Papers › DFAC Framework: Factorizing the Value Function via Quantile Mixture for Multi-Agent...

DFAC Framework: Factorizing the Value Function via Quantile Mixture for Multi-Agent Distributional Q-Learning

16 Feb 2021arXiv:2102.07936archive 2025-07-28

Wei-Fang Sun, Cheng-Kuang Lee, Chun-Yi Lee

In fully cooperative multi-agent reinforcement learning (MARL) settings, the environments are highly stochastic due to the partial observability of each agent and the continuously changing policies of the other agents. To address the above issues, we integrate distributional RL and value function factorization methods by proposing a Distributional Value Function Factorization (DFAC) framework to generalize expected value function factorization methods to their DFAC variants. DFAC extends the individual utility functions from deterministic variables to random variables, and models the quantile function of the total return as a quantile mixture. To validate DFAC, we demonstrate DFAC's ability to factorize a simple two-step matrix game with stochastic rewards and perform experiments on all Super Hard tasks of StarCraft Multi-Agent Challenge, showing that DFAC is able to outperform expected value function factorization baselines.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 1 code sample harvested from 1 repository linked to this paper: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified; the one sample that ran constructed an object rather than computing a result. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 1 sample from 1 repository, 1 ran (1 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

j3soon/dfac officialmentioned in papermentioned on GitHubtfSyntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

1 sample harvested: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified; the one sample that ran constructed an object rather than computing a result. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran

Licence: 0 of the 1 sample are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from j3soon/dfac. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

j3soon/dfac official repository 1 sample, 1 ran (0 instrument-failed) · 0 unverified
DMixer pymarl/src/modules/mixers/dmix.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 38a1d2c8a56f1d27 · report fixture: returned DMixer · level recorded 2026-09-22

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it.

1 assembly on this paper: 1 harvested, 0 generated.

harvested Assembly for DFAC Framework (graph method: DFAC Framework). 1 brick; individually on a fixture: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified; the one sample that ran constructed an object rather than computing a result. The assembly has not been run end-to-end.

  1. DMixer entry · level 1 · ran · 38a1d2c8a56f1d27

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2102.07936") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Multi-agent Reinforcement LearningQ-LearningSMACSMAC+Starcraft

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
SMAC SMAC 27m_vs_30m DDN Average Score 19.71 #1 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m DDN Median Win Rate 91.48 #1 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m DMIX Average Score 19.43 #3 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m DMIX Median Win Rate 85.45 #3 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m QMIX Average Score 19.41 #4 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m QMIX Median Win Rate 84.77 #4 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m VDN Average Score 18.45 #6 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m VDN Median Win Rate 63.12 #6 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m DIQL Average Score 14.45 #9 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m DIQL Median Win Rate 6.02 #9 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m IQL Average Score 14.01 #10 of 11 Archive leaderboard report
SMAC SMAC 27m_vs_30m IQL Median Win Rate 2.27 #10 of 11 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DDN Average Score 20.94 #2 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DDN Median Win Rate 94.03 #2 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DMIX Average Score 19.7 #3 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DMIX Median Win Rate 91.08 #3 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z VDN Average Score 19.75 #5 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z VDN Median Win Rate 89.2 #5 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z QMIX Average Score 20.16 #7 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z QMIX Median Win Rate 67.22 #7 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DIQL Average Score 17.52 #8 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z DIQL Median Win Rate 62.22 #8 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z IQL Average Score 16.54 #9 of 13 Archive leaderboard report
SMAC SMAC 3s5z_vs_3s6z IQL Median Win Rate 29.83 #9 of 13 Archive leaderboard report
SMAC SMAC 6h_vs_8z DDN Average Score 19.4 #2 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z DDN Median Win Rate 83.92 #2 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z DMIX Average Score 17.14 #3 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z DMIX Median Win Rate 49.43 #3 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z QMIX Average Score 14.37 #5 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z QMIX Median Win Rate 12.78 #5 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z VDN Average Score 15.41 #8 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z VDN Median Win Rate 0 #8 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z DIQL Average Score 14.94 #9 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z DIQL Median Win Rate 0.00 #9 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z IQL Average Score 13.78 #10 of 14 Archive leaderboard report
SMAC SMAC 6h_vs_8z IQL Median Win Rate 0 #10 of 14 Archive leaderboard report
SMAC SMAC MMM2 DDN Average Score 20.9 #2 of 14 Archive leaderboard report
SMAC SMAC MMM2 DDN Median Win Rate 97.22 #2 of 14 Archive leaderboard report
SMAC SMAC MMM2 DMIX Average Score 19.87 #5 of 14 Archive leaderboard report
SMAC SMAC MMM2 DMIX Median Win Rate 95.11 #5 of 14 Archive leaderboard report
SMAC SMAC MMM2 QMIX Average Score 19.42 #6 of 14 Archive leaderboard report
SMAC SMAC MMM2 QMIX Median Win Rate 92.44 #6 of 14 Archive leaderboard report
SMAC SMAC MMM2 VDN Average Score 19.36 #7 of 14 Archive leaderboard report
SMAC SMAC MMM2 VDN Median Win Rate 89.2 #7 of 14 Archive leaderboard report
SMAC SMAC MMM2 DIQL Average Score 19.21 #8 of 14 Archive leaderboard report
SMAC SMAC MMM2 DIQL Median Win Rate 85.23 #8 of 14 Archive leaderboard report
SMAC SMAC MMM2 IQL Average Score 17.5 #11 of 14 Archive leaderboard report
SMAC SMAC MMM2 IQL Median Win Rate 68.92 #11 of 14 Archive leaderboard report
SMAC SMAC corridor DDN Average Score 20 #2 of 13 Archive leaderboard report
SMAC SMAC corridor DDN Median Win Rate 95.4 #2 of 13 Archive leaderboard report
SMAC SMAC corridor DIQL Average Score 19.68 #3 of 13 Archive leaderboard report
SMAC SMAC corridor DIQL Median Win Rate 91.62 #3 of 13 Archive leaderboard report
SMAC SMAC corridor DMIX Average Score 19.66 #4 of 13 Archive leaderboard report
SMAC SMAC corridor DMIX Median Win Rate 90.45 #4 of 13 Archive leaderboard report
SMAC SMAC corridor VDN Average Score 19.47 #5 of 13 Archive leaderboard report
SMAC SMAC corridor VDN Median Win Rate 85.34 #5 of 13 Archive leaderboard report
SMAC SMAC corridor IQL Average Score 19.42 #6 of 13 Archive leaderboard report
SMAC SMAC corridor IQL Median Win Rate 84.87 #6 of 13 Archive leaderboard report
SMAC SMAC corridor QMIX Average Score 15.07 #9 of 13 Archive leaderboard report
SMAC SMAC corridor QMIX Median Win Rate 37.61 #9 of 13 Archive leaderboard report
SMAC+ Def_Armored_parallel DMIX Median Win Rate 90.0 #1 of 10 Archive leaderboard report
SMAC+ Def_Armored_parallel DDN Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Def_Armored_parallel DIQL Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Def_Armored_sequential DMIX Median Win Rate 81.3 #5 of 11 Archive leaderboard report
SMAC+ Def_Armored_sequential DDN Median Win Rate 71.9 #6 of 11 Archive leaderboard report
SMAC+ Def_Armored_sequential DIQL Median Win Rate 53.1 #7 of 11 Archive leaderboard report
SMAC+ Def_Infantry_parallel DMIX Median Win Rate 90.0 #5 of 10 Archive leaderboard report
SMAC+ Def_Infantry_parallel DDN Median Win Rate 20.0 #10 of 10 Archive leaderboard report
SMAC+ Def_Infantry_sequential DMIX Median Win Rate 100 #4 of 11 Archive leaderboard report
SMAC+ Def_Infantry_sequential DIQL Median Win Rate 93.8 #8 of 11 Archive leaderboard report
SMAC+ Def_Infantry_sequential DDN Median Win Rate 90.6 #9 of 11 Archive leaderboard report
SMAC+ Def_Outnumbered_parallel DMIX Median Win Rate 5.0 #3 of 10 Archive leaderboard report
SMAC+ Def_Outnumbered_parallel DDN Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Def_Outnumbered_parallel DIQL Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Def_Outnumbered_sequential DDN Median Win Rate 0.0 #9 of 11 Archive leaderboard report
SMAC+ Def_Outnumbered_sequential DIQL Median Win Rate 0.0 #10 of 11 Archive leaderboard report
SMAC+ Def_Outnumbered_sequential DMIX Median Win Rate 0.0 #11 of 11 Archive leaderboard report
SMAC+ Off_Complicated_parallel DDN Median Win Rate 0.0 #8 of 10 Archive leaderboard report
SMAC+ Off_Complicated_parallel DIQL Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Off_Complicated_parallel DMIX Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Off_Distant_parallel DDN Median Win Rate 0.0 #8 of 10 Archive leaderboard report
SMAC+ Off_Distant_parallel DIQL Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Off_Distant_parallel DMIX Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Off_Hard_parallel DDN Median Win Rate 0.0 #8 of 10 Archive leaderboard report
SMAC+ Off_Hard_parallel DIQL Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Off_Hard_parallel DMIX Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Off_Near_parallel DDN Median Win Rate 0.0 #8 of 10 Archive leaderboard report
SMAC+ Off_Near_parallel DIQL Median Win Rate 0.0 #9 of 10 Archive leaderboard report
SMAC+ Off_Near_parallel DMIX Median Win Rate 0.0 #10 of 10 Archive leaderboard report
SMAC+ Off_Superhard_parallel DDN Median Win Rate 0.0 #7 of 10 Archive leaderboard report
SMAC+ Off_Superhard_parallel DIQL Median Win Rate 0.0 #8 of 10 Archive leaderboard report
SMAC+ Off_Superhard_parallel DMIX Median Win Rate 0.0 #9 of 10 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Q-Learning

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections