!paste
#🔒 Help with mandelbrot set calculation
fathom relic
fathom relic
i am trying to make the mandelbrot set using a custom arbitrary floating point arithmetic c extension i wrote
here is the main code https://github.com/HRLO77/fractals/tree/main/tests
drifting scarab
Just from the shape, it looks like you're giving up waiting for inifinity too early.
fathom relic
Just from the shape, it looks like you're giving up waiting for inifinity too ea...
?
here is the main loop```py
cdef unsigned int mandelbrot( _cydecimal creal, _cydecimal cimag, const unsigned int maxiter, unsigned int testi, unsigned int testj) except *:
cdef _cydecimal temp1, real = creal, imag = cimag, real2, imag2
cdef unsigned int n
for n in range(maxiter):
real2 = _square_decimal(real) # x^2 + y^2
imag2 = _square_decimal(imag)
temp1 = _add_decimals(real2, imag2) # x^2 + y^2 > 4
if _true_greater_than(temp1,FOUR):
return n
imag = _add_decimals(_mult_decimals(TWO, _mult_decimals(real, imag)), cimag) # y = (2*x*y)+cimag
real = _add_decimals(_subtract_decimals(real2, imag2), creal) # x = (x^2 - y^2) + creal
#if _true_eq(real, creal):
# if _true_eq(imag, cimag):
# return maxiter
return maxiter```
That's a shape you tend to get if you assume a number won't go off into infinity iirc.
fathom relic
huh
drifting scarab
That should give a better shape. I use 200.
fathom relic
and sometimes the image just doesnt show up??
after the calculations are done, the data is fine but plt doesnt show it
drifting scarab
I have no experience with plt, so I can't help with that unfortunately.
fathom relic
its alr
drifting scarab
I don't know if this will help any, but here's the calculation routine I use: https://github.com/carcigenicate/mandelbrot_explorer/blob/master/src/logic/mandelbrot_iteration.rs
I don't think I've ever done this project proper in Python unfortunately.
test_point is comparable to your mandelbrot function.
fathom relic
i think i may have found the issue?
static inline char _close_zero(const _cydecimal_ptr first){
exponent_t temp;
if (first->exp < -(MAX_INDICE)){ // may lose a digit of precision :/
return 1; // if the exponent is super slow i.e practically zero
} // 1 means YES - NUMBER IS ZERO
else if (first->exp < -(N_DIGITS)){
_normalize_digits(first, false);
temp = ((first->exp)-_n_whole_digits(first)); // this is ignoring the decimal portion of the array - will check later
if (temp < -(N_DIGITS)){
return 0; // YES - NUMBER CLOSE TO ZERO, NO CALCULATIONS CAN BE FURTHER PERFORMED BUT THERE MAY STILL BE SPACE TO THE RIGHT
}
else{
_normalize_digits(first, true); // maybe pushy (i.e removing beginning digits)
temp = ((first->exp)-_n_precision(first));
if (temp < -(N_PRECISION)){
return 0; // YES - NUMBER CLOSE TO ZERO, NO CALCULATIONS CAN BE FURTHER PERFORMED BUT THERE MAY STILL BE SPACE TO THE RIGHT
}
}
}
return -1; // NO - NUMBER DID NOT FILL UP ENTIRELY OR THE N_DIGITS SPACE
}```this _close_zero function approximates if a number is close to zero (because if it is and i keep doing math on it, the exponent goes crazy)
this is probably rounding too much
halcyon mulchBOT
Python help channel closed