File size: 20,724 Bytes
f106663
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6abeb6
f106663
 
 
d6abeb6
 
 
 
 
 
 
f106663
 
 
d6abeb6
 
 
 
 
 
 
 
 
 
 
f106663
 
 
 
 
 
 
 
 
 
 
d6abeb6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f106663
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6abeb6
f106663
 
 
 
d6abeb6
 
 
 
 
 
 
 
f106663
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6abeb6
 
 
 
 
f106663
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e2e1a8c
 
f106663
e2e1a8c
 
f106663
e2e1a8c
f106663
 
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
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
import numpy as np
import math
from qiskit.circuit import QuantumCircuit, QuantumRegister
from qiskit.circuit.library import StatePreparation, QFTGate, RZGate
from qiskit.quantum_info import Statevector
import pyvista as pv

def create_impulse_state(grid_dims, impulse_pos):
    """
    Creates an initial state vector with a single delta impulse at a specified grid position.

    The 2D grid is flattened into a 1D vector in row-major order, and this
    vector is then padded to match the full simulation state space size (4x).

    Args:
        grid_dims (tuple): A tuple (width, height) defining the simulation grid dimensions.
                           For your original code, this would be (nx, nx).
        impulse_pos (tuple): A tuple (x, y) for the position of the impulse.
                             Coordinates are 0-indexed.

    Returns:
        numpy.ndarray: The full, padded initial state vector with a single 1.
    
    Raises:
        ValueError: If the impulse position is outside the grid dimensions.
    """
    grid_width, grid_height = grid_dims
    impulse_x, impulse_y = impulse_pos

    # --- Input Validation ---
    # Ensure the requested impulse position is actually on the grid.
    if not (0 <= impulse_x < grid_width and 0 <= impulse_y < grid_height):
        raise ValueError(f"Impulse position ({impulse_x}, {impulse_y}) is outside the "
                         f"grid dimensions ({grid_width}x{grid_height}).")

    # --- 1. Calculate the 1D Array Index ---
    # Convert the (x, y) coordinate to a single index in a flattened 1D array.
    # The formula for row-major order is: index = y_coord * width + x_coord
    flat_index = impulse_y * grid_width + impulse_x

    # --- 2. Create the Full, Padded State Vector ---
    grid_size = grid_width * grid_height
    total_size = 4 * grid_size  # The simulation space is 4x the grid size.
    initial_state = np.zeros(total_size)

    # --- 3. Set the Delta Impulse ---
    initial_state[flat_index] = 1

    return initial_state

def create_gaussian_state(grid_dims, mu, sigma):
    """
    Creates an initial state vector with a 2D Gaussian distribution.

    The state is normalized and padded to match the full simulation state space size (4x).

    Args:
        grid_dims (tuple): A tuple (width, height) defining the grid dimensions.
        mu (tuple): A tuple (mu_x, mu_y) for the center (mean) of the Gaussian.
        sigma (tuple): A tuple (sigma_x, sigma_y) for the standard deviation (spread).

    Returns:
        numpy.ndarray: The full, padded initial state vector for the Gaussian state.

    Raises:
        ValueError: If sigma values are not positive.
    """
    grid_width, grid_height = grid_dims
    mu_x, mu_y = mu
    sigma_x, sigma_y = sigma

    if sigma_x <= 0 or sigma_y <= 0:
        raise ValueError("Sigma values (spread) must be positive.")

    # --- 1. Create a Coordinate Grid ---
    x = np.arange(0, grid_width)
    y = np.arange(0, grid_height)
    X, Y = np.meshgrid(x, y)

    # --- 2. Calculate the 2D Gaussian Function ---
    gaussian_2d = np.exp(-((X - mu_x)**2 / (2 * sigma_x**2)) - 
                         ((Y - mu_y)**2 / (2 * sigma_y**2)))

    # --- 3. Normalize the State Vector ---
    # For a valid quantum state, the L2 norm (sum of squares of amplitudes) must be 1.
    norm = np.linalg.norm(gaussian_2d)
    if norm > 0:
        gaussian_2d = gaussian_2d / norm
    
    # --- 4. Flatten and Pad the Vector ---
    gaussian_flat = gaussian_2d.flatten()
    grid_size = grid_width * grid_height
    total_size = 4 * grid_size
    initial_state = np.pad(gaussian_flat, (0, total_size - grid_size), mode='constant')

    return initial_state





# --- New: Continuous-position helpers for excitation before meshing ---
def _normalize_to_unit(vec: np.ndarray) -> np.ndarray:
    n = np.linalg.norm(vec)
    return vec / n if n > 0 else vec




def create_impulse_state_from_pos(grid_dims, pos01, snap_to_grid=True):
    """
    Create a delta-like initial state from continuous position pos01=(x,y) in [0,1].

    Args:
        grid_dims: Tuple (nx, ny) defining the simulation grid dimensions.
        pos01: Tuple (x, y) with continuous position in [0, 1] range.
        snap_to_grid: If True (default), snaps the impulse to the nearest grid node
                      for an exact delta with peak value = 1.0. If False, uses
                      bilinear interpolation which distributes amplitude to 4 nodes.

    Why grid_dims?
    - Simulation runs on a discrete nx×ny lattice; the continuous position must be
      discretized onto that grid to produce the state vector fed into the solver.
    - grid_dims provides (nx, ny) so we can map (x,y)∈[0,1]→grid coordinates.

    When snap_to_grid=True (default):
    - The impulse is placed exactly on the nearest grid node.
    - Peak amplitude = 1.0 (exact delta function on the discrete grid).
    - This is recommended for visualization and accurate peak value display.

    When snap_to_grid=False:
    - Uses bilinear interpolation to distribute amplitude to the 4 neighboring nodes.
    - Peak amplitude depends on position (e.g., 0.5 when exactly between 4 nodes).
    - Total energy is preserved (L2 norm = 1).

    The preview uses create_impulse_preview_state(), which renders a smooth bump on a
    fixed unit-square grid independent of nx for visualization.
    """
    grid_width, grid_height = grid_dims
    px, py = pos01
    px = float(max(0.0, min(1.0, px)))
    py = float(max(0.0, min(1.0, py)))

    gx = px * (grid_width - 1)
    gy = py * (grid_height - 1)

    grid_size = grid_width * grid_height
    total_size = 4 * grid_size

    if snap_to_grid:
        # Snap to nearest grid node for exact delta with peak = 1.0
        i0 = int(round(gx))
        j0 = int(round(gy))
        i0 = max(0, min(i0, grid_width - 1))
        j0 = max(0, min(j0, grid_height - 1))

        field = np.zeros(grid_size)
        field[j0 * grid_width + i0] = 1.0  # Peak value = 1.0

        initial_state = np.zeros(total_size)
        initial_state[:grid_size] = field
        return initial_state

    # Bilinear interpolation mode (snap_to_grid=False)
    i0, j0 = int(np.floor(gx)), int(np.floor(gy))
    i1, j1 = min(i0 + 1, grid_width - 1), min(j0 + 1, grid_height - 1)
    dx, dy = gx - i0, gy - j0

    w00 = (1 - dx) * (1 - dy)
    w10 = dx * (1 - dy)
    w01 = (1 - dx) * dy
    w11 = dx * dy

    field = np.zeros(grid_size)
    field[j0 * grid_width + i0] += w00
    field[j0 * grid_width + i1] += w10
    field[j1 * grid_width + i0] += w01
    field[j1 * grid_width + i1] += w11
    field = _normalize_to_unit(field)

    initial_state = np.zeros(total_size)
    initial_state[:grid_size] = field
    return initial_state


def create_gaussian_state_from_pos(grid_dims, mu01, sigma01, snap_to_grid=True):
    """
    Create a Gaussian initial state with center mu01=(x,y) and spreads sigma01=(sx,sy)
    in [0,1] of the domain, then discretize to the solver grid given by grid_dims.

    Args:
        grid_dims: Tuple (nx, ny) defining the simulation grid dimensions.
        mu01: Tuple (x, y) with Gaussian center in [0, 1].
        sigma01: Tuple (sx, sy) with Gaussian std dev as fraction of domain in [0, 1].
        snap_to_grid: If True (default), snap the Gaussian center to the nearest grid
            node before discretization. This makes the simulator behavior consistent
            with the "nearest-node" semantics often used elsewhere in the app.

    Why grid_dims?
    - The quantum solver expects a vector aligned to the chosen nx×ny simulation grid.
      We convert normalized μ and σ (fractions of the domain) into grid units using
      (nx-1) and (ny-1). This step is necessary for the simulation, not for the preview.

    For preview-only rendering, use create_impulse_preview_state() to keep the visuals
    continuous and independent of nx.
    """
    grid_width, grid_height = grid_dims
    mu_x01, mu_y01 = mu01
    sig_x01, sig_y01 = sigma01

    mu_x01 = float(max(0.0, min(1.0, mu_x01)))
    mu_y01 = float(max(0.0, min(1.0, mu_y01)))
    sig_x01 = float(sig_x01)
    sig_y01 = float(sig_y01)
    if sig_x01 <= 0 or sig_y01 <= 0:
        raise ValueError("Sigma values (spread) must be positive.")

    mu_x = mu_x01 * (grid_width - 1)
    mu_y = mu_y01 * (grid_height - 1)
    if snap_to_grid:
        mu_x = float(int(round(mu_x)))
        mu_y = float(int(round(mu_y)))
        mu_x = float(max(0, min(int(mu_x), grid_width - 1)))
        mu_y = float(max(0, min(int(mu_y), grid_height - 1)))
    sigma_x = sig_x01 * (grid_width - 1)
    sigma_y = sig_y01 * (grid_height - 1)

    x = np.arange(0, grid_width)
    y = np.arange(0, grid_height)
    X, Y = np.meshgrid(x, y)
    gaussian_2d = np.exp(-((X - mu_x) ** 2) / (2 * sigma_x ** 2) - ((Y - mu_y) ** 2) / (2 * sigma_y ** 2))

    field = _normalize_to_unit(gaussian_2d.ravel())
    grid_size = grid_width * grid_height
    total_size = 4 * grid_size
    initial_state = np.zeros(total_size)
    initial_state[:grid_size] = field
    return initial_state

# --- Simulation Code (from previous context) ---
def Wj_block(j, n, ctrl_state, theta, lam, name='Wj_block', xgate=False):
    qc = QuantumCircuit(n + j, name=name)
    if j > 1: qc.cx(n + j - 1, range(n, n + j - 1))
    if lam != 0: qc.p(lam, n + j - 1)
    qc.h(n + j - 1)
    if xgate and j > 1:
        if isinstance(xgate, (list, tuple)):
            for idx, flag in enumerate(xgate):
                if flag: qc.x(n + idx)
        elif xgate is True: qc.x(range(n, n + j - 1))
    if j > 1:
        mcrz = RZGate(theta).control(len(ctrl_state) + j - 1, ctrl_state="1" * (j - 1) + ctrl_state)
        qc.append(mcrz, range(0, n + j))
    else:
        mcrz = RZGate(theta).control(len(ctrl_state), ctrl_state=ctrl_state)
        qc.append(mcrz, range(0, n + j))
    if xgate and j > 1:
        if isinstance(xgate, (list, tuple)):
            for idx, flag in enumerate(xgate):
                if flag: qc.x(n + idx)
        elif xgate is True: qc.x(range(n, n + j - 1))
    qc.h(n + j - 1)
    if lam != 0: qc.p(-lam, n + j - 1)
    if j > 1: qc.cx(n + j - 1, range(n, n + j - 1))
    return qc.to_gate(label=name)

def V1(nx, dt):
    n = int(np.ceil(np.log2(nx)))
    derivatives, blocks = QuantumRegister(2 * n), QuantumRegister(2)
    qc = QuantumCircuit(derivatives, blocks)
    qc.append(Wj_block(2, n, "0" * n, -dt, 0, xgate=True), list(derivatives[0:n]) + list(blocks[:]))
    qc.append(Wj_block(3, n - 1, "1" * (n - 1), dt, 0, xgate=[0, 1]), list(derivatives[1:n]) + [derivatives[0]] + list(blocks[:]))
    qc.append(Wj_block(1, n + 1, "0" * (n + 1), dt, 0, xgate=True), list(derivatives[n:2 * n]) + list(blocks[:]))
    qc.append(Wj_block(2, n, "0" + "1" * (n - 1), -dt, 0, xgate=False), list(derivatives[n + 1:2 * n]) + [blocks[0]] + [derivatives[n]] + [blocks[1]])
    return qc

def V2(nx, dt):
    n = int(np.ceil(np.log2(nx)))
    derivatives, blocks = QuantumRegister(2 * n), QuantumRegister(2)
    qc = QuantumCircuit(derivatives, blocks)
    qc.append(Wj_block(2, 0, "", -2 * dt, -np.pi / 2, xgate=True), blocks[:])
    for j in range(1, n + 1): qc.append(Wj_block(2 + j, 0, "", 2 * dt, -np.pi / 2, xgate=[1] * (j - 1) + [0, 1]), list(derivatives[0:j]) + list(blocks[:]))
    qc.append(Wj_block(2, n, "0" * n, -dt, -np.pi / 2, xgate=True), list(derivatives[0:n]) + list(blocks[:]))
    qc.append(Wj_block(2, n, "1" * n, 2 * dt, -np.pi / 2, xgate=True), list(derivatives[0:n]) + list(blocks[:]))
    qc.append(Wj_block(3, n - 1, "1" * (n - 1), dt, -np.pi / 2, xgate=[0, 1]), list(derivatives[1:n]) + [derivatives[0]] + list(blocks[:]))
    qc.append(Wj_block(1, 1, "0", 2 * dt, -np.pi / 2, xgate=False), blocks[:])
    for j in range(1, n + 1): qc.append(Wj_block(1 + j, 1, "0", -2 * dt, -np.pi / 2, xgate=[1] * (j - 1)), [blocks[0]] + list(derivatives[n:n + j]) + [blocks[1]])
    qc.append(Wj_block(1, n + 1, "0" * (n + 1), dt, -np.pi / 2, xgate=False), list(derivatives[n:2 * n]) + list(blocks[:]))
    qc.append(Wj_block(1, n + 1, "0" + "1" * n, -2 * dt, -np.pi / 2, xgate=False), list(derivatives[n:2 * n]) + list(blocks[:]))
    qc.append(Wj_block(2, n, "0" + "1" * (n - 1), -dt, -np.pi / 2, xgate=False), list(derivatives[n + 1:2 * n]) + [blocks[0]] + [derivatives[n]] + [blocks[1]])
    return qc

def run_sim(nx, na, R, initial_state, T, snapshot_dt=None, stop_check=None, progress_callback=None, print_callback=None):
    """
    Runs the quantum simulation for electromagnetic scattering with fixed dt=0.1.
    Captures frames only at user-defined snapshot times: [0, Δt, 2Δt, ..., ≤ T_eff],
    always including t=0 and the final solver-aligned T (T_eff = floor(T/dt)*dt).

    Returns:
        frames (np.ndarray), snapshot_times (np.ndarray)
    """
    def _log(msg):
        if print_callback:
            print_callback(msg)
        else:
            print(msg)

    dt = 0.1
    # Validate total time and compute solver-aligned end time
    try:
        T_val = float(T)
    except Exception:
        return np.array([]), np.array([])
    if T_val <= 0:
        return np.array([]), np.array([])

    steps = int(np.floor(T_val / dt))
    if steps <= 0:
        return np.array([]), np.array([])
    T_eff = steps * dt

    # Determine snapshot Δt on solver grid
    tol = 1e-12
    if snapshot_dt is None:
        snapshot_dt_val = dt
    else:
        try:
            snapshot_dt_val = float(snapshot_dt)
        except Exception:
            snapshot_dt_val = dt
    if snapshot_dt_val < dt - tol:
        snapshot_dt_val = dt
    k = max(1, int(round(snapshot_dt_val / dt)))
    snapshot_dt_eff = k * dt

    # Build requested snapshot times on solver grid
    target_times = [0.0]
    t = 0.0
    while t + snapshot_dt_eff <= T_eff + tol:
        t = round(t + snapshot_dt_eff, 12)
        if t <= T_eff + tol:
            target_times.append(min(t, T_eff))
    if abs(target_times[-1] - T_eff) > tol:
        target_times.append(T_eff)

    # Setup circuit
    nq = int(np.ceil(np.log2(nx)))
    dp = 2 * R * np.pi / 2 ** na
    p = np.arange(-R * np.pi, R * np.pi, step=dp)
    fp = np.exp(-np.abs(p))
    system, ancilla = QuantumRegister(2 * nq + 2), QuantumRegister(na)
    qc = QuantumCircuit(system, ancilla)
    qc.append(StatePreparation(initial_state), system)
    qc.append(StatePreparation(fp / np.linalg.norm(fp)), ancilla)
    expA1 = V1(nx, dt).to_gate()
    expA2 = V2(nx, dt)

    frames = []
    # Capture initial frame at t=0
    sv0 = np.real(Statevector(qc)).reshape(2 ** na, 2 ** (2 * nq + 2))
    frames.append(sv0[2 ** (na - 1)])
    next_idx = 1  # next target_times index to capture

    _log(f"Starting simulation: T={T_eff:.2f}s, steps={steps}, snapshot_dt={snapshot_dt_eff:.2f}s")

    for i in range(steps):
        if stop_check and stop_check():
            _log(f"Simulation interrupted at step {i}/{steps}")
            break
        # One solver step
        qc.append(QFTGate(na), ancilla)
        qc.x(ancilla[-1])
        for j in range(na - 1):
            qc.append(expA1.control().repeat(2 ** j), [ancilla[j]] + system[:])
        qc.append(expA1.inverse().control(ctrl_state="0").repeat(2 ** (na - 1)), [ancilla[na - 1]] + system[:])
        qc.append(expA2, system[:])
        qc.x(ancilla[-1])
        qc.append(QFTGate(na).inverse(), ancilla)

        current_time = (i + 1) * dt
        if next_idx < len(target_times) and abs(current_time - target_times[next_idx]) <= tol:
            u = np.real(Statevector(qc)).reshape(2 ** na, 2 ** (2 * nq + 2))
            frames.append(u[2 ** (na - 1)])
            next_idx += 1

        if progress_callback:
            try:
                progress = ((i + 1) / steps) * 100
                progress_callback(progress)
            except Exception:
                pass

    if progress_callback:
        try:
            progress_callback(100.0)
        except Exception:
            pass
    
    _log("Simulation completed.")

    # Ensure snapshot_times align with number of captured frames (covers early stop)
    frames_arr = np.asarray(frames)
    times_arr = np.asarray(target_times[: len(frames_arr)])
    return frames_arr, times_arr

def create_impulse_preview_state(preview_n: int, pos01, sigma01: float = 0.02):
    """
    Smooth delta-like preview on a unit square using a narrow Gaussian (sigma in [0,1]).
    Preview-only helper, independent of simulation grid size (nx). Use this for the
    Excitation preview; use the *_from_pos() variants for the actual simulation.
    """
    try:
        sx = float(sigma01) if sigma01 and sigma01 > 0 else 0.02
    except Exception:
        sx = 0.02
    return create_gaussian_state_from_pos((int(preview_n), int(preview_n)), (float(pos01[0]), float(pos01[1])), (sx, sx))






##### Statevector Estimator Simulation Code Below #####

from .base_functions import * 

def create_time_frames(total_time, snapshot_interval):
    dt = 0.1
    tol = 1e-9
    try:
        T_val = float(total_time)
    except (ValueError, TypeError):
        return []
    if T_val <= 0:
        return []
    steps = int(np.floor(T_val / dt))
    if steps <= 0:
        return [0.0]
    T_eff = steps * dt
    try:
        snapshot_dt_val = float(snapshot_interval)
    except (ValueError, TypeError):
        snapshot_dt_val = dt
    if snapshot_dt_val < dt:
        snapshot_dt_val = dt
    k = max(1, int(round(snapshot_dt_val / dt)))
    snapshot_dt_eff = k * dt
    times = np.arange(0, T_eff + tol, snapshot_dt_eff)
    if abs(times[-1] - T_eff) > tol:
        times = np.append(times, T_eff)
    times = np.round(times, 12)
    unique_times = []
    for t in times:
        if not unique_times or abs(t - unique_times[-1]) > tol:
            unique_times.append(float(t))
    return unique_times



def run_sve(field, x, y, T, snapshot_time, nx, initial_state, impulse_pos, progress_callback=None, print_callback=None):
    """Statevector Estimator for time-series field values.

    Supports both single-point and multi-point modes.

    - Single-point (backward compatible): x, y are integers; returns list[float].
    - Multi-point: x is a list/tuple of (ix, iy) integer pairs and y is None; returns dict[(ix,iy) -> list[float]].
    """
    def _log(msg):
        if print_callback:
            print_callback(msg)
        else:
            print(msg)

    na = 1
    dt = 0.1
    R = 4
    nq = int(np.ceil(np.log2(nx)))

    # Normalize monitor points input
    if isinstance(x, (list, tuple)) and y is None:
        points = [tuple(map(int, pt)) for pt in x]
        multi = True
    else:
        points = [(int(x), int(y))]
        multi = False

    xref, yref = impulse_pos

    offset = 0
    grid_dims = (nx, nx)
    initial_state = create_impulse_state(grid_dims, impulse_pos)
    
    dp = 2 * R * np.pi / 2**na
    p = np.arange(- R * np.pi, R * np.pi, step=dp)
    fp = np.exp(-np.abs(p))
    norm = np.linalg.norm(fp)

    time_frames = create_time_frames(T, snapshot_time)
    total_frames = len(time_frames)

    _log(f"Starting QPU simulation: T={T}s, frames={total_frames}, points={len(points)}")

    # Prepare outputs
    if multi:
        series_by_point = { (px, py): [] for (px, py) in points }
    else:
        series_single = []

    for idx, time in enumerate(time_frames):
        steps = int(math.ceil(time / dt))
        # Reference Ez field at impulse location for sign
        Eref = Eref_value(nx, nq, R, dt, na, steps, xref, yref, field_ref='Ez')

        for (px, py) in points:
            circ_magnitude = circ_for_magnitude(field, px, py, nx, na, R, dt, initial_state, steps)
            magnitude = get_absolute_field_value(circ_magnitude, nq, na, offset, norm)

            if field == 'Ez' and px == xref and py == yref:
                Field_value = -magnitude if Eref < 0 else magnitude
            else:
                circsum, circdiff = circuits_for_sign(field, px, py, nx, na, dt, R, initial_state, steps, xref, yref, field_ref='Ez')
                sign = get_relative_sign(circsum, circdiff, nq, na)
                if (sign == 'same' and Eref > 0) or (sign == 'different' and Eref < 0):
                    Field_value = magnitude
                else:
                    Field_value = -magnitude

            if multi:
                series_by_point[(px, py)].append(Field_value)
            else:
                series_single.append(Field_value)
        
        # Calculate and report progress
        pct = (idx + 1) / total_frames * 100
        if progress_callback:
            progress_callback(pct)
        _log(f"SVE Progress: {int(pct)}% (frame {idx + 1}/{total_frames})")
    
    _log("Statevector Estimator simulation completed.")

    return series_by_point if multi else series_single