Internet Research Task Force (IRTF)                          A. Davidson
Request for Comments: 9497                                Brave Software
Category: Informational                                 A. Faz-Hernandez
ISSN: 2070-1721                                              N. Sullivan
                                                             C. A. Wood
                                                       Cloudflare, Inc.
                                                          December 2023


  Oblivious Pseudorandom Functions (OPRFs) Using Prime-Order Groups

Abstract

  An Oblivious Pseudorandom Function (OPRF) is a two-party protocol
  between a client and a server for computing the output of a
  Pseudorandom Function (PRF).  The server provides the PRF private
  key, and the client provides the PRF input.  At the end of the
  protocol, the client learns the PRF output without learning anything
  about the PRF private key, and the server learns neither the PRF
  input nor output.  An OPRF can also satisfy a notion of
  'verifiability', called a VOPRF.  A VOPRF ensures clients can verify
  that the server used a specific private key during the execution of
  the protocol.  A VOPRF can also be partially oblivious, called a
  POPRF.  A POPRF allows clients and servers to provide public input to
  the PRF computation.  This document specifies an OPRF, VOPRF, and
  POPRF instantiated within standard prime-order groups, including
  elliptic curves.  This document is a product of the Crypto Forum
  Research Group (CFRG) in the IRTF.

Status of This Memo

  This document is not an Internet Standards Track specification; it is
  published for informational purposes.

  This document is a product of the Internet Research Task Force
  (IRTF).  The IRTF publishes the results of Internet-related research
  and development activities.  These results might not be suitable for
  deployment.  This RFC represents the consensus of the Crypto Forum
  Research Group of the Internet Research Task Force (IRTF).  Documents
  approved for publication by the IRSG are not candidates for any level
  of Internet Standard; see Section 2 of RFC 7841.

  Information about the current status of this document, any errata,
  and how to provide feedback on it may be obtained at
  https://www.rfc-editor.org/info/rfc9497.

Copyright Notice

  Copyright (c) 2023 IETF Trust and the persons identified as the
  document authors.  All rights reserved.

  This document is subject to BCP 78 and the IETF Trust's Legal
  Provisions Relating to IETF Documents
  (https://trustee.ietf.org/license-info) in effect on the date of
  publication of this document.  Please review these documents
  carefully, as they describe your rights and restrictions with respect
  to this document.

Table of Contents

  1.  Introduction
    1.1.  Requirements Language
    1.2.  Notation and Terminology
  2.  Preliminaries
    2.1.  Prime-Order Group
    2.2.  Discrete Logarithm Equivalence Proofs
      2.2.1.  Proof Generation
      2.2.2.  Proof Verification
  3.  Protocol
    3.1.  Configuration
    3.2.  Key Generation and Context Setup
      3.2.1.  Deterministic Key Generation
    3.3.  Online Protocol
      3.3.1.  OPRF Protocol
      3.3.2.  VOPRF Protocol
      3.3.3.  POPRF Protocol
  4.  Ciphersuites
    4.1.  OPRF(ristretto255, SHA-512)
    4.2.  OPRF(decaf448, SHAKE-256)
    4.3.  OPRF(P-256, SHA-256)
    4.4.  OPRF(P-384, SHA-384)
    4.5.  OPRF(P-521, SHA-512)
    4.6.  Future Ciphersuites
    4.7.  Random Scalar Generation
      4.7.1.  Rejection Sampling
      4.7.2.  Random Number Generation Using Extra Random Bits
  5.  Application Considerations
    5.1.  Input Limits
    5.2.  External Interface Recommendations
    5.3.  Error Considerations
    5.4.  POPRF Public Input
  6.  IANA Considerations
  7.  Security Considerations
    7.1.  Security Properties
    7.2.  Security Assumptions
      7.2.1.  OPRF and VOPRF Assumptions
      7.2.2.  POPRF Assumptions
      7.2.3.  Static Diffie-Hellman Attack and Security Limits
    7.3.  Domain Separation
    7.4.  Timing Leaks
  8.  References
    8.1.  Normative References
    8.2.  Informative References
  Appendix A.  Test Vectors
    A.1.  ristretto255-SHA512
      A.1.1.  OPRF Mode
      A.1.2.  VOPRF Mode
      A.1.3.  POPRF Mode
    A.2.  decaf448-SHAKE256
      A.2.1.  OPRF Mode
      A.2.2.  VOPRF Mode
      A.2.3.  POPRF Mode
    A.3.  P256-SHA256
      A.3.1.  OPRF Mode
      A.3.2.  VOPRF Mode
      A.3.3.  POPRF Mode
    A.4.  P384-SHA384
      A.4.1.  OPRF Mode
      A.4.2.  VOPRF Mode
      A.4.3.  POPRF Mode
    A.5.  P521-SHA512
      A.5.1.  OPRF Mode
      A.5.2.  VOPRF Mode
      A.5.3.  POPRF Mode
  Acknowledgements
  Authors' Addresses

1.  Introduction

  A Pseudorandom Function (PRF) F(k, x) is an efficiently computable
  function taking a private key k and a value x as input.  This
  function is pseudorandom if the keyed function K(_) = F(k, _) is
  indistinguishable from a randomly sampled function acting on the same
  domain and range as K().  An Oblivious PRF (OPRF) is a two-party
  protocol between a server and a client, wherein the server holds a
  PRF key k and the client holds some input x.  The protocol allows
  both parties to cooperate in computing F(k, x), such that the client
  learns F(k, x) without learning anything about k and the server does
  not learn anything about x or F(k, x).  A Verifiable OPRF (VOPRF) is
  an OPRF, wherein the server also proves to the client that F(k, x)
  was produced by the key k corresponding to the server's public key,
  which the client knows.  A Partially Oblivious PRF (POPRF) is a
  variant of a VOPRF, where the client and server interact in computing
  F(k, x, y), for some PRF F with server-provided key k, client-
  provided input x, and public input y, and the client receives proof
  that F(k, x, y) was computed using k corresponding to the public key
  that the client knows.  A POPRF with fixed input y is functionally
  equivalent to a VOPRF.

  OPRFs have a variety of applications, including password-protected
  secret sharing schemes [JKKX16], privacy-preserving password stores
  [SJKS17], and password-authenticated key exchange (PAKE) [OPAQUE].
  Verifiable OPRFs are necessary in some applications, such as Privacy
  Pass [PRIVACY-PASS].  Verifiable OPRFs have also been used for
  password-protected secret sharing schemes, such as that of [JKK14].

  This document specifies OPRF, VOPRF, and POPRF protocols built upon
  prime-order groups.  The document describes each protocol variant,
  along with application considerations, and their security properties.

  This document represents the consensus of the Crypto Forum Research
  Group (CFRG).  It is not an IETF product and is not a standard.

1.1.  Requirements Language

  The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
  "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
  "OPTIONAL" in this document are to be interpreted as described in
  BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
  capitals, as shown here.

1.2.  Notation and Terminology

  The following functions and notation are used throughout the
  document.

  *  For any object x, we write len(x) to denote its length in bytes.

  *  For two-byte arrays x and y, write x || y to denote their
     concatenation.

  *  I2OSP(x, xLen) converts a nonnegative integer x into a byte array
     of specified length xLen, as described in [RFC8017].  Note that
     this function returns a byte array in big-endian byte order.

  *  The notation T U[N] refers to an array called U, containing N
     items of type T.  The type opaque means one single byte of
     uninterpreted data.  Items of the array are zero-indexed and
     referred to as U[j], such that 0 <= j < N.

  All algorithms and procedures described in this document are laid out
  in a Python-like pseudocode.  Each function takes a set of inputs and
  parameters and produces a set of output values.  Parameters become
  constant values once the protocol variant and the ciphersuite are
  fixed.

  The PrivateInput data type refers to inputs that are known only to
  the client in the protocol, whereas the PublicInput data type refers
  to inputs that are known to both the client and server in the
  protocol.  Both PrivateInput and PublicInput are opaque byte strings
  of arbitrary length no larger than 2^16 - 1 bytes.  This length
  restriction exists because PublicInput and PrivateInput values are
  length-prefixed with two bytes before use throughout the protocol.

  String values, such as "DeriveKeyPair", "Seed-", and "Finalize", are
  ASCII string literals.

  The following terms are used throughout this document.

  PRF:  Pseudorandom Function

  OPRF:  Oblivious Pseudorandom Function

  VOPRF:  Verifiable Oblivious Pseudorandom Function

  POPRF:  Partially Oblivious Pseudorandom Function

  Client:  Protocol initiator.  Learns PRF evaluation as the output of
     the protocol.

  Server:  Computes the PRF using a private key.  Learns nothing about
     the client's input or output.

2.  Preliminaries

  The protocols in this document have two primary dependencies:

  Group:  A prime-order group implementing the API described below in
     Section 2.1.  See Section 4 for specific instances of groups.

  Hash:  A cryptographic hash function whose output length is Nh bytes.

  Section 4 specifies ciphersuites as combinations of Group and Hash.

2.1.  Prime-Order Group

  In this document, we assume the construction of an additive, prime-
  order group, denoted Group, for performing all mathematical
  operations.  In prime-order groups, any element (other than the
  identity) can generate the other elements of the group.  Usually, one
  element is fixed and defined as the group generator.  Such groups are
  uniquely determined by the choice of the prime p that defines the
  order of the group.  (However, different representations of the group
  for a single p may exist.  Section 4 lists specific groups that
  indicate both the order and representation.)

  The fundamental group operation is addition + with identity element
  I.  For any elements A and B of the group, A + B = B + A is also a
  member of the group.  Also, for any A in the group, there exists an
  element -A, such that A + (-A) = (-A) + A = I.  Scalar multiplication
  by r is equivalent to the repeated application of the group operation
  on an element A with itself r - 1 times; this is denoted as r * A = A
  + ... + A.  For any element A, p * A = I.  The case when the scalar
  multiplication is performed on the group generator is denoted as
  ScalarMultGen(r).  Given two elements A and B, the discrete logarithm
  problem is to find an integer k, such that B = k * A.  Thus, k is the
  discrete logarithm of B with respect to the base A.  The set of
  scalars corresponds to GF(p), a prime field of order p, and is
  represented as the set of integers defined by {0, 1, ..., p - 1}.
  This document uses types Element and Scalar to denote elements of the
  group and its set of scalars, respectively.

  We now detail a number of member functions that can be invoked on a
  prime-order group.

  Order():  Outputs the order of the group (i.e., p).

  Identity():  Outputs the identity element of the group (i.e., I).

  Generator():  Outputs the generator element of the group.

  HashToGroup(x):  Deterministically maps an array of bytes x to an
     element of Group.  The map must ensure that, for any adversary
     receiving R = HashToGroup(x), it is computationally difficult to
     reverse the mapping.  This function is optionally parameterized by
     a domain separation tag (DST); see Section 4.  Security properties
     of this function are described in [RFC9380].

  HashToScalar(x):  Deterministically maps an array of bytes x to an
     element in GF(p).  This function is optionally parameterized by a
     DST; see Section 4.  Security properties of this function are
     described in [RFC9380], Section 10.5.

  RandomScalar():  Chooses at random a nonzero element in GF(p).

  ScalarInverse(s):  Returns the inverse of input Scalar s on GF(p).

  SerializeElement(A):  Maps an Element A to a canonical byte array buf
     of fixed-length Ne.

  DeserializeElement(buf):  Attempts to map a byte array buf to an
     Element A and fails if the input is not the valid canonical byte
     representation of an element of the group.  This function can
     raise a DeserializeError if deserialization fails or A is the
     identity element of the group; see Section 4 for group-specific
     input validation steps.

  SerializeScalar(s):  Maps Scalar s to a canonical byte array buf of
     fixed-length Ns.

  DeserializeScalar(buf):  Attempts to map a byte array buf to Scalar
     s.  This function can raise a DeserializeError if deserialization
     fails; see Section 4 for group-specific input validation steps.

  Section 4 contains details for the implementation of this interface
  for different prime-order groups instantiated over elliptic curves.
  In particular, for some choices of elliptic curves, e.g., those
  detailed in [RFC7748], which require accounting for cofactors,
  Section 4 describes required steps necessary to ensure the resulting
  group is of prime order.

2.2.  Discrete Logarithm Equivalence Proofs

  A proof of knowledge allows a prover to convince a verifier that some
  statement is true.  If the prover can generate a proof without
  interaction with the verifier, the proof is noninteractive.  If the
  verifier learns nothing other than whether the statement claimed by
  the prover is true or false, the proof is zero-knowledge.

  This section describes a noninteractive, zero-knowledge proof for
  discrete logarithm equivalence (DLEQ), which is used in the
  construction of VOPRF and POPRF.  A DLEQ proof demonstrates that two
  pairs of group elements have the same discrete logarithm without
  revealing the discrete logarithm.

  The DLEQ proof resembles the Chaum-Pedersen [ChaumPedersen] proof,
  which is shown to be zero-knowledge by Jarecki, et al. [JKK14] and is
  noninteractive after applying the Fiat-Shamir transform [FS00].
  Furthermore, Davidson, et al. [DGSTV18] showed a proof system for
  batching DLEQ proofs that has constant-size proofs with respect to
  the number of inputs.  The specific DLEQ proof system presented below
  follows this latter construction with two modifications: (1) the
  transcript used to generate the seed includes more context
  information and (2) the individual challenges for each element in the
  proof is derived from a seed-prefixed hash-to-scalar invocation,
  rather than being sampled from a seeded Pseudorandom Number Generator
  (PRNG).  The description is split into two subsections: one for
  generating the proof, which is done by servers in the verifiable
  protocols, and another for verifying the proof, which is done by
  clients in the protocol.

2.2.1.  Proof Generation

  Generating a proof is done with the GenerateProof function, as
  defined below.  Given Element values A and B, two non-empty lists of
  Element values C and D of length m, and Scalar k, this function
  produces a proof that k * A == B and k * C[i] == D[i] for each i in
  [0, ..., m - 1].  The output is a value of type Proof, which is a
  tuple of two Scalar values.  We use the notation proof[0] and
  proof[1] to denote the first and second elements in this tuple,
  respectively.

  GenerateProof accepts lists of inputs to amortize the cost of proof
  generation.  Applications can take advantage of this functionality to
  produce a single, constant-sized proof for m DLEQ inputs, rather than
  m proofs for m DLEQ inputs.

  Input:

    Scalar k
    Element A
    Element B
    Element C[m]
    Element D[m]

  Output:

    Proof proof

  Parameters:

    Group G

  def GenerateProof(k, A, B, C, D):
    (M, Z) = ComputeCompositesFast(k, B, C, D)

    r = G.RandomScalar()
    t2 = r * A
    t3 = r * M

    Bm = G.SerializeElement(B)
    a0 = G.SerializeElement(M)
    a1 = G.SerializeElement(Z)
    a2 = G.SerializeElement(t2)
    a3 = G.SerializeElement(t3)

    challengeTranscript =
      I2OSP(len(Bm), 2) || Bm ||
      I2OSP(len(a0), 2) || a0 ||
      I2OSP(len(a1), 2) || a1 ||
      I2OSP(len(a2), 2) || a2 ||
      I2OSP(len(a3), 2) || a3 ||
      "Challenge"

    c = G.HashToScalar(challengeTranscript)
    s = r - c * k

    return [c, s]

  The helper function ComputeCompositesFast is as defined below and is
  an optimization of the ComputeComposites function for servers since
  they have knowledge of the private key.

  Input:

    Scalar k
    Element B
    Element C[m]
    Element D[m]

  Output:

    Element M
    Element Z

  Parameters:

    Group G
    PublicInput contextString

  def ComputeCompositesFast(k, B, C, D):
    Bm = G.SerializeElement(B)
    seedDST = "Seed-" || contextString
    seedTranscript =
      I2OSP(len(Bm), 2) || Bm ||
      I2OSP(len(seedDST), 2) || seedDST
    seed = Hash(seedTranscript)

    M = G.Identity()
    for i in range(m):
      Ci = G.SerializeElement(C[i])
      Di = G.SerializeElement(D[i])
      compositeTranscript =
        I2OSP(len(seed), 2) || seed || I2OSP(i, 2) ||
        I2OSP(len(Ci), 2) || Ci ||
        I2OSP(len(Di), 2) || Di ||
        "Composite"

      di = G.HashToScalar(compositeTranscript)
      M = di * C[i] + M

    Z = k * M

    return (M, Z)

  When used in the protocol described in Section 3, the parameter
  contextString is as defined in Section 3.2.

2.2.2.  Proof Verification

  Verifying a proof is done with the VerifyProof function, as defined
  below.  This function takes Element values A and B, two non-empty
  lists of Element values C and D of length m, and a Proof value output
  from GenerateProof.  It outputs a single boolean value indicating
  whether or not the proof is valid for the given DLEQ inputs.  Note
  this function can verify proofs on lists of inputs whenever the proof
  was generated as a batched DLEQ proof with the same inputs.

  Input:

    Element A
    Element B
    Element C[m]
    Element D[m]
    Proof proof

  Output:

    boolean verified

  Parameters:

    Group G

  def VerifyProof(A, B, C, D, proof):
    (M, Z) = ComputeComposites(B, C, D)
    c = proof[0]
    s = proof[1]

    t2 = ((s * A) + (c * B))
    t3 = ((s * M) + (c * Z))

    Bm = G.SerializeElement(B)
    a0 = G.SerializeElement(M)
    a1 = G.SerializeElement(Z)
    a2 = G.SerializeElement(t2)
    a3 = G.SerializeElement(t3)

    challengeTranscript =
      I2OSP(len(Bm), 2) || Bm ||
      I2OSP(len(a0), 2) || a0 ||
      I2OSP(len(a1), 2) || a1 ||
      I2OSP(len(a2), 2) || a2 ||
      I2OSP(len(a3), 2) || a3 ||
      "Challenge"

    expectedC = G.HashToScalar(challengeTranscript)
    verified = (expectedC == c)

    return verified

  The definition of ComputeComposites is given below.

  Input:

    Element B
    Element C[m]
    Element D[m]

  Output:

    Element M
    Element Z

  Parameters:

    Group G
    PublicInput contextString

  def ComputeComposites(B, C, D):
    Bm = G.SerializeElement(B)
    seedDST = "Seed-" || contextString
    seedTranscript =
      I2OSP(len(Bm), 2) || Bm ||
      I2OSP(len(seedDST), 2) || seedDST
    seed = Hash(seedTranscript)

    M = G.Identity()
    Z = G.Identity()
    for i in range(m):
      Ci = G.SerializeElement(C[i])
      Di = G.SerializeElement(D[i])
      compositeTranscript =
        I2OSP(len(seed), 2) || seed || I2OSP(i, 2) ||
        I2OSP(len(Ci), 2) || Ci ||
        I2OSP(len(Di), 2) || Di ||
        "Composite"

      di = G.HashToScalar(compositeTranscript)
      M = di * C[i] + M
      Z = di * D[i] + Z

    return (M, Z)

  When used in the protocol described in Section 3, the parameter
  contextString is as defined in Section 3.2.

3.  Protocol

  In this section, we define and describe three protocol variants
  referred to as the OPRF, VOPRF, and POPRF modes.  Each of these
  variants involves two messages between the client and server, but
  they differ slightly in terms of the security properties; see
  Section 7.1 for more information.  A high-level description of the
  functionality of each mode follows.

  In the OPRF mode, a client and server interact to compute output =
  F(skS, input), where input is the client's private input, skS is the
  server's private key, and output is the OPRF output.  After the
  execution of the protocol, the client learns the output and the
  server learns nothing.  This interaction is shown below.

      Client(input)                                        Server(skS)
    -------------------------------------------------------------------
    blind, blindedElement = Blind(input)

                               blindedElement
                                 ---------->

                  evaluatedElement = BlindEvaluate(skS, blindedElement)

                               evaluatedElement
                                 <----------

    output = Finalize(input, blind, evaluatedElement)

                     Figure 1: OPRF Protocol Overview

  In the VOPRF mode, the client additionally receives proof that the
  server used skS in computing the function.  To achieve verifiability,
  as in [JKK14], the server provides a zero-knowledge proof that the
  key provided as input by the server in the BlindEvaluate function is
  the same key as is used to produce the server's public key, pkS,
  which the client receives as input to the protocol.  This proof does
  not reveal the server's private key to the client.  This interaction
  is shown below.

      Client(input, pkS)       <---- pkS ------        Server(skS, pkS)
    -------------------------------------------------------------------
    blind, blindedElement = Blind(input)

                               blindedElement
                                 ---------->

                evaluatedElement, proof = BlindEvaluate(skS, pkS,
                                                        blindedElement)

                           evaluatedElement, proof
                                 <----------

    output = Finalize(input, blind, evaluatedElement,
                      blindedElement, pkS, proof)

         Figure 2: VOPRF Protocol Overview with Additional Proof

  The POPRF mode extends the VOPRF mode such that the client and server
  can additionally provide the public input info, which is used in
  computing the PRF.  That is, the client and server interact to
  compute output = F(skS, input, info), as is shown below.

      Client(input, pkS, info) <---- pkS ------  Server(skS, pkS, info)
    -------------------------------------------------------------------
    blind, blindedElement, tweakedKey = Blind(input, info, pkS)

                               blindedElement
                                 ---------->

           evaluatedElement, proof = BlindEvaluate(skS, blindedElement,
                                                   info)

                           evaluatedElement, proof
                                 <----------

    output = Finalize(input, blind, evaluatedElement,
                      blindedElement, proof, info, tweakedKey)

      Figure 3: POPRF Protocol Overview with Additional Public Input

  Each protocol consists of an offline setup phase and an online phase,
  as described in Sections 3.2 and 3.3, respectively.  Configuration
  details for the offline phase are described in Section 3.1.

3.1.  Configuration

  Each of the three protocol variants are identified with a one-byte
  value (in hexadecimal):

                           +===========+=======+
                           | Mode      | Value |
                           +===========+=======+
                           | modeOPRF  | 0x00  |
                           +-----------+-------+
                           | modeVOPRF | 0x01  |
                           +-----------+-------+
                           | modePOPRF | 0x02  |
                           +-----------+-------+

                            Table 1: Identifiers
                           for Protocol Variants

  Additionally, each protocol variant is instantiated with a
  ciphersuite or suite.  Each ciphersuite is identified with an ASCII
  string identifier, referred to as identifier; see Section 4 for the
  set of initial ciphersuite values.

  The mode and ciphersuite identifier values are combined to create a
  "context string" used throughout the protocol with the following
  function:

  def CreateContextString(mode, identifier):
    return "OPRFV1-" || I2OSP(mode, 1) || "-" || identifier

3.2.  Key Generation and Context Setup

  In the offline setup phase, the server generates a fresh, random key
  pair (skS, pkS).  There are two ways to generate this key pair.  The
  first of which is using the GenerateKeyPair function described below.

  Input: None

  Output:

    Scalar skS
    Element pkS

  Parameters:

    Group G

  def GenerateKeyPair():
    skS = G.RandomScalar()
    pkS = G.ScalarMultGen(skS)
    return skS, pkS

  The second way to generate the key pair is via the deterministic key
  generation function DeriveKeyPair, as described in Section 3.2.1.
  Applications and implementations can use either method in practice.

  Also during the offline setup phase, both the client and server
  create a context used for executing the online phase of the protocol
  after agreeing on a mode and ciphersuite identifier.  The context,
  such as OPRFServerContext, is an implementation-specific data
  structure that stores a context string and the relevant key material
  for each party.

  The OPRF variant server and client contexts are created as follows:

  def SetupOPRFServer(identifier, skS):
    contextString = CreateContextString(modeOPRF, identifier)
    return OPRFServerContext(contextString, skS)

  def SetupOPRFClient(identifier):
    contextString = CreateContextString(modeOPRF, identifier)
    return OPRFClientContext(contextString)

  The VOPRF variant server and client contexts are created as follows:

  def SetupVOPRFServer(identifier, skS):
    contextString = CreateContextString(modeVOPRF, identifier)
    return VOPRFServerContext(contextString, skS)

  def SetupVOPRFClient(identifier, pkS):
    contextString = CreateContextString(modeVOPRF, identifier)
    return VOPRFClientContext(contextString, pkS)

  The POPRF variant server and client contexts are created as follows:

  def SetupPOPRFServer(identifier, skS):
    contextString = CreateContextString(modePOPRF, identifier)
    return POPRFServerContext(contextString, skS)

  def SetupPOPRFClient(identifier, pkS):
    contextString = CreateContextString(modePOPRF, identifier)
    return POPRFClientContext(contextString, pkS)

3.2.1.  Deterministic Key Generation

  This section describes a deterministic key generation function,
  DeriveKeyPair.  It accepts a seed of 32 bytes generated from a
  cryptographically secure random number generator and an optional
  (possibly empty) info string.  Note that, by design, knowledge of
  seed and info is necessary to compute this function, which means that
  the secrecy of the output private key (skS) depends on the secrecy of
  seed (since the info string is public).

  Input:

    opaque seed[32]
    PublicInput info

  Output:

    Scalar skS
    Element pkS

  Parameters:

    Group G
    PublicInput contextString

  Errors: DeriveKeyPairError

  def DeriveKeyPair(seed, info):
    deriveInput = seed || I2OSP(len(info), 2) || info
    counter = 0
    skS = 0
    while skS == 0:
      if counter > 255:
        raise DeriveKeyPairError
      skS = G.HashToScalar(deriveInput || I2OSP(counter, 1),
                            DST = "DeriveKeyPair" || contextString)
      counter = counter + 1
    pkS = G.ScalarMultGen(skS)
    return skS, pkS

3.3.  Online Protocol

  In the online phase, the client and server engage in a two-message
  protocol to compute the protocol output.  This section describes the
  protocol details for each protocol variant.  Throughout each
  description, the following parameters are assumed to exist:

  G:  a prime-order group implementing the API described in Section 2.1

  contextString:  a PublicInput domain separation tag constructed
     during context setup, as created in Section 3.1

  skS and pkS:  a Scalar and Element representing the private and
     public keys configured for the client and server in Section 3.2

  Applications serialize protocol messages between the client and
  server for transmission.  Element values and Scalar values are
  serialized to byte arrays, and values of type Proof are serialized as
  the concatenation of two serialized Scalar values.  Deserializing
  these values can fail; in which case, the application MUST abort the
  protocol, raising a DeserializeError failure.

  Applications MUST check that input Element values received over the
  wire are not the group identity element.  This check is handled after
  deserializing Element values; see Section 4 for more information and
  requirements on input validation for each ciphersuite.

3.3.1.  OPRF Protocol

  The OPRF protocol begins with the client blinding its input, as
  described by the Blind function below.  Note that this function can
  fail with an InvalidInputError error for certain inputs that map to
  the group identity element.  Dealing with this failure is an
  application-specific decision; see Section 5.3.

  Input:

    PrivateInput input

  Output:

    Scalar blind
    Element blindedElement

  Parameters:

    Group G

  Errors: InvalidInputError

  def Blind(input):
    blind = G.RandomScalar()
    inputElement = G.HashToGroup(input)
    if inputElement == G.Identity():
      raise InvalidInputError
    blindedElement = blind * inputElement

    return blind, blindedElement

  Clients store blind locally and send blindedElement to the server for
  evaluation.  Upon receipt, servers process blindedElement using the
  BlindEvaluate function described below.

  Input:

    Scalar skS
    Element blindedElement

  Output:

    Element evaluatedElement

  def BlindEvaluate(skS, blindedElement):
    evaluatedElement = skS * blindedElement
    return evaluatedElement

  Servers send the output evaluatedElement to clients for processing.
  Recall that servers may process multiple client inputs by applying
  the BlindEvaluate function to each blindedElement received and
  returning an array with the corresponding evaluatedElement values.

  Upon receipt of evaluatedElement, clients process it to complete the
  OPRF evaluation with the Finalize function described below.

  Input:

    PrivateInput input
    Scalar blind
    Element evaluatedElement

  Output:

    opaque output[Nh]

  Parameters:

    Group G

  def Finalize(input, blind, evaluatedElement):
    N = G.ScalarInverse(blind) * evaluatedElement
    unblindedElement = G.SerializeElement(N)

    hashInput = I2OSP(len(input), 2) || input ||
                I2OSP(len(unblindedElement), 2) || unblindedElement ||
                "Finalize"
    return Hash(hashInput)

  An entity that knows both the private key and the input can compute
  the PRF result using the following Evaluate function.

  Input:

    Scalar skS
    PrivateInput input

  Output:

    opaque output[Nh]

  Parameters:

    Group G

  Errors: InvalidInputError

  def Evaluate(skS, input):
    inputElement = G.HashToGroup(input)
    if inputElement == G.Identity():
      raise InvalidInputError
    evaluatedElement = skS * inputElement
    issuedElement = G.SerializeElement(evaluatedElement)

    hashInput = I2OSP(len(input), 2) || input ||
                I2OSP(len(issuedElement), 2) || issuedElement ||
                "Finalize"
    return Hash(hashInput)

3.3.2.  VOPRF Protocol

  The VOPRF protocol begins with the client blinding its input, using
  the same Blind function as in Section 3.3.1.  Clients store the
  output blind locally and send blindedElement to the server for
  evaluation.  Upon receipt, servers process blindedElement to compute
  an evaluated element and a DLEQ proof using the following
  BlindEvaluate function.

  Input:

    Scalar skS
    Element pkS
    Element blindedElement

  Output:

    Element evaluatedElement
    Proof proof

  Parameters:

    Group G

  def BlindEvaluate(skS, pkS, blindedElement):
    evaluatedElement = skS * blindedElement
    blindedElements = [blindedElement]     // list of length 1
    evaluatedElements = [evaluatedElement] // list of length 1
    proof = GenerateProof(skS, G.Generator(), pkS,
                          blindedElements, evaluatedElements)
    return evaluatedElement, proof

  In the description above, inputs to GenerateProof are one-item lists.
  Using larger lists allows servers to batch the evaluation of multiple
  elements while producing a single batched DLEQ proof for them.

  The server sends both evaluatedElement and proof back to the client.
  Upon receipt, the client processes both values to complete the VOPRF
  computation using the Finalize function below.

  Input:

    PrivateInput input
    Scalar blind
    Element evaluatedElement
    Element blindedElement
    Element pkS
    Proof proof

  Output:

    opaque output[Nh]

  Parameters:

    Group G

  Errors: VerifyError

  def Finalize(input, blind, evaluatedElement,
               blindedElement, pkS, proof):
    blindedElements = [blindedElement]     // list of length 1
    evaluatedElements = [evaluatedElement] // list of length 1
    if VerifyProof(G.Generator(), pkS, blindedElements,
                   evaluatedElements, proof) == false:
      raise VerifyError

    N = G.ScalarInverse(blind) * evaluatedElement
    unblindedElement = G.SerializeElement(N)

    hashInput = I2OSP(len(input), 2) || input ||
                I2OSP(len(unblindedElement), 2) || unblindedElement ||
                "Finalize"
    return Hash(hashInput)

  As in BlindEvaluate, inputs to VerifyProof are one-item lists.
  Clients can verify multiple inputs at once whenever the server
  produced a batched DLEQ proof for them.

  Finally, an entity that knows both the private key and the input can
  compute the PRF result using the Evaluate function described in
  Section 3.3.1.

3.3.3.  POPRF Protocol

  The POPRF protocol begins with the client blinding its input, using
  the following modified Blind function.  In this step, the client also
  binds a public info value, which produces an additional tweakedKey to
  be used later in the protocol.  Note that this function can fail with
  an InvalidInputError error for certain private inputs that map to the
  group identity element, as well as certain public inputs that, if not
  detected at this point, will cause server evaluation to fail.
  Dealing with either failure is an application-specific decision; see
  Section 5.3.

  Input:

    PrivateInput input
    PublicInput info
    Element pkS

  Output:

    Scalar blind
    Element blindedElement
    Element tweakedKey

  Parameters:

    Group G

  Errors: InvalidInputError

  def Blind(input, info, pkS):
    framedInfo = "Info" || I2OSP(len(info), 2) || info
    m = G.HashToScalar(framedInfo)
    T = G.ScalarMultGen(m)
    tweakedKey = T + pkS
    if tweakedKey == G.Identity():
      raise InvalidInputError

    blind = G.RandomScalar()
    inputElement = G.HashToGroup(input)
    if inputElement == G.Identity():
      raise InvalidInputError

    blindedElement = blind * inputElement

    return blind, blindedElement, tweakedKey

  Clients store the outputs blind and tweakedKey locally and send
  blindedElement to the server for evaluation.  Upon receipt, servers
  process blindedElement to compute an evaluated element and a DLEQ
  proof using the following BlindEvaluate function.

  Input:

    Scalar skS
    Element blindedElement
    PublicInput info

  Output:

    Element evaluatedElement
    Proof proof

  Parameters:

    Group G

  Errors: InverseError

  def BlindEvaluate(skS, blindedElement, info):
    framedInfo = "Info" || I2OSP(len(info), 2) || info
    m = G.HashToScalar(framedInfo)
    t = skS + m
    if t == 0:
      raise InverseError

    evaluatedElement = G.ScalarInverse(t) * blindedElement

    tweakedKey = G.ScalarMultGen(t)
    evaluatedElements = [evaluatedElement] // list of length 1
    blindedElements = [blindedElement]     // list of length 1
    proof = GenerateProof(t, G.Generator(), tweakedKey,
                          evaluatedElements, blindedElements)

    return evaluatedElement, proof

  In the description above, inputs to GenerateProof are one-item lists.
  Using larger lists allows servers to batch the evaluation of multiple
  elements while producing a single batched DLEQ proof for them.

  BlindEvaluate triggers InverseError when the function is about to
  calculate the inverse of a zero scalar, which does not exist and
  therefore yields a failure in the protocol.  This only occurs for
  info values that map to the private key of the server.  Thus, clients
  that cause this error should be assumed to know the server private
  key.  Hence, this error can be a signal for the server to replace its
  private key.

  The server sends both evaluatedElement and proof back to the client.
  Upon receipt, the client processes both values to complete the POPRF
  computation using the Finalize function below.

  Input:

    PrivateInput input
    Scalar blind
    Element evaluatedElement
    Element blindedElement
    Proof proof
    PublicInput info
    Element tweakedKey

  Output:

    opaque output[Nh]

  Parameters:

    Group G

  Errors: VerifyError

  def Finalize(input, blind, evaluatedElement, blindedElement,
               proof, info, tweakedKey):
    evaluatedElements = [evaluatedElement] // list of length 1
    blindedElements = [blindedElement]     // list of length 1
    if VerifyProof(G.Generator(), tweakedKey, evaluatedElements,
                   blindedElements, proof) == false:
      raise VerifyError

    N = G.ScalarInverse(blind) * evaluatedElement
    unblindedElement = G.SerializeElement(N)

    hashInput = I2OSP(len(input), 2) || input ||
                I2OSP(len(info), 2) || info ||
                I2OSP(len(unblindedElement), 2) || unblindedElement ||
                "Finalize"
    return Hash(hashInput)

  As in BlindEvaluate, inputs to VerifyProof are one-item lists.
  Clients can verify multiple inputs at once whenever the server
  produced a batched DLEQ proof for them.

  Finally, an entity that knows both the private key and the input can
  compute the PRF result using the Evaluate function described below.

  Input:

    Scalar skS
    PrivateInput input
    PublicInput info

  Output:

    opaque output[Nh]

  Parameters:

    Group G

  Errors: InvalidInputError, InverseError

  def Evaluate(skS, input, info):
    inputElement = G.HashToGroup(input)
    if inputElement == G.Identity():
      raise InvalidInputError

    framedInfo = "Info" || I2OSP(len(info), 2) || info
    m = G.HashToScalar(framedInfo)
    t = skS + m
    if t == 0:
      raise InverseError
    evaluatedElement = G.ScalarInverse(t) * inputElement
    issuedElement = G.SerializeElement(evaluatedElement)

    hashInput = I2OSP(len(input), 2) || input ||
                I2OSP(len(info), 2) || info ||
                I2OSP(len(issuedElement), 2) || issuedElement ||
                "Finalize"
    return Hash(hashInput)

4.  Ciphersuites

  A ciphersuite (also referred to as 'suite' in this document) for the
  protocol wraps the functionality required for the protocol to take
  place.  The ciphersuite should be available to both the client and
  server, and agreement on the specific instantiation is assumed
  throughout.

  A ciphersuite contains instantiations of the following
  functionalities:

  Group:  A prime-order group exposing the API detailed in Section 2.1,
     with the generator element defined in the corresponding reference
     for each group.  Each group also specifies HashToGroup,
     HashToScalar, and serialization functionalities.  For HashToGroup,
     the domain separation tag (DST) is constructed in accordance with
     the recommendations in [RFC9380], Section 3.1.  For HashToScalar,
     each group specifies an integer order that is used in reducing
     integer values to a member of the corresponding scalar field.

  Hash:  A cryptographic hash function whose output length is Nh bytes
     long.

  This section includes an initial set of ciphersuites with supported
  groups and hash functions.  It also includes implementation details
  for each ciphersuite, focusing on input validation.  Future documents
  can specify additional ciphersuites as needed, provided they meet the
  requirements in Section 4.6.

  For each ciphersuite, contextString is that which is computed in the
  Setup functions.  Applications should take caution in using
  ciphersuites targeting P-256 and ristretto255.  See Section 7.2 for
  related discussion.

4.1.  OPRF(ristretto255, SHA-512)

  This ciphersuite uses ristretto255 [RFC9496] for the Group and
  SHA-512 for the hash function.  The value of the ciphersuite
  identifier is "ristretto255-SHA512".

  Group:  ristretto255 [RFC9496]

     Order():  Return 2^252 + 27742317777372353535851937790883648493
        (see [RFC9496]).

     Identity():  As defined in [RFC9496].

     Generator():  As defined in [RFC9496].

     HashToGroup():  Use hash_to_ristretto255 [RFC9380] with DST =
        "HashToGroup-" || contextString and expand_message =
        expand_message_xmd using SHA-512.

     HashToScalar():  Compute uniform_bytes using expand_message =
        expand_message_xmd, DST = "HashToScalar-" || contextString, and
        an output length of 64 bytes, interpret uniform_bytes as a
        512-bit integer in little-endian order, and reduce the integer
        modulo Group.Order().

     ScalarInverse(s):  Returns the multiplicative inverse of input
        Scalar s mod Group.Order().

     RandomScalar():  Implemented by returning a uniformly random
        Scalar in the range [0, G.Order() - 1].  Refer to Section 4.7
        for implementation guidance.

     SerializeElement(A):  Implemented using the Encode function from
        Section 4.3.2 of [RFC9496]; Ne = 32.

     DeserializeElement(buf):  Implemented using the Decode function
        from Section 4.3.1 of [RFC9496].  Additionally, this function
        validates that the resulting element is not the group identity
        element.  If these checks fail, deserialization returns an
        InputValidationError error.

     SerializeScalar(s):  Implemented by outputting the little-endian,
        32-byte encoding of the Scalar value with the top three bits
        set to zero; Ns = 32.

     DeserializeScalar(buf):  Implemented by attempting to deserialize
        a Scalar from a little-endian, 32-byte string.  This function
        can fail if the input does not represent a Scalar in the range
        [0, G.Order() - 1].  Note that this means the top three bits of
        the input MUST be zero.

  Hash:  SHA-512; Nh = 64.

4.2.  OPRF(decaf448, SHAKE-256)

  This ciphersuite uses decaf448 [RFC9496] for the Group and SHAKE-256
  for the hash function.  The value of the ciphersuite identifier is
  "decaf448-SHAKE256".

  Group:  decaf448 [RFC9496]

     Order():  Return 2^446 - 13818066809895115352007386748515426880336
        692474882178609894547503885.

     Identity():  As defined in [RFC9496].

     Generator():  As defined in [RFC9496].

     RandomScalar():  Implemented by returning a uniformly random
        Scalar in the range [0, G.Order() - 1].  Refer to Section 4.7
        for implementation guidance.

     HashToGroup():  Use hash_to_decaf448 [RFC9380] with DST =
        "HashToGroup-" || contextString and expand_message =
        expand_message_xof using SHAKE-256.

     HashToScalar():  Compute uniform_bytes using expand_message =
        expand_message_xof, DST = "HashToScalar-" || contextString, and
        output length 64, interpret uniform_bytes as a 512-bit integer
        in little-endian order, and reduce the integer modulo
        Group.Order().

     ScalarInverse(s):  Returns the multiplicative inverse of input
        Scalar s mod Group.Order().

     SerializeElement(A):  Implemented using the Encode function from
        Section 5.3.2 of [RFC9496]; Ne = 56.

     DeserializeElement(buf):  Implemented using the Decode function
        from Section 5.3.1 of [RFC9496].  Additionally, this function
        validates that the resulting element is not the group identity
        element.  If these checks fail, deserialization returns an
        InputValidationError error.

     SerializeScalar(s):  Implemented by outputting the little-endian,
        56-byte encoding of the Scalar value; Ns = 56.

     DeserializeScalar(buf):  Implemented by attempting to deserialize
        a Scalar from a little-endian, 56-byte string.  This function
        can fail if the input does not represent a Scalar in the range
        [0, G.Order() - 1].

  Hash:  SHAKE-256; Nh = 64.

4.3.  OPRF(P-256, SHA-256)

  This ciphersuite uses P-256 [NISTCurves] for the Group and SHA-256
  for the hash function.  The value of the ciphersuite identifier is
  "P256-SHA256".

  Group:  P-256 (secp256r1) [NISTCurves]

     Order():  Return 0xffffffff00000000ffffffffffffffffbce6faada7179e8
        4f3b9cac2fc632551.

     Identity():  As defined in [NISTCurves].

     Generator():  As defined in [NISTCurves].

     RandomScalar():  Implemented by returning a uniformly random
        Scalar in the range [0, G.Order() - 1].  Refer to Section 4.7
        for implementation guidance.

     HashToGroup():  Use hash_to_curve with suite P256_XMD:SHA-
        256_SSWU_RO_ [RFC9380] and DST = "HashToGroup-" ||
        contextString.

     HashToScalar():  Use hash_to_field from [RFC9380] using L = 48,
        expand_message_xmd with SHA-256, DST = "HashToScalar-" ||
        contextString, and a prime modulus equal to Group.Order().

     ScalarInverse(s):  Returns the multiplicative inverse of input
        Scalar s mod Group.Order().

     SerializeElement(A):  Implemented using the compressed Elliptic-
        Curve-Point-to-Octet-String method according to [SEC1]; Ne =
        33.

     DeserializeElement(buf):  Implemented by attempting to deserialize
        a 33-byte input string to a public key using the compressed
        Octet-String-to-Elliptic-Curve-Point method according to [SEC1]
        and then performing partial public-key validation, as defined
        in Section 5.6.2.3.4 of [KEYAGREEMENT].  This includes checking
        that the coordinates of the resulting point are in the correct
        range, that the point is on the curve, and that the point is
        not the group identity element.  If these checks fail,
        deserialization returns an InputValidationError error.

     SerializeScalar(s):  Implemented using the Field-Element-to-Octet-
        String conversion according to [SEC1]; Ns = 32.

     DeserializeScalar(buf):  Implemented by attempting to deserialize
        a Scalar from a 32-byte string using Octet-String-to-Field-
        Element from [SEC1].  This function can fail if the input does
        not represent a Scalar in the range [0, G.Order() - 1].

  Hash:  SHA-256; Nh = 32.

4.4.  OPRF(P-384, SHA-384)

  This ciphersuite uses P-384 [NISTCurves] for the Group and SHA-384
  for the hash function.  The value of the ciphersuite identifier is
  "P384-SHA384".

  Group:  P-384 (secp384r1) [NISTCurves]

     Order():  Return 0xfffffffffffffffffffffffffffffffffffffffffffffff
        fc7634d81f4372ddf581a0db248b0a77aecec196accc52973.

     Identity():  As defined in [NISTCurves].

     Generator():  As defined in [NISTCurves].

     RandomScalar():  Implemented by returning a uniformly random
        Scalar in the range [0, G.Order() - 1].  Refer to Section 4.7
        for implementation guidance.

     HashToGroup():  Use hash_to_curve with suite P384_XMD:SHA-
        384_SSWU_RO_ [RFC9380] and DST = "HashToGroup-" ||
        contextString.

     HashToScalar():  Use hash_to_field from [RFC9380] using L = 72,
        expand_message_xmd with SHA-384, DST = "HashToScalar-" ||
        contextString, and a prime modulus equal to Group.Order().

     ScalarInverse(s):  Returns the multiplicative inverse of input
        Scalar s mod Group.Order().

     SerializeElement(A):  Implemented using the compressed Elliptic-
        Curve-Point-to-Octet-String method according to [SEC1]; Ne =
        49.

     DeserializeElement(buf):  Implemented by attempting to deserialize
        a 49-byte array to a public key using the compressed Octet-
        String-to-Elliptic-Curve-Point method according to [SEC1] and
        then performing partial public-key validation, as defined in
        Section 5.6.2.3.4 of [KEYAGREEMENT].  This includes checking
        that the coordinates of the resulting point are in the correct
        range, that the point is on the curve, and that the point is
        not the point at infinity.  Additionally, this function
        validates that the resulting element is not the group identity
        element.  If these checks fail, deserialization returns an
        InputValidationError error.

     SerializeScalar(s):  Implemented using the Field-Element-to-Octet-
        String conversion according to [SEC1]; Ns = 48.

     DeserializeScalar(buf):  Implemented by attempting to deserialize
        a Scalar from a 48-byte string using Octet-String-to-Field-
        Element from [SEC1].  This function can fail if the input does
        not represent a Scalar in the range [0, G.Order() - 1].

  Hash:  SHA-384; Nh = 48.

4.5.  OPRF(P-521, SHA-512)

  This ciphersuite uses P-521 [NISTCurves] for the Group and SHA-512
  for the hash function.  The value of the ciphersuite identifier is
  "P521-SHA512".

  Group:  P-521 (secp521r1) [NISTCurves]

     Order():  Return 0x01fffffffffffffffffffffffffffffffffffffffffffff
        ffffffffffffffffffffa51868783bf2f966b7fcc0148f709a5d03bb5c9b889
        9c47aebb6fb71e91386409.

     Identity():  As defined in [NISTCurves].

     Generator():  As defined in [NISTCurves].

     RandomScalar():  Implemented by returning a uniformly random
        Scalar in the range [0, G.Order() - 1].  Refer to Section 4.7
        for implementation guidance.

     HashToGroup():  Use hash_to_curve with suite P521_XMD:SHA-
        512_SSWU_RO_ [RFC9380] and DST = "HashToGroup-" ||
        contextString.

     HashToScalar():  Use hash_to_field from [RFC9380] using L = 98,
        expand_message_xmd with SHA-512, DST = "HashToScalar-" ||
        contextString, and a prime modulus equal to Group.Order().

     ScalarInverse(s):  Returns the multiplicative inverse of input
        Scalar s mod Group.Order().

     SerializeElement(A):  Implemented using the compressed Elliptic-
        Curve-Point-to-Octet-String method according to [SEC1]; Ne =
        67.

     DeserializeElement(buf):  Implemented by attempting to deserialize
        a 67-byte input string to a public key using the compressed
        Octet-String-to-Elliptic-Curve-Point method according to [SEC1]
        and then performing partial public-key validation, as defined
        in Section 5.6.2.3.4 of [KEYAGREEMENT].  This includes checking
        that the coordinates of the resulting point are in the correct
        range, that the point is on the curve, and that the point is
        not the point at infinity.  Additionally, this function
        validates that the resulting element is not the group identity
        element.  If these checks fail, deserialization returns an
        InputValidationError error.

     SerializeScalar(s):  Implemented using the Field-Element-to-Octet-
        String conversion according to [SEC1]; Ns = 66.

     DeserializeScalar(buf):  Implemented by attempting to deserialize
        a Scalar from a 66-byte string using Octet-String-to-Field-
        Element from [SEC1].  This function can fail if the input does
        not represent a Scalar in the range [0, G.Order() - 1].

  Hash:  SHA-512; Nh = 64.

4.6.  Future Ciphersuites

  A critical requirement of implementing the prime-order group using
  elliptic curves is a method to instantiate the function HashToGroup,
  which maps inputs to group elements.  In the elliptic curve setting,
  this deterministically maps inputs (as byte arrays) to uniformly
  chosen points on the curve.

  In the security proof of the construction, Hash is modeled as a
  random oracle.  This implies that any instantiation of HashToGroup
  must be pre-image and collision resistant.  In Section 4, we give
  instantiations of this functionality based on the functions described
  in [RFC9380].  Consequently, any OPRF implementation must adhere to
  the implementation and security considerations discussed in [RFC9380]
  when instantiating the function.

  The DeserializeElement and DeserializeScalar functions instantiated
  for a particular prime-order group corresponding to a ciphersuite
  MUST adhere to the description in Section 2.1.  Future ciphersuites
  MUST describe how input validation is done for DeserializeElement and
  DeserializeScalar.

  Additionally, future ciphersuites must take care when choosing the
  security level of the group.  See Section 7.2.3 for additional
  details.

4.7.  Random Scalar Generation

  Two popular algorithms for generating a random integer uniformly
  distributed in the range [0, G.Order() - 1] are described in the
  following subsections.

4.7.1.  Rejection Sampling

  Generate a random byte array with Ns bytes and attempt to map to a
  Scalar by calling DeserializeScalar in constant time.  If it
  succeeds, return the result.  If it fails, try again with another
  random byte array until the procedure succeeds.  Failure to implement
  DeserializeScalar in constant time can leak information about the
  underlying corresponding Scalar.

  As an optimization, if the group order is very close to a power of 2,
  it is acceptable to omit the rejection test completely.  In
  particular, if the group order is p and there is an integer b such
  that |p - 2^b| is less than 2^(b/2), then RandomScalar can simply
  return a uniformly random integer of at most b bits.

4.7.2.  Random Number Generation Using Extra Random Bits

  Generate a random byte array with L = ceil(((3 *
  ceil(log2(G.Order()))) / 2) / 8) bytes, and interpret it as an
  integer; reduce the integer modulo G.Order(), and return the result.
  See [RFC9380], Section 5 for the underlying derivation of L.

5.  Application Considerations

  This section describes considerations for applications, including
  external interface recommendations, explicit error treatment, and
  public input representation for the POPRF protocol variant.

5.1.  Input Limits

  Application inputs, expressed as PrivateInput or PublicInput values,
  MUST be smaller than 2^16 - 1 bytes in length.  Applications that
  require longer inputs can use a cryptographic hash function to map
  these longer inputs to a fixed-length input that fits within the
  PublicInput or PrivateInput length bounds.  Note that some
  cryptographic hash functions have input length restrictions
  themselves, but these limits are often large enough to not be a
  concern in practice.  For example, SHA-256 has an input limit of 2^61
  bytes.

5.2.  External Interface Recommendations

  In Section 3.3, the interface of the protocol functions allows that
  some inputs (and outputs) to be group Element and Scalar values.
  However, implementations can instead operate over Element and Scalar
  values internally and only expose interfaces that operate with an
  application-specific format of messages.

5.3.  Error Considerations

  Some OPRF variants specified in this document have fallible
  operations.  For example, Finalize and BlindEvaluate can fail if any
  element received from the peer fails input validation.  The explicit
  errors generated throughout this specification, along with the
  conditions that lead to each error, are as follows:

  VerifyError:  Verifiable OPRF proof verification failed (Sections
     3.3.2 and 3.3.3).

  DeserializeError:  Group Element or Scalar deserialization failure
     (Sections 2.1 and 3.3).

  InputValidationError:  Validation of byte array inputs failed
     (Section 4).

  There are other explicit errors generated in this specification;
  however, they occur with negligible probability in practice.  We note
  them here for completeness.

  InvalidInputError:  OPRF Blind input produces an invalid output
     element (Sections 3.3.1 and 3.3.3).

  InverseError:  A tweaked private key is invalid, i.e., has no
     multiplicative inverse (Sections 2.1 and 3.3).

  In general, the errors in this document are meant as a guide to
  implementors.  They are not an exhaustive list of all the errors an
  implementation might emit.  For example, implementations might run
  out of memory and return a corresponding error.

5.4.  POPRF Public Input

  Functionally, the VOPRF and POPRF variants differ in that the POPRF
  variant admits public input, whereas the VOPRF variant does not.
  Public input allows clients and servers to cryptographically bind
  additional data to the POPRF output.  A POPRF with fixed public input
  is functionally equivalent to a VOPRF.  However, there are
  differences in the underlying security assumptions made about each
  variant; see Section 7.2 for more details.

  This public input is known to both parties at the start of the
  protocol.  It is RECOMMENDED that this public input be constructed
  with some type of higher-level domain separation to avoid cross
  protocol attacks or related issues.  For example, protocols using
  this construction might ensure that the public input uses a unique,
  prefix-free encoding.  See [RFC9380], Section 10.4 for further
  discussion on constructing domain separation values.

  Implementations of the POPRF may choose to not let applications
  control info in cases where this value is fixed or otherwise not
  useful to the application.  In this case, the resulting protocol is
  functionally equivalent to the VOPRF, which does not admit public
  input.

6.  IANA Considerations

  This document has no IANA actions.

7.  Security Considerations

  This section discusses the security of the protocols defined in this
  specification, along with some suggestions and trade-offs that arise
  from the implementation of the protocol variants in this document.
  Note that the syntax of the POPRF variant is different from that of
  the OPRF and VOPRF variants since it admits an additional public
  input, but the same security considerations apply.

7.1.  Security Properties

  The security properties of an OPRF protocol with functionality y =
  F(k, x) include those of a standard PRF.  Specifically:

  Pseudorandomness:  For a random sampling of k, F is pseudorandom if
     the output y = F(k, x) on any input x is indistinguishable from
     uniformly sampling any element in F's range.

  In other words, consider an adversary that picks inputs x from the
  domain of F and evaluates F on (k, x) (without knowledge of randomly
  sampled k).  Then, the output distribution F(k, x) is
  indistinguishable from the output distribution of a randomly chosen
  function with the same domain and range.

  A consequence of showing that a function is pseudorandom is that it
  is necessarily nonmalleable (i.e., we cannot compute a new evaluation
  of F from an existing evaluation).  A genuinely random function will
  be nonmalleable with high probability, so a pseudorandom function
  must be nonmalleable to maintain indistinguishability.

  Unconditional input secrecy:  The server does not learn anything
     about the client input x, even with unbounded computation.

  In other words, an attacker with infinite computing power cannot
  recover any information about the client's private input x from an
  invocation of the protocol.

  Essentially, input secrecy is the property that, even if the server
  learns the client's private input x at some point in the future, the
  server cannot link any particular PRF evaluation to x.  This property
  is also known as unlinkability [DGSTV18].

  Beyond client input secrecy, in the OPRF protocol, the server learns
  nothing about the output y of the function, nor does the client learn
  anything about the server's private key k.

  For the VOPRF and POPRF protocol variants, there is an additional
  security property:

  Verifiable:  The client must only complete execution of the protocol
     if it can successfully assert that the output it computes is
     correct.  This is taken with respect to the private key held by
     the server.

  Any VOPRF or POPRF that satisfies the 'verifiable' security property
  is known as 'verifiable'.  In practice, the notion of verifiability
  requires that the server commits to the key before the actual
  protocol execution takes place.  Then, the client verifies that the
  server has used the key in the protocol using this commitment.  In
  the following, we may also refer to this commitment as a public key.

  Finally, the POPRF variant also has the following security property:

  Partial obliviousness:  The client and server must be able to perform
     the PRF on the client's private and public input.  Both the client
     and server know the public input, but similar to the OPRF and
     VOPRF protocols, the server learns nothing about the client's
     private input or the output of the function, and the client learns
     nothing about the server's private key.

  This property becomes useful when dealing with key management
  operations, such as the rotation of the server's keys.  Note that
  partial obliviousness only applies to the POPRF variant because
  neither the OPRF nor VOPRF variants accept public input to the
  protocol.

  Since the POPRF variant has a different syntax than the OPRF and
  VOPRF variants, i.e., y = F(k, x, info), the pseudorandomness
  property is generalized:

  Pseudorandomness:  For a random sampling of k, F is pseudorandom if
     the output y = F(k, x, info) on any input pairs (x, info) is
     indistinguishable from uniformly sampling any element in F's
     range.

7.2.  Security Assumptions

  Below, we discuss the cryptographic security of each protocol variant
  from Section 3, relative to the necessary cryptographic assumptions
  that need to be made.

7.2.1.  OPRF and VOPRF Assumptions

  The OPRF and VOPRF protocol variants in this document are based on
  [JKK14].  In particular, the VOPRF construction is similar to the
  [JKK14] construction with the following distinguishing properties:

  1.  This document does not use session identifiers to differentiate
      different instances of the protocol.

  2.  This document supports batching so that multiple evaluations can
      happen at once whilst only constructing one DLEQ proof object.
      This is enabled using an established batching technique
      [DGSTV18].

  The pseudorandomness and input secrecy (and verifiability) of the
  OPRF (and VOPRF) protocols in [JKK14] are based on the One-More Gap
  Computational Diffie-Hellman assumption that is computationally
  difficult to solve in the corresponding prime-order group.  In
  [JKK14], these properties are proven for one instance (i.e., one key)
  of the VOPRF protocol and without batching.  There is currently no
  security analysis available for the VOPRF protocol described in this
  document in a setting with multiple server keys or batching.

7.2.2.  POPRF Assumptions

  The POPRF construction in this document is based on the construction
  known as 3HashSDHI, given by [TCRSTW21].  The construction is
  identical to 3HashSDHI, except that this design can optionally
  perform multiple POPRF evaluations in one batch, whilst only
  constructing one DLEQ proof object.  This is enabled using an
  established batching technique [DGSTV18].

  Pseudorandomness, input secrecy, verifiability, and partial
  obliviousness of the POPRF variant is based on the assumption that
  the One-More Gap Strong Diffie-Hellman Inversion (SDHI) assumption
  from [TCRSTW21] is computationally difficult to solve in the
  corresponding prime-order group.  Tyagi et al. [TCRSTW21] show that
  both the One-More Gap Computational Diffie-Hellman assumption and the
  One-More Gap SDHI assumption reduce to the q-DL (Discrete Log)
  assumption in the algebraic group model for some q number of
  BlindEvaluate queries.  (The One-More Gap Computational Diffie-
  Hellman assumption was the hardness assumption used to evaluate the
  OPRF and VOPRF designs based on [JKK14], which is a predecessor to
  the POPRF variant in Section 3.3.3.)

7.2.3.  Static Diffie-Hellman Attack and Security Limits

  A side effect of the OPRF protocol variants in this document is that
  they allow instantiation of an oracle for constructing static Diffie-
  Hellman (DH) samples; see [BG04] and [Cheon06].  These attacks are
  meant to recover (bits of) the server private key.  Best-known
  attacks reduce the security of the prime-order group instantiation by
  log_2(Q) / 2 bits, where Q is the number of BlindEvaluate calls made
  by the attacker.

  As a result of this class of attacks, choosing prime-order groups
  with a 128-bit security level instantiates an OPRF with a reduced
  security level of 128 - (log_2(Q) / 2) bits of security.  Moreover,
  such attacks are only possible for those certain applications where
  the adversary can query the OPRF directly.  Applications can mitigate
  against this problem in a variety of ways, e.g., by rate-limiting
  client queries to BlindEvaluate or by rotating private keys.  In
  applications where such an oracle is not made available, this
  security loss does not apply.

  In most cases, it would require an informed and persistent attacker
  to launch a highly expensive attack to reduce security to anything
  much below 100 bits of security.  Applications that admit the
  aforementioned oracle functionality and that cannot tolerate discrete
  logarithm security of lower than 128 bits are RECOMMENDED to choose
  groups that target a higher security level, such as decaf448 (used by
  ciphersuite decaf448-SHAKE256), P-384 (used by ciphersuite
  P384-SHA384), or P-521 (used by ciphersuite P521-SHA512).

7.3.  Domain Separation

  Applications SHOULD construct input to the protocol to provide domain
  separation.  Any system that has multiple OPRF applications should
  distinguish client inputs to ensure the OPRF results are separate.
  Guidance for constructing info can be found in [RFC9380],
  Section 3.1.

7.4.  Timing Leaks

  To ensure no information is leaked during protocol execution, all
  operations that use secret data MUST run in constant time.  This
  includes all prime-order group operations and proof-specific
  operations that operate on secret data, including GenerateProof and
  BlindEvaluate.

8.  References

8.1.  Normative References

  [KEYAGREEMENT]
             Barker, E., Chen, L., Roginsky, A., Vassilev, A., and R.
             Davis, "Recommendation for pair-wise key-establishment
             schemes using discrete logarithm cryptography", NIST
             SP 800-56A (Rev. 3), DOI 10.6028/nist.sp.800-56ar3, April
             2018, <https://doi.org/10.6028/nist.sp.800-56ar3>.

  [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
             Requirement Levels", BCP 14, RFC 2119,
             DOI 10.17487/RFC2119, March 1997,
             <https://www.rfc-editor.org/info/rfc2119>.

  [RFC8017]  Moriarty, K., Ed., Kaliski, B., Jonsson, J., and A. Rusch,
             "PKCS #1: RSA Cryptography Specifications Version 2.2",
             RFC 8017, DOI 10.17487/RFC8017, November 2016,
             <https://www.rfc-editor.org/info/rfc8017>.

  [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
             2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
             May 2017, <https://www.rfc-editor.org/info/rfc8174>.

  [RFC9380]  Faz-Hernandez, A., Scott, S., Sullivan, N., Wahby, R. S.,
             and C. A. Wood, "Hashing to Elliptic Curves", RFC 9380,
             DOI 10.17487/RFC9380, August 2023,
             <https://www.rfc-editor.org/info/rfc9380>.

  [RFC9496]  de Valence, H., Grigg, J., Hamburg, M., Lovecruft, I.,
             Tankersley, G., and F. Valsorda, "The ristretto255 and
             decaf448 Groups", RFC 9496, DOI 10.17487/RFC9496, December
             2023, <https://www.rfc-editor.org/info/rfc9496>.

8.2.  Informative References

  [BG04]     Brown, D. and R. Gallant, "The Static Diffie-Hellman
             Problem", November 2004,
             <https://eprint.iacr.org/2004/306>.

  [ChaumPedersen]
             Chaum, D. and T. Pedersen, "Wallet Databases with
             Observers", Advances in Cryptology - CRYPTO' 92, pp.
             89-105, DOI 10.1007/3-540-48071-4_7, August 1992,
             <https://doi.org/10.1007/3-540-48071-4_7>.

  [Cheon06]  Cheon, J., "Security Analysis of the Strong Diffie-Hellman
             Problem", Advances in Cryptology - EUROCRYPT 2006, pp.
             1-11, DOI 10.1007/11761679_1, 2006,
             <https://doi.org/10.1007/11761679_1>.

  [DGSTV18]  Davidson, A., Goldberg, I., Sullivan, N., Tankersley, G.,
             and F. Valsorda, "Privacy Pass: Bypassing Internet
             Challenges Anonymously", Proceedings on Privacy Enhancing
             Technologies, vol. 2018, no. 3, pp. 164-180, DOI
             10.1515/popets-2018-0026, April 2018,
             <https://doi.org/10.1515/popets-2018-0026>.

  [FS00]     Fiat, A. and A. Shamir, "How To Prove Yourself: Practical
             Solutions to Identification and Signature Problems",
             Advances in Cryptology - CRYPTO' 86, pp. 186-194,
             DOI 10.1007/3-540-47721-7_12, 1986,
             <https://doi.org/10.1007/3-540-47721-7_12>.

  [JKK14]    Jarecki, S., Kiayias, A., and H. Krawczyk, "Round-Optimal
             Password-Protected Secret Sharing and T-PAKE in the
             Password-Only Model", Lecture Notes in Computer Science,
             pp. 233-253, DOI 10.1007/978-3-662-45608-8_13, 2014,
             <https://doi.org/10.1007/978-3-662-45608-8_13>.

  [JKKX16]   Jarecki, S., Kiayias, A., Krawczyk, H., and J. Xu,
             "Highly-Efficient and Composable Password-Protected Secret
             Sharing (Or: How to Protect Your Bitcoin Wallet Online)",
             2016 IEEE European Symposium on Security and Privacy
             (EuroS&P), DOI 10.1109/eurosp.2016.30, March 2016,
             <https://doi.org/10.1109/eurosp.2016.30>.

  [NISTCurves]
             National Institute of Standards and Technology (NIST),
             "Digital Signature Standard (DSS)", FIPS PUB 186-5,
             DOI 10.6028/NIST.FIPS.186-5, February 2023,
             <https://doi.org/10.6028/NIST.FIPS.186-5>.

  [OPAQUE]   Bourdrez, D., Krawczyk, H., Lewi, K., and C. A. Wood, "The
             OPAQUE Asymmetric PAKE Protocol", Work in Progress,
             Internet-Draft, draft-irtf-cfrg-opaque-13, 18 December
             2023, <https://datatracker.ietf.org/doc/html/draft-irtf-
             cfrg-opaque-13>.

  [PRIVACY-PASS]
             Celi, S., Davidson, A., Valdez, S., and C. A. Wood,
             "Privacy Pass Issuance Protocol", Work in Progress,
             Internet-Draft, draft-ietf-privacypass-protocol-16, 3
             October 2023, <https://datatracker.ietf.org/doc/html/
             draft-ietf-privacypass-protocol-16>.

  [PrivacyPass]
             "Privacy Pass", commit 085380a, March 2018,
             <https://github.com/privacypass/team>.

  [RFC7748]  Langley, A., Hamburg, M., and S. Turner, "Elliptic Curves
             for Security", RFC 7748, DOI 10.17487/RFC7748, January
             2016, <https://www.rfc-editor.org/info/rfc7748>.

  [SEC1]     Standards for Efficient Cryptography Group (SECG), "SEC 1:
             Elliptic Curve Cryptography", May 2009,
             <https://www.secg.org/sec1-v2.pdf>.

  [SJKS17]   Shirvanian, M., Jarecki, S., Krawczyk, H., and N. Saxena,
             "SPHINX: A Password Store that Perfectly Hides Passwords
             from Itself", 2017 IEEE 37th International Conference on
             Distributed Computing Systems (ICDCS),
             DOI 10.1109/ICDCS.2017.64, June 2017,
             <https://doi.org/10.1109/ICDCS.2017.64>.

  [TCRSTW21] Tyagi, N., Celi, S., Ristenpart, T., Sullivan, N.,
             Tessaro, S., and C. A. Wood, "A Fast and Simple Partially
             Oblivious PRF, with Applications", Advances in Cryptology
             - EUROCRYPT 2022 pp. 674-705,
             DOI 10.1007/978-3-031-07085-3_23, May 2022,
             <https://doi.org/10.1007/978-3-031-07085-3_23>.

Appendix A.  Test Vectors

  This section includes test vectors for the protocol variants
  specified in this document.  For each ciphersuite specified in
  Section 4, there is a set of test vectors for the protocol when
  running the OPRF, VOPRF, and POPRF modes.  Each test vector lists the
  batch size for the evaluation.  Each test vector value is encoded as
  a hexadecimal byte string.  The fields of each test vector are
  described below.

  "Input":  The private client input, an opaque byte string.

  "Info":  The public info, an opaque byte string.  Only present for
     POPRF test vectors.

  "Blind":  The blind value output by Blind(), a serialized Scalar of
     Ns bytes long.

  "BlindedElement":  The blinded value output by Blind(), a serialized
     Element of Ne bytes long.

  "EvaluatedElement":  The evaluated element output by BlindEvaluate(),
     a serialized Element of Ne bytes long.

  "Proof":  The serialized Proof output from GenerateProof() composed
     of two serialized Scalar values, each Ns bytes long.  Only present
     for VOPRF and POPRF test vectors.

  "ProofRandomScalar":  The random Scalar r computed in
     GenerateProof(), a serialized Scalar of Ns bytes long.  Only
     present for VOPRF and POPRF test vectors.

  "Output":  The protocol output, an opaque byte string of Nh bytes
     long.

  Test vectors with batch size B > 1 have inputs separated by a comma
  ",".  Applicable test vectors will have B different values for the
  "Input", "Blind", "BlindedElement", "EvaluationElement", and "Output"
  fields.

  The server key material, pkSm and skSm, are listed under the mode for
  each ciphersuite.  Both pkSm and skSm are the serialized values of
  pkS and skS, respectively, as used in the protocol.  Each key pair is
  derived from a seed, denoted Seed, and info string, denoted KeyInfo,
  which are listed as well, using the DeriveKeyPair function from
  Section 3.2.

A.1.  ristretto255-SHA512

A.1.1.  OPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 5ebcea5ee37023ccb9fc2d2019f9d7737be85591ae8652ffa9ef0f4d37063
  b0e

A.1.1.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = 609a0ae68c15a3cf6903766461307e5c8bb2f95e7e6550e1ffa
  2dc99e412803c
  EvaluationElement = 7ec6578ae5120958eb2db1745758ff379e77cb64fe77b0b2
  d8cc917ea0869c7e
  Output = 527759c3d9366f277d8c6020418d96bb393ba2afb20ff90df23fb770826
  4e2f3ab9135e3bd69955851de4b1f9fe8a0973396719b7912ba9ee8aa7d0b5e24bcf
  6

A.1.1.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = da27ef466870f5f15296299850aa088629945a17d1f5b7f5ff0
  43f76b3c06418
  EvaluationElement = b4cbf5a4f1eeda5a63ce7b77c7d23f461db3fcab0dd28e4e
  17cecb5c90d02c25
  Output = f4a74c9c592497375e796aa837e907b1a045d34306a749db9f34221f7e7
  50cb4f2a6413a6bf6fa5e19ba6348eb673934a722a7ede2e7621306d18951e7cf2c7
  3

A.1.2.  VOPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = e6f73f344b79b379f1a0dd37e07ff62e38d9f71345ce62ae3a9bc60b04ccd
  909
  pkSm = c803e2cc6b05fc15064549b5920659ca4a77b2cca6f04f6b357009335476a
  d4e

A.1.2.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = 863f330cc1a1259ed5a5998a23acfd37fb4351a793a5b3c090b
  642ddc439b945
  EvaluationElement = aa8fa048764d5623868679402ff6108d2521884fa138cd7f
  9c7669a9a014267e
  Proof = ddef93772692e535d1a53903db24367355cc2cc78de93b3be5a8ffcc6985
  dd066d4346421d17bf5117a2a1ff0fcb2a759f58a539dfbe857a40bce4cf49ec600d
  ProofRandomScalar = 222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d98
  81aa6f61d645fc0e
  Output = b58cfbe118e0cb94d79b5fd6a6dafb98764dff49c14e1770b566e42402d
  a1a7da4d8527693914139caee5bd03903af43a491351d23b430948dd50cde10d32b3
  c

A.1.2.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = cc0b2a350101881d8a4cba4c80241d74fb7dcbfde4a61fde2f9
  1443c2bf9ef0c
  EvaluationElement = 60a59a57208d48aca71e9e850d22674b611f752bed48b36f
  7a91b372bd7ad468
  Proof = 401a0da6264f8cf45bb2f5264bc31e109155600babb3cd4e5af7d181a2c9
  dc0a67154fabf031fd936051dec80b0b6ae29c9503493dde7393b722eafdf5a50b02
  ProofRandomScalar = 222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d98
  81aa6f61d645fc0e
  Output = 8a9a2f3c7f085b65933594309041fc1898d42d0858e59f90814ae90571a
  6df60356f4610bf816f27afdd84f47719e480906d27ecd994985890e5f539e7ea74b
  6

A.1.2.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706,222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d9881aa6f61d645fc0
  e
  BlindedElement = 863f330cc1a1259ed5a5998a23acfd37fb4351a793a5b3c090b
  642ddc439b945,90a0145ea9da29254c3a56be4fe185465ebb3bf2a1801f7124bbba
  dac751e654
  EvaluationElement = aa8fa048764d5623868679402ff6108d2521884fa138cd7f
  9c7669a9a014267e,cc5ac221950a49ceaa73c8db41b82c20372a4c8d63e5dded2db
  920b7eee36a2a
  Proof = cc203910175d786927eeb44ea847328047892ddf8590e723c37205cb7460
  0b0a5ab5337c8eb4ceae0494c2cf89529dcf94572ed267473d567aeed6ab873dee08
  ProofRandomScalar = 419c4f4f5052c53c45f3da494d2b67b220d02118e0857cdb
  cf037f9ea84bbe0c
  Output = b58cfbe118e0cb94d79b5fd6a6dafb98764dff49c14e1770b566e42402d
  a1a7da4d8527693914139caee5bd03903af43a491351d23b430948dd50cde10d32b3
  c,8a9a2f3c7f085b65933594309041fc1898d42d0858e59f90814ae90571a6df6035
  6f4610bf816f27afdd84f47719e480906d27ecd994985890e5f539e7ea74b6

A.1.3.  POPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 145c79c108538421ac164ecbe131942136d5570b16d8bf41a24d4337da981
  e07
  pkSm = c647bef38497bc6ec077c22af65b696efa43bff3b4a1975a3e8e0a1c5a79d
  631

A.1.3.1.  Test Vector 1, Batch Size 1

  Input = 00
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = c8713aa89241d6989ac142f22dba30596db635c772cbf25021f
  dd8f3d461f715
  EvaluationElement = 1a4b860d808ff19624731e67b5eff20ceb2df3c3c03b906f
  5693e2078450d874
  Proof = 41ad1a291aa02c80b0915fbfbb0c0afa15a57e2970067a602ddb9e8fd6b7
  100de32e1ecff943a36f0b10e3dae6bd266cdeb8adf825d86ef27dbc6c0e30c52206
  ProofRandomScalar = 222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d98
  81aa6f61d645fc0e
  Output = ca688351e88afb1d841fde4401c79efebb2eb75e7998fa9737bd5a82a15
  2406d38bd29f680504e54fd4587eddcf2f37a2617ac2fbd2993f7bdf45442ace7d22
  1

A.1.3.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706
  BlindedElement = f0f0b209dd4d5f1844dac679acc7761b91a2e704879656cb7c2
  01e82a99ab07d
  EvaluationElement = 8c3c9d064c334c6991e99f286ea2301d1bde170b54003fb9
  c44c6d7bd6fc1540
  Proof = 4c39992d55ffba38232cdac88fe583af8a85441fefd7d1d4a8d0394cd1de
  77018bf135c174f20281b3341ab1f453fe72b0293a7398703384bed822bfdeec8908
  ProofRandomScalar = 222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d98
  81aa6f61d645fc0e
  Output = 7c6557b276a137922a0bcfc2aa2b35dd78322bd500235eb6d6b6f91bc5b
  56a52de2d65612d503236b321f5d0bebcbc52b64b92e426f29c9b8b69f52de98ae50
  7

A.1.3.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec4c1f
  6706,222a5e897cf59db8145db8d16e597e8facb80ae7d4e26d9881aa6f61d645fc0
  e
  BlindedElement = c8713aa89241d6989ac142f22dba30596db635c772cbf25021f
  dd8f3d461f715,423a01c072e06eb1cce96d23acce06e1ea64a609d7ec9e9023f304
  9f2d64e50c
  EvaluationElement = 1a4b860d808ff19624731e67b5eff20ceb2df3c3c03b906f
  5693e2078450d874,aa1f16e903841036e38075da8a46655c94fc92341887eb5819f
  46312adfc0504
  Proof = 43fdb53be399cbd3561186ae480320caa2b9f36cca0e5b160c4a677b8bbf
  4301b28f12c36aa8e11e5a7ef551da0781e863a6dc8c0b2bf5a149c9e00621f02006
  ProofRandomScalar = 419c4f4f5052c53c45f3da494d2b67b220d02118e0857cdb
  cf037f9ea84bbe0c
  Output = ca688351e88afb1d841fde4401c79efebb2eb75e7998fa9737bd5a82a15
  2406d38bd29f680504e54fd4587eddcf2f37a2617ac2fbd2993f7bdf45442ace7d22
  1,7c6557b276a137922a0bcfc2aa2b35dd78322bd500235eb6d6b6f91bc5b56a52de
  2d65612d503236b321f5d0bebcbc52b64b92e426f29c9b8b69f52de98ae507

A.2.  decaf448-SHAKE256

A.2.1.  OPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = e8b1375371fd11ebeb224f832dcc16d371b4188951c438f751425699ed29e
  cc80c6c13e558ccd67634fd82eac94aa8d1f0d7fee990695d1e

A.2.1.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = e0ae01c4095f08e03b19baf47ffdc19cb7d98e583160522a3c7
  d6a0b2111cd93a126a46b7b41b730cd7fc943d4e28e590ed33ae475885f6c
  EvaluationElement = 50ce4e60eed006e22e7027454b5a4b8319eb2bc8ced609eb
  19eb3ad42fb19e06ba12d382cbe7ae342a0cad6ead0ef8f91f00bb7f0cd9c0a2
  Output = 37d3f7922d9388a15b561de5829bbf654c4089ede89c0ce0f3f85bcdba0
  9e382ce0ab3507e021f9e79706a1798ffeac68ebd5cf62e5eb9838c7068351d97ae3
  7

A.2.1.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = 86a88dc5c6331ecfcb1d9aacb50a68213803c462e377577cacc
  00af28e15f0ddbc2e3d716f2f39ef95f3ec1314a2c64d940a9f295d8f13bb
  EvaluationElement = 162e9fa6e9d527c3cd734a31bf122a34dbd5bcb7bb23651f
  1768a7a9274cc116c03b58afa6f0dede3994a60066c76370e7328e7062fd5819
  Output = a2a652290055cb0f6f8637a249ee45e32ef4667db0b4c80c0a70d2a6416
  4d01525cfdad5d870a694ec77972b9b6ec5d2596a5223e5336913f945101f0137f55
  e

A.2.2.  VOPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = e3c01519a076a326a0eb566343e9b21c115fa18e6e85577ddbe890b33104f
  cc2835ddfb14a928dc3f5d79b936e17c76b99e0bf6a1680930e
  pkSm = 945fc518c47695cf65217ace04b86ac5e4cbe26ca649d52854bb16c494ce0
  9069d6add96b20d4b0ae311a87c9a73e3a146b525763ab2f955

A.2.2.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = 7261bbc335c664ba788f1b1a1a4cd5190cc30e787ef277665ac
  1d314f8861e3ec11854ce3ddd42035d9e0f5cddde324c332d8c880abc00eb
  EvaluationElement = ca1491a526c28d880806cf0fb0122222392cf495657be6e4
  c9d203bceffa46c86406caf8217859d3fb259077af68e5d41b3699410781f467
  Proof = f84bbeee47aedf43558dae4b95b3853635a9fc1a9ea7eac9b454c64c66c4
  f49cd1c72711c7ac2e06c681e16ea693d5500bbd7b56455df52f69e00b76b4126961
  e1562fdbaaac40b7701065cbeece3febbfe09e00160f81775d36daed99d8a2a10be0
  759e01b7ee81217203416c9db208
  ProofRandomScalar = b1b748135d405ce48c6973401d9455bb8ccd18b01d0295c0
  627f67661200dbf9569f73fbb3925daa043a070e5f953d80bb464ea369e5522b
  Output = e2ac40b634f36cccd8262b285adff7c9dcc19cd308564a5f4e581d1a853
  5773b86fa4fc9f2203c370763695c5093aea4a7aedec4488b1340ba3bf663a23098c
  1

A.2.2.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = 88287e553939090b888ddc15913e1807dc4757215555e1c3a79
  488ef311594729c7fa74c772a732b78440b7d66d0aa35f3bb316f1d93e1b2
  EvaluationElement = c00978c73e8e4ee1d447ab0d3ad1754055e72cc85c08e3a0
  db170909a9c61cbff1f1e7015f289e3038b0f341faea5d7780c130106065c231
  Proof = 7a2831a6b237e11ac1657d440df93bc5ce00f552e6020a99d5c956ffc4d0
  7b5ade3e82ecdc257fd53d76239e733e0a1313e84ce16cc0d82734806092a693d7e8
  d3c420c2cb6ccd5d0ca32514fb78e9ad0973ebdcb52eba438fc73948d76339ee7101
  21d83e2fe6f001cfdf551aff9f36
  ProofRandomScalar = b1b748135d405ce48c6973401d9455bb8ccd18b01d0295c0
  627f67661200dbf9569f73fbb3925daa043a070e5f953d80bb464ea369e5522b
  Output = 862952380e07ec840d9f6e6f909c5a25d16c3dacb586d89a181b4aa7380
  c959baa8c480fe8e6c64e089d68ea7aeeb5817bd524d7577905b5bab487690048c94
  1

A.2.2.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112,b1b748135d405ce
  48c6973401d9455bb8ccd18b01d0295c0627f67661200dbf9569f73fbb3925daa043
  a070e5f953d80bb464ea369e5522b
  BlindedElement = 7261bbc335c664ba788f1b1a1a4cd5190cc30e787ef277665ac
  1d314f8861e3ec11854ce3ddd42035d9e0f5cddde324c332d8c880abc00eb,2e15f3
  93c035492a1573627a3606e528c6294c767c8d43b8c691ef70a52cc7dc7d1b53fe45
  8350a270abb7c231b87ba58266f89164f714d9
  EvaluationElement = ca1491a526c28d880806cf0fb0122222392cf495657be6e4
  c9d203bceffa46c86406caf8217859d3fb259077af68e5d41b3699410781f467,8ec
  68e9871b296e81c55647ce64a04fe75d19932f1400544cd601468c60f998408bbb54
  6601d4a636e8be279e558d70b95c8d4a4f61892be
  Proof = 167d922f0a6ffa845eed07f8aa97b6ac746d902ecbeb18f49c009adc0521
  eab1e4d275b74a2dc266b7a194c854e85e7eb54a9a36376dfc04ec7f3bd55fc9618c
  3970cb548e064f8a2f06183a5702933dbc3e4c25a73438f2108ee1981c306181003c
  7ea92fce963ec7b4ba4f270e6d38
  ProofRandomScalar = 63798726803c9451ba405f00ef3acb633ddf0c420574a2ec
  6cbf28f840800e355c9fbaac10699686de2724ed22e797a00f3bd93d105a7f23
  Output = e2ac40b634f36cccd8262b285adff7c9dcc19cd308564a5f4e581d1a853
  5773b86fa4fc9f2203c370763695c5093aea4a7aedec4488b1340ba3bf663a23098c
  1,862952380e07ec840d9f6e6f909c5a25d16c3dacb586d89a181b4aa7380c959baa
  8c480fe8e6c64e089d68ea7aeeb5817bd524d7577905b5bab487690048c941

A.2.3.  POPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 792a10dcbd3ba4a52a054f6f39186623208695301e7adb9634b74709ab22d
  e402990eb143fd7c67ac66be75e0609705ecea800992aac8e19
  pkSm = 6c9d12723a5bbcf305522cc04b4a34d9ced2e12831826018ea7b5dcf54526
  47ad262113059bf0f6e4354319951b9d513c74f29cb0eec38c1

A.2.3.1.  Test Vector 1, Batch Size 1

  Input = 00
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = 161183c13c6cb33b0e4f9b7365f8c5c12d13c72f8b62d276ca0
  9368d093dce9b42198276b9e9d870ac392dda53efd28d1b7e6e8c060cdc42
  EvaluationElement = 06ec89dfde25bb2a6f0145ac84b91ac277b35de39ad1d6f4
  02a8e46414952ce0d9ea1311a4ece283e2b01558c7078b040cfaa40dd63b3e6c
  Proof = 66caee75bf2460429f620f6ad3e811d524cb8ddd848a435fc5d89af48877
  abf6506ee341a0b6f67c2d76cd021e5f3d1c9abe5aa9f0dce016da746135fedba2af
  41ed1d01659bfd6180d96bc1b7f320c0cb6926011ce392ecca748662564892bae665
  16acaac6ca39aadf6fcca95af406
  ProofRandomScalar = b1b748135d405ce48c6973401d9455bb8ccd18b01d0295c0
  627f67661200dbf9569f73fbb3925daa043a070e5f953d80bb464ea369e5522b
  Output = 4423f6dcc1740688ea201de57d76824d59cd6b859e1f9884b7eebc49b0b
  971358cf9cb075df1536a8ea31bcf55c3e31c2ba9cfa8efe54448d17091daeb9924e
  d

A.2.3.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112
  BlindedElement = 12082b6a381c6c51e85d00f2a3d828cdeab3f5cb19a10b9c014
  c33826764ab7e7cfb8b4ff6f411bddb2d64e62a472af1cd816e5b712790c6
  EvaluationElement = f2919b7eedc05ab807c221fce2b12c4ae9e19e6909c47845
  64b690d1972d2994ca623f273afc67444d84ea40cbc58fcdab7945f321a52848
  Proof = a295677c54d1bc4286330907fc2490a7de163da26f9ce03a462a452fea42
  2b19ade296ba031359b3b6841e48455d20519ad01b4ac4f0b92e76d3cf16fbef0a3f
  72791a8401ef2d7081d361e502e96b2c60608b9fa566f43d4611c2f161d83aabef7f
  8017332b26ed1daaf80440772022
  ProofRandomScalar = b1b748135d405ce48c6973401d9455bb8ccd18b01d0295c0
  627f67661200dbf9569f73fbb3925daa043a070e5f953d80bb464ea369e5522b
  Output = 8691905500510843902c44bdd9730ab9dc3925aa58ff9dd42765a2baf63
  3126de0c3adb93bef5652f38e5827b6396e87643960163a560fc4ac9738c8de4e4a8
  d

A.2.3.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 64d37aed22a27f5191de1c1d69fadb899d8862b58eb4220029e036ec65fa
  3833a26e9388336361686ff1f83df55046504dfecad8549ba112,b1b748135d405ce
  48c6973401d9455bb8ccd18b01d0295c0627f67661200dbf9569f73fbb3925daa043
  a070e5f953d80bb464ea369e5522b
  BlindedElement = 161183c13c6cb33b0e4f9b7365f8c5c12d13c72f8b62d276ca0
  9368d093dce9b42198276b9e9d870ac392dda53efd28d1b7e6e8c060cdc42,fc8847
  d43fb4cea4e408f585661a8f2867533fa91d22155d3127a22f18d3b007add480f7d3
  00bca93fa47fe87ae06a57b7d0f0d4c30b12f0
  EvaluationElement = 06ec89dfde25bb2a6f0145ac84b91ac277b35de39ad1d6f4
  02a8e46414952ce0d9ea1311a4ece283e2b01558c7078b040cfaa40dd63b3e6c,2e7
  4c626d07de49b1c8c21d87120fd78105f485e36816af9bde3e3efbeef76815326062
  fd333925b66c5ce5a20f100bf01770c16609f990a
  Proof = fd94db736f97ea4efe9d0d4ad2933072697a6bbeb32834057b23edf7c700
  9f011dfa72157f05d2a507c2bbf0b54cad99ab99de05921c021fda7d70e65bcecdb0
  5f9a30154127ace983c74d10fd910b554c5e95f6bd1565fd1f3dbbe3c523ece5c72d
  57a559b7be1368c4786db4a3c910
  ProofRandomScalar = 63798726803c9451ba405f00ef3acb633ddf0c420574a2ec
  6cbf28f840800e355c9fbaac10699686de2724ed22e797a00f3bd93d105a7f23
  Output = 4423f6dcc1740688ea201de57d76824d59cd6b859e1f9884b7eebc49b0b
  971358cf9cb075df1536a8ea31bcf55c3e31c2ba9cfa8efe54448d17091daeb9924e
  d,8691905500510843902c44bdd9730ab9dc3925aa58ff9dd42765a2baf633126de0
  c3adb93bef5652f38e5827b6396e87643960163a560fc4ac9738c8de4e4a8d

A.3.  P256-SHA256

A.3.1.  OPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 159749d750713afe245d2d39ccfaae8381c53ce92d098a9375ee70739c7ac
  0bf

A.3.1.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 03723a1e5c09b8b9c18d1dcbca29e8007e95f14f4732d9346d4
  90ffc195110368d
  EvaluationElement = 030de02ffec47a1fd53efcdd1c6faf5bdc270912b8749e78
  3c7ca75bb412958832
  Output = a0b34de5fa4c5b6da07e72af73cc507cceeb48981b97b7285fc375345fe
  495dd

A.3.1.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 03cc1df781f1c2240a64d1c297b3f3d16262ef5d4cf10273488
  2675c26231b0838
  EvaluationElement = 03a0395fe3828f2476ffcd1f4fe540e5a8489322d398be3c
  4e5a869db7fcb7c52c
  Output = c748ca6dd327f0ce85f4ae3a8cd6d4d5390bbb804c9e12dcf94f853fece
  3dcce

A.3.2.  VOPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = ca5d94c8807817669a51b196c34c1b7f8442fde4334a7121ae4736364312f
  ca6
  pkSm = 03e17e70604bcabe198882c0a1f27a92441e774224ed9c702e51dd17038b1
  02462

A.3.2.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 02dd05901038bb31a6fae01828fd8d0e49e35a486b5c5d4b499
  4013648c01277da
  EvaluationElement = 0209f33cab60cf8fe69239b0afbcfcd261af4c1c5632624f
  2e9ba29b90ae83e4a2
  Proof = e7c2b3c5c954c035949f1f74e6bce2ed539a3be267d1481e9ddb178533df
  4c2664f69d065c604a4fd953e100b856ad83804eb3845189babfa5a702090d6fc5fa
  ProofRandomScalar = f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 0412e8f78b02c415ab3a288e228978376f99927767ff37c5718d420010a
  645a1

A.3.2.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 03cd0f033e791c4d79dfa9c6ed750f2ac009ec46cd4195ca6fd
  3800d1e9b887dbd
  EvaluationElement = 030d2985865c693bf7af47ba4d3a3813176576383d19aff0
  03ef7b0784a0d83cf1
  Proof = 2787d729c57e3d9512d3aa9e8708ad226bc48e0f1750b0767aaff73482c4
  4b8d2873d74ec88aebd3504961acea16790a05c542d9fbff4fe269a77510db00abab
  ProofRandomScalar = f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 771e10dcd6bcd3664e23b8f2a710cfaaa8357747c4a8cbba03133967b5c
  24f18

A.3.2.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364,f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b
  1
  BlindedElement = 02dd05901038bb31a6fae01828fd8d0e49e35a486b5c5d4b499
  4013648c01277da,03462e9ae64cae5b83ba98a6b360d942266389ac369b923eb3d5
  57213b1922f8ab
  EvaluationElement = 0209f33cab60cf8fe69239b0afbcfcd261af4c1c5632624f
  2e9ba29b90ae83e4a2,02bb24f4d838414aef052a8f044a6771230ca69c0a5677540
  fff738dd31bb69771
  Proof = bdcc351707d02a72ce49511c7db990566d29d6153ad6f8982fad2b435d6c
  e4d60da1e6b3fa740811bde34dd4fe0aa1b5fe6600d0440c9ddee95ea7fad7a60cf2
  ProofRandomScalar = 350e8040f828bf6ceca27405420cdf3d63cb3aef005f40ba
  51943c8026877963
  Output = 0412e8f78b02c415ab3a288e228978376f99927767ff37c5718d420010a
  645a1,771e10dcd6bcd3664e23b8f2a710cfaaa8357747c4a8cbba03133967b5c24f
  18

A.3.3.  POPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 6ad2173efa689ef2c27772566ad7ff6e2d59b3b196f00219451fb2c89ee4d
  ae2
  pkSm = 030d7ff077fddeec965db14b794f0cc1ba9019b04a2f4fcc1fa525dedf72e
  2a3e3

A.3.3.1.  Test Vector 1, Batch Size 1

  Input = 00
  Info = 7465737420696e666f
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 031563e127099a8f61ed51eeede05d747a8da2be329b40ba1f0
  db0b2bd9dd4e2c0
  EvaluationElement = 02c5e5300c2d9e6ba7f3f4ad60500ad93a0157e6288eb04b
  67e125db024a2c74d2
  Proof = f8a33690b87736c854eadfcaab58a59b8d9c03b569110b6f31f8bf7577f3
  fbb85a8a0c38468ccde1ba942be501654adb106167c8eb178703ccb42bccffb9231a
  ProofRandomScalar = f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 193a92520bd8fd1f37accb918040a57108daa110dc4f659abe212636d24
  5c592

A.3.3.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 021a440ace8ca667f261c10ac7686adc66a12be31e3520fca31
  7643a1eee9dcd4d
  EvaluationElement = 0208ca109cbae44f4774fc0bdd2783efdcb868cb4523d521
  96f700210e777c5de3
  Proof = 043a8fb7fc7fd31e35770cabda4753c5bf0ecc1e88c68d7d35a62bf2631e
  875af4613641be2d1875c31d1319d191c4bbc0d04875f4fd03c31d3d17dd8e069b69
  ProofRandomScalar = f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 1e6d164cfd835d88a31401623549bf6b9b306628ef03a7962921d62bc5f
  fce8c

A.3.3.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 3338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364,f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b
  1
  BlindedElement = 031563e127099a8f61ed51eeede05d747a8da2be329b40ba1f0
  db0b2bd9dd4e2c0,03ca4ff41c12fadd7a0bc92cf856732b21df652e01a3abdf0fa8
  847da053db213c
  EvaluationElement = 02c5e5300c2d9e6ba7f3f4ad60500ad93a0157e6288eb04b
  67e125db024a2c74d2,02f0b6bcd467343a8d8555a99dc2eed0215c71898c5edb77a
  3d97ddd0dbad478e8
  Proof = 8fbd85a32c13aba79db4b42e762c00687d6dbf9c8cb97b2a225645ccb00d
  9d7580b383c885cdfd07df448d55e06f50f6173405eee5506c0ed0851ff718d13e68
  ProofRandomScalar = 350e8040f828bf6ceca27405420cdf3d63cb3aef005f40ba
  51943c8026877963
  Output = 193a92520bd8fd1f37accb918040a57108daa110dc4f659abe212636d24
  5c592,1e6d164cfd835d88a31401623549bf6b9b306628ef03a7962921d62bc5ffce
  8c

A.4.  P384-SHA384

A.4.1.  OPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = dfe7ddc41a4646901184f2b432616c8ba6d452f9bcd0c4f75a5150ef2b2ed
  02ef40b8b92f60ae591bcabd72a6518f188

A.4.1.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 02a36bc90e6db34096346eaf8b7bc40ee1113582155ad379700
  3ce614c835a874343701d3f2debbd80d97cbe45de6e5f1f
  EvaluationElement = 03af2a4fc94770d7a7bf3187ca9cc4faf3732049eded2442
  ee50fbddda58b70ae2999366f72498cdbc43e6f2fc184afe30
  Output = ed84ad3f31a552f0456e58935fcc0a3039db42e7f356dcb32aa6d487b6b
  815a07d5813641fb1398c03ddab5763874357

A.4.1.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 02def6f418e3484f67a124a2ce1bfb19de7a4af568ede6a1ebb
  2733882510ddd43d05f2b1ab5187936a55e50a847a8b900
  EvaluationElement = 034e9b9a2960b536f2ef47d8608b21597ba400d5abfa1825
  fd21c36b75f927f396bf3716c96129d1fa4a77fa1d479c8d7b
  Output = dd4f29da869ab9355d60617b60da0991e22aaab243a3460601e48b07585
  9d1c526d36597326f1b985778f781a1682e75

A.4.2.  VOPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 051646b9e6e7a71ae27c1e1d0b87b4381db6d3595eeeb1adb41579adbf992
  f4278f9016eafc944edaa2b43183581779d
  pkSm = 031d689686c611991b55f1a1d8f4305ccd6cb719446f660a30db61b7aa87b
  46acf59b7c0d4a9077b3da21c25dd482229a0

A.4.2.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 02d338c05cbecb82de13d6700f09cb61190543a7b7e2c6cd4fc
  a56887e564ea82653b27fdad383995ea6d02cf26d0e24d9
  EvaluationElement = 02a7bba589b3e8672aa19e8fd258de2e6aae20101c8d7612
  46de97a6b5ee9cf105febce4327a326255a3c604f63f600ef6
  Proof = bfc6cf3859127f5fe25548859856d6b7fa1c7459f0ba5712a806fc091a30
  00c42d8ba34ff45f32a52e40533efd2a03bc87f3bf4f9f58028297ccb9ccb18ae718
  2bcd1ef239df77e3be65ef147f3acf8bc9cbfc5524b702263414f043e3b7ca2e
  ProofRandomScalar = 803d955f0e073a04aa5d92b3fb739f56f9db001266677f62
  c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  Output = 3333230886b562ffb8329a8be08fea8025755372817ec969d114d1203d0
  26b4a622beab60220bf19078bca35a529b35c

A.4.2.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 02f27469e059886f221be5f2cca03d2bdc61e55221721c3b3e5
  6fc012e36d31ae5f8dc058109591556a6dbd3a8c69c433b
  EvaluationElement = 03f16f903947035400e96b7f531a38d4a07ac89a80f89d86
  a1bf089c525a92c7f4733729ca30c56ce78b1ab4f7d92db8b4
  Proof = d005d6daaad7571414c1e0c75f7e57f2113ca9f4604e84bc90f9be52da89
  6fff3bee496dcde2a578ae9df315032585f801fb21c6080ac05672b291e575a40295
  b306d967717b28e08fcc8ad1cab47845d16af73b3e643ddcc191208e71c64630
  ProofRandomScalar = 803d955f0e073a04aa5d92b3fb739f56f9db001266677f62
  c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  Output = b91c70ea3d4d62ba922eb8a7d03809a441e1c3c7af915cbc2226f485213
  e895942cd0f8580e6d99f82221e66c40d274f

A.4.2.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364,803d955f0e073a04aa5d92b3fb739f5
  6f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  BlindedElement = 02d338c05cbecb82de13d6700f09cb61190543a7b7e2c6cd4fc
  a56887e564ea82653b27fdad383995ea6d02cf26d0e24d9,02fa02470d7f151018b4
  1e82223c32fad824de6ad4b5ce9f8e9f98083c9a726de9a1fc39d7a0cb6f4f188dd9
  cea01474cd
  EvaluationElement = 02a7bba589b3e8672aa19e8fd258de2e6aae20101c8d7612
  46de97a6b5ee9cf105febce4327a326255a3c604f63f600ef6,028e9e115625ff4c2
  f07bf87ce3fd73fc77994a7a0c1df03d2a630a3d845930e2e63a165b114d98fe34e6
  1b68d23c0b50a
  Proof = 6d8dcbd2fc95550a02211fb78afd013933f307d21e7d855b0b1ed0af7807
  6d8137ad8b0a1bfa05676d325249c1dbb9a52bd81b1c2b7b0efc77cf7b278e1c947f
  6283f1d4c513053fc0ad19e026fb0c30654b53d9cea4b87b037271b5d2e2d0ea
  ProofRandomScalar = a097e722ed2427de86966910acba9f5c350e8040f828bf6c
  eca27405420cdf3d63cb3aef005f40ba51943c8026877963
  Output = 3333230886b562ffb8329a8be08fea8025755372817ec969d114d1203d0
  26b4a622beab60220bf19078bca35a529b35c,b91c70ea3d4d62ba922eb8a7d03809
  a441e1c3c7af915cbc2226f485213e895942cd0f8580e6d99f82221e66c40d274f

A.4.3.  POPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 5b2690d6954b8fbb159f19935d64133f12770c00b68422559c65431942d72
  1ff79d47d7a75906c30b7818ec0f38b7fb2
  pkSm = 02f00f0f1de81e5d6cf18140d4926ffdc9b1898c48dc49657ae36eb1e45de
  b8b951aaf1f10c82d2eaa6d02aafa3f10d2b6

A.4.3.1.  Test Vector 1, Batch Size 1

  Input = 00
  Info = 7465737420696e666f
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 03859b36b95e6564faa85cd3801175eda2949707f6aa0640ad0
  93cbf8ad2f58e762f08b56b2a1b42a64953aaf49cbf1ae3
  EvaluationElement = 0220710e2e00306453f5b4f574cb6a512453f35c45080d09
  373e190c19ce5b185914fbf36582d7e0754bb7c8b683205b91
  Proof = 82a17ef41c8b57f1e3122311b4d5cd39a63df0f67443ef18d961f9b659c1
  601ced8d3c64b294f604319ca80230380d437a49c7af0d620e22116669c008ebb767
  d90283d573b49cdb49e3725889620924c2c4b047a2a6225a3ba27e640ebddd33
  ProofRandomScalar = 803d955f0e073a04aa5d92b3fb739f56f9db001266677f62
  c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  Output = 0188653cfec38119a6c7dd7948b0f0720460b4310e40824e048bf82a165
  27303ed449a08caf84272c3bbc972ede797df

A.4.3.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364
  BlindedElement = 03f7efcb4aaf000263369d8a0621cb96b81b3206e99876de2a0
  0699ed4c45acf3969cd6e2319215395955d3f8d8cc1c712
  EvaluationElement = 034993c818369927e74b77c400376fd1ae29b6ac6c6ddb77
  6cf10e4fbc487826531b3cf0b7c8ca4d92c7af90c9def85ce6
  Proof = 693471b5dff0cd6a5c00ea34d7bf127b2795164e3bdb5f39a1e5edfbd13e
  443bc516061cd5b8449a473c2ceeccada9f3e5b57302e3d7bc5e28d38d6e3a3056e1
  e73b6cc030f5180f8a1ffa45aa923ee66d2ad0a07b500f2acc7fb99b5506465c
  ProofRandomScalar = 803d955f0e073a04aa5d92b3fb739f56f9db001266677f62
  c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  Output = ff2a527a21cc43b251a567382677f078c6e356336aec069dea8ba369953
  43ca3b33bb5d6cf15be4d31a7e6d75b30d3f5

A.4.3.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 504650f53df8f16f6861633388936ea23338fa65ec36e0290022b48eb562
  889d89dbfa691d1cde91517fa222ed7ad364,803d955f0e073a04aa5d92b3fb739f5
  6f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b1
  BlindedElement = 03859b36b95e6564faa85cd3801175eda2949707f6aa0640ad0
  93cbf8ad2f58e762f08b56b2a1b42a64953aaf49cbf1ae3,021a65d618d645f1a20b
  c33b06deaa7e73d6d634c8a56a3d02b53a732b69a5c53c5a207ea33d5afdcde9a22d
  59726bce51
  EvaluationElement = 0220710e2e00306453f5b4f574cb6a512453f35c45080d09
  373e190c19ce5b185914fbf36582d7e0754bb7c8b683205b91,02017657b315ec65e
  f861505e596c8645d94685dd7602cdd092a8f1c1c0194a5d0485fe47d071d972ab51
  4370174cc23f5
  Proof = 4a0b2fe96d5b2a046a0447fe079b77859ef11a39a3520d6ff7c626aad9b4
  73b724fb0cf188974ec961710a62162a83e97e0baa9eeada73397032d928b3e97b1e
  a92ad9458208302be3681b8ba78bcc17745bac00f84e0fdc98a6a8cba009c080
  ProofRandomScalar = a097e722ed2427de86966910acba9f5c350e8040f828bf6c
  eca27405420cdf3d63cb3aef005f40ba51943c8026877963
  Output = 0188653cfec38119a6c7dd7948b0f0720460b4310e40824e048bf82a165
  27303ed449a08caf84272c3bbc972ede797df,ff2a527a21cc43b251a567382677f0
  78c6e356336aec069dea8ba36995343ca3b33bb5d6cf15be4d31a7e6d75b30d3f5

A.5.  P521-SHA512

A.5.1.  OPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 0153441b8faedb0340439036d6aed06d1217b34c42f17f8db4c5cc610a4a9
  55d698a688831b16d0dc7713a1aa3611ec60703bffc7dc9c84e3ed673b3dbe1d5fcc
  ea6

A.5.1.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 0300e78bf846b0e1e1a3c320e353d758583cd876df56100a3a1
  e62bacba470fa6e0991be1be80b721c50c5fd0c672ba764457acc18c6200704e9294
  fbf28859d916351
  EvaluationElement = 030166371cf827cb2fb9b581f97907121a16e2dc5d8b10ce
  9f0ede7f7d76a0d047657735e8ad07bcda824907b3e5479bd72cdef6b839b967ba5c
  58b118b84d26f2ba07
  Output = 26232de6fff83f812adadadb6cc05d7bbeee5dca043dbb16b03488abb99
  81d0a1ef4351fad52dbd7e759649af393348f7b9717566c19a6b8856284d69375c80
  9

A.5.1.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 0300c28e57e74361d87e0c1874e5f7cc1cc796d61f9cad50427
  cf54655cdb455613368d42b27f94bf66f59f53c816db3e95e68e1b113443d66a99b3
  693bab88afb556b
  EvaluationElement = 0301ad453607e12d0cc11a3359332a40c3a254eaa1afc642
  96528d55bed07ba322e72e22cf3bcb50570fd913cb54f7f09c17aff8787af75f6a7f
  af5640cbb2d9620a6e
  Output = ad1f76ef939042175e007738906ac0336bbd1d51e287ebaa66901abdd32
  4ea3ffa40bfc5a68e7939c2845e0fd37a5a6e76dadb9907c6cc8579629757fd4d04b
  a

A.5.2.  VOPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 015c7fc1b4a0b1390925bae915bd9f3d72009d44d9241b962428aad5d13f2
  2803311e7102632a39addc61ea440810222715c9d2f61f03ea424ec9ab1fe5e31cf9
  238
  pkSm = 0301505d646f6e4c9102451eb39730c4ba1c4087618641edbdba4a60896b0
  7fd0c9414ce553cbf25b81dfcca50a8f6724ab7a2bc4d0cf736967a287bb6084cc06
  78ac0

A.5.2.1.  Test Vector 1, Batch Size 1

  Input = 00
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 0301d6e4fb545e043ddb6aee5d5ceeee1b44102615ab04430c2
  7dd0f56988dedcb1df32ef384f160e0e76e718605f14f3f582f9357553d153b99679
  5b4b3628a4f6380
  EvaluationElement = 03013fdeaf887f3d3d283a79e696a54b66ff0edcb559265e
  204a958acf840e0930cc147e2a6835148d8199eebc26c03e9394c9762a1c991dde40
  bca0f8ca003eefb045
  Proof = 0077fcc8ec6d059d7759b0a61f871e7c1dadc65333502e09a51994328f79
  e5bda3357b9a4f410a1760a3612c2f8f27cb7cb032951c047cc66da60da583df7b24
  7edd0188e5eb99c71799af1d80d643af16ffa1545acd9e9233fbb370455b10eb257e
  a12a1667c1b4ee5b0ab7c93d50ae89602006960f083ca9adc4f6276c0ad60440393c
  ProofRandomScalar = 015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e07
  3a04aa5d92b3fb739f56f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 5e003d9b2fb540b3d4bab5fedd154912246da1ee5e557afd8f56415faa1
  a0fadff6517da802ee254437e4f60907b4cda146e7ba19e249eef7be405549f62954
  b

A.5.2.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 03005b05e656cb609ce5ff5faf063bb746d662d67bbd07c0626
  38396f52f0392180cf2365cabb0ece8e19048961d35eeae5d5fa872328dce98df076
  ee154dd191c615e
  EvaluationElement = 0301b19fcf482b1fff04754e282292ed736c5f0aa080d4f4
  2663cd3a416c6596f03129e8e096d8671fe5b0d19838312c511d2ce08d431e43e3ef
  06199d8cab7426238d
  Proof = 01ec9fece444caa6a57032e8963df0e945286f88fbdf233fb5101f0924f7
  ea89c47023f5f72f240e61991fd33a299b5b38c45a5e2dd1a67b072e59dfe86708a3
  59c701e38d383c60cf6969463bcf13251bedad47b7941f52e409a3591398e2792441
  0b18a301c0e19f527cad504fa08388050ac634e1b05c5216d337742f2754e1fc502f
  ProofRandomScalar = 015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e07
  3a04aa5d92b3fb739f56f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = fa15eebba81ecf40954f7135cb76f69ef22c6bae394d1a4362f9b03066b
  54b6604d39f2e53369ca6762a3d9787e230e832aa85955af40ecb8deebb009a8cf47
  4

A.5.2.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364,015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e073a04aa5d92b3fb7
  39f56f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b
  1
  BlindedElement = 0301d6e4fb545e043ddb6aee5d5ceeee1b44102615ab04430c2
  7dd0f56988dedcb1df32ef384f160e0e76e718605f14f3f582f9357553d153b99679
  5b4b3628a4f6380,0301403b597538b939b450c93586ba275f9711ba07e42364bac1
  d5769c6824a8b55be6f9a536df46d952b11ab2188363b3d6737635d9543d4dba14a6
  e19421b9245bf5
  EvaluationElement = 03013fdeaf887f3d3d283a79e696a54b66ff0edcb559265e
  204a958acf840e0930cc147e2a6835148d8199eebc26c03e9394c9762a1c991dde40
  bca0f8ca003eefb045,03001f96424497e38c46c904978c2fa1636c5c3dd2e634a85
  d8a7265977c5dce1f02c7e6c118479f0751767b91a39cce6561998258591b5d7c1bb
  02445a9e08e4f3e8d
  Proof = 00b4d215c8405e57c7a4b53398caf55f1f1623aaeb22408ddb9ea2913090
  9b3f95dbb1ff366e81e86e918f9f2fd8b80dbb344cd498c9499d112905e585417e00
  68c600fe5dea18b389ef6c4cc062935607b8ccbbb9a84fba3143868a3e8a58efa0bf
  6ca642804d09dc06e980f64837811227c4267b217f1099a4e28b0854f4e5ee659796
  ProofRandomScalar = 01ec21c7bb69b0734cb48dfd68433dd93b0fa097e722ed24
  27de86966910acba9f5c350e8040f828bf6ceca27405420cdf3d63cb3aef005f40ba
  51943c8026877963
  Output = 5e003d9b2fb540b3d4bab5fedd154912246da1ee5e557afd8f56415faa1
  a0fadff6517da802ee254437e4f60907b4cda146e7ba19e249eef7be405549f62954
  b,fa15eebba81ecf40954f7135cb76f69ef22c6bae394d1a4362f9b03066b54b6604
  d39f2e53369ca6762a3d9787e230e832aa85955af40ecb8deebb009a8cf474

A.5.3.  POPRF Mode

  Seed = a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a3a
  3a3
  KeyInfo = 74657374206b6579
  skSm = 014893130030ce69cf714f536498a02ff6b396888f9bb507985c32928c442
  7d6d39de10ef509aca4240e8569e3a88debc0d392e3361bcd934cb9bdd59e339dff7
  b27
  pkSm = 0301de8ceb9ffe9237b1bba87c320ea0bebcfc3447fe6f278065c6c69886d
  692d1126b79b6844f829940ace9b52a5e26882cf7cbc9e57503d4cca3cd834584729
  f812a

A.5.3.1.  Test Vector 1, Batch Size 1

  Input = 00
  Info = 7465737420696e666f
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 020095cff9d7ecf65bdfee4ea92d6e748d60b02de34ad98094f
  82e25d33a8bf50138ccc2cc633556f1a97d7ea9438cbb394df612f041c485a515849
  d5ebb2238f2f0e2
  EvaluationElement = 0301408e9c5be3ffcc1c16e5ae8f8aa68446223b0804b119
  62e856af5a6d1c65ebbb5db7278c21db4e8cc06d89a35b6804fb1738a295b691638a
  f77aa1327253f26d01
  Proof = 0106a89a61eee9dd2417d2849a8e2167bc5f56e3aed5a3ff23e22511fa1b
  37a29ed44d1bbfd6907d99cfbc558a56aec709282415a864a281e49dc53792a4a638
  a0660034306d64be12a94dcea5a6d664cf76681911c8b9a84d49bf12d4893307ec14
  436bd05f791f82446c0de4be6c582d373627b51886f76c4788256e3da7ec8fa18a86
  ProofRandomScalar = 015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e07
  3a04aa5d92b3fb739f56f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 808ae5b87662eaaf0b39151dd85991b94c96ef214cb14a68bf5c1439548
  82d330da8953a80eea20788e552bc8bbbfff3100e89f9d6e341197b122c46a208733
  b

A.5.3.2.  Test Vector 2, Batch Size 1

  Input = 5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364
  BlindedElement = 030112ea89cf9cf589496189eafc5f9eb13c9f9e170d6ecde7c
  5b940541cb1a9c5cfeec908b67efe16b81ca00d0ce216e34b3d5f46a658d3fd8573d
  671bdb6515ed508
  EvaluationElement = 0200ebc49df1e6fa61f412e6c391e6f074400ecdd2f56c4a
  8c03fe0f91d9b551f40d4b5258fd891952e8c9b28003bcfa365122e54a5714c8949d
  5d202767b31b4bf1f6
  Proof = 0082162c71a7765005cae202d4bd14b84dae63c29067e886b82506992bd9
  94a1c3aac0c1c5309222fe1af8287b6443ed6df5c2e0b0991faddd3564c73c7597ae
  cd9a003b1f1e3c65f28e58ab4e767cfb4adbcaf512441645f4c2aed8bf67d132d966
  006d35fa71a34145414bf3572c1de1a46c266a344dd9e22e7fb1e90ffba1caf556d9
  ProofRandomScalar = 015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e07
  3a04aa5d92b3fb739f56f9db001266677f62c095021db018cd8cbb55941d4073698c
  e45c405d1348b7b1
  Output = 27032e24b1a52a82ab7f4646f3c5df0f070f499db98b9c5df33972bd5af
  5762c3638afae7912a6c1acdb1ae2ab2fa670bd5486c645a0e55412e08d33a4a0d6e
  3

A.5.3.3.  Test Vector 3, Batch Size 2

  Input = 00,5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a5a
  Info = 7465737420696e666f
  Blind = 00d1dccf7a51bafaf75d4a866d53d8cafe4d504650f53df8f16f68616333
  88936ea23338fa65ec36e0290022b48eb562889d89dbfa691d1cde91517fa222ed7a
  d364,015e80ae32363b32cb76ad4b95a5a34e46bb803d955f0e073a04aa5d92b3fb7
  39f56f9db001266677f62c095021db018cd8cbb55941d4073698ce45c405d1348b7b
  1
  BlindedElement = 020095cff9d7ecf65bdfee4ea92d6e748d60b02de34ad98094f
  82e25d33a8bf50138ccc2cc633556f1a97d7ea9438cbb394df612f041c485a515849
  d5ebb2238f2f0e2,0201a328cf9f3fdeb86b6db242dd4cbb436b3a488b70b72d2fbb
  d1e5f50d7b0878b157d6f278c6a95c488f3ad52d6898a421658a82fe7ceb000b01ae
  dea7967522d525
  EvaluationElement = 0301408e9c5be3ffcc1c16e5ae8f8aa68446223b0804b119
  62e856af5a6d1c65ebbb5db7278c21db4e8cc06d89a35b6804fb1738a295b691638a
  f77aa1327253f26d01,020062ab51ac3aa829e0f5b7ae50688bcf5f63a18a83a6e0d
  a538666b8d50c7ea2b4ef31f4ac669302318dbebe46660acdda695da30c22cee7ca2
  1f6984a720504502e
  Proof = 00731738844f739bca0cca9d1c8bea204bed4fd00285785738b985763741
  de5cdfa275152d52b6a2fdf7792ef3779f39ba34581e56d62f78ecad5b7f8083f384
  961501cd4b43713253c022692669cf076b1d382ecd8293c1de69ea569737f37a2477
  2ab73517983c1e3db5818754ba1f008076267b8058b6481949ae346cdc17a8455fe2
  ProofRandomScalar = 01ec21c7bb69b0734cb48dfd68433dd93b0fa097e722ed24
  27de86966910acba9f5c350e8040f828bf6ceca27405420cdf3d63cb3aef005f40ba
  51943c8026877963
  Output = 808ae5b87662eaaf0b39151dd85991b94c96ef214cb14a68bf5c1439548
  82d330da8953a80eea20788e552bc8bbbfff3100e89f9d6e341197b122c46a208733
  b,27032e24b1a52a82ab7f4646f3c5df0f070f499db98b9c5df33972bd5af5762c36
  38afae7912a6c1acdb1ae2ab2fa670bd5486c645a0e55412e08d33a4a0d6e3

Acknowledgements

  This document resulted from the work of the Privacy Pass team
  [PrivacyPass].  The authors would also like to acknowledge helpful
  conversations with Hugo Krawczyk.  Eli-Shaoul Khedouri provided
  additional review and comments on key consistency.  Daniel Bourdrez,
  Tatiana Bradley, Sofia Celi, Frank Denis, Julia Hesse, Russ Housley,
  Kevin Lewi, Christopher Patton, and Bas Westerbaan also provided
  helpful input and contributions to the document.

Authors' Addresses

  Alex Davidson
  Brave Software
  Email: [email protected]


  Armando Faz-Hernandez
  Cloudflare, Inc.
  101 Townsend St
  San Francisco, CA
  United States of America
  Email: [email protected]


  Nick Sullivan
  Cloudflare, Inc.
  101 Townsend St
  San Francisco, CA
  United States of America
  Email: [email protected]


  Christopher A. Wood
  Cloudflare, Inc.
  101 Townsend St
  San Francisco, CA
  United States of America
  Email: [email protected]

You are viewing proxied material from gopher.fnord.one. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.