Common Errors

Common Errors#

The following comprises a list of common errors that might occur during development and their solutions. If the error occurs frequently, please raise a custom error in the respective code.

RuntimeError: The linear operator A must be Hermitian#

cond = False, msg = 'The linear operator A must be Hermitian'

    def assert_runtime(cond, msg=""):
        if not cond:
>           raise RuntimeError(msg)
E           RuntimeError: The linear operator A must be Hermitian

src/dxtb/_src/exlibs/xitorch/_utils/assertfuncs.py:49: RuntimeError

This error usually occurs in the later stages of the SCF for atoms. Although the error message claims that there is a problem with the Hamiltonian matrix, the error is actually raised because the mixer in xitorch produces a NaN value. In fact, the offending code is the following:

def update(self, x, y):
    dy = y - self.y_prev
    dx = x - self.x_prev # <-- can become exactly zero
    # update Gm
    self._update(x, y, dx, dy, dx.norm(), dy.norm())

    self.y_prev = y
    self.x_prev = x

def _update(self, x, y, dx, dy, dxnorm, dynorm):
    # keep the rank small
    self._reduce()

    v = self.Gm.rmv(dx)
    c = dx - self.Gm.mv(dy)
    d = v / torch.dot(dy, v) # <-- yields NaN
    self.Gm = self.Gm.append(c, d)

Apparently, this occurs if the convergence criteria are too strict in single precision. The solution is to increase the convergence criteria to more than 1e-6.

RuntimeError: clone is not supported by NestedIntSymNode#

This is a bug in PyTorch 2.3.0 and 2.3.1 (see PyTorch #128607). To avoid this error, manually import torch._dynamo in the code. For example:

from tad_mctc._version import __tversion__

if __tversion__ in ((2, 3, 0), (2, 3, 1)):
    import torch._dynamo

TimerError: Timer ‘<interaction>’ is running. Use .stop() to stop it.#

This error occurs when a calculation is launched again without resetting. If you don’t need the timer, you can disable it. The timer lives in the global space and always starts when importing dxtb.

from dxtb import kill_timer

kill_timer()

If you only want to disable the timer temporarily, you can use the following code:

from dxtb import timer

timer.disable()
Contents