1 / 53

Randomizing Quantum States

Randomizing Quantum States. Charles Bennett Patrick Hayden Debbie Leung Peter Shor Andreas Winter RSP: quant-ph/0307100 RAND: quant-ph/0307104. 1 bit of key per bit of message necessary and sufficient [Shannon]. One-time pad. Message. Shared key. Eavesdropper learns nothing:.

bruis
Download Presentation

Randomizing Quantum States

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Randomizing Quantum States Charles Bennett Patrick Hayden Debbie Leung Peter Shor Andreas Winter RSP: quant-ph/0307100 RAND: quant-ph/0307104

  2. 1 bit of key per bit of message necessary and sufficient [Shannon] One-time pad Message Shared key

  3. Eavesdropper learns nothing: Lower bound on key length: [BR,AMTW 2000] Private quantum channels

  4. MAIN RESULT: Given >0, there exists a choice of unitaries {Uk}, k=1,…,n such that the map is ε-randomizing, with Relax security criterion A CPTP map is ε-randomizing if for all states ,

  5. Security: 1 bit of key/qubit (1+2α) bits of key/qubit Approximate PQC Can encrypt a quantum state using 1 secret random bit per encrypted qubit asymptotically.

  6. Remote state preparation:Non-oblivious teleportation A circuit that needs no introduction:

  7. Remote state preparation:Non-oblivious teleportation Allow Alice to perform an arbitrary φ-dependent measurement

  8. From randomization to RSP Circuit for teleportation: l qubits k: 2l bits Before receiving k Bob knows nothing: (Private quantum channel!)

  9. From randomization to RSP Circuit for remote state preparation: l qubits k:l+o(l) bits Before receiving k, Bob knows nothing: (Approximate private quantum channel!)

  10. Knowledge is power All you need in this life is ignorance and confidence, and then success is sure. -Mark Twain Oblivious Alice (Teleportation): 1 ebit + 2 cbits per 1 qubit sent There is no knowledge that is not power. -Ralph Waldo Emerson Non-Oblivious Alice (RSP): 1 ebit + 1 cbit per 1 qubit sent Knowledge is power, if you know it about the right person. -Ethel Watts Mumford

  11. Randomization and nonlocality Does an ε-randomizing R destroy all correlations with the environment? True for separable ρ (easy). Not true for entangled inputs!

  12. Fidelity with maximally mixed state small: Rank argument Recall ε-randomizing map: Act on half of a maximally entangled state: has rank no more than n~d log d ~

  13. X Y Alice and Bob implement an LOCC measurement. Characterizing leftover correlations What does randomization map do to entangled inputs? R Charlie prepares maximally entangled state k then randomizes it.

  14. Separable state! k info randomized Conclusion: LOCC cannot distinguish strong and weak randomization Characterizing leftover correlations What does randomization map do to entangled inputs? R X Y Charlie prepares maximally entangled state k then randomizes it. Alice and Bob implement an LOCC measurement.

  15. (conjectured) Separable states Distillable states Product states Non-distillable states Detecting entanglement

  16. Strategy: • Select the unitaries {Uk} at random from Haar measure. • Show they are ε-randomizing with probability >0. Ingredients: • Large deviation estimate • Discretization Randomization proof ideas

  17. Will be concerned with distribution of: where In particular, (Exponential convergence to the mean) Large deviation estimate

  18. Will allow replacement of “all states” by “all states in M” in definition of ε-randomization. Discretization Find anε-net M: For all pure states φ, there exists a state φ’ in M such that Cd

  19. defn ε-net union bound large deviations Combine (Not as bad at it looks)

  20. ? ? Quantum data hiding GOAL: Charlie hides a bit from Alice and Bob, secure against LOCC RESULT: There exist bipartite n-qubit states hiding a bit with security 2-(n-1). [DLT, 2001]

  21. Security: Decodability: Hiding qubits GOAL: Charlie hides qubits from Alice and Bob, secure against LOCC [dV H T, 2002]

  22. Strategy

  23. Strategy

  24. Similar calculation: Choose Security Hinges on showing (Rank one projectors)

  25. Negligible overlaps between subspaces UkS if Decodability

  26. It is possible to hide a space of dimension in a bipartite system where each of the subsystem has dimension d. Asymptotically: 1 hidden qubit per 2 physical qubits. Previous best: ~ 1 hidden qubit per ~75 physical qubits. Data hiding summary

  27. Conclusion • Encryption of quantum states using 1 bit of shared key per encrypted qubit • Remote state preparation using 1 bit of communication and 1 ebit per qubit sent • Physical operation that destroys classical correlations but not quantum • Data hiding of 1 qubit in 2 physical qubits

  28. The End

  29. Act I LOCC data hiding for quantum states (quant-ph/0207147) Act II From RSP to PQC to data hiding Overview

  30. A few years ago in a lab moderately far away...

  31. ? ? 0 1 2 0 1 2 Not always possible: Nonlocality without entanglement [BDFMRSSW, 1999]

  32. Prepare superpositions of well hidden states? Hiding a qubit: First attempt TASK: Hide an arbitrary quantum state

  33. Hiding a qubit: First attempt TASK: Hide an arbitrary quantum state PROBLEM: Data hiding with pure states is impossible! (So much for superpositions.) [WSHV,2000]

  34. 2nd simplest idea THE PLAN: Use classical hidden bits as key to randomize a qubit PROPERTIES: 1) can be recovered using quantum communication 2) Naïve attacks fail (AB1 to find key then rotate B2) PROBLEM: Alice and Bob can attack AB1B2

  35. Actually, not a problem Any method to learn about by LOCC will provide a method to defeat the original cbit hiding scheme. Will argue the contrapositive: Assume there is an LOCC operation L (with output on Bob’s system alone) and two input states to the hiding map E such that

  36. CLAIM: Not all Li can be the same TPCP map. If they were, then by linearity: This says that L would never reveal any information about the input state, violating the hypothesis that L defeats the qubit hiding scheme. Minor algebra

  37. The attack: 1) Bob prepares a local maximally entangled state on B2B3 2) Alice and Bob apply L to AB1B2 3) Bob performs a measurement on B2B3 By Choi, there is a measurement that can partially distinguish L0 andLk Defeating the cbit hiding Conclusion from previous slide: there is a k such that L0Lk

  38. Imperfect hiding Wish to limit distinguishability through LOCC: For all input states and attacks. If the original 2n bit hiding scheme has security , then  <2n+1 . Not so bad: security of classical hiding schemes appears to improve exponentially with number of qubits used.

  39.  Authorized sets can recover the secret using quantum communication Multipartite cbit hiding A E B D C  With LOCC alone the five parties cannot learn i All monotonic access structures are possible: [Eggeling, Werner 2002]

  40. This authorization structure is impossible due to no-cloning! Multipartite qubit hiding A E B D C  With LOCC alone the five parties cannot learn  Authorized sets can recover the secret using quantum communication

  41. Problem for generalizing construction: B must be in all authorized sets Multipartite qubit hiding A E B D C  With LOCC alone the five parties cannot learn  Authorized sets can recover the secret using quantum communication

  42. Quantum secret sharing A e.g. ((2,3)) threshold scheme E B D C Secure against quantum communication in unauthorized sets but secret can be recovered by quantum communication in authorized sets. All monotonic threshold schemes not violating no-cloning [CGL,1999] All monotonic schemes not violating no-cloning [G,2000]

  43. Hiding distributed quantum data A Logical Pauli operators Multipartite cbit hiding states E B D C Resulting state provides strengthening of quantum secret sharing:  Secure against classical communication between all parties  Secure against quantum communication in unauthorized sets  Secret can be recovered only by quantum communication in authorized sets. All monotonic schemes not violating no-cloning

  44. Act II Fig. 1: Glimpse of a master magician’s workshop

  45. On the meaning of “ “ For any probability density P() on states in Cd and >0 there exists a choice of unitaries {Us}, s=1,…,S such that and Compare to the perfect private quantum channel: To achieve =0 requires log M = 2 log d.

  46. Another version There exists a choice of unitaries {Ups}, p=1,…,P, s=1,…,S such that for all states  in Cd and Can randomize every n-qubit state using 1 secret random bit and 1 public random bit per qubit.

  47. Consequences • Universal remote state preparation with only 1 ebit + 1 cbit per qubit • (No shared random bits necessary) • Weakly randomized maximally entangled states indistinguishable from maximally mixed states using LOCC • (Not just 1-way LOCC as sketched earlier)

  48. Application to data hiding R Consider ensemble of randomized states with V chosen using Haar measure Rank bound on entropy So we can do coding to get about n hidden bits using nxn bipartite states!

  49. Glyph collection

  50. Faction 1 Destroying classical correlations requires only 1 rbit per qubit Destroying quantum correlations requires 2 rbits per qubit Faction 2 Randomizing an arbitrary pure quantum state requires 1 public rbit and 1 secret rbit per qubit Competing visions

More Related