exponential decay of correlations implies area law: single-shot techniques
DESCRIPTION
Fernando G.S.L. Brand ão ETH Zürich Based on joint work with Michał Horodecki Arxiv:1206.2947 Beyond i.i.d . in Info. Theory, Cambridge 08/01/2012. Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques. This talk:. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/1.jpg)
Exponential Decay of Correlations Implies Area Law:
Single-Shot Techniques
Fernando G.S.L. BrandãoETH Zürich
Based on joint work with Michał HorodeckiArxiv:1206.2947
Beyond i.i.d. in Info. Theory, Cambridge 08/01/2012
![Page 2: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/2.jpg)
This talk:• Go beyond the i.i.d. simplification used in the
previous talk. • Contrast two views of single-shot info theory: 1. It’s the right info theory, i.i.d.ness is an abstraction 2. It only gives a reformulation of the problem e.g. Hmax gives the compression rate or is it defined as the compression rate?
![Page 3: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/3.jpg)
This talk:• Go beyond the i.i.d. simplification used in the
previous talk. • Contrast two views of single-shot info theory: 1. It’s the right info theory, iid-ness is an abstraction 2. It only gives a reformulation of the problems
(e.g. Hmax gives the compression rate or is it defined as the compression rate?)
![Page 4: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/4.jpg)
This talk:• Go beyond the i.i.d. simplification used in the
previous talk. • Contrast two views of single-shot info theory: 1. It’s the right info theory, iid-ness is an abstraction 2. It only gives a trivial reformulation of the problems
(e.g. Hmax gives the compression rate or is it defined as the compression rate?)
![Page 5: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/5.jpg)
The result
Thm 1 (B., Horodecki ‘12) If has ξ-EDC, then for every X = [1, r],
X Y
ξ-EDC (Expontial Decay of Correlations): For all regions A and C separated by l sites,
![Page 6: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/6.jpg)
The proof sketch……was based on two ingredients:
1. State Merging Protocol (Horodecki, Oppenheim, Winter ‘05)
2. “Saturation of Mutual Information” (based on Hastings ‘07)
But we only have one copy of the state…Single-shot info theory to rescue?
![Page 7: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/7.jpg)
1. Conditional min-entropy (Renner ‘05)
2. Smooth conditional min entropy
3. Smooth cond. max entropy (Tomamichel, Colbeck, Renner ‘09)
4. Smooth max mutual info
Single-Shot Entropies
![Page 8: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/8.jpg)
1. Conditional min-entropy (Renner ‘05)
2. Smooth conditional min entropy
3. Smooth cond. max entropy (Tomamichel, Colbeck, Renner ‘09)
4. Smooth max mutual info
Single-Shot Entropies
![Page 9: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/9.jpg)
1. Conditional min-entropy (Renner ‘05)
2. Smooth conditional min entropy
3. Smooth cond. max entropy (Tomamichel, Colbeck, Renner ‘09)
4. Smooth max mutual info
Single-Shot Entropies
![Page 10: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/10.jpg)
1. Conditional min-entropy (Renner ‘05)
2. Smooth conditional min entropy
3. Smooth cond. max entropy (Tomamichel, Colbeck, Renner ‘09)
4. Smooth max mutual info
Single-Shot Entropies
![Page 11: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/11.jpg)
Single-Shot Counterparts 1: Correlations vs Entropies
From i.i.d. merging: For
Lemma: For and δ, ν > 0,
1.
2.
![Page 12: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/12.jpg)
Single-Shot Counterparts 1: Correlations vs Entropies
From i.i.d. merging: For
Lemma: For and δ, ν > 0,
1.
2.
![Page 13: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/13.jpg)
Single-Shot Counterparts 1: Correlations vs Entropies
From i.i.d. merging: For
Lemma: For and δ, ν > 0,
1.
2.
Independent bounds since
![Page 14: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/14.jpg)
Single-Shot Merging
Follows from single-shot state merging: (Dupuis, Berta, Wullschleger, Renner ‘10 )
A B C
“Environment”
EPR rate:
C. Com. Cost:
![Page 15: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/15.jpg)
Beyond Single-Shot Merging
Does not follows from single shot-merging. Might have
We operate in a regime where one cannot get EPR pairs. Still one can get correlations by random measurements
It only happens in the single-shot regime, asymptoticallyeither one gets EPR pairs between A and C or decouples A from C
but !!
![Page 16: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/16.jpg)
Noisy Correlations by Random Measurements
Perfect correlationsEPR distillation
no correlations(decoupling)
iid
Perfect correlationsEPR distillation
no correlations(decoupling)
Single-shot
Noisy correlations
![Page 17: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/17.jpg)
Single-Shot Counterparts 2: Saturation Mutual Info
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
Given a site s, for all l0, ε > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < 2O(1/ε) at a distance < l02O(1/ε) from s s.t.
I(BC,l:BL,l/2BR,l/2) < εl
![Page 18: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/18.jpg)
Single-Shot Counterparts 2: Saturation Mutual Info
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
Single-shot Let Ψ have ξ-EDC. Given a site s, for all l0, ε, δ > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < exp(log(1/ε)/ε) at a distance < l0exp(log(1/ε)/ε) from s s.t.
Iδmax(BC,l:BL,l/2BR,l/2) < 8εl/δ + δ
Given a site s, for all l0, ε > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < 2O(1/ε) at a distance < l02O(1/ε) from s s.t.
I(BC,l:BL,l/2BR,l/2) < εl
![Page 19: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/19.jpg)
Saturation von Neumann Mutual Information
Given a site s, for all l0, ε > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < 2O(1/ε) at a distance < l02O(1/ε) from s s.t.
I(BC,l:BL,l/2BR,l/2) < εl
Proof by contradiction: If it doesn’t hold, for all l:
![Page 20: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/20.jpg)
Saturation von Neumann Mutual Information
Given a site s, for all l0, ε > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < 2O(1/ε) at a distance < l02O(1/ε) from s s.t.
I(BC,l:BL,l/2BR,l/2) < εl
Proof by contradiction: If it doesn’t hold, for all l:
Interating:
H(2l) < 2H(l) – εl < 4 H(l/2) – 2εl < 8 H(l/4) – 3εl < …
… which implies H(l) < l H(2) - εl log(l)
![Page 21: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/21.jpg)
Saturation max Mutual InformationSingle-shot Let Ψ have ξ-EDC. Given a site s, for all l0, ε, δ > 0 there is a region B2l := BL,l/2BC,lBR,l/2 of size 2l with 1 < l/l0 < exp(log(1/ε)/ε) at a distance < l0exp(log(1/ε)/ε) from s s.t.
Iδmax(BC,l:BL,l/2BR,l/2) < 8εl/δ + δ
We do not know how to prove it without exponential decay of correlations.
Proof breaks down completely:
1. Imax not linear combination of entropies
2. Maybe use bound: Imax(A:B) < Hmax(A)+Hmax(B)-Hmin(AB) But arbitrary gaps between Hmax and Hmin
![Page 22: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/22.jpg)
Saturation max Mutual InformationThe idea: Use EDC to relate single-shot and i.i.d.
1. From the q. substate theorem (Jain, Radhakrishnan, Sen ’07)
Suffices to prove:
![Page 23: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/23.jpg)
Saturation max Mutual InformationThe idea: Use EDC to relate single-shot and i.i.d.
1. From the q. substate theorem (Jain, Radhakrishnan, Sen ’07)
Suffices to prove:
![Page 24: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/24.jpg)
Saturation max Mutual InformationThe idea: Use EDC to relate single-shot and i.i.d.
1. From the q. substate theorem (Jain, Radhakrishnan, Sen ’07)
Suffices to prove:
2. Suppose not: For all l,
We must relate to . In general not possible
But we have EDC…
![Page 25: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/25.jpg)
Relating max to vNeumann entropy
But by EDC, is small.
Thus
By q. equipartition property (Tomamichel, Colbeck, Renner ’08)
By subadditivity:
![Page 26: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/26.jpg)
Using Single-Shot Lemmas to show the main result
1. Area law from subvolume law
2. Getting subvolume law
![Page 27: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/27.jpg)
Area Law from Subvolume Law (cheating)
l
A B C
![Page 28: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/28.jpg)
Area Law from Subvolume Law (cheating)
l
A B C
![Page 29: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/29.jpg)
Area Law from Subvolume Law (cheating)
l
A B C
Suppose H(B) < l/(4ξ). Since I(A:B) < 2H(B) < l/(2ξ), if state has ξ-EDC then Cor(A:C) < 2-l/ξ < 2-I(A:B)
Thus: H(C) < H(B) : Area Law for C!
![Page 30: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/30.jpg)
Area Law from Subvolume Law (cheating)
l
A B C
Suppose H(B) < l/(4ξ). Since I(A:B) < 2H(B) < l/(2ξ), if state has ξ-EDC then Cor(A:C) < 2-l/ξ < 2-I(A:B)
Thus: H(C) < H(B) : Area Law for C!
It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)
![Page 31: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/31.jpg)
Area Law from Subvolume Law (cheating)
l
A B C
Suppose H(B) < l/(4ξ). Since I(A:B) < 2H(B) < l/(2ξ), if state has ξ-EDC then Cor(A:C) < 2-l/ξ < 2-I(A:B)
Thus: H(C) < H(B) : Area Law for C!
It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)
Could consider single-shot merging. But remember
![Page 32: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/32.jpg)
Area Law from Subvolume Lawl
A B C
![Page 33: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/33.jpg)
Area Law from Subvolume Lawl
A B C
![Page 34: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/34.jpg)
Area Law from Subvolume Law l
A B C
Suppose Hmax(B) < l/(4ξ). If state has ξ-EDC then Cor(A:C) < 2-l/ξ < 2-3Hmax(B)
Thus: Hmax(C) < Hmax(B) : Area Law for C!
![Page 35: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/35.jpg)
Area Law from Subvolume Lawl
A B C
Suppose Hmax(B) < l/(4ξ). If state has ξ-EDC then Cor(A:C) < 2-l/ξ < 2-3Hmax(B)
Thus: Hmax(C) < Hmax(B) : Area Law for C!
It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with max entropy < l/(4ξ)
![Page 36: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/36.jpg)
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
< l02O(1/ε) < l02O(1/ε)
Getting Subvolume Law (cheating)
A sBL BC BR
R := all except BLBCBR :
Hastings lemma with ε = 1/(4ξ), gives I(BC:BLBR) < l/(4ξ)Then from state merging protocol and ξ-EDC:
![Page 37: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/37.jpg)
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
R := all except BLBCBR :
Finally: S(BC) ≤ S(BC) + S(BLBR) – S(R) = I(BC:BLBR) ≤ l/(4ξ)
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
Getting Subvolume Law (cheating)
Hastings lemma with ε = 1/(4ξ), gives I(BC:BLBR) < l/(4ξ)Then from state merging protocol and ξ-EDC:
![Page 38: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/38.jpg)
Single-shot Hastings lemma with ε = 1/(8ξ) gives
Iδmax(BC:BLBR) < l/(2ξ)
Then from single-shot merging and ξ-EDC:
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
R := all except BLBCBR :
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
Getting Subvolume Law
![Page 39: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/39.jpg)
Single-shot Hastings lemma with ε = 1/(8ξ) gives
Iδmax(BC:BLBR) < l/(2ξ)
Then from single-shot merging and ξ-EDC:
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
R := all except BLBCBR :
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
Getting Subvolume Law
![Page 40: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/40.jpg)
Single-shot Hastings lemma with ε = 1/(8ξ) gives
Iδmax(BC:BLBR) < l/(2ξ)
Then from single-shot merging and ξ-EDC:
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
R := all except BLBCBR :
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
Getting Subvolume Law
for
![Page 41: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/41.jpg)
Single-shot Hastings lemma with ε = 1/(8ξ) gives
Iδmax(BC:BLBR) < l/(2ξ)
Then from single-shot merging and ξ-EDC:
< l02O(1/ε) < l02O(1/ε)
A sBL BC BR
R := all except BLBCBR :
“It suffices to prove that nearby the boundary of C there is a region of size < l02O(ξ) with entropy < l/(4ξ)”
Getting Subvolume Law
for
![Page 42: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/42.jpg)
Getting Subvolume Law
By duality of min and max entropies(Tomamichel, Colbeck, Renner ‘09)
![Page 43: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/43.jpg)
Conclusions• EDC implies Area Law in 1D.
• Single-shot info theory useful in physics, here condensed-matter physics/quantum many-body theory
• Single-shot is more than to replace von Neumann by either min or max entropy: new regimes
• Single-shot theory is less structured, but gives a unified way of addressing different types of structures (iid, permutation invariance, ergodicity, exponential decay of correlations)
![Page 44: Exponential Decay of Correlations Implies Area Law: Single-Shot Techniques](https://reader035.vdocuments.site/reader035/viewer/2022062501/568164b8550346895dd6c360/html5/thumbnails/44.jpg)
Thanks!