# SPDX-License-Identifier: LGPL-3.0-or-later
import logging
import warnings
from typing import (
List,
Optional,
Set,
Tuple,
)
import numpy as np
from packaging.version import (
Version,
)
from deepmd.common import (
cast_precision,
get_np_precision,
)
from deepmd.env import (
GLOBAL_NP_FLOAT_PRECISION,
GLOBAL_TF_FLOAT_PRECISION,
TF_VERSION,
default_tf_session_config,
op_module,
tf,
)
from deepmd.nvnmd.descriptor.se_atten import (
build_davg_dstd,
build_op_descriptor,
check_switch_range,
descrpt2r4,
filter_GR2D,
filter_lower_R42GR,
)
from deepmd.nvnmd.utils.config import (
nvnmd_cfg,
)
from deepmd.utils.compress import (
get_extra_side_embedding_net_variable,
get_two_side_type_embedding,
make_data,
)
from deepmd.utils.graph import (
get_attention_layer_variables_from_graph_def,
get_extra_embedding_net_suffix,
get_extra_embedding_net_variables_from_graph_def,
get_pattern_nodes_from_graph_def,
get_tensor_by_name_from_graph,
)
from deepmd.utils.network import (
embedding_net,
one_layer,
)
from deepmd.utils.sess import (
run_sess,
)
from deepmd.utils.tabulate import (
DPTabulate,
)
from .descriptor import (
Descriptor,
)
from .se_a import (
DescrptSeA,
)
log = logging.getLogger(__name__)
[docs]@Descriptor.register("se_atten")
class DescrptSeAtten(DescrptSeA):
r"""Smooth version descriptor with attention.
Parameters
----------
rcut
The cut-off radius :math:`r_c`
rcut_smth
From where the environment matrix should be smoothed :math:`r_s`
sel : int
sel[i] specifies the maxmum number of type i atoms in the cut-off radius
neuron : list[int]
Number of neurons in each hidden layers of the embedding net :math:`\mathcal{N}`
axis_neuron
Number of the axis neuron :math:`M_2` (number of columns of the sub-matrix of the embedding matrix)
resnet_dt
Time-step `dt` in the resnet construction:
y = x + dt * \phi (Wx + b)
trainable
If the weights of embedding net are trainable.
seed
Random seed for initializing the network parameters.
type_one_side
Try to build N_types embedding nets. Otherwise, building N_types^2 embedding nets
exclude_types : List[List[int]]
The excluded pairs of types which have no interaction with each other.
For example, `[[0, 1]]` means no interaction between type 0 and type 1.
set_davg_zero
Set the shift of embedding net input to zero.
activation_function
The activation function in the embedding net. Supported options are |ACTIVATION_FN|
precision
The precision of the embedding net parameters. Supported options are |PRECISION|
uniform_seed
Only for the purpose of backward compatibility, retrieves the old behavior of using the random seed
attn
The length of hidden vector during scale-dot attention computation.
attn_layer
The number of layers in attention mechanism.
attn_dotr
Whether to dot the relative coordinates on the attention weights as a gated scheme.
attn_mask
Whether to mask the diagonal in the attention weights.
multi_task
If the model has multi fitting nets to train.
stripped_type_embedding
Whether to strip the type embedding into a separated embedding network.
Default value will be True in `se_atten_v2` descriptor.
smooth_type_embedding
When using stripped type embedding, whether to dot smooth factor on the network output of type embedding
to keep the network smooth, instead of setting `set_davg_zero` to be True.
Default value will be True in `se_atten_v2` descriptor.
Raises
------
ValueError
if ntypes is 0.
"""
def __init__(
self,
rcut: float,
rcut_smth: float,
sel: int,
ntypes: int,
neuron: List[int] = [24, 48, 96],
axis_neuron: int = 8,
resnet_dt: bool = False,
trainable: bool = True,
seed: Optional[int] = None,
type_one_side: bool = True,
set_davg_zero: bool = True,
exclude_types: List[List[int]] = [],
activation_function: str = "tanh",
precision: str = "default",
uniform_seed: bool = False,
attn: int = 128,
attn_layer: int = 2,
attn_dotr: bool = True,
attn_mask: bool = False,
multi_task: bool = False,
stripped_type_embedding: bool = False,
smooth_type_embedding: bool = False,
**kwargs,
) -> None:
if not set_davg_zero and not (
stripped_type_embedding and smooth_type_embedding
):
warnings.warn(
"Set 'set_davg_zero' False in descriptor 'se_atten' "
"may cause unexpected incontinuity during model inference!"
)
DescrptSeA.__init__(
self,
rcut,
rcut_smth,
[sel],
neuron=neuron,
axis_neuron=axis_neuron,
resnet_dt=resnet_dt,
trainable=trainable,
seed=seed,
type_one_side=type_one_side,
exclude_types=exclude_types,
set_davg_zero=set_davg_zero,
activation_function=activation_function,
precision=precision,
uniform_seed=uniform_seed,
multi_task=multi_task,
)
"""
Constructor
"""
if not (nvnmd_cfg.enable and (nvnmd_cfg.version == 1)):
assert Version(TF_VERSION) > Version(
"2"
), "se_atten only support tensorflow version 2.0 or higher."
if ntypes == 0:
raise ValueError("`model/type_map` is not set or empty!")
self.stripped_type_embedding = stripped_type_embedding
self.smooth = smooth_type_embedding
self.ntypes = ntypes
self.att_n = attn
self.attn_layer = attn_layer
self.attn_mask = attn_mask
self.attn_dotr = attn_dotr
self.filter_np_precision = get_np_precision(precision)
self.two_side_embeeding_net_variables = None
self.layer_size = len(neuron)
# descrpt config
self.sel_all_a = [sel]
self.sel_all_r = [0]
avg_zero = np.zeros([self.ntypes, self.ndescrpt]).astype(
GLOBAL_NP_FLOAT_PRECISION
)
std_ones = np.ones([self.ntypes, self.ndescrpt]).astype(
GLOBAL_NP_FLOAT_PRECISION
)
self.beta = np.zeros([self.attn_layer, self.filter_neuron[-1]]).astype(
GLOBAL_NP_FLOAT_PRECISION
)
self.gamma = np.ones([self.attn_layer, self.filter_neuron[-1]]).astype(
GLOBAL_NP_FLOAT_PRECISION
)
self.attention_layer_variables = None
sub_graph = tf.Graph()
with sub_graph.as_default():
name_pfx = "d_sea_"
for ii in ["coord", "box"]:
self.place_holders[ii] = tf.placeholder(
GLOBAL_NP_FLOAT_PRECISION, [None, None], name=name_pfx + "t_" + ii
)
self.place_holders["type"] = tf.placeholder(
tf.int32, [None, None], name=name_pfx + "t_type"
)
self.place_holders["natoms_vec"] = tf.placeholder(
tf.int32, [self.ntypes + 2], name=name_pfx + "t_natoms"
)
self.place_holders["default_mesh"] = tf.placeholder(
tf.int32, [None], name=name_pfx + "t_mesh"
)
(
self.stat_descrpt,
self.descrpt_deriv_t,
self.rij_t,
self.nlist_t,
self.nei_type_vec_t,
self.nmask_t,
) = op_module.prod_env_mat_a_mix(
self.place_holders["coord"],
self.place_holders["type"],
self.place_holders["natoms_vec"],
self.place_holders["box"],
self.place_holders["default_mesh"],
tf.constant(avg_zero),
tf.constant(std_ones),
rcut_a=self.rcut_a,
rcut_r=self.rcut_r,
rcut_r_smth=self.rcut_r_smth,
sel_a=self.sel_all_a,
sel_r=self.sel_all_r,
)
if len(self.exclude_types):
# exclude types applied to data stat
mask = self.build_type_exclude_mask_mixed(
self.exclude_types,
self.ntypes,
self.sel_a,
self.ndescrpt,
# for data stat, nloc == nall
self.place_holders["type"],
tf.size(self.place_holders["type"]),
self.nei_type_vec_t, # extra input for atten
)
self.stat_descrpt *= tf.reshape(mask, tf.shape(self.stat_descrpt))
self.sub_sess = tf.Session(graph=sub_graph, config=default_tf_session_config)
[docs] def enable_compression(
self,
min_nbor_dist: float,
graph: tf.Graph,
graph_def: tf.GraphDef,
table_extrapolate: float = 5,
table_stride_1: float = 0.01,
table_stride_2: float = 0.1,
check_frequency: int = -1,
suffix: str = "",
) -> None:
"""Reveive the statisitcs (distance, max_nbor_size and env_mat_range) of the training data.
Parameters
----------
min_nbor_dist
The nearest distance between atoms
graph : tf.Graph
The graph of the model
graph_def : tf.GraphDef
The graph_def of the model
table_extrapolate
The scale of model extrapolation
table_stride_1
The uniform stride of the first table
table_stride_2
The uniform stride of the second table
check_frequency
The overflow check frequency
suffix : str, optional
The suffix of the scope
"""
# do some checks before the mocel compression process
assert (
not self.filter_resnet_dt
), "Model compression error: descriptor resnet_dt must be false!"
for tt in self.exclude_types:
if (tt[0] not in range(self.ntypes)) or (tt[1] not in range(self.ntypes)):
raise RuntimeError(
"exclude types"
+ str(tt)
+ " must within the number of atomic types "
+ str(self.ntypes)
+ "!"
)
if self.ntypes * self.ntypes - len(self.exclude_types) == 0:
raise RuntimeError(
"empty embedding-net are not supported in model compression!"
)
if self.attn_layer != 0:
raise RuntimeError("can not compress model when attention layer is not 0.")
ret = get_pattern_nodes_from_graph_def(
graph_def,
f"filter_type_all{suffix}/.+{get_extra_embedding_net_suffix(type_one_side=False)}",
)
if len(ret) == 0:
raise RuntimeError(
f"can not find variables of embedding net `*{get_extra_embedding_net_suffix(type_one_side=False)}` from graph_def, maybe it is not a compressible model."
)
self.compress = True
self.table = DPTabulate(
self,
self.filter_neuron,
graph,
graph_def,
True,
self.exclude_types,
self.compress_activation_fn,
suffix=suffix,
)
self.table_config = [
table_extrapolate,
table_stride_1,
table_stride_2,
check_frequency,
]
self.lower, self.upper = self.table.build(
min_nbor_dist, table_extrapolate, table_stride_1, table_stride_2
)
self.final_type_embedding = get_two_side_type_embedding(self, graph)
type_side_suffix = get_extra_embedding_net_suffix(type_one_side=False)
self.matrix = get_extra_side_embedding_net_variable(
self, graph_def, type_side_suffix, "matrix", suffix
)
self.bias = get_extra_side_embedding_net_variable(
self, graph_def, type_side_suffix, "bias", suffix
)
self.two_embd = make_data(self, self.final_type_embedding)
self.davg = get_tensor_by_name_from_graph(graph, f"descrpt_attr{suffix}/t_avg")
self.dstd = get_tensor_by_name_from_graph(graph, f"descrpt_attr{suffix}/t_std")
[docs] def build(
self,
coord_: tf.Tensor,
atype_: tf.Tensor,
natoms: tf.Tensor,
box_: tf.Tensor,
mesh: tf.Tensor,
input_dict: dict,
reuse: Optional[bool] = None,
suffix: str = "",
) -> tf.Tensor:
"""Build the computational graph for the descriptor.
Parameters
----------
coord_
The coordinate of atoms
atype_
The type of atoms
natoms
The number of atoms. This tensor has the length of Ntypes + 2
natoms[0]: number of local atoms
natoms[1]: total number of atoms held by this processor
natoms[i]: 2 <= i < Ntypes+2, number of type i atoms
box_ : tf.Tensor
The box of the system
mesh
For historical reasons, only the length of the Tensor matters.
if size of mesh == 6, pbc is assumed.
if size of mesh == 0, no-pbc is assumed.
input_dict
Dictionary for additional inputs
reuse
The weights in the networks should be reused when get the variable.
suffix
Name suffix to identify this descriptor
Returns
-------
descriptor
The output descriptor
"""
davg = self.davg
dstd = self.dstd
if nvnmd_cfg.enable:
nvnmd_cfg.set_ntype(self.ntypes)
if nvnmd_cfg.restore_descriptor:
davg, dstd = build_davg_dstd()
check_switch_range(davg, dstd)
with tf.variable_scope("descrpt_attr" + suffix, reuse=reuse):
if davg is None:
davg = np.zeros([self.ntypes, self.ndescrpt])
if dstd is None:
dstd = np.ones([self.ntypes, self.ndescrpt])
t_rcut = tf.constant(
np.max([self.rcut_r, self.rcut_a]),
name="rcut",
dtype=GLOBAL_TF_FLOAT_PRECISION,
)
t_ntypes = tf.constant(self.ntypes, name="ntypes", dtype=tf.int32)
t_ndescrpt = tf.constant(self.ndescrpt, name="ndescrpt", dtype=tf.int32)
t_sel = tf.constant(self.sel_a, name="sel", dtype=tf.int32)
t_original_sel = tf.constant(
self.original_sel if self.original_sel is not None else self.sel_a,
name="original_sel",
dtype=tf.int32,
)
self.t_avg = tf.get_variable(
"t_avg",
davg.shape,
dtype=GLOBAL_TF_FLOAT_PRECISION,
trainable=False,
initializer=tf.constant_initializer(davg),
)
self.t_std = tf.get_variable(
"t_std",
dstd.shape,
dtype=GLOBAL_TF_FLOAT_PRECISION,
trainable=False,
initializer=tf.constant_initializer(dstd),
)
with tf.control_dependencies([t_sel, t_original_sel]):
coord = tf.reshape(coord_, [-1, natoms[1] * 3])
box = tf.reshape(box_, [-1, 9])
atype = tf.reshape(atype_, [-1, natoms[1]])
self.attn_weight = [None for i in range(self.attn_layer)]
self.angular_weight = [None for i in range(self.attn_layer)]
self.attn_weight_final = [None for i in range(self.attn_layer)]
op_descriptor = (
build_op_descriptor() if nvnmd_cfg.enable else op_module.prod_env_mat_a_mix
)
(
self.descrpt,
self.descrpt_deriv,
self.rij,
self.nlist,
self.nei_type_vec,
self.nmask,
) = op_descriptor(
coord,
atype,
natoms,
box,
mesh,
self.t_avg,
self.t_std,
rcut_a=self.rcut_a,
rcut_r=self.rcut_r,
rcut_r_smth=self.rcut_r_smth,
sel_a=self.sel_all_a,
sel_r=self.sel_all_r,
)
self.nei_type_vec = tf.reshape(self.nei_type_vec, [-1])
self.nmask = tf.cast(
tf.reshape(self.nmask, [-1, 1, self.sel_all_a[0]]),
self.filter_precision,
)
self.negative_mask = -(2 << 32) * (1.0 - self.nmask)
# hard coding the magnitude of attention weight shift
self.smth_attn_w_shift = 20.0
# only used when tensorboard was set as true
tf.summary.histogram("descrpt", self.descrpt)
tf.summary.histogram("rij", self.rij)
tf.summary.histogram("nlist", self.nlist)
self.descrpt_reshape = tf.reshape(self.descrpt, [-1, self.ndescrpt])
# prevent lookup error; the actual atype already used for nlist
atype = tf.clip_by_value(atype, 0, self.ntypes - 1)
self.atype_nloc = tf.reshape(
tf.slice(atype, [0, 0], [-1, natoms[0]]), [-1]
) ## lammps will have error without this
self._identity_tensors(suffix=suffix)
if self.smooth:
self.sliced_avg = tf.reshape(
tf.slice(
tf.reshape(self.t_avg, [self.ntypes, -1, 4]), [0, 0, 0], [-1, 1, 1]
),
[self.ntypes, 1],
)
self.sliced_std = tf.reshape(
tf.slice(
tf.reshape(self.t_std, [self.ntypes, -1, 4]), [0, 0, 0], [-1, 1, 1]
),
[self.ntypes, 1],
)
self.avg_looked_up = tf.reshape(
tf.nn.embedding_lookup(self.sliced_avg, self.atype_nloc),
[-1, natoms[0], 1],
)
self.std_looked_up = tf.reshape(
tf.nn.embedding_lookup(self.sliced_std, self.atype_nloc),
[-1, natoms[0], 1],
)
self.recovered_r = (
tf.reshape(
tf.slice(
tf.reshape(self.descrpt_reshape, [-1, 4]), [0, 0], [-1, 1]
),
[-1, natoms[0], self.sel_all_a[0]],
)
* self.std_looked_up
+ self.avg_looked_up
)
uu = 1 - self.rcut_r_smth * self.recovered_r
self.recovered_switch = -uu * uu * uu + 1
self.recovered_switch = tf.clip_by_value(self.recovered_switch, 0.0, 1.0)
self.recovered_switch = tf.cast(
self.recovered_switch, self.filter_precision
)
self.dout, self.qmat = self._pass_filter(
self.descrpt_reshape,
self.atype_nloc,
natoms,
input_dict,
suffix=suffix,
reuse=reuse,
trainable=self.trainable,
)
# only used when tensorboard was set as true
tf.summary.histogram("embedding_net_output", self.dout)
return self.dout
def _pass_filter(
self, inputs, atype, natoms, input_dict, reuse=None, suffix="", trainable=True
):
assert (
input_dict is not None
and input_dict.get("type_embedding", None) is not None
), "se_atten desctiptor must use type_embedding"
type_embedding = input_dict.get("type_embedding", None)
inputs = tf.reshape(inputs, [-1, natoms[0], self.ndescrpt])
output = []
output_qmat = []
inputs_i = inputs
inputs_i = tf.reshape(inputs_i, [-1, self.ndescrpt])
type_i = -1
if len(self.exclude_types):
mask = self.build_type_exclude_mask_mixed(
self.exclude_types,
self.ntypes,
self.sel_a,
self.ndescrpt,
self.atype_nloc, # when nloc != nall, pass nloc to mask
tf.shape(inputs_i)[0],
self.nei_type_vec, # extra input for atten
)
if self.smooth:
inputs_i = tf.where(
tf.cast(mask, tf.bool),
inputs_i,
# (nframes * nloc, 1) -> (nframes * nloc, ndescrpt)
tf.tile(
tf.reshape(self.avg_looked_up, [-1, 1]), [1, self.ndescrpt]
),
)
self.recovered_switch *= tf.reshape(
tf.slice(
tf.reshape(tf.cast(mask, self.filter_precision), [-1, 4]),
[0, 0],
[-1, 1],
),
[-1, natoms[0], self.sel_all_a[0]],
)
else:
inputs_i *= mask
if nvnmd_cfg.enable and nvnmd_cfg.quantize_descriptor:
inputs_i = descrpt2r4(inputs_i, atype)
layer, qmat = self._filter(
inputs_i,
type_i,
natoms,
name="filter_type_all" + suffix,
suffix=suffix,
reuse=reuse,
trainable=trainable,
activation_fn=self.filter_activation_fn,
type_embedding=type_embedding,
atype=atype,
)
layer = tf.reshape(layer, [tf.shape(inputs)[0], natoms[0], self.get_dim_out()])
qmat = tf.reshape(
qmat, [tf.shape(inputs)[0], natoms[0], self.get_dim_rot_mat_1() * 3]
)
output.append(layer)
output_qmat.append(qmat)
output = tf.concat(output, axis=1)
output_qmat = tf.concat(output_qmat, axis=1)
return output, output_qmat
def _compute_dstats_sys_smth(
self,
data_coord,
data_box,
data_atype,
natoms_vec,
mesh,
mixed_type=False,
real_natoms_vec=None,
):
dd_all, descrpt_deriv_t, rij_t, nlist_t, nei_type_vec_t, nmask_t = run_sess(
self.sub_sess,
[
self.stat_descrpt,
self.descrpt_deriv_t,
self.rij_t,
self.nlist_t,
self.nei_type_vec_t,
self.nmask_t,
],
feed_dict={
self.place_holders["coord"]: data_coord,
self.place_holders["type"]: data_atype,
self.place_holders["natoms_vec"]: natoms_vec,
self.place_holders["box"]: data_box,
self.place_holders["default_mesh"]: mesh,
},
)
if mixed_type:
nframes = dd_all.shape[0]
sysr = [0.0 for i in range(self.ntypes)]
sysa = [0.0 for i in range(self.ntypes)]
sysn = [0 for i in range(self.ntypes)]
sysr2 = [0.0 for i in range(self.ntypes)]
sysa2 = [0.0 for i in range(self.ntypes)]
for ff in range(nframes):
natoms = real_natoms_vec[ff]
dd_ff = np.reshape(dd_all[ff], [-1, self.ndescrpt * natoms_vec[0]])
start_index = 0
for type_i in range(self.ntypes):
end_index = (
start_index + self.ndescrpt * natoms[2 + type_i]
) # center atom split
dd = dd_ff[:, start_index:end_index]
dd = np.reshape(
dd, [-1, self.ndescrpt]
) # nframes * typen_atoms , nnei * 4
start_index = end_index
# compute
dd = np.reshape(dd, [-1, 4]) # nframes * typen_atoms * nnei, 4
ddr = dd[:, :1]
dda = dd[:, 1:]
sumr = np.sum(ddr)
suma = np.sum(dda) / 3.0
sumn = dd.shape[0]
sumr2 = np.sum(np.multiply(ddr, ddr))
suma2 = np.sum(np.multiply(dda, dda)) / 3.0
sysr[type_i] += sumr
sysa[type_i] += suma
sysn[type_i] += sumn
sysr2[type_i] += sumr2
sysa2[type_i] += suma2
else:
natoms = natoms_vec
dd_all = np.reshape(dd_all, [-1, self.ndescrpt * natoms[0]])
start_index = 0
sysr = []
sysa = []
sysn = []
sysr2 = []
sysa2 = []
for type_i in range(self.ntypes):
end_index = (
start_index + self.ndescrpt * natoms[2 + type_i]
) # center atom split
dd = dd_all[:, start_index:end_index]
dd = np.reshape(
dd, [-1, self.ndescrpt]
) # nframes * typen_atoms , nnei * 4
start_index = end_index
# compute
dd = np.reshape(dd, [-1, 4]) # nframes * typen_atoms * nnei, 4
ddr = dd[:, :1]
dda = dd[:, 1:]
sumr = np.sum(ddr)
suma = np.sum(dda) / 3.0
sumn = dd.shape[0]
sumr2 = np.sum(np.multiply(ddr, ddr))
suma2 = np.sum(np.multiply(dda, dda)) / 3.0
sysr.append(sumr)
sysa.append(suma)
sysn.append(sumn)
sysr2.append(sumr2)
sysa2.append(suma2)
return sysr, sysr2, sysa, sysa2, sysn
def _lookup_type_embedding(
self,
xyz_scatter,
natype,
type_embedding,
):
"""Concatenate `type_embedding` of neighbors and `xyz_scatter`.
If not self.type_one_side, concatenate `type_embedding` of center atoms as well.
Parameters
----------
xyz_scatter:
shape is [nframes*natoms[0]*self.nnei, 1]
natype:
neighbor atom type
type_embedding:
shape is [self.ntypes, Y] where Y=jdata['type_embedding']['neuron'][-1]
Returns
-------
embedding:
environment of each atom represented by embedding.
"""
te_out_dim = type_embedding.get_shape().as_list()[-1]
self.test_type_embedding = type_embedding
self.test_nei_embed = tf.nn.embedding_lookup(
type_embedding, self.nei_type_vec
) # shape is [self.nnei, 1+te_out_dim]
# nei_embed = tf.tile(nei_embed, (nframes * natoms[0], 1)) # shape is [nframes*natoms[0]*self.nnei, te_out_dim]
nei_embed = tf.reshape(self.test_nei_embed, [-1, te_out_dim])
self.embedding_input = tf.concat(
[xyz_scatter, nei_embed], 1
) # shape is [nframes*natoms[0]*self.nnei, 1+te_out_dim]
if not self.type_one_side:
self.atm_embed = tf.nn.embedding_lookup(
type_embedding, natype
) # shape is [nframes*natoms[0], te_out_dim]
self.atm_embed = tf.tile(
self.atm_embed, [1, self.nnei]
) # shape is [nframes*natoms[0], self.nnei*te_out_dim]
self.atm_embed = tf.reshape(
self.atm_embed, [-1, te_out_dim]
) # shape is [nframes*natoms[0]*self.nnei, te_out_dim]
self.embedding_input_2 = tf.concat(
[self.embedding_input, self.atm_embed], 1
) # shape is [nframes*natoms[0]*self.nnei, 1+te_out_dim+te_out_dim]
return self.embedding_input_2
return self.embedding_input
def _feedforward(self, input_xyz, d_in, d_mid):
residual = input_xyz
input_xyz = tf.nn.relu(
one_layer(
input_xyz,
d_mid,
name="c_ffn1",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=True,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
)
input_xyz = one_layer(
input_xyz,
d_in,
name="c_ffn2",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=True,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
input_xyz += residual
input_xyz = tf.keras.layers.LayerNormalization()(input_xyz)
return input_xyz
def _scaled_dot_attn(
self,
Q,
K,
V,
temperature,
input_r,
dotr=False,
do_mask=False,
layer=0,
save_weights=True,
):
attn = tf.matmul(Q / temperature, K, transpose_b=True)
if self.smooth:
# (nb x nloc) x nsel
nsel = self.sel_all_a[0]
attn = (attn + self.smth_attn_w_shift) * tf.reshape(
self.recovered_switch, [-1, 1, nsel]
) * tf.reshape(
self.recovered_switch, [-1, nsel, 1]
) - self.smth_attn_w_shift
else:
attn *= self.nmask
attn += self.negative_mask
attn = tf.nn.softmax(attn, axis=-1)
if self.smooth:
attn = (
attn
* tf.reshape(self.recovered_switch, [-1, 1, nsel])
* tf.reshape(self.recovered_switch, [-1, nsel, 1])
)
else:
attn *= tf.reshape(self.nmask, [-1, attn.shape[-1], 1])
if save_weights:
self.attn_weight[layer] = attn[0] # atom 0
if dotr:
angular_weight = tf.matmul(input_r, input_r, transpose_b=True) # normalized
attn *= angular_weight
if save_weights:
self.angular_weight[layer] = angular_weight[0] # atom 0
self.attn_weight_final[layer] = attn[0] # atom 0
if do_mask:
nei = int(attn.shape[-1])
mask = tf.cast(tf.ones((nei, nei)) - tf.eye(nei), self.filter_precision)
attn *= mask
output = tf.matmul(attn, V)
return output
def _attention_layers(
self,
input_xyz,
layer_num,
shape_i,
outputs_size,
input_r,
dotr=False,
do_mask=False,
trainable=True,
suffix="",
):
sd_k = tf.sqrt(tf.cast(1.0, dtype=self.filter_precision))
for i in range(layer_num):
name = f"attention_layer_{i}{suffix}"
with tf.variable_scope(name, reuse=tf.AUTO_REUSE):
# input_xyz_in = tf.nn.l2_normalize(input_xyz, -1)
Q_c = one_layer(
input_xyz,
self.att_n,
name="c_query",
scope=name + "/",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
if not self.uniform_seed and self.seed is not None:
self.seed += 1
K_c = one_layer(
input_xyz,
self.att_n,
name="c_key",
scope=name + "/",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
if not self.uniform_seed and self.seed is not None:
self.seed += 1
V_c = one_layer(
input_xyz,
self.att_n,
name="c_value",
scope=name + "/",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
if not self.uniform_seed and self.seed is not None:
self.seed += 1
# # natom x nei_type_i x out_size
# xyz_scatter = tf.reshape(xyz_scatter, (-1, shape_i[1] // 4, outputs_size[-1]))
# natom x nei_type_i x att_n
Q_c = tf.nn.l2_normalize(
tf.reshape(Q_c, (-1, shape_i[1] // 4, self.att_n)), -1
)
K_c = tf.nn.l2_normalize(
tf.reshape(K_c, (-1, shape_i[1] // 4, self.att_n)), -1
)
V_c = tf.nn.l2_normalize(
tf.reshape(V_c, (-1, shape_i[1] // 4, self.att_n)), -1
)
input_att = self._scaled_dot_attn(
Q_c, K_c, V_c, sd_k, input_r, dotr=dotr, do_mask=do_mask, layer=i
)
input_att = tf.reshape(input_att, (-1, self.att_n))
# (natom x nei_type_i) x out_size
input_xyz += one_layer(
input_att,
outputs_size[-1],
name="c_out",
scope=name + "/",
reuse=tf.AUTO_REUSE,
seed=self.seed,
activation_fn=None,
precision=self.filter_precision,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.attention_layer_variables,
)
if not self.uniform_seed and self.seed is not None:
self.seed += 1
input_xyz = tf.keras.layers.LayerNormalization(
beta_initializer=tf.constant_initializer(self.beta[i]),
gamma_initializer=tf.constant_initializer(self.gamma[i]),
dtype=self.filter_precision,
)(input_xyz)
# input_xyz = self._feedforward(input_xyz, outputs_size[-1], self.att_n)
return input_xyz
def _filter_lower(
self,
type_i,
type_input,
start_index,
incrs_index,
inputs,
type_embedding=None,
atype=None,
is_exclude=False,
activation_fn=None,
bavg=0.0,
stddev=1.0,
trainable=True,
suffix="",
name="filter_",
reuse=None,
):
"""Input env matrix, returns R.G."""
outputs_size = [1, *self.filter_neuron]
# cut-out inputs
# with natom x (nei_type_i x 4)
inputs_i = tf.slice(inputs, [0, start_index * 4], [-1, incrs_index * 4])
shape_i = inputs_i.get_shape().as_list()
natom = tf.shape(inputs_i)[0]
# with (natom x nei_type_i) x 4
inputs_reshape = tf.reshape(inputs_i, [-1, 4])
# with (natom x nei_type_i) x 1
xyz_scatter = tf.reshape(tf.slice(inputs_reshape, [0, 0], [-1, 1]), [-1, 1])
assert atype is not None, "atype must exist!!"
type_embedding = tf.cast(type_embedding, self.filter_precision) # ntypes * Y
# natom x 4 x outputs_size
if not is_exclude:
with tf.variable_scope(name, reuse=reuse):
# with (natom x nei_type_i) x out_size
if not self.stripped_type_embedding:
log.info("use the previous se_atten model")
xyz_scatter = self._lookup_type_embedding(
xyz_scatter, atype, type_embedding
)
xyz_scatter = embedding_net(
xyz_scatter,
self.filter_neuron,
self.filter_precision,
activation_fn=activation_fn,
resnet_dt=self.filter_resnet_dt,
name_suffix="",
stddev=stddev,
bavg=bavg,
seed=self.seed,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.embedding_net_variables,
mixed_prec=self.mixed_prec,
)
if (not self.uniform_seed) and (self.seed is not None):
self.seed += self.seed_shift
else:
if self.attn_layer == 0:
log.info(
"use the compressible model with stripped type embedding"
)
else:
log.info(
"use the non-compressible model with stripped type embedding"
)
if nvnmd_cfg.enable:
if nvnmd_cfg.quantize_descriptor:
return filter_lower_R42GR(
inputs_i,
atype,
self.nei_type_vec,
)
elif nvnmd_cfg.restore_descriptor:
self.embedding_net_variables = (
nvnmd_cfg.get_dp_init_weights()
)
self.two_side_embeeding_net_variables = (
nvnmd_cfg.get_dp_init_weights()
)
if not self.compress:
xyz_scatter = embedding_net(
xyz_scatter,
self.filter_neuron,
self.filter_precision,
activation_fn=activation_fn,
resnet_dt=self.filter_resnet_dt,
name_suffix="",
stddev=stddev,
bavg=bavg,
seed=self.seed,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.embedding_net_variables,
mixed_prec=self.mixed_prec,
)
if (not self.uniform_seed) and (self.seed is not None):
self.seed += self.seed_shift
else:
net = "filter_net"
info = [
self.lower[net],
self.upper[net],
self.upper[net] * self.table_config[0],
self.table_config[1],
self.table_config[2],
self.table_config[3],
]
padding_ntypes = type_embedding.shape[
0
] # this must be self.ntypes + 1
atype_expand = tf.reshape(atype, [-1, 1])
idx_i = tf.tile(atype_expand * padding_ntypes, [1, self.nnei])
idx_j = tf.reshape(self.nei_type_vec, [-1, self.nnei])
idx = idx_i + idx_j
index_of_two_side = tf.reshape(idx, [-1])
if self.compress:
two_embd = tf.nn.embedding_lookup(
self.two_embd, index_of_two_side
)
else:
type_embedding_nei = tf.tile(
tf.reshape(type_embedding, [1, padding_ntypes, -1]),
[padding_ntypes, 1, 1],
) # (ntypes) * ntypes * Y
type_embedding_center = tf.tile(
tf.reshape(type_embedding, [padding_ntypes, 1, -1]),
[1, padding_ntypes, 1],
) # ntypes * (ntypes) * Y
two_side_type_embedding = tf.concat(
[type_embedding_nei, type_embedding_center], -1
) # ntypes * ntypes * (Y+Y)
two_side_type_embedding = tf.reshape(
two_side_type_embedding,
[-1, two_side_type_embedding.shape[-1]],
)
embedding_of_two_side_type_embedding = embedding_net(
two_side_type_embedding,
self.filter_neuron,
self.filter_precision,
activation_fn=activation_fn,
resnet_dt=self.filter_resnet_dt,
name_suffix=get_extra_embedding_net_suffix(
type_one_side=False
),
stddev=stddev,
bavg=bavg,
seed=self.seed,
trainable=trainable,
uniform_seed=self.uniform_seed,
initial_variables=self.two_side_embeeding_net_variables,
mixed_prec=self.mixed_prec,
)
if (not self.uniform_seed) and (self.seed is not None):
self.seed += self.seed_shift
two_embd = tf.nn.embedding_lookup(
embedding_of_two_side_type_embedding, index_of_two_side
)
if self.smooth:
two_embd = two_embd * tf.reshape(self.recovered_switch, [-1, 1])
if not self.compress:
xyz_scatter = xyz_scatter * two_embd + xyz_scatter
else:
return op_module.tabulate_fusion_se_atten(
tf.cast(self.table.data[net], self.filter_precision),
info,
xyz_scatter,
tf.reshape(inputs_i, [natom, shape_i[1] // 4, 4]),
two_embd,
last_layer_size=outputs_size[-1],
is_sorted=len(self.exclude_types) == 0,
)
input_r = tf.slice(
tf.reshape(inputs_i, (-1, shape_i[1] // 4, 4)), [0, 0, 1], [-1, -1, 3]
)
input_r = tf.nn.l2_normalize(input_r, -1)
# natom x nei_type_i x out_size
xyz_scatter_att = tf.reshape(
self._attention_layers(
xyz_scatter,
self.attn_layer,
shape_i,
outputs_size,
input_r,
dotr=self.attn_dotr,
do_mask=self.attn_mask,
trainable=trainable,
suffix=suffix,
),
(-1, shape_i[1] // 4, outputs_size[-1]),
)
# xyz_scatter = tf.reshape(xyz_scatter, (-1, shape_i[1] // 4, outputs_size[-1]))
else:
raise RuntimeError("this should not be touched")
# When using tf.reshape(inputs_i, [-1, shape_i[1]//4, 4]) below
# [588 24] -> [588 6 4] correct
# but if sel is zero
# [588 0] -> [147 0 4] incorrect; the correct one is [588 0 4]
# So we need to explicitly assign the shape to tf.shape(inputs_i)[0] instead of -1
return tf.matmul(
tf.reshape(inputs_i, [natom, shape_i[1] // 4, 4]),
xyz_scatter_att,
transpose_a=True,
)
@cast_precision
def _filter(
self,
inputs,
type_input,
natoms,
type_embedding=None,
atype=None,
activation_fn=tf.nn.tanh,
stddev=1.0,
bavg=0.0,
suffix="",
name="linear",
reuse=None,
trainable=True,
):
nframes = tf.shape(tf.reshape(inputs, [-1, natoms[0], self.ndescrpt]))[0]
# natom x (nei x 4)
shape = inputs.get_shape().as_list()
outputs_size = [1, *self.filter_neuron]
outputs_size_2 = self.n_axis_neuron
start_index = 0
type_i = 0
# natom x 4 x outputs_size
xyz_scatter_1 = self._filter_lower(
type_i,
type_input,
start_index,
np.cumsum(self.sel_a)[-1],
inputs,
type_embedding=type_embedding,
is_exclude=False,
activation_fn=activation_fn,
stddev=stddev,
bavg=bavg,
trainable=trainable,
suffix=suffix,
name=name,
reuse=reuse,
atype=atype,
)
if nvnmd_cfg.enable:
return filter_GR2D(xyz_scatter_1)
# natom x nei x outputs_size
# xyz_scatter = tf.concat(xyz_scatter_total, axis=1)
# natom x nei x 4
# inputs_reshape = tf.reshape(inputs, [-1, shape[1]//4, 4])
# natom x 4 x outputs_size
# xyz_scatter_1 = tf.matmul(inputs_reshape, xyz_scatter, transpose_a = True)
if self.original_sel is None:
# shape[1] = nnei * 4
nnei = shape[1] / 4
else:
nnei = tf.cast(
tf.Variable(
np.sum(self.original_sel),
dtype=tf.int32,
trainable=False,
name="nnei",
),
self.filter_precision,
)
xyz_scatter_1 = xyz_scatter_1 / nnei
# natom x 4 x outputs_size_2
xyz_scatter_2 = tf.slice(xyz_scatter_1, [0, 0, 0], [-1, -1, outputs_size_2])
# # natom x 3 x outputs_size_2
# qmat = tf.slice(xyz_scatter_2, [0,1,0], [-1, 3, -1])
# natom x 3 x outputs_size_1
qmat = tf.slice(xyz_scatter_1, [0, 1, 0], [-1, 3, -1])
# natom x outputs_size_1 x 3
qmat = tf.transpose(qmat, perm=[0, 2, 1])
# natom x outputs_size x outputs_size_2
result = tf.matmul(xyz_scatter_1, xyz_scatter_2, transpose_a=True)
# natom x (outputs_size x outputs_size_2)
result = tf.reshape(result, [-1, outputs_size_2 * outputs_size[-1]])
return result, qmat
[docs] def init_variables(
self,
graph: tf.Graph,
graph_def: tf.GraphDef,
suffix: str = "",
) -> None:
"""Init the embedding net variables with the given dict.
Parameters
----------
graph : tf.Graph
The input frozen model graph
graph_def : tf.GraphDef
The input frozen model graph_def
suffix : str, optional
The suffix of the scope
"""
super().init_variables(graph=graph, graph_def=graph_def, suffix=suffix)
self.attention_layer_variables = get_attention_layer_variables_from_graph_def(
graph_def, suffix=suffix
)
if self.attn_layer > 0:
self.beta[0] = self.attention_layer_variables[
f"attention_layer_0{suffix}/layer_normalization/beta"
]
self.gamma[0] = self.attention_layer_variables[
f"attention_layer_0{suffix}/layer_normalization/gamma"
]
for i in range(1, self.attn_layer):
self.beta[i] = self.attention_layer_variables[
f"attention_layer_{i}{suffix}/layer_normalization_{i}/beta"
]
self.gamma[i] = self.attention_layer_variables[
f"attention_layer_{i}{suffix}/layer_normalization_{i}/gamma"
]
if self.stripped_type_embedding:
self.two_side_embeeding_net_variables = (
get_extra_embedding_net_variables_from_graph_def(
graph_def,
suffix,
get_extra_embedding_net_suffix(type_one_side=False),
self.layer_size,
)
)
[docs] def build_type_exclude_mask_mixed(
self,
exclude_types: Set[Tuple[int, int]],
ntypes: int,
sel: List[int],
ndescrpt: int,
atype: tf.Tensor,
shape0: tf.Tensor,
nei_type_vec: tf.Tensor,
) -> tf.Tensor:
r"""Build the type exclude mask for the attention descriptor.
Notes
-----
This method has the similiar way to build the type exclude mask as
:meth:`deepmd.descriptor.descriptor.Descriptor.build_type_exclude_mask`.
The mathmatical expression has been explained in that method.
The difference is that the attention descriptor has provided the type of
the neighbors (idx_j) that is not in order, so we use it from an extra
input.
Parameters
----------
exclude_types : List[Tuple[int, int]]
The list of excluded types, e.g. [(0, 1), (1, 0)] means the interaction
between type 0 and type 1 is excluded.
ntypes : int
The number of types.
sel : List[int]
The list of the number of selected neighbors for each type.
ndescrpt : int
The number of descriptors for each atom.
atype : tf.Tensor
The type of atoms, with the size of shape0.
shape0 : tf.Tensor
The shape of the first dimension of the inputs, which is equal to
nsamples * natoms.
nei_type_vec : tf.Tensor
The type of neighbors, with the size of (shape0, nnei).
Returns
-------
tf.Tensor
The type exclude mask, with the shape of (shape0, ndescrpt), and the
precision of GLOBAL_TF_FLOAT_PRECISION. The mask has the value of 1 if the
interaction between two types is not excluded, and 0 otherwise.
See Also
--------
deepmd.descriptor.descriptor.Descriptor.build_type_exclude_mask
"""
# generate a mask
# op returns ntypes when the neighbor doesn't exist, so we need to add 1
type_mask = np.array(
[
[
1 if (tt_i, tt_j) not in exclude_types else 0
for tt_i in range(ntypes + 1)
]
for tt_j in range(ntypes)
],
dtype=bool,
)
type_mask = tf.convert_to_tensor(type_mask, dtype=GLOBAL_TF_FLOAT_PRECISION)
type_mask = tf.reshape(type_mask, [-1])
# (nsamples * natoms, 1)
atype_expand = tf.reshape(atype, [-1, 1])
# (nsamples * natoms, ndescrpt)
idx_i = tf.tile(atype_expand * (ntypes + 1), (1, ndescrpt))
# idx_j has been provided by atten op
# (nsamples * natoms, nnei, 1)
idx_j = tf.reshape(nei_type_vec, [shape0, sel[0], 1])
# (nsamples * natoms, nnei, ndescrpt // nnei)
idx_j = tf.tile(idx_j, (1, 1, ndescrpt // sel[0]))
# (nsamples * natoms, ndescrpt)
idx_j = tf.reshape(idx_j, [shape0, ndescrpt])
idx = idx_i + idx_j
idx = tf.reshape(idx, [-1])
mask = tf.nn.embedding_lookup(type_mask, idx)
# same as inputs_i, (nsamples * natoms, ndescrpt)
mask = tf.reshape(mask, [-1, ndescrpt])
return mask
@property
def explicit_ntypes(self) -> bool:
"""Explicit ntypes with type embedding."""
return True
[docs] @classmethod
def update_sel(cls, global_jdata: dict, local_jdata: dict):
"""Update the selection and perform neighbor statistics.
Parameters
----------
global_jdata : dict
The global data, containing the training section
local_jdata : dict
The local data refer to the current class
"""
from deepmd.entrypoints.train import (
update_one_sel,
)
local_jdata_cpy = local_jdata.copy()
return update_one_sel(global_jdata, local_jdata_cpy, True)