search

eltrompetero / coniii

Convenient Interface to Inverse Ising (ConIII)
MIT License
20 stars 11 forks source link

RegularizedMeanField sometimes fails #16

Open artemmaksov opened 5 years ago

artemmaksov commented 5 years ago

I am trying to run RMF solver with synthetic data, but sometimes it results in an error. Here is the traceback:

File "python3.7/site-packages/coniii/solvers.py", line 1964, in solve solution = minimize_scalar(func) File "python3.7/site-packages/scipy/optimize/_minimize.py", line 770, in minimize_scalar return _minimize_scalar_brent(fun, bracket, args, **options) File "python3.7/site-packages/scipy/optimize/optimize.py", line 2141, in _minimize_scalar_brent brent.optimize() File "python3.7/site-packages/scipy/optimize/optimize.py", line 1925, in optimize xa, xb, xc, fa, fb, fc, funcalls = self.get_bracket_info() File "python3.7/site-packages/scipy/optimize/optimize.py", line 1899, in get_bracket_info xa, xb, xc, fa, fb, fc, funcalls = bracket(func, args=args) File "python3.7/site-packages/scipy/optimize/optimize.py", line 2324, in bracket fa = func(*(xa,) + args) File "python3.7/site-packages/coniii/solvers.py", line 1935, in func isingSamples = samples(J) File "python3.7/site-packages/coniii/solvers.py", line 1917, in samples self.multipliers = np.concatenate([J.diagonal(), squareform(mean_field_ising.zeroDiag(-J))]) File "python3.7/site-packages/scipy/spatial/distance.py", line 2193, in squareform is_valid_dm(X, throw=True, name='X') File "python3.7/site-packages/scipy/spatial/distance.py", line 2269, in is_valid_dm 'symmetric.') % name) ValueError: Distance matrix 'X' must be symmetric.

eltrompetero commented 5 years ago

Thanks for pointing this out.

My first guess is that there is some numerical precision error because the matrix J is symmetrized in the preceding line. Do you have a sense for what range of values your coupling matrix will be? For example, do you have nearly perfectly correlated spins?

artemmaksov commented 5 years ago

Thanks for the quick response! That's something I haven't really looked into. I have been using randomly generated matrix following an example in one of the notebooks and just changing value of n: h,J = np.random.normal(scale=.1,size=n),np.random.normal(scale=.1,size=n*(n-1)//2)

eltrompetero commented 5 years ago

I'm guessing that this problem occurs more frequently at smaller n. If that's the case, you may want to regularize the problem by disallowing correlations that are close to -1 or 1. That's where the parameters diverge.

Probably we should include a warning for such situations.

bcdaniels commented 4 years ago

Yes, I'm guessing that there's an issue with loss of precision or probably nans. Note squareform gives the same error for symmetric nans:

In [20]: scipy.spatial.distance.squareform([[0,1],[1,0]])
Out[20]: array([1])

In [21]: scipy.spatial.distance.squareform([[0,scipy.nan],[scipy.nan,0]])
ValueError: Distance matrix 'X' must be symmetric.

Maybe we need to check for nans in the J matrix?

eltrompetero commented 4 years ago

@bcdaniels The RMF code is a bit opaque to me right now. Could you point out where nans could appear?