mindthegap-erc / lma-matlab Goto Github PK

Another discrepancy identified by @HannoSpreeuw:

Equation (6) from L'Heureux is not propagated into the Matlab implementation of equations (42) while this is propagated in the Python code.
So in the Matlab code we have

dCa=DCa/D0Ca=1
dCO3=DCO3/D0Ca=272.6/131.9

In Python there is an extra division by 1 − log φ2.

Convert all parameters into CMS units

As identified by @HannoSpreeuw:

Matlab has

U=(presum + rhorat*Phi^3*(1-exp(10-10/Phi))/(1-Phi))

while Python has

F = 1 - np.exp(10 - 10 / Phi)
one_minus_Phi = 1 - Phi
U = self.presum + self.rhorat * Phi ** 3 * F /one_minus_Ph

So those two extra brackets in Matlab make a difference: presum is also divided by 1-Phi in Matlab.

For the manuscript, we need a minimal release:

• complete the README
• include documentation as needed
• release and publish on Zenodo

Use constant scalar as defined in L'Heureux (2018) as diffusion coefficient

For cross-comparison, save output after a specific number of time steps or years, make it clear what end time is saved (dts or years)