub-problem
return dp[i][j];;
}
We start with the big problem as well:
solve_r(p, n, 1, n - 1)
Iterative solution is only to, well, iterate all the states, instead of starting from the top:
/* Solve the problem iteratively
p - matrix dimensions
n - size of p
We don't need to pass state, because we iterate the states. */
int solve_i(int p[], int n) {
// But we do need our table, just like before
static int dp[MAXN][MAXN];
// Multiplying a matrix by itself needs no operations
for (int i = 1; i < n; ++i)
dp[i][i] = 0;
// L represents the length of the chain. We go from smallest, to
// biggest. Made L capital to distinguish letter l from number 1
for (int L = 2; L < n; ++L) {
// This double loop goes through all the states in the current
// chain length.
for (int i = 1; i <= n - L + 1; ++i) {
int j = i + L - 1;
dp[i][j] = std::numeric_limits<int>::max();
for (int k = i; k <= j - 1; ++k) {
int tmp = dp[i][k] + dp[k+1][j] + p[i-1] * p[k] * p[j];
if (tmp < dp[i][j])
dp[i][j] = tmp;
}
}
}
// Return the result of the biggest problem
return dp[1][n-1];
}
To compute the result, just call it:
solve_i(p, n)
Explanation of the loop counters in the last example:
Let's say we need to optimize the multiplication of 4 matrices: A B C D. We are doing an iterative approach, so we will first compute the chains with the length of two: (A B) C D, A (B C) D, and A B (C D). And then chains of three: (A B C) D, and A (B C D). That is what L, i and j are for.
L represents the chain length, it goes from 2 to n - 1 (n is 4 in this case, so that is 3).
i and j represent the starting and ending position of the chain. In case L = 2, i goes from 1 to 3, and j goes from 2 to 4:
(A B) C D A (B C) D A B (C D)
^ ^ ^ ^ ^ ^
i j i j i j
In case L = 3, i goes from 1 to 2, and j goes from 3 to 4:
(A B C) D A (B C D)
^ ^ ^ ^
i j i j
So generally, i goes from 1 to n - L + 1, and j is i + L - 1.
Now, let's continue with the algorithm assuming that we are at the step where we have (A B C) D. We now need to take into account the sub-problems (which are already calculated): ((A B) C) D and (A (B C)) D. That is what k is for. It goes through all the positions between i and j and computes the sub problems.
I hope I helped.
The problem with recursion is the high number of stack frames that need to be pushed/popped. This can quickly become the bottle-neck.
The Fibonacci Series can be calculated with iterative DP or recursion with memoization. If we calculate F(100) in DP all we need is an array of length 100 e.g. int[100] and that's the guts of our used memory. We calculate all entries of the array pre-filling f[0] and f[1] as they are defined to be 1. and each value just depends on the previous two.
If we use a recursive solution we start at fib(100) and work down. Every method call from 100 down to 0 is pushed onto the stack, AND checked if it's memoized. These operations add up and iteration doesn't suffer from either of these. In iteration (bottom-up) we already know all of the previous answers are valid. The bigger impact is probably the stack frames; and given a larger input you may get a StackOverflowException for what was otherwise trivial with an iterative DP approach.
Given positive-integer inputs x and y, is there a mathematical formula that will return 1 if x==y and 0 otherwise? I am in the unfortunate position of having to use a tool that only allows me to use the following symbols: numerals 0-9; decimal point .; parentheses ( and ); and the four basic arithmetic operations +, -, /, and *.
Currently I am relying on the fact that the tool that evaluates division by zero to be zero. (I can't tell if this is a bug or a feature.) Because of this, I have been able to use ((x-y)/(y-x))+1. Obviously, this is ugly and unideal, especially in the case that it is a bug and they fix it in a future version.
Taking advantage of integer division in C truncates toward 0, the follows works well. No multiplication overflow. Well defined for all "positive-integer inputs x and y".
(x/y) * (y/x)
#include <stdio.h>
#include <limits.h>
void etest(unsigned x, unsigned y) {
unsigned ref = x == y;
unsigned z = (x/y) * (y/x);
if (ref != z) {
printf("%u %u %u %u\n", x,y,z,ref);
}
}
void etests(void) {
unsigned list[] = { 1,2,3,4,5,6,7,8,9,10,100,1000, UINT_MAX/2 , UINT_MAX - 1, UINT_MAX };
for (unsigned x = 0; x < sizeof list/sizeof list[0]; x++) {
for (unsigned y = 0; y < sizeof list/sizeof list[0]; y++) {
etest(list[x], list[y]);
}
}
}
int main(void) {
etests();
printf("Done\n");
return 0;
}
Output (No difference from x == y)
Done
If division is truncating and the numbers are not too big, then:
((x - y) ^ 2 + 2) / ((x - y) ^ 2 + 1) - 1
The division has the value 2 if x = y and otherwise truncates to 1.
(Here x^2 is an abbreviation for x*x.)
This will fail if (x-y)^2 overflows. In that case, you need to independently check x/k = y/k and x%k = y%k where (k-1)*(k-1) doesn't overflow (which will work if k is ceil(sqrt(INT_MAX))). x%k can be computed as x-k*(x/k) and A&&B is simply A*B.
That will work for any x and y in the range [-k*k, k*k].
A slightly incorrect computation, using lots of intermediate values, which assumes that x - y won't overflow (or at least that the overflow won't produce a false 0).
int delta = x - y;
int delta_hi = delta / K;
int delta_lo = delta - K * delta_hi;
int equal_hi = (delta_hi * delta_hi + 2) / (delta_hi * delta_hi + 1) - 1;
int equal_lo = (delta_lo * delta_lo + 2) / (delta_lo * delta_lo + 1) - 1;
int equals = equal_hi * equal_lo;
or written out in full:
((((x-y)/K)*((x-y)/K)+2)/(((x-y)/K)*((x-y)/K)+1)-1)*
((((x-y)-K*((x-y)/K))*((x-y)-K*((x-y)/K))+2)/
(((x-y)-K*((x-y)/K))*((x-y)-K*((x-y)/K))+1)-1)
(For signed 31-bit integers, use K=46341; for unsigned 32-bit integers, 65536.)
Checked with #chux's test harness, adding the 0 case: live on coliru and with negative values also on coliru.
On a platform where integer subtraction might produce something other than the 2s-complement wraparound, a similar technique could be used, but dividing the numbers into three parts instead of two.
So the problem is that if they fix division by zero, it means that you cannot use any divisor that contains input variables anymore (you'd have to check that the divisor != 0, and implementing that check would solve the original x-y == 0 problem!); hence, division cannot be used at all.
Ergo, only +, -, * and the association operator () can be used. It's not hard to see that with only these operators, the desired behaviour cannot be implemented.
I'm trying to implement 2D Perlin noise to create Minecraft-like terrain (Minecraft doesn't actually use 2D Perlin noise) without overhangs or caves and stuff.
The way I'm doing it, is by creating a [50][20][50] array of cubes, where [20] will be the maximum height of the array, and its values will be determined with Perlin noise. I will then fill that array with arrays of cube.
I've been reading from this article and I don't understand, how do I compute the 4 gradient vector and use it in my code? Does every adjacent 2D array such as [2][3] and [2][4] have a different 4 gradient vector?
Also, I've read that the general Perlin noise function also takes a numeric value that will be used as seed, where do I put that in this case?
I'm going to explain Perlin noise using working code, and without relying on other explanations. First you need a way to generate a pseudo-random float at a 2D point. Each point should look random relative to the others, but the trick is that the same coordinates should always produce the same float. We can use any hash function to do that - not just the one that Ken Perlin used in his code. Here's one:
static float noise2(int x, int y) {
int n = x + y * 57;
n = (n << 13) ^ n;
return (float) (1.0-((n*(n*n*15731+789221)+1376312589)&0x7fffffff)/1073741824.0);
}
I use this to generate a "landscape" landscape[i][j] = noise2(i,j); (which I then convert to an image) and it always produces the same thing:
...
But that looks too random - like the hills and valleys are too densely packed. We need a way of "stretching" each random point over, say, 5 points. And for the values between those "key" points, you want a smooth gradient:
static float stretchedNoise2(float x_float, float y_float, float stretch) {
// stretch
x_float /= stretch;
y_float /= stretch;
// the whole part of the coordinates
int x = (int) Math.floor(x_float);
int y = (int) Math.floor(y_float);
// the decimal part - how far between the two points yours is
float fractional_X = x_float - x;
float fractional_Y = y_float - y;
// we need to grab the 4x4 nearest points to do cubic interpolation
double[] p = new double[4];
for (int j = 0; j < 4; j++) {
double[] p2 = new double[4];
for (int i = 0; i < 4; i++) {
p2[i] = noise2(x + i - 1, y + j - 1);
}
// interpolate each row
p[j] = cubicInterp(p2, fractional_X);
}
// and interpolate the results each row's interpolation
return (float) cubicInterp(p, fractional_Y);
}
public static double cubicInterp(double[] p, double x) {
return cubicInterp(p[0],p[1],p[2],p[3], x);
}
public static double cubicInterp(double v0, double v1, double v2, double v3, double x) {
double P = (v3 - v2) - (v0 - v1);
double Q = (v0 - v1) - P;
double R = v2 - v0;
double S = v1;
return P * x * x * x + Q * x * x + R * x + S;
}
If you don't understand the details, that's ok - I don't know how Math.cos() is implemented, but I still know what it does. And this function gives us stretched, smooth noise.
->
The stretchedNoise2 function generates a "landscape" at a certain scale (big or small) - a landscape of random points with smooth slopes between them. Now we can generate a sequence of landscapes on top of each other:
public static double perlin2(float xx, float yy) {
double noise = 0;
noise += stretchedNoise2(xx, yy, 5) * 1; // sample 1
noise += stretchedNoise2(xx, yy, 13) * 2; // twice as influential
// you can keep repeating different variants of the above lines
// some interesting variants are included below.
return noise / (1+2); // make sure you sum the multipliers above
}
To put it more accurately, we get the weighed average of the points from each sample.
( + 2 * ) / 3 =
When you stack a bunch of smooth noise together, usually about 5 samples of increasing "stretch", you get Perlin noise. (If you understand the last sentence, you understand Perlin noise.)
There are other implementations that are faster because they do the same thing in different ways, but because it is no longer 1983 and because you are getting started with writing a landscape generator, you don't need to know about all the special tricks and terminology they use to understand Perlin noise or do fun things with it. For example:
1) 2) 3)
// 1
float smearX = interpolatedNoise2(xx, yy, 99) * 99;
float smearY = interpolatedNoise2(xx, yy, 99) * 99;
ret += interpolatedNoise2(xx + smearX, yy + smearY, 13)*1;
// 2
float smearX2 = interpolatedNoise2(xx, yy, 9) * 19;
float smearY2 = interpolatedNoise2(xx, yy, 9) * 19;
ret += interpolatedNoise2(xx + smearX2, yy + smearY2, 13)*1;
// 3
ret += Math.cos( interpolatedNoise2(xx , yy , 5)*4) *1;
About perlin noise
Perlin noise was developed to generate a random continuous surfaces (actually, procedural textures). Its main feature is that the noise is always continuous over space.
From the article:
Perlin noise is function for generating coherent noise over a space. Coherent noise means that for any two points in the space, the value of the noise function changes smoothly as you move from one point to the other -- that is, there are no discontinuities.
Simply, a perlin noise looks like this:
_ _ __
\ __/ \__/ \__
\__/
But this certainly is not a perlin noise, because there are gaps:
_ _
\_ __/
___/ __/
Calculating the noise (or crushing gradients!)
As #markspace said, perlin noise is mathematically hard. Lets simplify by generating 1D noise.
Imagine the following 1D space:
________________
Firstly, we define a grid (or points in 1D space):
1 2 3 4
________________
Then, we randomly chose a noise value to each grid point (This value is equivalent to the gradient in the 2D noise):
1 2 3 4
________________
-1 0 0.5 1 // random noise value
Now, calculating the noise value for a grid point it is easy, just pick the value:
noise(3) => 0.5
But the noise value for a arbitrary point p needs to be calculated based in the closest grid points p1 and p2 using their value and influence:
// in 1D the influence is just the distance between the points
noise(p) => noise(p1) * influence(p1) + noise(p2) * influence(p2)
noise(2.5) => noise(2) * influence(2, 2.5) + noise(3) * influence(3, 2.5)
=> 0 * 0.5 + 0.5 * 0.5 => 0.25
The end! Now we are able to calculate 1D noise, just add one dimension for 2D. :-)
Hope it helps you understand! Now read #mk.'s answer for working code and have happy noises!
Edit:
Follow up question in the comments:
I read in wikipedia article that the gradient vector in 2d perlin should be length of 1 (unit circle) and random direction. since vector has X and Y, how do I do that exactly?
This could be easily lifted and adapted from the original perlin noise code. Find bellow a pseudocode.
gradient.x = random()*2 - 1;
gradient.y = random()*2 - 1;
normalize_2d( gradient );
Where normalize_2d is:
// normalizes a 2d vector
function normalize_2d(v)
size = square_root( v.x * v.x + v.y * v.y );
v.x = v.x / size;
v.y = v.y / size;
Compute Perlin noise at coordinates x, y
function perlin(float x, float y) {
// Determine grid cell coordinates
int x0 = (x > 0.0 ? (int)x : (int)x - 1);
int x1 = x0 + 1;
int y0 = (y > 0.0 ? (int)y : (int)y - 1);
int y1 = y0 + 1;
// Determine interpolation weights
// Could also use higher order polynomial/s-curve here
float sx = x - (double)x0;
float sy = y - (double)y0;
// Interpolate between grid point gradients
float n0, n1, ix0, ix1, value;
n0 = dotGridGradient(x0, y0, x, y);
n1 = dotGridGradient(x1, y0, x, y);
ix0 = lerp(n0, n1, sx);
n0 = dotGridGradient(x0, y1, x, y);
n1 = dotGridGradient(x1, y1, x, y);
ix1 = lerp(n0, n1, sx);
value = lerp(ix0, ix1, sy);
return value;
}
I tried to implement bessel function using that formula, this is the code:
function result=Bessel(num);
if num==0
result=bessel(0,1);
elseif num==1
result=bessel(1,1);
else
result=2*(num-1)*Bessel(num-1)-Bessel(num-2);
end;
But if I use MATLAB's bessel function to compare it with this one, I get too high different values.
For example if I type Bessel(20) it gives me 3.1689e+005 as result, if instead I type bessel(20,1) it gives me 3.8735e-025 , a totally different result.
such recurrence relations are nice in mathematics but numerically unstable when implementing algorithms using limited precision representations of floating-point numbers.
Consider the following comparison:
x = 0:20;
y1 = arrayfun(#(n)besselj(n,1), x); %# builtin function
y2 = arrayfun(#Bessel, x); %# your function
semilogy(x,y1, x,y2), grid on
legend('besselj','Bessel')
title('J_\nu(z)'), xlabel('\nu'), ylabel('log scale')
So you can see how the computed values start to differ significantly after 9.
According to MATLAB:
BESSELJ uses a MEX interface to a Fortran library by D. E. Amos.
and gives the following as references for their implementation:
D. E. Amos, "A subroutine package for Bessel functions of a complex
argument and nonnegative order", Sandia National Laboratory Report,
SAND85-1018, May, 1985.
D. E. Amos, "A portable package for Bessel functions of a complex
argument and nonnegative order", Trans. Math. Software, 1986.
The forward recurrence relation you are using is not stable. To see why, consider that the values of BesselJ(n,x) become smaller and smaller by about a factor 1/2n. You can see this by looking at the first term of the Taylor series for J.
So, what you're doing is subtracting a large number from a multiple of a somewhat smaller number to get an even smaller number. Numerically, that's not going to work well.
Look at it this way. We know the result is of the order of 10^-25. You start out with numbers that are of the order of 1. So in order to get even one accurate digit out of this, we have to know the first two numbers with at least 25 digits precision. We clearly don't, and the recurrence actually diverges.
Using the same recurrence relation to go backwards, from high orders to low orders, is stable. When you start with correct values for J(20,1) and J(19,1), you can calculate all orders down to 0 with full accuracy as well. Why does this work? Because now the numbers are getting larger in each step. You're subtracting a very small number from an exact multiple of a larger number to get an even larger number.
You can just modify the code below which is for the Spherical bessel function. It is well tested and works for all arguments and order range. I am sorry it is in C#
public static Complex bessel(int n, Complex z)
{
if (n == 0) return sin(z) / z;
if (n == 1) return sin(z) / (z * z) - cos(z) / z;
if (n <= System.Math.Abs(z.real))
{
Complex h0 = bessel(0, z);
Complex h1 = bessel(1, z);
Complex ret = 0;
for (int i = 2; i <= n; i++)
{
ret = (2 * i - 1) / z * h1 - h0;
h0 = h1;
h1 = ret;
if (double.IsInfinity(ret.real) || double.IsInfinity(ret.imag)) return double.PositiveInfinity;
}
return ret;
}
else
{
double u = 2.0 * abs(z.real) / (2 * n + 1);
double a = 0.1;
double b = 0.175;
int v = n - (int)System.Math.Ceiling((System.Math.Log(0.5e-16 * (a + b * u * (2 - System.Math.Pow(u, 2)) / (1 - System.Math.Pow(u, 2))), 2)));
Complex ret = 0;
while (v > n - 1)
{
ret = z / (2 * v + 1.0 - z * ret);
v = v - 1;
}
Complex jnM1 = ret;
while (v > 0)
{
ret = z / (2 * v + 1.0 - z * ret);
jnM1 = jnM1 * ret;
v = v - 1;
}
return jnM1 * sin(z) / z;
}
}
I have 2 tables of values and want to scale the first one so that it matches the 2nd one as good as possible. Both have the same length. If both are drawn as graphs in a diagram they should be as close to each other as possible. But I do not want quadratic, but simple linear weights.
My problem is, that I have no idea how to actually compute the best scaling factor because of the Abs function.
Some pseudocode:
//given:
float[] table1= ...;
float[] table2= ...;
//wanted:
float factor= ???; // I have no idea how to compute this
float remainingDifference=0;
for(int i=0; i<length; i++)
{
float scaledValue=table1[i] * factor;
//Sum up the differences. I use the Abs function because negative differences are differences too.
remainingDifference += Abs(scaledValue - table2[i]);
}
I want to compute the scaling factor so that the remainingDifference is minimal.
Simple linear weights is hard like you said.
a_n = first sequence
b_n = second sequence
c = scaling factor
Your residual function is (sums are from i=1 to N, the number of points):
SUM( |a_i - c*b_i| )
Taking the derivative with respect to c yields:
d/dc SUM( |a_i - c*b_i| )
= SUM( b_i * (a_i - c*b_i)/|a_i - c*b_i| )
Setting to 0 and solving for c is hard. I don't think there's an analytic way of doing that. You may want to try https://math.stackexchange.com/ to see if they have any bright ideas.
However if you work with quadratic weights, it becomes significantly simpler:
d/dc SUM( (a_i - c*b_i)^2 )
= SUM( 2*(a_i - c*b_i)* -c )
= -2c * SUM( a_i - c*b_i ) = 0
=> SUM(a_i) - c*SUM(b_i) = 0
=> c = SUM(a_i) / SUM(b_i)
I strongly suggest the latter approach if you can.
I would suggest trying some sort of variant on Newton Raphson.
Construct a function Diff(k) that looks at the difference in area between your two graphs between fixed markers A and B.
mathematically I guess it would be integral ( x = A to B ){ f(x) - k * g(x) }dx
anyway realistically you could just subtract the values,
like if you range from X = -10 to 10, and you have a data point for f(i) and g(i) on each integer i in [-10, 10], (ie 21 datapoints )
then you just sum( i = -10 to 10 ){ f(i) - k * g(i) }
basically you would expect this function to look like a parabola -- there will be an optimum k, and deviating slightly from it in either direction will increase the overall area difference
and the bigger the difference, you would expect the bigger the gap
so, this should be a pretty smooth function ( if you have a lot of data points )
so you want to minimise Diff(k)
so you want to find whether derivative ie d/dk Diff(k) = 0
so just do Newton Raphson on this new function D'(k)
kick it off at k=1 and it should zone in on a solution pretty fast
that's probably going to give you an optimal computation time
if you want something simpler, just start with some k1 and k2 that are either side of 0
so say Diff(1.5) = -3 and Diff(2.9) = 7
so then you would pick a k say 3/10 of the way (10 = 7 - -3) between 1.5 and 2.9
and depending on whether that yields a positive or negative value, use it as the new k1 or k2, rinse and repeat
In case anyone stumbles upon this in the future, here is some code (c++)
The trick is to first sort the samples by the scaling factor that would result in the best fit for the 2 samples each. Then start at both ends iterate to the factor that results in the minimum absolute deviation (L1-norm).
Everything except for the sort has a linear run time => Runtime is O(n*log n)
/*
* Find x so that the sum over std::abs(pA[i]-pB[i]*x) from i=0 to (n-1) is minimal
* Then return x
*/
float linearFit(const float* pA, const float* pB, int n)
{
/*
* Algebraic solution is not possible for the general case
* => iterative algorithm
*/
if (n < 0)
throw "linearFit has invalid argument: expected n >= 0";
if (n == 0)
return 0;//If there is nothing to fit, any factor is a perfect fit (sum is always 0)
if (n == 1)
return pA[0] / pB[0];//return x so that pA[0] = pB[0]*x
//If you don't like this , use a std::vector :P
std::unique_ptr<float[]> targetValues_(new float[n]);
std::unique_ptr<int[]> indices_(new int[n]);
//Get proper pointers:
float* targetValues = targetValues_.get();//The value for x that would cause pA[i] = pB[i]*x
int* indices = indices_.get(); //Indices of useful (not nan and not infinity) target values
//The code above guarantees n > 1, so it is safe to get these pointers:
int m = 0;//Number of useful target values
for (int i = 0; i < n; i++)
{
float a = pA[i];
float b = pB[i];
float targetValue = a / b;
targetValues[i] = targetValue;
if (std::isfinite(targetValue))
{
indices[m++] = i;
}
}
if (m <= 0)
return 0;
if (m == 1)
return targetValues[indices[0]];//If there is only one target value, then it has to be the best one.
//sort the indices by target value
std::sort(indices, indices + m, [&](int ia, int ib){
return targetValues[ia] < targetValues[ib];
});
//Start from the extremes and meet at the optimal solution somewhere in the middle:
int l = 0;
int r = m - 1;
// m >= 2 is guaranteed => l > r
float penaltyFactorL = std::abs(pB[indices[l]]);
float penaltyFactorR = std::abs(pB[indices[r]]);
while (l < r)
{
if (l == r - 1 && penaltyFactorL == penaltyFactorR)
{
break;
}
if (penaltyFactorL < penaltyFactorR)
{
l++;
if (l < r)
{
penaltyFactorL += std::abs(pB[indices[l]]);
}
}
else
{
r--;
if (l < r)
{
penaltyFactorR += std::abs(pB[indices[r]]);
}
}
}
//return the best target value
if (l == r)
return targetValues[indices[l]];
else
return (targetValues[indices[l]] + targetValues[indices[r]])*0.5;
}