Skip to content

MFRM Desktop 0.2.19-beta — macOS (Apple Silicon) & Windows 11 (x64) Preview

Pre-release
Pre-release

Choose a tag to compare

View all tags
@Ryuya-dot-com Ryuya-dot-com released this 30 Oct 19:39
desktop-20261031
ac98697

Separate examinee ability from rater severity, task difficulty, and rubric behavior — standalone offline app for macOS & Windows with Python bundled inside.

MFRM Desktop (0.2.19-beta, source commit ac986978673dfaa8bcbed6ff2c8735b1e9ff342c) integrates Many-Facet Rasch calibration (RSM / PCM × JMLE / fixed-SD MML), Wright / Pathway / Category probability plots, conditional bias screening, fixed-SD specified intervals, fixed-N(0,1) joint interaction, and univariate balanced G/D-Study reliability planning into one offline window (Data → Settings → Results) with instant English / 日本語 switching.

1. Download (Python Bundled — No Extra Setup Needed)

Operating System Direct Application Download Notes
macOS (Apple Silicon, M1 or later) Download MFRM-Desktop-macOS-arm64.zip (87.8 MiB) Targets macOS 14.0+ (arm64). Intel Macs can run from Python source.
Windows 11 (x64) Download MFRM-Desktop-Windows-x64.zip (106.6 MiB) Hosted directly on GitHub Releases with GitHub Actions Sigstore Build Provenance.
Optional Sample Projects Download MFRM-Desktop-examples.zip (54 KiB) Ready-to-open PCM_MML.mfrm, GD_ANOVA.mfrm, and desktop_writing.csv.

(Note: GitHub's automatic Source code (zip) at the very bottom of this page contains raw Python source code rather than the standalone desktop app.)

2. Opening on First Launch (3 Steps)

Because MFRM Desktop is distributed 100% free of charge as open-source academic software without paid commercial Apple Developer Program notarization or commercial Windows code-signing subscriptions, macOS Gatekeeper and Windows SmartScreen show a standard "unverified publisher / could not verify free of malware" notice on first launch—not because malicious code was detected, but because no commercial certificate is attached. All estimation, plotting, and saving run 100% offline with zero network calls or telemetry.

macOS (Sequoia 15+ / Sonoma 14+, Apple Silicon)

  1. Double-click MFRM-Desktop-macOS-arm64.zip in Finder to extract MFRM-Desktop.app and move it to a writable folder such as Applications.
  2. Double-click MFRM-Desktop.app. When "MFRM-Desktop" Not Opened appears, click Done (do NOT click the blue Move to Trash button).
  3. Open System Settings → Privacy & Security, scroll down to Security, and click Open Anyway (required once only; never disable Gatekeeper globally or strip quarantine attributes).

Windows 11 (x64)

  1. Right-click MFRM-Desktop-Windows-x64.zip in File Explorer and choose Extract All, keeping MFRM-Desktop.exe and the _internal folder together in the same directory.
  2. Double-click MFRM-Desktop.exe.
  3. If Microsoft Defender SmartScreen (Windows protected your PC) appears, click More info → Run anyway (required once only; never disable Windows Defender protection).

3. 日本語でのご案内（macOS・Windows 公式直接配布版）

macOS（Apple Silicon / M1以降）版・Windows 11（x64）版ともに、GitHub Releases から直接ダウンロードできるようになりました（Python同梱・追加環境構築不要）。

初回起動の3ステップ（未公証・未署名アプリの確認画面について）

無償の学術オープンソースソフトウェアとして有償の Apple Developer Program 公証や商用コード署名を付与していないため、初回起動時のみOS標準の確認画面が表示されます（全ソースコードをGitHubで公開しており、Wi-Fiを切った完全オフライン状態でもすべての分析・作図・保存が動作します）。

  • macOS（Sequoia 15以降 / Sonoma 14以降）：
    1. MFRM-Desktop-macOS-arm64.zip を展開し、MFRM-Desktop.app を Applications などの書き込み可能な場所へ置きます。
    2. 開いた際に 「“MFRM-Desktop” は開いていません」 と表示されたら、青い「ゴミ箱に入れる」は押さず、下の「完了」をクリックします。
    3. **「システム設定」→「プライバシーとセキュリティ」**の下部にある 「このまま開く」 を押して許可します（初回のみ。Gatekeeper自体は絶対に無効化しないでください）。
  • Windows 11（x64）：
    1. MFRM-Desktop-Windows-x64.zip を右クリックして 「すべて展開」 を選び、MFRM-Desktop.exe と _internal フォルダーを同じ場所に保ちます。
    2. MFRM-Desktop.exe をダブルクリックします。
    3. SmartScreen（「Windows によって PC が保護されました」）が表示された場合は、「詳細情報」→「実行」 を選択します（初回のみ。Windows Defender自体は絶対に無効化しないでください）。

4. Highlights in desktop-20261031 (ac98697)

  • Unified GitHub Release & package hygiene: Both macOS (arm64) and Windows (x64) standalone ZIPs are now hosted directly as GitHub Release assets with custom MFRM logit-scale application icons (.icns / .ico), Windows VSVersionInfo metadata, uncompressed binaries (upx=False), and GitHub Actions Sigstore SLSA build provenance on the Windows archive.
  • 55.8× faster vectorized estimation engine: Vectorized sufficient-statistic and Hessian assembly in mfrm_app/engine.py with machine-precision numerical equivalence across RSM, PCM, JMLE, and fixed-SD MML.
  • Refined G/D-Study workflow & mathematics: Built-in 2-facet balanced sample (GD_ANOVA), automatic exclusion of Person/Score from facet pickers, preflight cell-balance diagnostics, fixed-facet conditioning for ≥3-facet balanced data, closed-form projections for arbitrary positive integer D-Study designs, centered-tensor orthogonal projection for ANOVA residual sums of squares, and concise table-specific bilingual notes instead of persistent warning banners on normal runs.
  • Unified bilingual terminology: Standardized ability / severity in English and 相 (facet) / G/D-Study in Japanese across UI, plots, reports, and documentation.

Verification & Scope Boundary

  • Build & CI verification: Built from ac986978673dfaa8bcbed6ff2c8735b1e9ff342c (app_version = 0.2.19-beta). Automated test suites and frozen 6-workflow + zero-re-estimation .mfrm reopen checks passed on Linux Python 3.11/3.12 (Run 41064085430), Windows Server 2025 x64 (Run 41064085216), and macOS arm64. Representative interactive Windows 11 Home 26H2 GUI checks were recorded on the preceding preview candidate, whereas automated CI executed the ac98697 frozen .exe workflows; clean multi-DPI OS acceptance remains scheduled for stable release. Keep backups of older .mfrm files when updating preview builds, as schema-1 releases cannot open newer extended projects.
  • Statistical & license scope: Read the note attached to each figure/table (SE or interval cells left blank mean withheld, never zero; see Interpretation & Reporting Limits). Provided free of charge under the MIT License "AS IS" without warranty; all responsibility for data protection and substantive interpretation rests with the user. Cite as: Komuro, R. (2026). MFRM Desktop (Version 0.2.19-beta) [Computer software]. https://github.com/Ryuya-dot-com/MFRM_Python_Application (recording tag desktop-20261031 and archive name).
Optional: SHA-256 Checksums & Cryptographic Build Provenance Verification

GitHub displays the verified sha256 digest directly next to each asset below, matching SHA256SUMS.txt:

6a357542fac69335b0aefed398e7bcc8bd3ed14e7e2fdf04675c5af1bcaefd00  MFRM-Desktop-macOS-arm64.zip
f40c6ecb9785b8eb0467616437bf6747f25152e311c4a3af06c483353913d722  MFRM-Desktop-Windows-x64.zip
11257abb483a9fe56a2bf3c09eb7b62b335ca4b95d130abfb6efe9ba9165df48  MFRM-Desktop-examples.zip
  • macOS Terminal (optional): shasum -a 256 ~/Downloads/MFRM-Desktop-macOS-arm64.zip
  • Windows PowerShell (optional): Get-FileHash "$HOME\Downloads\MFRM-Desktop-Windows-x64.zip" -Algorithm SHA256
  • Windows GitHub Actions SLSA Provenance (optional, via GitHub CLI): 
    gh attestation verify ~/Downloads/MFRM-Desktop-Windows-x64.zip --repo Ryuya-dot-com/MFRM_Python_Application
Assets 6
Loading