Wrong local energy for Non-interactive numerical interaction on gaussian

copperwire commented 6 years ago

When running the current code the local energy does not stabilize to an equilibrium value (i.e. it changes for each run). I suspect that the problem lies in the computation of the metropolis hastings test.

I propose that we follow this sign convention for all MH implementations, let:

$P \equiv \frac{\Psi _t(\bold{R})}{\Psi_t(\bold{R}_p)}$

then the MH test should read:

${}\text{if } \epsilon \geq P \text{ then:} \\ \bold{R} \leftarrow \bold{R_p} \\ \text{else:}\\ \bold{R} \leftarrow \bold{R}$

for a random uniformly distributed number epsilon in the interval [0, 1].

EDIT: WRONG MH TEST - SIGN SHOULD BE REVERSED AND NEW POSITION SHOULD BE IN THE DIVISOR

copperwire commented 6 years ago

At least I think that's what it should read.. fuck now I'm uncertain again

copperwire commented 6 years ago

Issue closed by commit previous but I was just testing closing by commit :)

copperwire / comp_phys_2

Wrong local energy for Non-interactive numerical interaction on gaussian #1