-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathed_core.jl
334 lines (281 loc) · 10.4 KB
/
ed_core.jl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
# KeldyshED.jl
#
# Copyright (C) 2019-2025 Igor Krivenko
# Copyright (C) 2015 P. Seth, I. Krivenko, M. Ferrero and O. Parcollet
#
# KeldyshED.jl is free software: you can redistribute it and/or modify it under the
# terms of the GNU General Public License as published by the Free Software
# Foundation, either version 3 of the License, or (at your option) any later
# version.
#
# KeldyshED.jl is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
# details.
#
# You should have received a copy of the GNU General Public License along with
# KeldyshED.jl. If not, see <http://www.gnu.org/licenses/>.
#
# Author: Igor Krivenko
using Test
using LinearAlgebra: I, Diagonal
using KeldyshED.Operators
using KeldyshED.Hilbert
using KeldyshED
function make_hamiltonian(n_orb, mu, U, J)
soi = SetOfIndices([[s,o] for s in ("up","dn") for o = 1:n_orb])
H = OperatorExpr{Float64}()
for o=1:n_orb; H += -mu * (n("up", o) + n("dn", o)) end
for o=1:n_orb; H += U * n("up", o) * n("dn", o) end
for o1=1:n_orb, o2=1:n_orb
o1 == o2 && continue
H += (U - 2 * J) * n("up", o1) * n("dn", o2)
end
for o1=1:n_orb, o2=1:n_orb
o2 >= o1 && continue
H += (U - 3 * J) * n("up", o1) * n("up", o2)
H += (U - 3 * J) * n("dn", o1) * n("dn", o2)
end
for o1=1:n_orb, o2=1:n_orb
o1 == o2 && continue
H += -J * c_dag("up", o1) * c_dag("dn", o1) * c("up", o2) * c("dn", o2)
H += -J * c_dag("up", o1) * c_dag("dn", o2) * c("up", o2) * c("dn", o1)
end
(soi, H)
end
@testset "EDCore: 3-orbital Hubbard-Kanamori atom" begin
n_orb = 3
U = 1.0
J = 0.2
mu = 0.5*U
beta = 5
soi, H = make_hamiltonian(n_orb, mu, U, J)
ed = EDCore(H, soi)
n_subspaces = 44
@test length(ed.subspaces) == n_subspaces
@test isapprox(ed.gs_energy, -0.6, atol=1e-10)
# Thorough testing
basis_ref = Vector{FockState}[
[3],[5],[6],[10,17],[12,33],[20,34],[24],[40],[48],[1],[2],[4],[8],[16],[32],[7],
[14, 21, 35],[28, 42, 49],[56],[9, 18, 36],[0],[11, 38],[13, 22],[19, 37],
[25, 52],[26, 44],[41, 50],[15],[23],[29, 43],[30, 51],[39],[46, 53],[57],[58],
[60],[27, 45, 54],[31],[47],[55],[59],[61],[62],[63]]
# fock_states()
basis = fock_states(ed)
@test sort(basis) == sort(basis_ref) # Equal up to a permutation of subspaces
# energies()
en_ref = Vector{Float64}[[0],[0],[0],[0,0.4],[0,0.4],[0,0.4],[0],[0],[0],
[0.1],[0.1],[0.1],[0.1],[0.1],[0.1],[0.3],[0.3,0.9,0.9],[0.3,0.9,0.9],[0.3],
[0.4,0.4,1],[0.6],[0.9,1.3],[0.9,1.3],[0.9,1.3],[0.9,1.3],[0.9,1.3],[0.9,1.3],
[2],[2],[2,2.4],[2,2.4],[2],[2,2.4],[2],[2],[2],[2.4,2.4,3],[4.1],[4.1],[4.1],
[4.1],[4.1],[4.1],[6.6]]
en = energies(ed)
for i=1:n_subspaces
i_ref = findfirst(x->x==basis[i], basis_ref)
@test isapprox(en[i], en_ref[i_ref], atol = 1e-10)
end
# unitary_matrices()
Id = hcat(1) # identity matrix 1x1
u_mat_ref = Matrix{Float64}[Id, Id, Id,
[0.7071067812 0.7071067812; -0.7071067812 0.7071067812],
[0.7071067812 0.7071067812; -0.7071067812 0.7071067812],
[0.7071067812 0.7071067812; -0.7071067812 0.7071067812],
Id, Id, Id, Id, Id, Id, Id, Id, Id, Id,
[-0.5773502692 0.7071067812 0.4082482905;
0.5773502692 0.7071067812 -0.4082482905;
-0.5773502692 0 -0.8164965809],
[-0.5773502692 0.7071067812 0.4082482905;
0.5773502692 0.7071067812 -0.4082482905;
-0.5773502692 0 -0.8164965809],
Id,
[-0.1685467429 -0.7989109225 -0.5773502692;
-0.6076037828 0.5454212223 -0.5773502692;
0.7761505257 0.2534897002 -0.5773502692],
Id,
[-0.7071067812 -0.7071067812; -0.7071067812 0.7071067812],
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
[-0.7071067812 -0.7071067812; -0.7071067812 0.7071067812],
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
Id, Id,
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
Id,
[-0.7071067812 0.7071067812; 0.7071067812 0.7071067812],
Id, Id, Id,
[ 0.4082482905 0.7071067812 0.5773502692;
0.4082482905 -0.7071067812 0.5773502692;
-0.8164965809 0 0.5773502692],
Id, Id, Id, Id, Id, Id, Id]
u_mat = unitary_matrices(ed)
for i=1:n_subspaces
i_ref = findfirst(x->x==basis[i], basis_ref)
h = u_mat[i] * Diagonal(en[i]) * (u_mat[i]')
h_ref = u_mat_ref[i_ref] * Diagonal(en_ref[i_ref]) * (u_mat_ref[i_ref]')
@test isapprox(h, h_ref, atol = 1e-8)
end
# cdag_connection()
cdag_conn_ref = [Dict([3=>16,4=>22,5=>23,6=>17,7=>25,8=>27,9=>18,11=>1,
12=>2,13=>20,14=>4,15=>5,17=>28,18=>30,19=>34,20=>24,21=>10,22=>32,23=>29,
25=>33,26=>37,27=>31,31=>38,33=>39,35=>41,36=>42,37=>40,43=>44]),
Dict([2=>16,4=>24,5=>17,6=>23,7=>26,8=>18,9=>27,10=>1,12=>3,13=>4,14=>20,15=>6,
17=>29,18=>31,19=>35,20=>22,21=>11,23=>28,24=>32,25=>37,26=>33,27=>30,30=>38,
33=>40,34=>41,36=>43,37=>39,42=>44]),
Dict([1=>16,4=>17,5=>24,6=>22,7=>18,8=>26,9=>25,10=>2,11=>3,13=>5,14=>6,15=>20,
17=>32,18=>33,19=>36,20=>23,21=>12,22=>28,24=>29,25=>30,26=>31,27=>37,30=>39,
31=>40,34=>42,35=>43,37=>38,41=>44]),
Dict([1=>22,2=>23,3=>17,4=>25,5=>27,6=>18,9=>19,10=>20,11=>4,12=>5,14=>7,15=>8,
16=>28,17=>30,18=>34,20=>26,21=>13,22=>33,23=>31,24=>37,25=>36,27=>35,29=>38,
31=>41,32=>39,33=>42,37=>43,40=>44]),
Dict([1=>24,2=>17,3=>23,4=>26,5=>18,6=>27,8=>19,10=>4,11=>20,12=>6,13=>7,15=>9,
16=>29,17=>31,18=>35,20=>25,21=>14,22=>37,23=>30,24=>33,26=>36,27=>34,28=>38,
30=>41,32=>40,33=>43,37=>42,39=>44]),
Dict([1=>17,2=>24,3=>22,4=>18,5=>26,6=>25,7=>19,10=>5,11=>6,12=>20,13=>8,14=>9,
16=>32,17=>33,18=>36,20=>27,21=>15,22=>30,23=>37,24=>31,25=>34,26=>35,28=>39,
29=>40,30=>42,31=>43,37=>41,38=>44])]
function check_connection(conn_ref, i, j)
# Account for a possible difference in subspace order in basis and basis_ref
i_ref = findfirst(x->x==basis[i], basis_ref)
j_ref = get(conn_ref, i_ref, nothing)
(isnothing(j) && isnothing(j_ref)) || (basis[j] == basis_ref[j_ref])
end
for (indices, n) in soi
for i=1:n_subspaces
@test check_connection(cdag_conn_ref[n], i, cdag_connection(ed, n, i))
@test check_connection(cdag_conn_ref[n], i, cdag_connection(ed, indices, i))
end
cdag_conn_mat = cdag_connection(ed, n)
@test cdag_conn_mat == cdag_connection(ed, indices)
@test cdag_conn_mat ==
[(i == cdag_connection(ed, n, j)) for i=1:n_subspaces, j=1:n_subspaces]
end
# c_connection()
# 'Transpose' cdag_conn_ref
c_conn_ref = map(d -> Dict(j => i for (i, j) in d), cdag_conn_ref)
for (indices, n) in soi
for i=1:n_subspaces
@test check_connection(c_conn_ref[n], i, c_connection(ed, n, i))
@test check_connection(c_conn_ref[n], i, c_connection(ed, indices, i))
end
c_conn_mat = c_connection(ed, n)
@test c_conn_mat == c_connection(ed, indices)
@test c_conn_mat ==
[(i == c_connection(ed, n, j)) for i=1:n_subspaces, j=1:n_subspaces]
end
# monomial_connection()
# Constant monomial
let m = Operators.Monomial()
for i=1:n_subspaces
@test monomial_connection(ed, m, i) == i
end
@test monomial_connection(ed, m) ==
BitArray([i == j for i in 1:n_subspaces, j in 1:n_subspaces])
end
# Quadratic monomial
for (indices1, n1) in soi
for (indices2, n2) in soi
let m = Operators.Monomial([Operators.CanonicalOperator(true, indices1),
Operators.CanonicalOperator(false, indices2)])
for i=1:n_subspaces
conn_ref = cdag_connection(ed, indices1, c_connection(ed, indices2, i))
@test monomial_connection(ed, m, i) == conn_ref
end
conn_mat_ref = cdag_connection(ed, indices1) * c_connection(ed, indices2)
@test monomial_connection(ed, m) == conn_mat_ref
end
end
end
# cdag_matrix() and c_matrix()
# Check that C† * C is the number of particles
for (indices, n) in soi
for i=1:n_subspaces
j = c_connection(ed, n, i)
isnothing(j) && continue
cdag_mat = cdag_matrix(ed, n, j)
c_mat = c_matrix(ed, n, i)
@test cdag_mat == cdag_matrix(ed, indices, j)
@test c_mat == c_matrix(ed, indices, i)
n_mat = cdag_mat * c_mat
n_mat = u_mat[i] * n_mat * (u_mat[i]')
n_mat_ref = Diagonal([digits(fs,base=2,pad=64)[n] for fs in basis[i]])
@test isapprox(n_mat, n_mat_ref, atol = 1e-8)
end
end
# monomial_matrix()
# Constant monomial
let m = Operators.Monomial()
for i=1:n_subspaces
sp_dim = length(ed.subspaces[i])
@test monomial_matrix(ed, m, i) == Matrix{Float64}(I, sp_dim, sp_dim)
end
end
# Quadratic monomial
for (indices1, n1) in soi
for (indices2, n2) in soi
let m = Operators.Monomial([Operators.CanonicalOperator(true, indices1),
Operators.CanonicalOperator(false, indices2)])
for i=1:n_subspaces
j = c_connection(ed, indices2, i)
k = cdag_connection(ed, indices1, j)
if j === nothing || k === nothing
@test_throws DomainError monomial_matrix(ed, m, i)
else
mat_ref = cdag_matrix(ed, indices1, j) * c_matrix(ed, indices2, i)
mat = monomial_matrix(ed, m, i)
@test isapprox(mat, mat_ref, atol = 1e-8)
end
end
end
end
end
# operator_blocks()
# Check the case of one-monomial operators
let op_one_term = RealOperatorExpr()
for o1=1:n_orb, o2=1:n_orb
op_one_term = (o1 + o2) * c_dag("up", o1) * c("dn", o2)
mon = Operators.Monomial([
Operators.CanonicalOperator(true, IndicesType(["up", o1])),
Operators.CanonicalOperator(false, IndicesType(["dn", o2]))
])
for j=1:n_subspaces
blocks = operator_blocks(ed, op_one_term, j)
i = monomial_connection(ed, mon, j)
if i === nothing
@test isempty(blocks)
else
@test length(blocks) == 1
@test blocks[i] ≈ (o1 + o2) * monomial_matrix(ed, mon, j)
end
end
end
end
# Check that both methods of operator_blocks() give equivalent results
let op = ComplexOperatorExpr()
for o1=1:n_orb, o2=1:n_orb
op += 1im * (o1 + o2) * c_dag("up", o1; scalar_type = ComplexF64) *
c("dn", o2; scalar_type = ComplexF64)
end
blocks_ref2 = Dict{Tuple{Int64,Int64},Matrix{ComplexF64}}()
for j=1:n_subspaces
blocks = operator_blocks(ed, op, j)
# Fill blocks_ref2
for (i, mat) in blocks
blocks_ref2[(i, j)] = mat
end
end
@test operator_blocks(ed, op) == blocks_ref2
end
end
@testset "EDCore: 7-orbital Hubbard-Kanamori atom" begin
n_orb = 7
U = 1.0
J = 0.2
mu = 0.5*U
beta = 5
soi, H = make_hamiltonian(n_orb, mu, U, J)
ed = EDCore(H, soi)
# Only basic tests
@test length(ed.subspaces) == 2368
@test isapprox(ed.gs_energy, -0.6, atol=1e-10)
end