.md .pdf

repository open issue

Contents

Postprocessing

1. Apply Spin Reversal Transformation

2. Postprocess (Before Chain Resolution)

   • 2000Q: optimization mode/sampling mode, user-defined

   • Advantage: user-defined

3. Chain Resolution

   • majority_vote/random_weighted/discard/MinimizeEnergy

4. Postprocess samples (After Chain Resolution)

   • user-defined

D-Wave 2000Q

Chain Resolution -> Postprocess (None/Optimization/Sampling) -> Postprocess (User-defined)

sampler = EmbeddingComposite(DWaveSampler())
# or
sampler = FixedEmbeddingComposite(DWaveSampler())

# Embed Problem
# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
# 1. Chain Resolution (majority_vote/random_weighted/discard)
# 2. Postprocess (None/Optimization/Sampling)
sampleset = sampler.sample(
    bqm, num_spin_reversal_transforms, postprocess
)

# 3. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=sampleset
)

Postprocess (None/Optimization/Sampling) -> Chain Resolution -> Postprocess (User-defined)

# Embed Problem
bqm_embed = embed_bqm(bqm)
sampler = DWaveSampler()

# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
# 1. Postprocess (None/Optimization/Sampling)
raw_sampleset = sampler.sample(
    bqm_embed, num_spin_reversal_transforms, postprocess
)

# 2. Chain Resolution (majority_vote/random_weighted/discard/MinimizeEnergy)
sampleset = unembed_sampleset(raw_sampleset)

# 3. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=sampleset
)

Postprocess (None/Optimization/Sampling) -> Postprocess (User-defined) -> Chain Resolution

# Embed Problem
bqm_embed = embed_bqm(bqm)
sampler = DWaveSampler()

# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
# 1. Postprocess (None/Optimization/Sampling)
raw_sampleset = sampler.sample(
    bqm_embed, num_spin_reversal_transforms, postprocess
)

# 2. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

# 3. Chain Resolution (majority_vote/random_weighted/discard/MinimizeEnergy)
sampleset = unembed_sampleset(sampleset)

Postprocess (None/Optimization/Sampling) -> Postprocess (User-defined) -> Chain Resolution -> Postprocess (User-defined)

# Embed Problem
bqm_embed = embed_bqm(bqm)
sampler = DWaveSampler()

# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
# 1. Postprocess (None/Optimization/Sampling)
raw_sampleset = sampler.sample(
    bqm_embed, num_spin_reversal_transforms, postprocess
)

# 2. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

# 3. Chain Resolution (majority_vote/random_weighted/discard/MinimizeEnergy)
sampleset = unembed_sampleset(sampleset)

# 4. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

D-Wave Advantage

Chain Resolution -> Postprocess (User-defined)

sampler = EmbeddingComposite(DWaveSampler())
# or
sampler = FixedEmbeddingComposite(DWaveSampler())

# Embed Problem
# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
# 1. Chain Resolution (majority_vote/random_weighted/discard)
sampleset = sampler.sample(bqm, num_spin_reversal_transforms)

# 2. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=sampleset
)

Postprocess (User-defined) -> Chain Resolution

# Embed Problem
bqm_embed = embed_bqm(bqm)
sampler = DWaveSampler()

# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
raw_sampleset = sampler.sample(bqm_embed, num_spin_reversal_transforms)

# 1. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

# 2. Chain Resolution (majority_vote/random_weighted/discard/MinimizeEnergy)
sampleset = unembed_sampleset(sampleset)

Postprocess (User-defined) -> Chain Resolution -> Postprocess (User-defined)

# Embed Problem
bqm_embed = embed_bqm(bqm)
sampler = DWaveSampler()

# Receive Problem
# QPU Sampling
# Apply Spin Reversal Transformation
raw_sampleset = sampler.sample(bqm_embed, num_spin_reversal_transforms)

# 1. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

# 2. Chain Resolution (majority_vote/random_weighted/discard/MinimizeEnergy)
sampleset = unembed_sampleset(sampleset)

# 3. Postprocess (Optional, User-defined, ex.greedy)
sampleset = SteepestDescentSampler().sample(
    bqm, initial_states=raw_sampleset
)

Reference

• https://docs.dwavesys.com/docs/latest/c_post-processing_1.html

previous

Greedy descent as a post-processing

next

QPU access time