A Unifying Tool for Bounding the Quality of Non-Cooperative Solutions in Weighted Congestion Games View Full Text


Ontology type: schema:ScholarlyArticle     


Article Info

DATE

2018-07

AUTHORS

Vittorio Bilò

ABSTRACT

We present a general technique, based on a primal-dual formulation, for analyzing the quality of self-emerging solutions in weighted congestion games. With respect to traditional combinatorial approaches, the primal-dual schema has at least three advantages: first, it provides an analytic tool which can always be used to prove tight upper bounds for all the cases in which we are able to characterize exactly the polyhedron of the solutions under analysis; secondly, in each such a case, the complementary slackness conditions give us a hint on how to construct matching lower bounding instances; thirdly, proofs become simpler and easy to check. For the sake of exposition, we first apply our technique to the problems of bounding the price of anarchy and stability of exact and approximate pure Nash equilibria, as well as the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state, in the case of affine latency functions and we show how all the known upper bounds for these measures (and some of their generalizations) can be easily reobtained under a unified approach. Then, we use the technique to attack the more challenging setting of polynomial latency functions. In particular, we obtain the first known upper bounds on the price of stability of pure Nash equilibria and on the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state for unweighted players in the cases of quadratic and cubic latency functions. More... »

PAGES

1288-1317

Journal

TITLE

Theory of Computing Systems

ISSUE

5

VOLUME

62

Author Affiliations

Identifiers

URI

http://scigraph.springernature.com/pub.10.1007/s00224-017-9826-1

DOI

http://dx.doi.org/10.1007/s00224-017-9826-1

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1093123382


Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
Incoming Citations Browse incoming citations for this publication using opencitations.net

JSON-LD is the canonical representation for SciGraph data.

TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

[
  {
    "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
    "about": [
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0101", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Pure Mathematics", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/01", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Mathematical Sciences", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "University of Salento", 
          "id": "https://www.grid.ac/institutes/grid.9906.6", 
          "name": [
            "Department of Mathematics and Physics \u201cEnnio De Giorgi\u201d, University of Salento, Provinciale Lecce-Arnesano, P.O. Box 193, 73100, Lecce, Italy"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Bil\u00f2", 
        "givenName": "Vittorio", 
        "id": "sg:person.012652036261.27", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.012652036261.27"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "https://doi.org/10.1145/950620.950621", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1000944981"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/3-540-49116-3_38", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1002207134", 
          "https://doi.org/10.1007/3-540-49116-3_38"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/3-540-49116-3_38", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1002207134", 
          "https://doi.org/10.1007/3-540-49116-3_38"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-540-27821-4_17", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1003507443", 
          "https://doi.org/10.1007/978-3-540-27821-4_17"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-540-27821-4_17", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1003507443", 
          "https://doi.org/10.1007/978-3-540-27821-4_17"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/1993636.1993716", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1007979705"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-662-53354-3_8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1011215800", 
          "https://doi.org/10.1007/978-3-662-53354-3_8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/1868237.1868251", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1012378343"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-02927-1_24", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1014032202", 
          "https://doi.org/10.1007/978-3-642-02927-1_24"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-02927-1_24", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1014032202", 
          "https://doi.org/10.1007/978-3-642-02927-1_24"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s10878-015-9898-2", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1015171432", 
          "https://doi.org/10.1007/s10878-015-9898-2"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00453-010-9427-8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1016706126", 
          "https://doi.org/10.1007/s00453-010-9427-8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-15640-3_12", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1017021240", 
          "https://doi.org/10.1007/978-3-642-15640-3_12"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-15640-3_12", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1017021240", 
          "https://doi.org/10.1007/978-3-642-15640-3_12"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1016/j.tcs.2012.02.033", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1018901539"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00453-006-1211-4", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1022756805", 
          "https://doi.org/10.1007/s00453-006-1211-4"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/1187436.1216584", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1025772968"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-33996-7_22", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1026445824", 
          "https://doi.org/10.1007/978-3-642-33996-7_22"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00453-010-9449-2", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1027923807", 
          "https://doi.org/10.1007/s00453-010-9449-2"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-008-9112-3", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1029504498", 
          "https://doi.org/10.1007/s00224-008-9112-3"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-008-9112-3", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1029504498", 
          "https://doi.org/10.1007/s00224-008-9112-3"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/2209249.2209274", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1029940041"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/2841229", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1030920497"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/2629666", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1031469011"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/2940716.2940750", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1032682251"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-31104-8_13", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1036296323", 
          "https://doi.org/10.1007/978-3-642-31104-8_13"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00453-013-9806-z", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1036677546", 
          "https://doi.org/10.1007/s00453-013-9806-z"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-010-9309-0", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1040895760", 
          "https://doi.org/10.1007/s00224-010-9309-0"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-17572-5_26", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1041684511", 
          "https://doi.org/10.1007/978-3-642-17572-5_26"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-642-17572-5_26", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1041684511", 
          "https://doi.org/10.1007/978-3-642-17572-5_26"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-008-9152-8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1042669741", 
          "https://doi.org/10.1007/s00224-008-9152-8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-008-9152-8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1042669741", 
          "https://doi.org/10.1007/s00224-008-9152-8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-319-08783-2_47", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1044149470", 
          "https://doi.org/10.1007/978-3-319-08783-2_47"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/bf01737559", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1044775936", 
          "https://doi.org/10.1007/bf01737559"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-011-9315-x", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1046753659", 
          "https://doi.org/10.1007/s00224-011-9315-x"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/s00224-013-9529-1", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1047944956", 
          "https://doi.org/10.1007/s00224-013-9529-1"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/11561071_8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1049866450", 
          "https://doi.org/10.1007/11561071_8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/11561071_8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1049866450", 
          "https://doi.org/10.1007/11561071_8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/11561071_8", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1049866450", 
          "https://doi.org/10.1007/11561071_8"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/1060590.1060600", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1050456256"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-662-48995-6_10", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1053291738", 
          "https://doi.org/10.1007/978-3-662-48995-6_10"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/11671411_13", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1053385843", 
          "https://doi.org/10.1007/11671411_13"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/11671411_13", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1053385843", 
          "https://doi.org/10.1007/11671411_13"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1017/s0960129515000079", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1054905039"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1137/06067660x", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1062849940"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1137/070680096", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1062850841"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1137/090748986", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1062855983"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1287/moor.1120.0543", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1064723231"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "2018-07", 
    "datePublishedReg": "2018-07-01", 
    "description": "We present a general technique, based on a primal-dual formulation, for analyzing the quality of self-emerging solutions in weighted congestion games. With respect to traditional combinatorial approaches, the primal-dual schema has at least three advantages: first, it provides an analytic tool which can always be used to prove tight upper bounds for all the cases in which we are able to characterize exactly the polyhedron of the solutions under analysis; secondly, in each such a case, the complementary slackness conditions give us a hint on how to construct matching lower bounding instances; thirdly, proofs become simpler and easy to check. For the sake of exposition, we first apply our technique to the problems of bounding the price of anarchy and stability of exact and approximate pure Nash equilibria, as well as the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state, in the case of affine latency functions and we show how all the known upper bounds for these measures (and some of their generalizations) can be easily reobtained under a unified approach. Then, we use the technique to attack the more challenging setting of polynomial latency functions. In particular, we obtain the first known upper bounds on the price of stability of pure Nash equilibria and on the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state for unweighted players in the cases of quadratic and cubic latency functions.", 
    "genre": "research_article", 
    "id": "sg:pub.10.1007/s00224-017-9826-1", 
    "inLanguage": [
      "en"
    ], 
    "isAccessibleForFree": false, 
    "isPartOf": [
      {
        "id": "sg:journal.1052098", 
        "issn": [
          "1432-4350", 
          "1433-0490"
        ], 
        "name": "Theory of Computing Systems", 
        "type": "Periodical"
      }, 
      {
        "issueNumber": "5", 
        "type": "PublicationIssue"
      }, 
      {
        "type": "PublicationVolume", 
        "volumeNumber": "62"
      }
    ], 
    "name": "A Unifying Tool for Bounding the Quality of Non-Cooperative Solutions in Weighted Congestion Games", 
    "pagination": "1288-1317", 
    "productId": [
      {
        "name": "readcube_id", 
        "type": "PropertyValue", 
        "value": [
          "c6acd6f4961ed256e8f119b71ecfd92f2c01638e1b2f5b4c15a76e7cd2cea20b"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1007/s00224-017-9826-1"
        ]
      }, 
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1093123382"
        ]
      }
    ], 
    "sameAs": [
      "https://doi.org/10.1007/s00224-017-9826-1", 
      "https://app.dimensions.ai/details/publication/pub.1093123382"
    ], 
    "sdDataset": "articles", 
    "sdDatePublished": "2019-04-10T17:27", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-uberresearch-data-dimensions-target-20181106-alternative/cleanup/v134/2549eaecd7973599484d7c17b260dba0a4ecb94b/merge/v9/a6c9fde33151104705d4d7ff012ea9563521a3ce/jats-lookup/v90/0000000001_0000000264/records_8672_00000493.jsonl", 
    "type": "ScholarlyArticle", 
    "url": "http://link.springer.com/10.1007/s00224-017-9826-1"
  }
]
 

Download the RDF metadata as:  json-ld nt turtle xml License info

HOW TO GET THIS DATA PROGRAMMATICALLY:

JSON-LD is a popular format for linked data which is fully compatible with JSON.

curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1007/s00224-017-9826-1'

N-Triples is a line-based linked data format ideal for batch operations.

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1007/s00224-017-9826-1'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1007/s00224-017-9826-1'

RDF/XML is a standard XML format for linked data.

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1007/s00224-017-9826-1'


 

This table displays all metadata directly associated to this object as RDF triples.

198 TRIPLES      21 PREDICATES      65 URIs      19 LITERALS      7 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1007/s00224-017-9826-1 schema:about anzsrc-for:01
2 anzsrc-for:0101
3 schema:author Nb4a921c52b714520b64293358d392c55
4 schema:citation sg:pub.10.1007/11561071_8
5 sg:pub.10.1007/11671411_13
6 sg:pub.10.1007/3-540-49116-3_38
7 sg:pub.10.1007/978-3-319-08783-2_47
8 sg:pub.10.1007/978-3-540-27821-4_17
9 sg:pub.10.1007/978-3-642-02927-1_24
10 sg:pub.10.1007/978-3-642-15640-3_12
11 sg:pub.10.1007/978-3-642-17572-5_26
12 sg:pub.10.1007/978-3-642-31104-8_13
13 sg:pub.10.1007/978-3-642-33996-7_22
14 sg:pub.10.1007/978-3-662-48995-6_10
15 sg:pub.10.1007/978-3-662-53354-3_8
16 sg:pub.10.1007/bf01737559
17 sg:pub.10.1007/s00224-008-9112-3
18 sg:pub.10.1007/s00224-008-9152-8
19 sg:pub.10.1007/s00224-010-9309-0
20 sg:pub.10.1007/s00224-011-9315-x
21 sg:pub.10.1007/s00224-013-9529-1
22 sg:pub.10.1007/s00453-006-1211-4
23 sg:pub.10.1007/s00453-010-9427-8
24 sg:pub.10.1007/s00453-010-9449-2
25 sg:pub.10.1007/s00453-013-9806-z
26 sg:pub.10.1007/s10878-015-9898-2
27 https://doi.org/10.1016/j.tcs.2012.02.033
28 https://doi.org/10.1017/s0960129515000079
29 https://doi.org/10.1137/06067660x
30 https://doi.org/10.1137/070680096
31 https://doi.org/10.1137/090748986
32 https://doi.org/10.1145/1060590.1060600
33 https://doi.org/10.1145/1187436.1216584
34 https://doi.org/10.1145/1868237.1868251
35 https://doi.org/10.1145/1993636.1993716
36 https://doi.org/10.1145/2209249.2209274
37 https://doi.org/10.1145/2629666
38 https://doi.org/10.1145/2841229
39 https://doi.org/10.1145/2940716.2940750
40 https://doi.org/10.1145/950620.950621
41 https://doi.org/10.1287/moor.1120.0543
42 schema:datePublished 2018-07
43 schema:datePublishedReg 2018-07-01
44 schema:description We present a general technique, based on a primal-dual formulation, for analyzing the quality of self-emerging solutions in weighted congestion games. With respect to traditional combinatorial approaches, the primal-dual schema has at least three advantages: first, it provides an analytic tool which can always be used to prove tight upper bounds for all the cases in which we are able to characterize exactly the polyhedron of the solutions under analysis; secondly, in each such a case, the complementary slackness conditions give us a hint on how to construct matching lower bounding instances; thirdly, proofs become simpler and easy to check. For the sake of exposition, we first apply our technique to the problems of bounding the price of anarchy and stability of exact and approximate pure Nash equilibria, as well as the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state, in the case of affine latency functions and we show how all the known upper bounds for these measures (and some of their generalizations) can be easily reobtained under a unified approach. Then, we use the technique to attack the more challenging setting of polynomial latency functions. In particular, we obtain the first known upper bounds on the price of stability of pure Nash equilibria and on the approximation ratio of the strategy profiles achieved after a one-round walk starting from the empty state for unweighted players in the cases of quadratic and cubic latency functions.
45 schema:genre research_article
46 schema:inLanguage en
47 schema:isAccessibleForFree false
48 schema:isPartOf Ne48ed4f08d43414ba876d16da070a557
49 Ne73d3af7e10a46fabc175efe62a6bc7a
50 sg:journal.1052098
51 schema:name A Unifying Tool for Bounding the Quality of Non-Cooperative Solutions in Weighted Congestion Games
52 schema:pagination 1288-1317
53 schema:productId N12595b502aee4e64adc242d019e974a8
54 N1980d1a38f184c08859bcba0ff9a7d2c
55 Nb4642cf47d9e4409976b76cb38f3122d
56 schema:sameAs https://app.dimensions.ai/details/publication/pub.1093123382
57 https://doi.org/10.1007/s00224-017-9826-1
58 schema:sdDatePublished 2019-04-10T17:27
59 schema:sdLicense https://scigraph.springernature.com/explorer/license/
60 schema:sdPublisher Na068770a0e4c4385b43f055d65b83d03
61 schema:url http://link.springer.com/10.1007/s00224-017-9826-1
62 sgo:license sg:explorer/license/
63 sgo:sdDataset articles
64 rdf:type schema:ScholarlyArticle
65 N12595b502aee4e64adc242d019e974a8 schema:name dimensions_id
66 schema:value pub.1093123382
67 rdf:type schema:PropertyValue
68 N1980d1a38f184c08859bcba0ff9a7d2c schema:name doi
69 schema:value 10.1007/s00224-017-9826-1
70 rdf:type schema:PropertyValue
71 Na068770a0e4c4385b43f055d65b83d03 schema:name Springer Nature - SN SciGraph project
72 rdf:type schema:Organization
73 Nb4642cf47d9e4409976b76cb38f3122d schema:name readcube_id
74 schema:value c6acd6f4961ed256e8f119b71ecfd92f2c01638e1b2f5b4c15a76e7cd2cea20b
75 rdf:type schema:PropertyValue
76 Nb4a921c52b714520b64293358d392c55 rdf:first sg:person.012652036261.27
77 rdf:rest rdf:nil
78 Ne48ed4f08d43414ba876d16da070a557 schema:issueNumber 5
79 rdf:type schema:PublicationIssue
80 Ne73d3af7e10a46fabc175efe62a6bc7a schema:volumeNumber 62
81 rdf:type schema:PublicationVolume
82 anzsrc-for:01 schema:inDefinedTermSet anzsrc-for:
83 schema:name Mathematical Sciences
84 rdf:type schema:DefinedTerm
85 anzsrc-for:0101 schema:inDefinedTermSet anzsrc-for:
86 schema:name Pure Mathematics
87 rdf:type schema:DefinedTerm
88 sg:journal.1052098 schema:issn 1432-4350
89 1433-0490
90 schema:name Theory of Computing Systems
91 rdf:type schema:Periodical
92 sg:person.012652036261.27 schema:affiliation https://www.grid.ac/institutes/grid.9906.6
93 schema:familyName Bilò
94 schema:givenName Vittorio
95 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.012652036261.27
96 rdf:type schema:Person
97 sg:pub.10.1007/11561071_8 schema:sameAs https://app.dimensions.ai/details/publication/pub.1049866450
98 https://doi.org/10.1007/11561071_8
99 rdf:type schema:CreativeWork
100 sg:pub.10.1007/11671411_13 schema:sameAs https://app.dimensions.ai/details/publication/pub.1053385843
101 https://doi.org/10.1007/11671411_13
102 rdf:type schema:CreativeWork
103 sg:pub.10.1007/3-540-49116-3_38 schema:sameAs https://app.dimensions.ai/details/publication/pub.1002207134
104 https://doi.org/10.1007/3-540-49116-3_38
105 rdf:type schema:CreativeWork
106 sg:pub.10.1007/978-3-319-08783-2_47 schema:sameAs https://app.dimensions.ai/details/publication/pub.1044149470
107 https://doi.org/10.1007/978-3-319-08783-2_47
108 rdf:type schema:CreativeWork
109 sg:pub.10.1007/978-3-540-27821-4_17 schema:sameAs https://app.dimensions.ai/details/publication/pub.1003507443
110 https://doi.org/10.1007/978-3-540-27821-4_17
111 rdf:type schema:CreativeWork
112 sg:pub.10.1007/978-3-642-02927-1_24 schema:sameAs https://app.dimensions.ai/details/publication/pub.1014032202
113 https://doi.org/10.1007/978-3-642-02927-1_24
114 rdf:type schema:CreativeWork
115 sg:pub.10.1007/978-3-642-15640-3_12 schema:sameAs https://app.dimensions.ai/details/publication/pub.1017021240
116 https://doi.org/10.1007/978-3-642-15640-3_12
117 rdf:type schema:CreativeWork
118 sg:pub.10.1007/978-3-642-17572-5_26 schema:sameAs https://app.dimensions.ai/details/publication/pub.1041684511
119 https://doi.org/10.1007/978-3-642-17572-5_26
120 rdf:type schema:CreativeWork
121 sg:pub.10.1007/978-3-642-31104-8_13 schema:sameAs https://app.dimensions.ai/details/publication/pub.1036296323
122 https://doi.org/10.1007/978-3-642-31104-8_13
123 rdf:type schema:CreativeWork
124 sg:pub.10.1007/978-3-642-33996-7_22 schema:sameAs https://app.dimensions.ai/details/publication/pub.1026445824
125 https://doi.org/10.1007/978-3-642-33996-7_22
126 rdf:type schema:CreativeWork
127 sg:pub.10.1007/978-3-662-48995-6_10 schema:sameAs https://app.dimensions.ai/details/publication/pub.1053291738
128 https://doi.org/10.1007/978-3-662-48995-6_10
129 rdf:type schema:CreativeWork
130 sg:pub.10.1007/978-3-662-53354-3_8 schema:sameAs https://app.dimensions.ai/details/publication/pub.1011215800
131 https://doi.org/10.1007/978-3-662-53354-3_8
132 rdf:type schema:CreativeWork
133 sg:pub.10.1007/bf01737559 schema:sameAs https://app.dimensions.ai/details/publication/pub.1044775936
134 https://doi.org/10.1007/bf01737559
135 rdf:type schema:CreativeWork
136 sg:pub.10.1007/s00224-008-9112-3 schema:sameAs https://app.dimensions.ai/details/publication/pub.1029504498
137 https://doi.org/10.1007/s00224-008-9112-3
138 rdf:type schema:CreativeWork
139 sg:pub.10.1007/s00224-008-9152-8 schema:sameAs https://app.dimensions.ai/details/publication/pub.1042669741
140 https://doi.org/10.1007/s00224-008-9152-8
141 rdf:type schema:CreativeWork
142 sg:pub.10.1007/s00224-010-9309-0 schema:sameAs https://app.dimensions.ai/details/publication/pub.1040895760
143 https://doi.org/10.1007/s00224-010-9309-0
144 rdf:type schema:CreativeWork
145 sg:pub.10.1007/s00224-011-9315-x schema:sameAs https://app.dimensions.ai/details/publication/pub.1046753659
146 https://doi.org/10.1007/s00224-011-9315-x
147 rdf:type schema:CreativeWork
148 sg:pub.10.1007/s00224-013-9529-1 schema:sameAs https://app.dimensions.ai/details/publication/pub.1047944956
149 https://doi.org/10.1007/s00224-013-9529-1
150 rdf:type schema:CreativeWork
151 sg:pub.10.1007/s00453-006-1211-4 schema:sameAs https://app.dimensions.ai/details/publication/pub.1022756805
152 https://doi.org/10.1007/s00453-006-1211-4
153 rdf:type schema:CreativeWork
154 sg:pub.10.1007/s00453-010-9427-8 schema:sameAs https://app.dimensions.ai/details/publication/pub.1016706126
155 https://doi.org/10.1007/s00453-010-9427-8
156 rdf:type schema:CreativeWork
157 sg:pub.10.1007/s00453-010-9449-2 schema:sameAs https://app.dimensions.ai/details/publication/pub.1027923807
158 https://doi.org/10.1007/s00453-010-9449-2
159 rdf:type schema:CreativeWork
160 sg:pub.10.1007/s00453-013-9806-z schema:sameAs https://app.dimensions.ai/details/publication/pub.1036677546
161 https://doi.org/10.1007/s00453-013-9806-z
162 rdf:type schema:CreativeWork
163 sg:pub.10.1007/s10878-015-9898-2 schema:sameAs https://app.dimensions.ai/details/publication/pub.1015171432
164 https://doi.org/10.1007/s10878-015-9898-2
165 rdf:type schema:CreativeWork
166 https://doi.org/10.1016/j.tcs.2012.02.033 schema:sameAs https://app.dimensions.ai/details/publication/pub.1018901539
167 rdf:type schema:CreativeWork
168 https://doi.org/10.1017/s0960129515000079 schema:sameAs https://app.dimensions.ai/details/publication/pub.1054905039
169 rdf:type schema:CreativeWork
170 https://doi.org/10.1137/06067660x schema:sameAs https://app.dimensions.ai/details/publication/pub.1062849940
171 rdf:type schema:CreativeWork
172 https://doi.org/10.1137/070680096 schema:sameAs https://app.dimensions.ai/details/publication/pub.1062850841
173 rdf:type schema:CreativeWork
174 https://doi.org/10.1137/090748986 schema:sameAs https://app.dimensions.ai/details/publication/pub.1062855983
175 rdf:type schema:CreativeWork
176 https://doi.org/10.1145/1060590.1060600 schema:sameAs https://app.dimensions.ai/details/publication/pub.1050456256
177 rdf:type schema:CreativeWork
178 https://doi.org/10.1145/1187436.1216584 schema:sameAs https://app.dimensions.ai/details/publication/pub.1025772968
179 rdf:type schema:CreativeWork
180 https://doi.org/10.1145/1868237.1868251 schema:sameAs https://app.dimensions.ai/details/publication/pub.1012378343
181 rdf:type schema:CreativeWork
182 https://doi.org/10.1145/1993636.1993716 schema:sameAs https://app.dimensions.ai/details/publication/pub.1007979705
183 rdf:type schema:CreativeWork
184 https://doi.org/10.1145/2209249.2209274 schema:sameAs https://app.dimensions.ai/details/publication/pub.1029940041
185 rdf:type schema:CreativeWork
186 https://doi.org/10.1145/2629666 schema:sameAs https://app.dimensions.ai/details/publication/pub.1031469011
187 rdf:type schema:CreativeWork
188 https://doi.org/10.1145/2841229 schema:sameAs https://app.dimensions.ai/details/publication/pub.1030920497
189 rdf:type schema:CreativeWork
190 https://doi.org/10.1145/2940716.2940750 schema:sameAs https://app.dimensions.ai/details/publication/pub.1032682251
191 rdf:type schema:CreativeWork
192 https://doi.org/10.1145/950620.950621 schema:sameAs https://app.dimensions.ai/details/publication/pub.1000944981
193 rdf:type schema:CreativeWork
194 https://doi.org/10.1287/moor.1120.0543 schema:sameAs https://app.dimensions.ai/details/publication/pub.1064723231
195 rdf:type schema:CreativeWork
196 https://www.grid.ac/institutes/grid.9906.6 schema:alternateName University of Salento
197 schema:name Department of Mathematics and Physics “Ennio De Giorgi”, University of Salento, Provinciale Lecce-Arnesano, P.O. Box 193, 73100, Lecce, Italy
198 rdf:type schema:Organization
 




Preview window. Press ESC to close (or click here)


...