Slow implementation of round to significant digits function

[ValentinGebhart]

Describe the problem
The functions util.value_representation.sig_dig and util.value_representation.sig_dig_list seem to be a bit slow (they convert the float input to a string and back).

def sig_dig(x, n_sig_dig=16):
    num_of_digits = len(str(x).replace(".", ""))
    if n_sig_dig >= num_of_digits:
        return x
    n = math.floor(math.log10(abs(x)) + 1 - n_sig_dig)
    result = decimal.Decimal(str(np.round(x * 10 ** (-n)))) * decimal.Decimal(
        str(10**n)
    )
    return float(result)

E.g., this function seemed to be more than 100 times faster. It created some floating point issues though and is thus not implemented.

def round_to_sig_digits(x, n_sig_dig):
    x = np.asarray(x, dtype=np.float64)  # Ensure consistent precision
    x_positive = np.where(np.isfinite(x) & (x != 0), np.abs(x), 10 ** (n_sig_dig - 1))
    mags = 10 ** (n_sig_dig - 1 - np.floor(np.log10(x_positive)))
    return np.around(x * mags) / mags

The function does not seem to be used much so this might not be urgent.

[peanutfun]

@ValentinGebhart Was this solved by #1012?

[ValentinGebhart]

No, we created this issue to document a problem with an existing method. The suggested solution also has its downs, which is why we did not change the method, and we currently do not plan to do so.

If such issues should rather be documented somewhere else, happy to move this issue and close it here.

[peanutfun]

I'll convert this to a Discussion, then.