to_jdftxinfile method for JDFTXOutfile (#4408)

* `strip_structure_tags` method and testing for `JDFTXInfile`

* `to_jdftxinfile` method for JDFTXOutfile and JDFTXOutfileSlice, JDFTXOutfile inherits its most recent slice's `infile`, TODO's for cleaning up outputs module

* Although its default is printed as a float, providing a float instead of an int for this parameter will raise an error

* Minimum value field for `FloatTag`s, which changes `FloatTag.write` to return an empty string if the requested value is not greater than the set minimum value. An empty string is chosen instead of raising an error as error-raising values are dumped for the internal infile by JDFTx, so raising an error here would gaurantee an initialization error for every out file without these fields specified in the in file. There are likely more float tags that need this field set.

* todo

Author: benrich37

src/pymatgen/io/jdftx/generic_tags.py

@@ -430,6 +430,7 @@ class FloatTag(AbstractTag):
     """
 
     prec: int | None = None
+    minval: float | None = None
 
     def validate_value_type(self, tag: str, value: Any, try_auto_type_fix: bool = False) -> tuple[str, bool, Any]:
         """Validate the type of the value for this tag.
@@ -472,6 +473,11 @@ def write(self, tag: str, value: Any) -> str:
         Returns:
             str: The tag and its value as a string.
         """
+        # Returning an empty string instead of raising an error as value == self.minval
+        # will cause JDFTx to throw an error, but the internal infile dumps the value as
+        # as the minval if not set by the user.
+        if (self.minval is not None) and (not value > self.minval):
+            return ""
         # pre-convert to string: self.prec+3 is minimum room for:
         # - sign, 1 integer left of decimal, decimal, and precision.
         # larger numbers auto add places to left of decimal

src/pymatgen/io/jdftx/inputs.py

@@ -573,6 +573,17 @@ def validate_tags(
                     warnmsg += "(Check earlier warnings for more details)\n"
                 warnings.warn(warnmsg, stacklevel=2)
 
+    def strip_structure_tags(self) -> None:
+        """Strip all structural tags from the JDFTXInfile.
+
+        Strip all structural tags from the JDFTXInfile. This is useful for preparing a JDFTXInfile object
+        from a pre-existing calculation for a new structure.
+        """
+        strucural_tags = ["lattice", "ion", "lattice-scale", "coords-type"]
+        for tag in strucural_tags:
+            if tag in self:
+                del self[tag]
+
     def __setitem__(self, key: str, value: Any) -> None:
         """Set an item in the JDFTXInfile.
 

src/pymatgen/io/jdftx/jdftxinfile_ref_options.py

@@ -792,15 +792,15 @@
     "knormThreshold": FloatTag(),
     "linminMethod": StrTag(options=["CubicWolfe", "DirUpdateRecommended", "Quad", "Relax"]),
     "maxThreshold": BoolTag(),
-    "nAlphaAdjustMax": FloatTag(),
+    "nAlphaAdjustMax": IntTag(),
     "nEnergyDiff": IntTag(),
     "nIterations": IntTag(),
     "updateTestStepSize": BoolTag(),
     "wolfeEnergy": FloatTag(),
     "wolfeGradient": FloatTag(),
 }
 jdftxfluid_subtagdict = {
-    "epsBulk": FloatTag(),
+    "epsBulk": FloatTag(minval=1.0),
     "epsInf": FloatTag(),
     "epsLJ": FloatTag(),
     "Nnorm": FloatTag(),
@@ -814,12 +814,12 @@
             "A0": FloatTag(write_tagname=False, optional=False),
         },
     ),
-    "Pvap": FloatTag(),
+    "Pvap": FloatTag(minval=0.0),
     "quad_nAlpha": FloatTag(),
     "quad_nBeta": FloatTag(),
     "quad_nGamma": FloatTag(),
     "representation": TagContainer(subtags={"MuEps": FloatTag(), "Pomega": FloatTag(), "PsiAlpha": FloatTag()}),
-    "Res": FloatTag(),
+    "Res": FloatTag(minval=0.0),
     "Rvdw": FloatTag(),
     "s2quadType": StrTag(
         options=[
@@ -844,7 +844,7 @@
             "Tetrahedron",
         ]
     ),
-    "sigmaBulk": FloatTag(),
+    "sigmaBulk": FloatTag(minval=0.0),
     "tauNuc": FloatTag(),
     "translation": StrTag(options=["ConstantSpline", "Fourier", "LinearSpline"]),
 }

src/pymatgen/io/jdftx/jdftxoutfileslice.py

@@ -391,7 +391,10 @@ def _set_internal_infile(self, text: list[str]) -> None:
         Args:
             text (list[str]): Output of read_file for out file.
         """
-        start_line_idx = find_key("Input parsed successfully", text) + 2
+        start_line_idx = find_key("Input parsed successfully", text)
+        if start_line_idx is None:
+            raise ValueError("JDFTx input parsing failed on most recent call.")
+        start_line_idx += 2
         end_line_idx = None
         for i in range(start_line_idx, len(text)):
             if not len(text[i].strip()):
@@ -1157,6 +1160,24 @@ def determine_is_metal(self) -> bool | None:
             raise ValueError("Cannot determine if system is metal - self.nspin undefined")
         return None
 
+    def to_jdftxinfile(self) -> JDFTXInfile:
+        """
+        Convert the JDFTXOutfile object to a JDFTXInfile object with the most recent structure.
+        If the input structure is desired, simply fetch JDFTXOutfile.infile
+
+        Returns:
+            JDFTXInfile: A JDFTXInfile object representing the input parameters of the JDFTXOutfile.
+        """
+        # Use internal infile as a reference for calculation parameters
+        base_infile = self.infile.copy()
+        # Strip references to the input
+        base_infile.strip_structure_tags()
+        if self.structure is None:
+            return base_infile
+        infile = JDFTXInfile.from_structure(self.structure)
+        infile += base_infile
+        return infile
+
     def _check_solvation(self) -> bool:
         """Check for implicit solvation.
 

src/pymatgen/io/jdftx/outputs.py

@@ -27,6 +27,7 @@ class is written.
 if TYPE_CHECKING:
     from pymatgen.core.structure import Structure
     from pymatgen.core.trajectory import Trajectory
+    from pymatgen.io.jdftx.inputs import JDFTXInfile
     from pymatgen.io.jdftx.jelstep import JElSteps
     from pymatgen.io.jdftx.jminsettings import (
         JMinSettingsElectronic,
@@ -287,8 +288,43 @@ def _store_eigenvals(self):
     "electronic_output",
     "t_s",
     "ecomponents",
+    "infile",
 )
 
+# TODO: Remove references to the deprecated 'jsettings_*' attributes in `JDFTXOutfile` and `JDFTXOutfileSlice`
+# TODO: JDFTXOutfile and JDFTXOutfileSlice are VERY bloated. The following attributes are redundant
+# to fields of the `Structure` object and their removal likely won't cause any confusion.
+# - lattice
+# - lattice_final
+# - atom_coords
+# - atom_coords_final
+# - a/b/c
+# The following attributes are redundant to the internal `JDFTXInfile` object, but since the JDFTXInfile
+# object is not a standard pymatgen object, it may be better to decorate them with a deprecated decorator
+# until end of year. Additionally, it should be made certain which of these are dumped in the outfile's
+# input section regardless of what was set in the input file - for those that are not, they need to be kept.
+# - lattice_initial
+# - atom_coords_initial
+# - broadening
+# - broadening_type
+# - kgrid
+# - truncation_type
+# - truncation_radius
+# - fluid
+# - pwcut
+# - rhocut
+# The following are attributes that come from the electronic/fluid/ionic/lattice optimization step
+# logs. I am not sure if it is actually helpful to have access to these from the `JDFTXOutfile` object,
+# as they are only informative for analyzing optimization convergence. Additionally, the fields are
+# less universal than I thought, and can change depending on the optimization settings, so it might
+# be smarter to store them in a dictionary of arbitrary keys instead within the contained substructures.
+# - grad_k
+# - alpha
+# - linmin
+# - elec_grad_k
+# - elec_alpha
+# - elec_linmin
+
 
 @dataclass
 class JDFTXOutfile:
@@ -420,6 +456,7 @@ class JDFTXOutfile:
         elec_alpha (float): The step size of the final electronic step in the most recent JDFTx call.
         elec_linmin (float): The final normalized projection of the electronic step direction onto the gradient for the
             most recent JDFTx call.
+        infile (JDFTXInfile): The JDFTXInfile object representing the input parameters of the JDFTXOutfile.
 
     Magic Methods:
         __getitem__(key: str | int) -> Any: Decides behavior of how JDFTXOutfile objects are indexed. If the key is a
@@ -506,6 +543,7 @@ class JDFTXOutfile:
     elec_alpha: float = field(init=False)
     elec_linmin: float = field(init=False)
     electronic_output: float = field(init=False)
+    infile: JDFTXInfile = field(init=False)
 
     @classmethod
     def from_calc_dir(
@@ -558,6 +596,17 @@ def from_file(
         ]
         return cls(slices=slices)
 
+    # TODO: Write testing for this function
+    def to_jdftxinfile(self) -> JDFTXInfile:
+        """
+        Convert the JDFTXOutfile object to a JDFTXInfile object with the most recent structure.
+        If the input structure is desired, simply fetch JDFTXOutfile.infile
+
+        Returns:
+            JDFTXInfile: A JDFTXInfile object representing the input parameters of the JDFTXOutfile.
+        """
+        return self.slices[-1].to_jdftxinfile()
+
     def __post_init__(self):
         last_slice = None
         for slc in self.slices[::-1]:

tests/io/jdftx/test_jdftxinfile.py

@@ -186,6 +186,14 @@ def test_JDFTXInfile_expected_exceptions():
         jif[1.2] = 3.4
 
 
+def test_JDFTXInfile_strip_structure():
+    jif = JDFTXInfile.from_file(ex_infile1_fname)
+    structural_tags = ["lattice", "ion", "coords-type"]
+    assert all(tag in jif for tag in structural_tags)
+    jif.strip_structure_tags()
+    assert all(tag not in jif for tag in structural_tags)
+
+
 def test_JDFTXInfile_niche_cases():
     jif = JDFTXInfile.from_file(ex_infile1_fname)
     tag_object, tag, value = jif._preprocess_line("dump-only")