resolve poisson and gamma distributions between _commonspeckle and skbeam

[ambarb]

Neet to address differences in two distribution functions.

• gamma distribution
• poisson distribution

Gamma Distribution

_commonspeckle

(lines 472-474 are completely different)

pyCHX/pyCHX/v2/_commonspeckle/chx_speckle.py

Lines 471 to 475 in 2ecdacf

	# gamma_dist = (stats.gamma(M, 0., K/M)).pdf(bin_values)
	x = bin_values
	coeff = np.exp(M * np.log(M) + (M - 1) * np.log(x) - gammaln(M) - M * np.log(K))
	gamma_dist = coeff * np.exp(-M * x / K)
	return gamma_dist

skbeam

https://github.com/scikit-beam/scikit-beam/blob/655d55f6396524f012bd5d1e2dc3b3fa9d6d6512/skbeam/core/fitting/lineshapes.py#L414-L415

Poisson Distribution

_commonspeckle

(make sure K is a float)

pyCHX/pyCHX/v2/_commonspeckle/chx_speckle.py

Lines 545 to 550 in 2ecdacf

	P(K) = \frac{<K>^K}{K!}\exp(-<K>)
	"""
	# poisson_dist = stats.poisson.pmf(K, bin_values)
	K = float(K)
	poisson_dist = np.exp(-K) * np.power(K, bin_values) / gamma(bin_values + 1)
	return poisson_dist

skbeam

https://github.com/scikit-beam/scikit-beam/blob/655d55f6396524f012bd5d1e2dc3b3fa9d6d6512/skbeam/core/fitting/lineshapes.py#L489-L493

[ambarb]

@danielballan @afluerasu would the ultimate plan be to import these distributions INTO skbeam or to import FROM skbeam?

[ambarb]

For gamma distribution @yugangzhang told us that the original scipy function is wrong and he had to correct it manually. @xmchenGH checked that the latest scipy function performs as expected. It sounds like we can convert back to the original skbeam code for the gamma distribution. For this we will simply import.

Thanks @xmchenGH for your help on this one.