14995306316

Committed 13 May 2025 11:21AM UTC coverage: 83.844%. Remained the same

Build # 14995306316

Build Type

Pull #194

github

Committed by

web-flow

Commit Message

Merge d8ac4723f into dd0c5094e

Pull Request Pull Request #194: apply black to code base

Run Details

955 of 1334 branches covered (71.59%)

Branch coverage included in aggregate %.

240 of 268 new or added lines in 50 files covered. (89.55%)

5 existing lines in 4 files now uncovered.

4515 of 5190 relevant lines covered (86.99%)

0.87 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

74.74

/qmctorch/wavefunction/orbitals/backflow/backflow_transformation.py

import torch
from torch import nn
from typing import Dict, Optional, Callable, Tuple
from ....scf import Molecule
from .kernels.backflow_kernel_base import BackFlowKernelBase
from ...jastrows.distance.electron_electron_distance import ElectronElectronDistance


class BackFlowTransformation(nn.Module):
    def __init__(
        self,
        mol: Molecule,
        backflow_kernel: BackFlowKernelBase,
        backflow_kernel_kwargs: Optional[Dict] = {},
        cuda: Optional[bool] = False,
    ):
        """Transform the electorn coordinates into backflow coordinates.
        see : Orbital-dependent backflow wave functions for real-space quantum Monte Carlo
        https://arxiv.org/abs/1910.07167

        .. math:
            \\bold{q}_i = \\bold{r}_i + \\sum_{j\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)
        """
        super().__init__()
        self.nao = mol.basis.nao
        self.nelec = mol.nelec
        self.ndim = 3

        self.backflow_kernel = backflow_kernel(mol, cuda, **backflow_kernel_kwargs)

        self.edist = ElectronElectronDistance(mol.nelec)

        self.cuda = cuda
        self.device = torch.device("cpu")
        if self.cuda:
            self.device = torch.device("cuda")

    def forward(self, pos: torch.Tensor, derivative: Optional[int] = 0) -> torch.Tensor:
        if derivative == 0:
            return self._get_backflow(pos)

        elif derivative == 1:
            return self._get_backflow_derivative(pos)

        elif derivative == 2:
            return self._get_backflow_second_derivative(pos)

        else:
            raise ValueError(
                "derivative of the backflow transformation must be 0, 1 or 2"
            )

    def _get_backflow(self, pos: torch.Tensor) -> torch.Tensor:
        """Computes the backflow transformation

        .. math:
            \\bold{q}_i = \\bold{r}_i + \\sum_{j\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)

        Args:
            pos(torch.tensor): original positions Nbatch x[Nelec*Ndim]

        Returns:
            torch.tensor: transformed positions Nbatch x[Nelec*Ndim]
        """
        # compute the difference
        # Nbatch x Nelec x Nelec x 3
        delta_ee = self.edist.get_difference(pos.reshape(-1, self.nelec, self.ndim))

        # compute the backflow function
        # Nbatch x Nelec x Nelec
        bf_kernel = self.backflow_kernel(self.edist(pos))

        # update pos
        pos = pos.reshape(-1, self.nelec, self.ndim) + (
            bf_kernel.unsqueeze(-1) * delta_ee
        ).sum(2)

        return pos.reshape(-1, self.nelec * self.ndim)

    def _get_backflow_derivative(self, pos: torch.Tensor) -> torch.Tensor:
        r"""Computes the derivative of the backflow transformation
           wrt the original positions of the electrons

        .. math::
            \\bold{q}_i = \\bold{r}_i + \\sum_{j\\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)

        .. math::
            \\frac{d q_i}{d x_k} = \\delta_{ik}(1 + \\sum_{j\\neq i} \\frac{d \\eta(r_ij)}{d x_i}(x_i-x_j) + \\eta(r_ij)) +
                                   \\delta_{i\\neq k}(-\\frac{d \\eta(r_ik)}{d x_k}(x_i-x_k) - \\eta(r_ik))

        Args:
            pos(torch.tensor): orginal positions of the electrons Nbatch x[Nelec*Ndim]

        Returns:
            torch.tensor: d q_{i}/d x_k with:
                          q_{i} bf position of elec i
                          x_k original coordinate of the kth elec
                          Nelec x  Nbatch x Nelec x Norb x Ndim
        """

        # ee dist matrix : Nbatch x  Nelec x Nelec
        ree = self.edist(pos)
        nbatch, nelec, _ = ree.shape

        # derivative ee dist matrix : Nbatch x 3 x Nelec x Nelec
        # dr_ij / dx_i = - dr_ij / dx_j
        dree = self.edist(pos, derivative=1)

        # difference between elec pos
        # Nbatch, 3, Nelec, Nelec
        delta_ee = self.edist.get_difference(pos.reshape(nbatch, nelec, 3)).permute(
            0, 3, 1, 2
        )

        # backflow kernel : Nbatch x 1 x Nelec x Nelec
        bf = self.backflow_kernel(ree)

        # (d eta(r_ij) / d r_ij) (d r_ij/d beta_i)
        # derivative of the back flow kernel : Nbatch x 3 x Nelec x Nelec
        dbf = self.backflow_kernel(ree, derivative=1).unsqueeze(1)
        dbf = dbf * dree

        # (d eta(r_ij) / d beta_i) (alpha_i - alpha_j)
        # Nbatch x 3 x 3 x Nelec x Nelec
        dbf_delta_ee = dbf.unsqueeze(1) * delta_ee.unsqueeze(2)

        # compute the delta_ij * (1 + sum k \neq i eta(rik))
        # Nbatch x Nelec x Nelec (diagonal matrix)
        delta_ij_bf = torch.diag_embed(1 + bf.sum(-1), dim1=-1, dim2=-2)

        # eye 3x3 in 1x3x3x1x1
        eye_mat = torch.eye(3, 3).view(1, 3, 3, 1, 1).to(self.device)

        # compute the delta_ab * delta_ij * (1 + sum k \neq i eta(rik))
        # Nbatch x Ndim x Ndim x Nelec x Nelec (diagonal matrix)
        delta_ab_delta_ij_bf = eye_mat * delta_ij_bf.view(nbatch, 1, 1, nelec, nelec)

        # compute sum_k df(r_ik)/dbeta_i (alpha_i - alpha_k)
        # Nbatch x Ndim x Ndim x Nelec x Nelec
        delta_ij_sum = torch.diag_embed(dbf_delta_ee.sum(-1), dim1=-1, dim2=-2)

        # compute delta_ab * f(rij)
        delta_ab_bf = eye_mat * bf.view(nbatch, 1, 1, nelec, nelec)

        # return Nbatch x Ndim(alpha) x Ndim(beta) x Nelec(i) x Nelec(j)
        # nbatch d alpha_i / d beta_j
        out = delta_ab_delta_ij_bf + delta_ij_sum - dbf_delta_ee - delta_ab_bf

        return out.unsqueeze(-1)

    def _get_backflow_second_derivative(self, pos: torch.Tensor) -> torch.Tensor:
        r"""Computes the second derivative of the backflow transformation
           wrt the original positions of the electrons

        .. math::
            \\bold{q}_i = \\bold{r}_i + \\sum_{j\\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)

        .. math::
            \\frac{d q_i}{d x_k} = \\delta_{ik}(1 + \\sum_{j\\neqi} \\frac{d \\eta(r_ij)}{d x_i} + \\eta(r_ij)) +
                                   \\delta_{i\\neq k}(-\\frac{d \\eta(r_ik)}{d x_k} - \\eta(r_ik))

        .. math::
            \\frac{d ^ 2 q_i}{d x_k ^ 2} = \\delta_{ik}(\\sum_{j\\neqi} \\frac{d ^ 2 \\eta(r_ij)}{d x_i ^ 2} + 2 \\frac{d \\eta(r_ij)}{d x_i}) +
                                       - \\delta_{i\\neq k}(\\frac{d ^ 2 \\eta(r_ik)}{d x_k ^ 2} + \\frac{d \\eta(r_ik)}{d x_k})

        Args:
            pos(torch.tensor): orginal positions of the electrons Nbatch x[Nelec*Ndim]

        Returns:
            torch.tensor: d q_{i}/d x_k with:
                          q_{i} bf position of elec i
                          x_k original coordinate of the kth elec
                          Nelec x  Nbatch x Nelec x Norb x Ndim
        """
        # ee dist matrix :
        # Nbatch x  Nelec x Nelec
        ree = self.edist(pos)
        nbatch, nelec, _ = ree.shape

        # difference between elec pos
        # Nbatch, 3, Nelec, Nelec
        delta_ee = self.edist.get_difference(pos.reshape(nbatch, nelec, 3)).permute(
            0, 3, 1, 2
        )

        # derivative ee dist matrix  d r_{ij} / d x_i
        # Nbatch x 3 x Nelec x Nelec
        dree = self.edist(pos, derivative=1)

        # derivative ee dist matrix :  d2 r_{ij} / d2 x_i
        # Nbatch x 3 x Nelec x Nelec
        d2ree = self.edist(pos, derivative=2)

        # derivative of the back flow kernel : d eta(r_ij)/d r_ij
        # Nbatch x 1 x Nelec x Nelec
        dbf = self.backflow_kernel(ree, derivative=1).unsqueeze(1)

        # second derivative of the back flow kernel : d2 eta(r_ij)/d2 r_ij
        # Nbatch x 1 x Nelec x Nelec
        d2bf = self.backflow_kernel(ree, derivative=2).unsqueeze(1)

        # (d^2 eta(r_ij) / d r_ij^2) (d r_ij/d x_i)^2
        # + (d eta(r_ij) / d r_ij) (d^2 r_ij/d x_i^2)
        # Nbatch x 3 x Nelec x Nelec
        d2bf = (d2bf * dree * dree) + (dbf * d2ree)

        # (d eta(r_ij) / d r_ij) (d r_ij/d x_i)
        # Nbatch x 3 x Nelec x Nelec
        dbf = dbf * dree

        # eye matrix in dim x dim
        eye_mat = torch.eye(3, 3).reshape(1, 3, 3, 1, 1).to(self.device)

        # compute delta_ij delta_ab 2 sum_k dbf(ik) / dbeta_i
        term1 = (
            2
            * eye_mat
            * torch.diag_embed(dbf.sum(-1), dim1=-1, dim2=-2).reshape(
                nbatch, 1, 3, nelec, nelec
            )
        )

        # (d2 eta(r_ij) / d2 beta_i) (alpha_i - alpha_j)
        # Nbatch x 3 x 3 x Nelec x Nelec
        d2bf_delta_ee = d2bf.unsqueeze(1) * delta_ee.unsqueeze(2)

        # compute sum_k d2f(r_ik)/d2beta_i (alpha_i - alpha_k)
        # Nbatch x Ndim x Ndim x Nelec x Nelec
        term2 = torch.diag_embed(d2bf_delta_ee.sum(-1), dim1=-1, dim2=-2)

        # compute delta_ab * df(rij)/dbeta_j
        term3 = 2 * eye_mat * dbf.reshape(nbatch, 1, 3, nelec, nelec)

        # return Nbatch x Ndim(alpha) x Ndim(beta) x Nelec(i) x Nelec(j)
        # nbatch d2 alpha_i / d2 beta_j
        out = term1 + term2 + d2bf_delta_ee + term3

        return out.unsqueeze(-1)

    def fit_kernel(
        self,
        lambda_func: Callable,
        xmin: float = 0.01,
        xmax: float = 1.0,
        npts: int = 100,
        lr: float = 0.001,
        num_epochs: int = 1000,
    ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
        """
        Fit the backflow kernel to a given function.

        Args:
            lambda_func (Callable): function to be fit
            xmin (float): minimum x value
            xmax (float): maximum x value
            npts (int): number of points to sample in the interval [xmin, xmax]
            lr (float): learning rate
            num_epochs (int): number of epochs to run the optimization

        Returns:
            xpts (torch.tensor): x values used for fitting
            ground_truth (torch.tensor): y values of the given function
            fit_values (torch.tensor): y values of the fit function
        """
        xpts = torch.linspace(xmin, xmax, npts)
        ground_truth = lambda_func(xpts)

        criterion = torch.nn.MSELoss()
        optimizer = torch.optim.Adam(self.backflow_kernel.parameters(), lr=lr)

        for epoch in range(num_epochs):
            running_loss = 0.0
            optimizer.zero_grad()
            outputs = self.backflow_kernel(xpts.unsqueeze(1))
            loss = criterion(outputs.squeeze(), ground_truth)
            loss.backward()
            optimizer.step()
            running_loss += loss.item()

            if epoch % 100 == 0:
                print("Epoch {}: Loss = {}".format(epoch, loss.detach().numpy()))

        fit_values = self.backflow_kernel(xpts.unsqueeze(1)).squeeze()
        return xpts, ground_truth, fit_values

    def __repr__(self):
        """representation of the backflow transformation"""
        return self.backflow_kernel.__class__.__name__

1	import torch	1✔
2	from torch import nn	1✔
3	from typing import Dict, Optional, Callable, Tuple	1✔
4	from ....scf import Molecule	1✔
5	from .kernels.backflow_kernel_base import BackFlowKernelBase	1✔
6	from ...jastrows.distance.electron_electron_distance import ElectronElectronDistance	1✔
7
8
9	class BackFlowTransformation(nn.Module):	1✔
10	def __init__(	1✔
11	self,
12	mol: Molecule,
13	backflow_kernel: BackFlowKernelBase,
14	backflow_kernel_kwargs: Optional[Dict] = {},
15	cuda: Optional[bool] = False,
16	):
17	"""Transform the electorn coordinates into backflow coordinates.
18	see : Orbital-dependent backflow wave functions for real-space quantum Monte Carlo
19	https://arxiv.org/abs/1910.07167
20
21	.. math:
22	\\bold{q}_i = \\bold{r}_i + \\sum_{j\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)
23	"""
24	super().__init__()	1✔
25	self.nao = mol.basis.nao	1✔
26	self.nelec = mol.nelec	1✔
27	self.ndim = 3	1✔
28
29	self.backflow_kernel = backflow_kernel(mol, cuda, **backflow_kernel_kwargs)	1✔
30
31	self.edist = ElectronElectronDistance(mol.nelec)	1✔
32
33	self.cuda = cuda	1✔
34	self.device = torch.device("cpu")	1✔
35	if self.cuda:	1!
36	self.device = torch.device("cuda")	×
37
38	def forward(self, pos: torch.Tensor, derivative: Optional[int] = 0) -> torch.Tensor:	1✔
39	if derivative == 0:	1✔
40	return self._get_backflow(pos)	1✔
41
42	elif derivative == 1:	1✔
43	return self._get_backflow_derivative(pos)	1✔
44
45	elif derivative == 2:	1!
46	return self._get_backflow_second_derivative(pos)	1✔
47
48	else:
49	raise ValueError(	×
50	"derivative of the backflow transformation must be 0, 1 or 2"
51	)
52
53	def _get_backflow(self, pos: torch.Tensor) -> torch.Tensor:	1✔
54	"""Computes the backflow transformation
55
56	.. math:
57	\\bold{q}_i = \\bold{r}_i + \\sum_{j\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)
58
59	Args:
60	pos(torch.tensor): original positions Nbatch x[Nelec*Ndim]
61
62	Returns:
63	torch.tensor: transformed positions Nbatch x[Nelec*Ndim]
64	"""
65	# compute the difference
66	# Nbatch x Nelec x Nelec x 3
67	delta_ee = self.edist.get_difference(pos.reshape(-1, self.nelec, self.ndim))	1✔
68
69	# compute the backflow function
70	# Nbatch x Nelec x Nelec
71	bf_kernel = self.backflow_kernel(self.edist(pos))	1✔
72
73	# update pos
74	pos = pos.reshape(-1, self.nelec, self.ndim) + (	1✔
75	bf_kernel.unsqueeze(-1) * delta_ee
76	).sum(2)
77
78	return pos.reshape(-1, self.nelec * self.ndim)	1✔
79
80	def _get_backflow_derivative(self, pos: torch.Tensor) -> torch.Tensor:	1✔
81	r"""Computes the derivative of the backflow transformation
82	wrt the original positions of the electrons
83
84	.. math::
85	\\bold{q}_i = \\bold{r}_i + \\sum_{j\\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)
86
87	.. math::
88	\\frac{d q_i}{d x_k} = \\delta_{ik}(1 + \\sum_{j\\neq i} \\frac{d \\eta(r_ij)}{d x_i}(x_i-x_j) + \\eta(r_ij)) +
89	\\delta_{i\\neq k}(-\\frac{d \\eta(r_ik)}{d x_k}(x_i-x_k) - \\eta(r_ik))
90
91	Args:
92	pos(torch.tensor): orginal positions of the electrons Nbatch x[Nelec*Ndim]
93
94	Returns:
95	torch.tensor: d q_{i}/d x_k with:
96	q_{i} bf position of elec i
97	x_k original coordinate of the kth elec
98	Nelec x Nbatch x Nelec x Norb x Ndim
99	"""
100
101	# ee dist matrix : Nbatch x Nelec x Nelec
102	ree = self.edist(pos)	1✔
103	nbatch, nelec, _ = ree.shape	1✔
104
105	# derivative ee dist matrix : Nbatch x 3 x Nelec x Nelec
106	# dr_ij / dx_i = - dr_ij / dx_j
107	dree = self.edist(pos, derivative=1)	1✔
108
109	# difference between elec pos
110	# Nbatch, 3, Nelec, Nelec
111	delta_ee = self.edist.get_difference(pos.reshape(nbatch, nelec, 3)).permute(	1✔
112	0, 3, 1, 2
113	)
114
115	# backflow kernel : Nbatch x 1 x Nelec x Nelec
116	bf = self.backflow_kernel(ree)	1✔
117
118	# (d eta(r_ij) / d r_ij) (d r_ij/d beta_i)
119	# derivative of the back flow kernel : Nbatch x 3 x Nelec x Nelec
120	dbf = self.backflow_kernel(ree, derivative=1).unsqueeze(1)	1✔
121	dbf = dbf * dree	1✔
122
123	# (d eta(r_ij) / d beta_i) (alpha_i - alpha_j)
124	# Nbatch x 3 x 3 x Nelec x Nelec
125	dbf_delta_ee = dbf.unsqueeze(1) * delta_ee.unsqueeze(2)	1✔
126
127	# compute the delta_ij * (1 + sum k \neq i eta(rik))
128	# Nbatch x Nelec x Nelec (diagonal matrix)
129	delta_ij_bf = torch.diag_embed(1 + bf.sum(-1), dim1=-1, dim2=-2)	1✔
130
131	# eye 3x3 in 1x3x3x1x1
132	eye_mat = torch.eye(3, 3).view(1, 3, 3, 1, 1).to(self.device)	1✔
133
134	# compute the delta_ab * delta_ij * (1 + sum k \neq i eta(rik))
135	# Nbatch x Ndim x Ndim x Nelec x Nelec (diagonal matrix)
136	delta_ab_delta_ij_bf = eye_mat * delta_ij_bf.view(nbatch, 1, 1, nelec, nelec)	1✔
137
138	# compute sum_k df(r_ik)/dbeta_i (alpha_i - alpha_k)
139	# Nbatch x Ndim x Ndim x Nelec x Nelec
140	delta_ij_sum = torch.diag_embed(dbf_delta_ee.sum(-1), dim1=-1, dim2=-2)	1✔
141
142	# compute delta_ab * f(rij)
143	delta_ab_bf = eye_mat * bf.view(nbatch, 1, 1, nelec, nelec)	1✔
144
145	# return Nbatch x Ndim(alpha) x Ndim(beta) x Nelec(i) x Nelec(j)
146	# nbatch d alpha_i / d beta_j
147	out = delta_ab_delta_ij_bf + delta_ij_sum - dbf_delta_ee - delta_ab_bf	1✔
148
149	return out.unsqueeze(-1)	1✔
150
151	def _get_backflow_second_derivative(self, pos: torch.Tensor) -> torch.Tensor:	1✔
152	r"""Computes the second derivative of the backflow transformation
153	wrt the original positions of the electrons
154
155	.. math::
156	\\bold{q}_i = \\bold{r}_i + \\sum_{j\\neq i} \\eta(r_{ij})(\\bold{r}_i - \\bold{r}_j)
157
158	.. math::
159	\\frac{d q_i}{d x_k} = \\delta_{ik}(1 + \\sum_{j\\neqi} \\frac{d \\eta(r_ij)}{d x_i} + \\eta(r_ij)) +
160	\\delta_{i\\neq k}(-\\frac{d \\eta(r_ik)}{d x_k} - \\eta(r_ik))
161
162	.. math::
163	\\frac{d ^ 2 q_i}{d x_k ^ 2} = \\delta_{ik}(\\sum_{j\\neqi} \\frac{d ^ 2 \\eta(r_ij)}{d x_i ^ 2} + 2 \\frac{d \\eta(r_ij)}{d x_i}) +
164	- \\delta_{i\\neq k}(\\frac{d ^ 2 \\eta(r_ik)}{d x_k ^ 2} + \\frac{d \\eta(r_ik)}{d x_k})
165
166	Args:
167	pos(torch.tensor): orginal positions of the electrons Nbatch x[Nelec*Ndim]
168
169	Returns:
170	torch.tensor: d q_{i}/d x_k with:
171	q_{i} bf position of elec i
172	x_k original coordinate of the kth elec
173	Nelec x Nbatch x Nelec x Norb x Ndim
174	"""
175	# ee dist matrix :
176	# Nbatch x Nelec x Nelec
177	ree = self.edist(pos)	1✔
178	nbatch, nelec, _ = ree.shape	1✔
179
180	# difference between elec pos
181	# Nbatch, 3, Nelec, Nelec
182	delta_ee = self.edist.get_difference(pos.reshape(nbatch, nelec, 3)).permute(	1✔
183	0, 3, 1, 2
184	)
185
186	# derivative ee dist matrix d r_{ij} / d x_i
187	# Nbatch x 3 x Nelec x Nelec
188	dree = self.edist(pos, derivative=1)	1✔
189
190	# derivative ee dist matrix : d2 r_{ij} / d2 x_i
191	# Nbatch x 3 x Nelec x Nelec
192	d2ree = self.edist(pos, derivative=2)	1✔
193
194	# derivative of the back flow kernel : d eta(r_ij)/d r_ij
195	# Nbatch x 1 x Nelec x Nelec
196	dbf = self.backflow_kernel(ree, derivative=1).unsqueeze(1)	1✔
197
198	# second derivative of the back flow kernel : d2 eta(r_ij)/d2 r_ij
199	# Nbatch x 1 x Nelec x Nelec
200	d2bf = self.backflow_kernel(ree, derivative=2).unsqueeze(1)	1✔
201
202	# (d^2 eta(r_ij) / d r_ij^2) (d r_ij/d x_i)^2
203	# + (d eta(r_ij) / d r_ij) (d^2 r_ij/d x_i^2)
204	# Nbatch x 3 x Nelec x Nelec
205	d2bf = (d2bf * dree * dree) + (dbf * d2ree)	1✔
206
207	# (d eta(r_ij) / d r_ij) (d r_ij/d x_i)
208	# Nbatch x 3 x Nelec x Nelec
209	dbf = dbf * dree	1✔
210
211	# eye matrix in dim x dim
212	eye_mat = torch.eye(3, 3).reshape(1, 3, 3, 1, 1).to(self.device)	1✔
213
214	# compute delta_ij delta_ab 2 sum_k dbf(ik) / dbeta_i
215	term1 = (	1✔
216	2
217	* eye_mat
218	* torch.diag_embed(dbf.sum(-1), dim1=-1, dim2=-2).reshape(
219	nbatch, 1, 3, nelec, nelec
220	)
221	)
222
223	# (d2 eta(r_ij) / d2 beta_i) (alpha_i - alpha_j)
224	# Nbatch x 3 x 3 x Nelec x Nelec
225	d2bf_delta_ee = d2bf.unsqueeze(1) * delta_ee.unsqueeze(2)	1✔
226
227	# compute sum_k d2f(r_ik)/d2beta_i (alpha_i - alpha_k)
228	# Nbatch x Ndim x Ndim x Nelec x Nelec
229	term2 = torch.diag_embed(d2bf_delta_ee.sum(-1), dim1=-1, dim2=-2)	1✔
230
231	# compute delta_ab * df(rij)/dbeta_j
232	term3 = 2 * eye_mat * dbf.reshape(nbatch, 1, 3, nelec, nelec)	1✔
233
234	# return Nbatch x Ndim(alpha) x Ndim(beta) x Nelec(i) x Nelec(j)
235	# nbatch d2 alpha_i / d2 beta_j
236	out = term1 + term2 + d2bf_delta_ee + term3	1✔
237
238	return out.unsqueeze(-1)	1✔
239
240	def fit_kernel(	1✔
241	self,
242	lambda_func: Callable,
243	xmin: float = 0.01,
244	xmax: float = 1.0,
245	npts: int = 100,
246	lr: float = 0.001,
247	num_epochs: int = 1000,
248	) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor]:
249	"""
250	Fit the backflow kernel to a given function.
251
252	Args:
253	lambda_func (Callable): function to be fit
254	xmin (float): minimum x value
255	xmax (float): maximum x value
256	npts (int): number of points to sample in the interval [xmin, xmax]
257	lr (float): learning rate
258	num_epochs (int): number of epochs to run the optimization
259
260	Returns:
261	xpts (torch.tensor): x values used for fitting
262	ground_truth (torch.tensor): y values of the given function
263	fit_values (torch.tensor): y values of the fit function
264	"""
265	xpts = torch.linspace(xmin, xmax, npts)	×
266	ground_truth = lambda_func(xpts)	×
267
268	criterion = torch.nn.MSELoss()	×
269	optimizer = torch.optim.Adam(self.backflow_kernel.parameters(), lr=lr)	×
270
271	for epoch in range(num_epochs):	×
272	running_loss = 0.0	×
273	optimizer.zero_grad()	×
274	outputs = self.backflow_kernel(xpts.unsqueeze(1))	×
275	loss = criterion(outputs.squeeze(), ground_truth)	×
276	loss.backward()	×
277	optimizer.step()	×
278	running_loss += loss.item()	×
279
280	if epoch % 100 == 0:	×
281	print("Epoch {}: Loss = {}".format(epoch, loss.detach().numpy()))	×
282
283	fit_values = self.backflow_kernel(xpts.unsqueeze(1)).squeeze()	×
NEW 284	return xpts, ground_truth, fit_values	×
285
286	def __repr__(self):	1✔
287	"""representation of the backflow transformation"""
288	return self.backflow_kernel.__class__.__name__	1✔

NLESC-JCER / QMCTorch / 14995306316

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous