<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jake Jang on Medium]]></title>
        <description><![CDATA[Stories by Jake Jang on Medium]]></description>
        <link>https://medium.com/@jake.j_23848?source=rss-ed01137b8bdc------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*KNMko-PXsyN7MOsO</url>
            <title>Stories by Jake Jang on Medium</title>
            <link>https://medium.com/@jake.j_23848?source=rss-ed01137b8bdc------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 20:21:34 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jake.j_23848/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Tokamak Network ZKP, 2025 Q3 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-network-zkp-2025-q3-report-a4fc62513e76?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/a4fc62513e76</guid>
            <category><![CDATA[research]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Tue, 25 Nov 2025 07:31:25 GMT</pubDate>
            <atom:updated>2025-11-26T23:15:39.755Z</atom:updated>
            <content:encoded><![CDATA[<h4>Leap Forward to “Tokamak Private App Channels”</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SrrLpouFTsiXtcqBKegP_g.png" /></figure><p>Hi, this is Jake, leading the Tokamak Network ZKP project. We’re excited to share our progress for 2025 Q3.</p><h4>Useful Related Articles</h4><ul><li><em>Introduction to </em><a href="https://medium.com/tokamak-network/project-tokamak-zk-evm-67483656fd21"><em>Project “Tokamak N</em>etwork ZKP”</a> (formerly, “Tokamak zk-EVM”)</li><li><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2025-q2-report-58179a44d27d"><em>Tokamak zk-EVM 2025 Q2 report</em></a></li></ul><h3>Tokamak zk-SNARK Optimization</h3><p>We’ve further reduced proof generation time with our solution, Tokamak zk-SNARK. Now you can turn your Ethereum transaction batch into a zero-knowledge proof on your home laptop — faster than you can cook instant ramen.</p><ul><li>Test box: CPU (<strong>14‑core M4 pro with 48 GB RAM) </strong><em>or</em> GPU (<strong>RTX 5060 ti GPU with 16 GB RAM)</strong></li><li>Circuit capacity: 1–8<strong> Ethereum transactions </strong>depends on transaction complexity (in case of TON transfers, 5 transactions)</li><li>Result: <strong>144 seconds</strong> with CPU <em>or</em> <strong>37 seconds</strong> with GPU</li></ul><blockquote>Source code: <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/dev">https://github.com/tokamak-network/Tokamak-zk-EVM/tree/dev</a></blockquote><h3>Tokamak zk-EVM App</h3><p>We have provided a standalone app with a graphical interface that handles the entire Tokamak zk-EVM pipeline for you. You can apply a ZKP to your Ethereum transaction without writing code or needing technical expertise. In short, it makes the entire transaction-to-ZKP process understandable and repeatable for anyone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F5qLtw1sMMuyE9HoNDDtSA.png" /></figure><p>We’ve significantly simplified the app’s graphical user interface. You can convert transactions you want to convert to ZKP. You type (or paste) a transaction hash and press a button. The app then fetches the transaction, builds a circuit, generates a ZK proof, and lets you verify it — end to end, in one place.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FPdbzobSY-ME%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DPdbzobSY-ME&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FPdbzobSY-ME%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/af6f1882bb52fefddf8bcc90cf3ede9b/href">https://medium.com/media/af6f1882bb52fefddf8bcc90cf3ede9b/href</a></iframe><p><em>App download: </em><a href="https://github.com/tokamak-network/Tokamak-zk-EVM-playgrounds/releases/tag/1.0.0"><em>https://github.com/tokamak-network/Tokamak-zk-EVM-playgrounds/releases/tag/1.0.0</em></a></p><h3><strong>Successful Conclusion of the TON Airdrop Event</strong></h3><p>We held a TON airdrop for about two weeks last September. The prize pool was awarded to participants who used the Tokamak zk-EVM app and shared brief feedback. Through this event, we received dozens of meaningful and constructive responses. All rewards were distributed fairly and transparently via Ethereum transactions.</p><p>Thank you for your interest and participation — and please look forward to our next airdrops.</p><p><em>Official Airdrop announcement: </em><a href="https://medium.com/tokamak-network/tokamak-giving-away-4-500-ton-tokens-heres-how-to-get-yours-73c4d181a802"><em>https://medium.com/tokamak-network/tokamak-giving-away-4-500-ton-tokens-heres-how-to-get-yours-73c4d181a802</em></a></p><h3>Leap Forward to Tokamak Private App Channels</h3><p>Now that all of our core building blocks have been developed, we’ve entered the final phase of our long journey. The reason we’ve been so committed to researching and developing ZKPs has been to create <strong>Tokamak Private App Channels</strong>.</p><p>User privacy isn’t guaranteed on Ethereum. While transparency and fairness are the very reasons blockchains like Ethereum exist, there are moments when we simply want a break — much like a teenager locking their bedroom door to enjoy a bit of privacy.</p><p>Tokamak Private App Channels are a tool for exactly that. They let you and your friends create your own “room” whenever you need one. <strong>Tokamak Private App Channels allow you to open a private channel dedicated to a specific app and exchange transactions within it.</strong> Only the final result is revealed to Ethereum. We don’t operate the channel for you; you create, manage, operate, and close it on your own. We simply provide the tools that make it easy.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F6m3H6wZEDzw%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D6m3H6wZEDzw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F6m3H6wZEDzw%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/d44e033f484a2995239bc29a76cad994/href">https://medium.com/media/d44e033f484a2995239bc29a76cad994/href</a></iframe><p>Similar concepts have been explored on Ethereum before — most notably state channels. Tokamak Private App Channels share many similarities with traditional state channels: activities inside the channel can remain private, membership is tightly controlled, and the channel’s lifetime is short.</p><p><em>Get more info about state channels: </em><a href="https://ethereum.org/developers/docs/scaling/state-channels/"><em>https://ethereum.org/developers/docs/scaling/state-channels</em></a></p><p>So what’s innovative about Tokamak Private App Channels? The key difference is this: while you and your friends can lock the room so that “your parents” (Ethereum) can’t hear your conversations, they can still verify what your group is gathering for and whether you are fulfilling a promised purpose (the app-specific). ZKPs make this possible.</p><p>Let’s get slightly more technical. <strong>A major distinction between legacy state channels and Tokamak Private App Channels is the verifiable computation of off-chain activity.</strong> Put differently: can the correctness of off-chain activity be verified?</p><p>Legacy state channels are signature-based. Correctness and consensus rely entirely on user signatures — there is no additional security model. Ideally, they work well only when every participant behaves honestly and no disputes arise. But, in practice, when disputes occur or participant liveness isn’t guaranteed, resolving issues becomes extremely difficult.</p><p><strong>Tokamak Private App Channels, on the other hand, are ZKP-based.</strong> The correctness of off-chain activity is mathematically guaranteed through ZKPs. As a result, they are highly robust against disputes and do not rely on participant liveness when reflecting the final channel state on-chain.</p><p>Within the channel, <strong>each participant is responsible for their own transactions and for generating their own ZKPs</strong>. No one else sequences or proves transactions on your behalf — if they did, your privacy would no longer be guaranteed. To make this model possible, we spent years developing our own ZK-SNARK technology. <strong>This is where we fundamentally differ from today’s mainstream rollup designs.</strong></p><h4>Join Us!</h4><p>We’re looking for passionate researchers and developers to help shape the future of ZKP-enhanced Ethereum. If you’re interested in collaborating with us, reach out:</p><ul><li><strong>General inquiries:</strong> hr@tokamak.network</li><li><strong>Technical discussions:</strong> jake@tokamak.network</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a4fc62513e76" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-network-zkp-2025-q3-report-a4fc62513e76">Tokamak Network ZKP, 2025 Q3 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2025 Q2 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2025-q2-report-58179a44d27d?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/58179a44d27d</guid>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[ethereum]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Tue, 29 Jul 2025 04:13:36 GMT</pubDate>
            <atom:updated>2025-07-29T04:13:36.669Z</atom:updated>
            <content:encoded><![CDATA[<h4>Tokamak zero-knowledge proof system is ready to use</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bANHypyYPBfpIkBiV6Uq9A.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We’re excited to share our progress for 2025 Q2.</p><h4>Useful Related Articles</h4><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>Introduction to Project Tokamak zk-EVM</em></a><em> (last updated in Apr. 2025)</em></li><li><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2025-q1-report-99e1219fb9fe"><em>Tokamak zk-EVM 2025 Q1 report</em></a></li></ul><h3>Tokamak zk-SNARK Implementation</h3><p>We finished the first production-ready version of <strong>Tokamak zk-SNARK</strong> and plugged it into Tokamak zk-EVM. With the lightweight frontend we shipped earlier, you can already generate a zero-knowledge proof that the operations inside a transaction executed correctly (assuming the signature and input state are valid).</p><p><strong>Performance</strong></p><ul><li>Test box: <strong>32 GB RAM</strong>, <strong>14‑core M4 CPU</strong> <em>or</em> <strong>RTX 3070 Ti GPU</strong></li><li>Circuit capacity: <strong>1–8 Ethereum transactions</strong></li><li>Result: <strong>≈ 7 minutes</strong> to generate a proof</li></ul><p>In short: <strong>you don’t need specialized hardware</strong>. A decent home PC is enough to try this ZK proof generation today.</p><blockquote>Github: <a href="https://github.com/tokamak-network/Tokamak-zk-EVM">https://github.com/tokamak-network/Tokamak-zk-EVM</a></blockquote><h3>Easy-to-use Interface</h3><p>We’ve developed a <strong>Playground</strong>, a standalone app with a graphical interface that handles the entire Tokamak zk-EVM pipeline for you. You type (or paste) a transaction hash, press a button, and the app fetches the transaction, builds the circuit, generates the ZK proof, and lets you verify it — end to end, in one place.</p><p>Because everything runs through this UI, you can apply a ZKP to your Ethereum transaction without writing code or needing technical expertise. In short, it makes the entire transaction-to-ZKP process understandable and repeatable for anyone.</p><blockquote>Github: <a href="https://github.com/tokamak-network/Tokamak-zk-EVM-playgrounds">https://github.com/tokamak-network/Tokamak-zk-EVM-playgrounds</a></blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FnECTxMTgcx0%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DnECTxMTgcx0&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FnECTxMTgcx0%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8b3d4deeb03cfbda7146f92568358a7a/href">https://medium.com/media/8b3d4deeb03cfbda7146f92568358a7a/href</a></iframe><h3>Pre-SNARK Consensus for Decentralized Setup</h3><p>Tokamak zk-SNARK needs a setup that runs only once. This setup is the core of any SNARK system: it guarantees security while making verification drastically cheaper. You could let a trusted third party run it, but that creates a single point of failure. One common alternative is to use secure multi‑party computation (MPC), where many participants jointly execute the setup so that no one can learn or control the secret parameters.</p><p>We applied a well-established MPC protocol to Tokamak zk-SNARK and named it ‘<strong>pre-SNARK consensus’</strong>. Its output (the parameters) is reusable; one successful ceremony is enough.</p><blockquote>Github: <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/backend/setup/mpc-setup">https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/backend/setup</a></blockquote><blockquote>Technical Doc: <a href="https://eprint.iacr.org/2024/1671">https://eprint.iacr.org/2024/1671</a></blockquote><h3>“ZKP for Everyone” — Talks at EthCluj &amp; EthCC</h3><p>We presented <strong>“ZKP for Everyone: Apply ZKP to your Ethereum Transactions in a Few Clicks”</strong> twice this quarter: <strong>June 26 at EthCluj (Cluj, Romania)</strong> and <strong>June 30 at EthCC (Cannes, France)</strong>. In both demos, we showcased the Playground live — we pasted a real Ethereum transaction hash, generated the proof, and verified it on stage.</p><blockquote>EthCC talk video: <a href="https://www.youtube.com/live/aNDTH-JqXYY?si=uA0VW8Io_K2NP32d">https://www.youtube.com/live/aNDTH-JqXYY?si=uA0VW8Io_K2NP32d</a></blockquote><blockquote>Presentation slides (full version): <a href="https://docs.google.com/presentation/d/1D49fRElwkZYbEvQXB_rp5DEy22HFsabnXyeMQdNgjRw/edit?usp=sharing">https://docs.google.com/presentation/d/1D49fRElwkZYbEvQXB_rp5DEy22HFsabnXyeMQdNgjRw/edit?usp=sharing</a></blockquote><h3>What’s Next?</h3><p>Our next goal is to show that a <strong>batch of transactions can be replaced entirely by a single ZKP</strong>. We will move Tokamak zk-EVM from an execution-only proof to a full transaction proof, add a proof aggregation scheme, and update the Playground accordingly.</p><p>To make a batch proof complete, it must cover: (1) signature &amp; sender verification, (2) input state integrity, (3) execution correctness, and (4) post‑state update.</p><p>Today, Tokamak zk-SNARK still assumes (1), (2), and (4); it only proves (3). Next, we remove those assumptions and include all four requirements in the proof.</p><p>Once implemented, the transaction itself becomes optional — the proof will serve its replacement. That enables our target scenario: small and mid-sized user groups spin up short‑lived Layer 2 channels, run a limited set of pre‑agreed transaction types, generate proofs, close the channel, and submit those proofs to L1 for verification and state updates. In short, <strong>transactions are swapped out for proofs</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sEQX59lZciFueuv8dYAVSQ.png" /></figure><h4>Join Us!</h4><p>We’re looking for passionate researchers and developers to help shape the future of zk-EVM technology. If you’re interested in collaborating with us, reach out:</p><ul><li><strong>General inquiries:</strong> hr@tokamak.network</li><li><strong>Technical discussions:</strong> jake@tokamak.network</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=58179a44d27d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2025-q2-report-58179a44d27d">Tokamak zk-EVM 2025 Q2 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2025 Q1 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2025-q1-report-99e1219fb9fe?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/99e1219fb9fe</guid>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[layer-2-solution]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Thu, 10 Apr 2025 13:02:19 GMT</pubDate>
            <atom:updated>2025-04-13T08:40:39.914Z</atom:updated>
            <content:encoded><![CDATA[<h4>Upgrade Complete! Zk-SNARK Efficiency Improvements</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*T5kMIF62gryXSu0YpPXcEA.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We’re excited to share our progress for 2025 Q1.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>Introduction to Project Tokamak zk-EVM</em></a><em> (last updated in Apr. 2025)</em></li></ul><h3>Backend Optimization of Tokamak zk-SNARK</h3><p>We have identified significant room for improvement in the efficiency of the <a href="https://eprint.iacr.org/2024/507">Tokamak zk-SNARK protocol</a>. Our latest optimization dramatically simplifies computation within the verifier contract, substantially reducing computational overhead. The trade-off was a modest increase in proof size. More importantly, zk proofs are now linearized, allowing aggregation. This means multiple independent proving problems can now be efficiently verified with a single verification operation.</p><p>Specifically, the changes are:</p><ul><li><strong>Reference string size:</strong> reduced significantly — by approximately 3–4x (dependent on environmental parameters).</li><li><strong>Proof length:</strong> slightly increased by 608 bytes (to 1.9KB) when using BLS12–381 curve.</li><li><strong>Proof aggregation:</strong> all proof components can now be linearly combined.</li><li><strong>Gas efficiency:</strong> minimum gas consumption for BLS12–381 elliptic curve operations in the verifier contract is reduced by approximately 6.3x (to 592K gas).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qqTSQ3ptam3c1wI5fgTYCw.png" /></figure><h3>Implementation of Backend Algorithms</h3><h4>Algebra Library</h4><p>The backend algorithms of Tokamak zk-SNARK involve manipulating bivariate polynomials. We’ve successfully extended all algebraic functions — originally provided by ICICLE for univariate polynomials — to efficiently support bivariate polynomials without sacrificing computational efficiency. Functions include<strong> 2D-NTT</strong> and <strong>efficient multiplications and divisions on bivariate polynomials</strong>. You can review our implementations in <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/blob/dev/packages/backend/libs/src/polynomials/mod.rs">our repository</a>.</p><h4>Current Implementation Progress (as of March 2025)</h4><ul><li><strong>Setup:</strong> Implementation in progress, supporting both <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/blob/dev/packages/backend/setup/trusted-setup/src/main.rs">trusted</a> and <a href="https://github.com/tokamak-network/SNARK-MPC-setup/tree/ceremony-demo/Tokamak_zkSNARK">secure multi-party computation (MPC)</a> setups.</li><li><a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/jake-backend/packages/backend"><strong>Prove</strong></a><strong>:</strong> Implementation is planned for the next quarter.</li><li><a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/jake-backend/packages/backend"><strong>Verify</strong></a><strong>:</strong> Implementation is planned for the next quarter.</li></ul><h3>Updates on Frontend Implementation</h3><p>The Synthesizer’s <a href="https://github.com/tokamak-network/Tokamak-zk-EVM-playgrounds/tree/dev/packages/synthesizer-playground">graphical user interface (GUI)</a> now supports ERC-20 standard transactions for TON, USDT, and USDC contracts. You can easily test the Synthesizer using random transactions from <a href="https://etherscan.io/">Etherscan.io</a> — just provide the transaction hash.</p><p>However, we emphasize that the <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/frontend/synthesizer">Synthesizer</a>’s capability is not limited to TON, USDT, USDC, or ERC-20 transactions. It can process <strong>any transaction and contract </strong>that runs on an EVM, as long as the initial state of the contract being invoked is properly prepared. For testing purposes, the GUI currently provides dummy contract states for those contracts. (In the future, we plan to update the Frontend to support ZKP-style retrieval of initial contract states as well.)</p><h3>What’s Next?</h3><p>Our key priority for the second quarter of 2025 is demonstrating how easily anyone can convert random Ethereum transactions into ZK proofs — just by clicking a few buttons. We believe this demonstration will provide tangible proof of the project’s practical value.</p><h3>Join Us!</h3><p>We’re looking for passionate researchers and developers to help shape the future of zk-EVM technology. If you’re interested in collaborating with us, reach out:</p><ul><li><strong>General inquiries:</strong> hr@tokamak.network</li><li><strong>Technical discussions:</strong> jake@tokamak.network</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99e1219fb9fe" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2025-q1-report-99e1219fb9fe">Tokamak zk-EVM 2025 Q1 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2024 Q4 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2024-q4-report-01d91e90c3d8?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/01d91e90c3d8</guid>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[layer-2]]></category>
            <category><![CDATA[rollup]]></category>
            <category><![CDATA[research]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Fri, 07 Feb 2025 07:50:34 GMT</pubDate>
            <atom:updated>2025-02-10T06:46:06.764Z</atom:updated>
            <content:encoded><![CDATA[<h4>User-friendly front-end for Tokamak zk-EVM</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2QLMx8wr3-Gv4nGmZaHDJg.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We’re excited to share our 2024 Q4 progress.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>About the Tokamak zk-EVM project</em></a></li><li><a href="https://drive.google.com/drive/folders/1szmsbN0hKgzaHZ24fQ3tk7WOVH_yq3Ms?usp=sharing"><em>Presentation slides</em></a><em> (ETHTOKYO24, ETHKL24)</em></li></ul><h3>Tokamak zk-EVM’s front-end compilers</h3><p>During the fourth quarter, we finally released the front-end compilers for Tokamak zk-EVM. These include:</p><ul><li><a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/frontend/qap-compiler"><strong>QAP Compiler</strong></a><strong>: </strong>Takes EVM’s specifications as input and generates <a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/frontend/qap-compiler/subcircuits/library"><em>a library of subcircuits</em></a> for fundamental EVM operations.</li><li><a href="https://github.com/tokamak-network/Tokamak-zk-EVM/tree/main/packages/frontend/synthesizer"><strong>Synthesizer</strong></a>: Uses the subcircuit library to convert Ethereum transactions into <em>transaction-specific circuits</em>.</li></ul><p>These front-end compilers serve as a bridge between <a href="https://eprint.iacr.org/2024/507">Tokamak zk-SNARK</a> and Ethereum by providing optimized ZKP circuits for Ethereum transactions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*e_v_31sugNGzI0Pq.png" /><figcaption>Signal processing flow of Tokamak zk-EVM. The front-end algorithms (blue) bridges Ethereum applications with the back-end algorithms (green) of Tokamak zk-SNARK.</figcaption></figure><h4><strong>User-friendly compiler for everyone</strong></h4><p>Unlike previous compilers, our front-end compiler doesn’t require users to have any technical knowledge of ZKP. Traditional compilers often required users to manually write ZKP circuits for their target program in unfamiliar languages like Circom or Noir, placing the burden of optimization and security on them.</p><p>In contrast, our compilers automate the circuit generation process. Synthesizer simplifies this process by requiring only one user input: <em>the hash of a transaction to be convert into a circuit.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sWYUyFg0GG013InrnWOM2g.png" /></figure><h4><strong>Current development status</strong></h4><p>As of Feb. 7 2025, the release is a demo version, with Synthesizer currently supporting transactions based on the <a href="https://etherscan.io/token/0x2be5e8c109e2197d077d13a82daead6a9b3433c5">TON contract</a>.</p><p>We are actively working on expanding support to almost contracts and transaction types in the near future.</p><h4><strong>Synergy with Tokamak zk-EVM’s back-end algorithm</strong></h4><p>The primary goal of Tokamak zk-EVM’s back-end algorithm is to create a <em>fast and lightweight back-end</em>. Specifically, we aim to enable ZKP proof generation on a standard personal computer, eliminating the need for high-performance equipment or expensive servers. This encourages broad participation in decentralized networks as ZKP provers.</p><p>This back-end optimization aligns with the front-end compiler’s objective: <em>to allow anyone to convert Ethereum transaction validation problems into ZKP problems</em>. In the Tokamak zk-rollup we are developing, <strong>anyone will be able to launch and operate their own specialized rollup network — without requiring technical expertise or complex infrastructure.</strong></p><p>This will lead to a truly decentralized rollup ecosystem, both in purpose and quantity.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/825/1*tW9h0sNndOHB7iIxnSMm6A.png" /><figcaption>Rollup networks decentralized in purpose and quantity.</figcaption></figure><h3>What’s Next?</h3><p>Our key priorities for 2025 Q1 include:</p><ul><li>Expanding Synthesizer’s compatibility and stability for broader transaction support,</li><li>Releasing demo versions of Tokamak zk-EVM’s back-end algorithms.</li></ul><h3>Join Us!</h3><p>We’re looking for passionate researchers and developers to help shape the future of zk-EVM technology. If you’re interested in collaborating with us, reach out:</p><ul><li><strong>General inquiries:</strong> hr@tokamak.network</li><li><strong>Technical discussions:</strong> jake@tokamak.network</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=01d91e90c3d8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2024-q4-report-01d91e90c3d8">Tokamak zk-EVM 2024 Q4 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2024 Q3 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2024-q3-report-e3a082fabc93?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/e3a082fabc93</guid>
            <category><![CDATA[rollup]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[layer-2-solution]]></category>
            <category><![CDATA[research]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Tue, 22 Oct 2024 18:01:59 GMT</pubDate>
            <atom:updated>2025-02-05T04:27:03.758Z</atom:updated>
            <content:encoded><![CDATA[<h4>Zk-EVM, more feasible, more decentralized</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JoZoFOsOIXzJrsP7GS4ktA.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We’re excited to share our Q3 2024 progress.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>About the Tokamak zk-EVM project</em></a></li><li><a href="https://drive.google.com/drive/folders/1szmsbN0hKgzaHZ24fQ3tk7WOVH_yq3Ms?usp=sharing"><em>Presentation slides</em></a><em> (ETHTOKYO24, ETHKL24)</em></li></ul><p>During the third quarter, we finally completed the design of the core <em>trinity</em> of our zk-EVM.</p><p>While <a href="https://diablo.fandom.com/wiki/Great_Evils">Blizzard’s Diablo II</a> has three Prime Evils and two Lesser Evils, Tokamak zk-EVM has three core protocols, called <em>the trinity</em>, and two front-end compilers.</p><p>The three core protocols are:</p><ul><li><a href="https://eprint.iacr.org/2024/507">Tokamak zk-SNARK</a></li><li><a href="https://eprint.iacr.org/2024/1671">A pre-SNARK consensus protocol</a> (also known as a secure multi-party computation of the Tokamak SNARK’s setup)</li><li><a href="https://eprint.iacr.org/2024/1664">A post-SNARK consensus protocol</a></li></ul><p>The two front-end compilers are:</p><ul><li><a href="https://github.com/tokamak-network/circom-ethereum-opcodes/tree/main">Verifiable sub-circuits for EVM arithmetic and logical operations</a></li><li><em>Synthesizer</em> that converts an Ethereum transaction into a wire map (also known as a permutation in PlonK’s terms)</li></ul><p>We just drafted the three protocol designs. In 2024 Q4, we will be focusing on the two front-end compilers.</p><h3>Recap of Tokamak zk-EVM</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*P486lvyvN7ukGafxzcfOgw.jpeg" /></figure><p>In our definition, zk-EVM means the execution of transactions where EVM is replaced with a SNARK. Thus, Tokamak zk-EVM performs <em>verifiable computation of contract transactions</em>. Next year, we will extend this concept to <em>verifiable computation of any state differences</em>, aiming forTokamak zk-rollup.</p><p>Tokamak zk-EVM takes a completely new approach that is technically different from the zk-EVMs of Scroll or Linea, or zk-VM of RISC Zero. Unlike their giant machine circuits that mimic the whole EVM or RISC-V, we adopt lightweight, transaction-specific circuits (or rollup-specific, later). This makes Tokamak zk-EVM feasible for anyone to generate zero-knowledge proofs anytime, anywhere.</p><p>Of course, in engineering, gains come with losses. We have faced many technical difficulties, and to overcome them, we configured <strong>the three core protocols</strong> and two front-end compilers as above. This contrasts the other zk-EVMs that can verify EVM transactions with only one circuit and one SNARK protocol.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8DoTxLWxtUA6TCqXwNyPQQ.jpeg" /></figure><h3>The trinity</h3><p>The three core protocols will interact with each other to complement their weaknesses. Let me explain their strengths and weaknesses one by one.</p><h4>Tokamak zk-SNARK</h4><p>Tokamak zk-SNARK is a hybrid of Groth16 and PlonK in the view that it uses both <strong>pre-defined sub-circuits</strong> (in the form of common-reference string) and <strong>transaction-specific wire maps</strong> (in the form of permutations).</p><p>Tokamak zk-SNARK adopts transaction-specific circuits. A circuit is a combination of the sub-circuits, where their input and output wires are connected according to a wire map. As with all other SNARKs, Tokamak zk-SNARK also relies on three assumptions:</p><ol><li>The prover and the verifier are using the same sub-circuits and wire map,</li><li>there is a random encoding that the prover doesn’t know the randomness,</li><li>the verifier can access pre-processed commitments to the sub-circuits and wire map based on the random encoding.</li></ol><p>These assumptions are necessary for succinct verification, preventing a malicious prover from deceiving the verifier.</p><p>To sum up, Tokamak zk-SNARK has the following strength and weakness:</p><ul><li>Strength: It provides algorithms for highly efficient verificable computation of transactions.</li><li>Weakness: How can the three assumptions above be realized?</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z7IoHC3arnUfbdTPdtrZWA.jpeg" /></figure><h4>Pre-SNARK consensus (a.k.a. MPC)</h4><p>In the pre-SNARK consensus paper, we propose a secure Multi-Party Computation (MPC) protocol for committing to the pre-defined sub-circuits. In simple terms, it explains how the Tokamak zk-SNARK verifier reliably takes over the random encoding and commitments to the <strong>pre-defined sub-circuits</strong> from distributed and untrusted participants.</p><p>MPC is a practical and well-established technology that enables untrusted participants to collaborate in building a random encoding and generating commitments to the sub-circuits. Under the assumption that at least one participant of the protocol is honest (which is quite practical), it ensures that none of the participants can exploit their random seeds to deceive the SNARK verifier. Additionally, all participants must input the same sub-circuits, ensuring no conflicts.</p><p>However, the consistency of the sub-circuits does not necessarily guarantee integrity of the sub-circuits between the MPC participants and the SNARK verifier. If the first MPC participant, say Alice, corrupts the sub-circuits, the MPC protocol has no explicit mechanism for others to detect or correct the corruption. If the other participants strictly follow the protocol and do nothing outside of it, they have to obey Alice. As a result, Alice can potentially become a malicious SNARK prover who could deceive the SNARK verifier.</p><p>Moreover, the cost of MPC verification performed by each participant is too high to run on the blockchain. If performed off-chain and the result is used on-chain for the SNARK verification, there is a serious security risk as this process relies on an oracle, which introduces trust issues.</p><p>To sum up, pre-SNARK consensus has the following strength and weakness:</p><ul><li>Strength: It is a practical way to realize Tokamak zk-SNARK’s assumptions 2 and 3.</li><li>Weakness: There is no guarantee for Tokamak zk-SNARK’s assumption 1. It is too expensive to run on a blockchain.</li></ul><h4>Post-SNARK consensus</h4><p>In the post-SNARK consensus paper, we propose a protocol to reach consensus on a <strong>wire map</strong> among distributed and untrusted participants. The protocol returns the wire map that is backed by the largest group of participants. The resulting wire map is considered honest and delivered to the SNARK verifier.</p><p>In this protocol, each participant claims a wire map, which may be different or identical across participants. Among them, the circuits claimed by honest nodes must be identical. The role of the protocol is to cluster participants that claim the same wire map. Each can support only the wire maps they fully know.</p><p>While this protocol cannot securely build a random encoding, the overall communication and computational complexity, including verifying participants’ knowledge of wire maps is cheap enough to run on a blockchain. Unfortunately, it has weaker security as it resists collusion only if there are at least 50% honest participants.</p><ul><li>Strength: Given a random encoding, it realizes Tokamak zk-SNARK’s assumptions 1 and 3. It is computationally cheap enough to be run on a blcockahin.</li><li>Weakness: It doesn’t provide a random encoding to realize Tokamak zk-SNARK’s assumption 2. At least 50% honest participants are needed to prevent collusion.</li></ul><h3>What’s next?</h3><p>Our goals for the fourth quarter will be</p><ul><li>Implementation of the three core protocols.</li><li>Development of the front-end compilers.</li></ul><h3>Contact us</h3><p>We’re actively seeking creative researchers and developers. Reach out to us at <a href="/service/mailto:%20hr@tokamak.network">hr@tokamak.network</a>. For technical discussions, contact <a href="/service/mailto:jake@tokamak.network">jake@tokamak.network</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e3a082fabc93" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2024-q3-report-e3a082fabc93">Tokamak zk-EVM 2024 Q3 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2024 Q2 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2024-q2-report-0e3355e0c8e8?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/0e3355e0c8e8</guid>
            <category><![CDATA[cryptography]]></category>
            <category><![CDATA[layer-2-solution]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Tue, 23 Jul 2024 19:38:39 GMT</pubDate>
            <atom:updated>2024-07-23T19:41:30.264Z</atom:updated>
            <content:encoded><![CDATA[<h4>Prepare for Takeoff</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lZhklry24HrisaYsmHxEbw.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We’re thrilled to share our progress for the second quarter of 2024.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>About the Tokamak zk-EVM project</em></a></li></ul><p>During the second quarter, we’ve been busy preparing to share our story with the world. While we’re not quite ready to launch a service or demo a product, our strategy has become more detailed and is starting to meet the needs of the Ethereum ecosystem. We’re also working hard on documenting and implementing our concepts.</p><h3>Events We’ll Speak At</h3><p>This November, <a href="https://devcon.org/en/">DEVCON</a>, the biggest conference for Ethereum contributors, will take place in Bangkok. In addition, various events are scheduled across Asian countries in the second half of the year. Here are events where we will share our story:</p><ul><li><a href="https://www.ethtokyo.com/">ETHTOKYO</a>: It is an Ethereum community hackaton held in Tokyo, Japan from August 23 to 26, 2024.</li><li><a href="https://www.2024.ethkl.org/">ETHKL</a>: It is another Ethereum community hackaton held in Kulal Lumpur, Malaysia from October 4 to 6, 2024.</li></ul><p>We’re also looking forward to seeing you at DEVCON.</p><h3>Progress in research and development</h3><p>Besides preparing for our speeches, we’ve made good progress in R&amp;D.</p><h4>Research</h4><p>As we mentioned earlier, <a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21">our zk-EVM won’t rely solely on ZKPs or SNARKs; it will also utilize multi-party computation and threshold cryptography</a>. We’ve reviewed extensive literature on existing protocols, ranging from traditional distributed key generation based on Shamir’s secret key sharing to Snarky ceremonies and threshold cryptography for verifying a signature of correct computation. With this background, we’ve set out to design our own protocols.</p><h4>Development</h4><p>We’ve improved the compatibility of Tokamak zk-EVM subcircuits with EVM arithmetic operations. We’re now developing a compiler that takes unrolled transaction bytecode as input and generates a <em>wire map</em>, guiding how to combine the subcircuits to derive an optimized transaction circuit. This compiler is the last unimplemented element in the front-end of the <a href="https://medium.com/tokamak-network/tokamak-zk-evm-2024-q1-report-5b31e4beb2f0">Tokamak zk-SNARK</a>.</p><h3>What’s next?</h3><ul><li>Documenting secure multiparty computation for Tokamak zk-SNARK’s setup.</li><li>Documenting threshold cryptography for distributed verification of wire maps.</li><li>Developing the Tokamak zk-SNARK compiler.</li></ul><h3>Contact us</h3><p>We’re actively seeking creative researchers and developers. Reach out to us at <a href="/service/mailto:%20hr@tokamak.network">hr@tokamak.network</a>. For technical discussions, contact <a href="/service/mailto:jake@tokamak.network">jake@tokamak.network</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0e3355e0c8e8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2024-q2-report-0e3355e0c8e8">Tokamak zk-EVM 2024 Q2 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Can SNARK compress blockchain data in practice?]]></title>
            <link>https://medium.com/tokamak-network/can-snark-compress-blockchain-data-in-practice-b763f5397332?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/b763f5397332</guid>
            <category><![CDATA[zkrollup]]></category>
            <category><![CDATA[layer-2-solution]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[information]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Wed, 03 Jul 2024 05:59:47 GMT</pubDate>
            <atom:updated>2024-07-03T06:20:14.950Z</atom:updated>
            <content:encoded><![CDATA[<ul><li><em>Note: I have experience writing </em><a href="https://ieeexplore.ieee.org/document/8334251"><em>an information theory-applied research paper</em></a><em> and </em><a href="https://eprint.iacr.org/2024/507"><em>a SNARK design research paper</em></a><em>. The conclusion of this article is only a personal opinion derived from past experiences and has no academic basis.</em></li><li><em>Background knowledge: This article requires an understanding of SNARK and verification circuits. Please refer to </em><a href="https://medium.com/tokamak-network/project-tokamak-zk-evm-67483656fd21"><em>my previous post</em></a><em> for a brief explanation.</em></li><li><em>한글 번역본은 </em><a href="https://medium.com/@jake.j_23848/snark%EB%8A%94-%EC%A0%95%EB%A7%90-%EB%B8%94%EB%A1%9D%EC%B2%B4%EC%9D%B8-%EB%8D%B0%EC%9D%B4%ED%84%B0%EB%A5%BC-%EC%95%95%EC%B6%95%ED%95%A0-%EC%88%98-%EC%9E%88%EC%9D%84%EA%B9%8C-4be8fd05aa97"><em>여기서</em></a><em> 찾을 수 있습니다.</em></li></ul><blockquote>Can SNARK compress blockchain data?</blockquote><p>This question was motivated by Ian Miers’ presentation ‘Scalability is Boring, Privacy is Dead: ZK-Proofs, What are They Good for?’ at Devcon6 held in Bogota, Colombia in 2022.</p><p><a href="https://archive.devcon.org/archive/watch/6/scalability-is-boring-privacy-is-dead-zk-proofs-what-are-they-good-for/?tab=YouTube">Scalability is Boring, Privacy is Dead: ZK-Proofs, What are They Good for?</a></p><p>To conclude, my answer to this question is as follows:</p><blockquote>SNARK can compress blockchain data, but it inevitably results in degeneration in blockchain security or decentralization.</blockquote><p>This article explains the reasons for the above answer from the perspective of information theory.</p><h3>Information Theory?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/284/0*fjaZJLx5C1CVU0Tj.jpg" /><figcaption>Claude E. Shannon (picture from Wikipedia)</figcaption></figure><p>Information theory is an applied mathematics that has been studied since the 1920s and is the study of the amount of information contained in random variables. Information theory begins with measuring the amount of information, called entropy, of variable <em>X</em>. Claude E. Shannon, the father of information theory, defined entropy as follows:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*pis6oBP493r_nIeuOUO5IA.png" /></figure><p>Here, the domain of addition is the set of all possible events that <em>X</em> can output. The unit of entropy is “bits,” which is the same unit we use in computer science to measure word size, bits (you can also use other units for entropy by changing the base of the logarithm function). The entropy is designed to be larger as the variable becomes more uncertain, that is, the more uniform the probability distribution of the events. Conversely, if the variable is deterministic, the entropy equals 0 bits.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/439/1*r05ElzVij1DN8SW-3kEXCA.png" /><figcaption>Let X be a random variable that outputs either 0 or 1. This graph shows the entropy of <em>X</em> according to the probability of X=1 (image from Wikipedia)</figcaption></figure><p>Information theory explains the theoretical limit on the compression rate of non-deterministic data: in order to represent data without losing any uncertainty implied by a non-deterministic source <em>X</em>, the theoretical result says that the dimension of the representation medium (e.g., a field or a vector) must be at least equal to or greater than the entropy of <em>X</em>.</p><h4>Examples of utilizing the entropy</h4><figure><img alt="" src="https://cdn-images-1.medium.com/proxy/1*nfd5a7DB6beQFp1jQD7dQw.png" /></figure><p>For example, suppose we throw a dice of 8 sides and observe the value on the top side. Let <em>X</em> be defined as a random variable of the value. By definition, the entropy of X is 3 bits. Information theory tells us that to record the observation of X, we need a binary string at least 3-bit long.</p><p>Let me give you another example. Suppose we are recording the observation of a random variable <em>Y</em> in a 3-bit long binary string, and the following rules are found from the observation: <em>1)</em> at least one of the three bits must be 1; <em>2)</em> If the first or second bit of the three bits is 1, the next bit must be 0. The only possible observed strings under these rules are 001, 010, 100, and 101. Assuming that these outputs occur equally likely, that is, each with a probability of 1/4, the entropy of <em>Y</em> equals 2 bits. If you want to reduce the capacity of an existing 3-bit string that is storing an observation of <em>Y</em> by any signal processing, the capacity of the result must be equal to or larger than 2-bit.</p><p>These results for one-dimensional random variables can be extended to multidimensional random vectors (vectors of random variables). The limit on lossless compression of a vector (<em>array</em> in programmers’ terminology) is determined by the entropy of the vector. In other words, compression of a vector into a vector with a dimension smaller than the entropy inevitably results in loss of information, which means that it cannot be returned to the original vector. Examples of lossless compression algorithms that we are familiar with are zip, rar, or 7z, and an example of lossy compression is converting a high-quality Tiff format image into a low-quality JPEG format.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/296/1*hU2uICE6gZeKxDQYSTtnSA.png" /></figure><h3>Are SNARKs compression algorithms?</h3><p>Let’s return to the question “Can SNARK compress blockchain data in practice?” To answer this question, let’s first assume that SNARKs’ provers are compression algorithms. So, which type does the SNARK prover fall into: lossy compression or lossless compression? To check this, let us first look into the uncertainty in “witness”, which is the data that the SNARK prover is trying to compress.</p><p>The witness is a private vector that a SNARK’s prover creates and discards during proof generation. Under the assumption that the SNARK prover is a compression algorithm, the witness can be said to be the original data before compression. From the perspective of SNARK’s verifier, the witness is non-deterministic and cannot be predicted at all without hints, so it can be said to have the maximum entropy. That means there is no room for lossless compression. According to the information theory we discussed above, it should be impossible for a SNARK verifier to restore a witness from the proof without loss.</p><h4>A goal of SNARKs</h4><p>Information theory was negative about the SNARKs’ compression capability. However, SNARK’s goals may be different, so let’s check them carefully. Which type of compression does the SNARK prover aim for: lossless or lossy? The answer can be found in the definition of knowledge soundness, the second of SNARK’s three security definitions: completeness, knowledge soundness, and zero-knowledge. The point of Knowledge soundness is “Is there a hypothetical extraction algorithm (extractor) that can restore a very large dimensional witness from a very small dimensional proof with overwhelming probability close to 1?”. In other words, the SNARK prover aims for lossless compression.</p><p>Since most SNARK protocols have proven knowledge soundness in their papers, at first glance, it seems as if they have designed a prover that can perform lossless compression even though it deals with data that cannot be losslessly compressed information-theoretically. And most of you who have heard of zk-rollup will agree with that. However, there is a fatal assumption hidden here.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/497/1*DlFGebrgs0TQWy6E8zpPjg.png" /></figure><h4>A fatal assumption behind SNARKs for blockchain data compression</h4><p>An implicit assumption behind SNARKs is that the verifier has access to Oracle polynomials. A set of these oracle polynomials is called the verification circuit and gives rules to the witness. The verification circuit is also random vectors or matrices having nonzero entropy. Thus, this assumption implies that the verifier already has a very large amount of information. The SNARKs’ hypothetical extractors that restore a witness from a proof could exist only under this assumption (we are only talking about the existence; in reality, the verifier does not perform the extraction).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QU03UH8Ta0WCF_jO.png" /><figcaption>Example of a verification circuit</figcaption></figure><p>The SNARKs’ knowledge soundness can also be explained in information theory. Although a witness without any hints has the maximum amount of information, if there is a condition that some additional information is given such as a verification circuit, the conditional entropy of the witness becomes 0 bits. This is because the verification circuit determines all the rules for the values ​​a witness can have. In other words, the SNARK prover is a lossless compression algorithm that utilizes the mutual information of a witness and a verification circuit.</p><p>In an application where the SNARK prover attempts to compress data for blockchain scalability, this assumption of prior information is very unrealistic. In blockchain scalability, one of the goals is to enable validator nodes to verify the correctness of large-dimensional original data with only very small-dimensional compressed data. Replacing a validator node with a SNARK verifier does not meet the goal of blockchain scalability, as the validator node already has large-dimensional data, a circuit.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/589/1*Y8mjs6GNnDVqKFRdEAhN_A.png" /></figure><h4>Practical ways to apply SNARK to blockchain scalability</h4><p>One practical way to overcome this problem is to add a new assumption that <strong>there exists a trusted third party</strong> who publishes the correct verification circuit. This assumption allows the validator nodes to perform their role simply by referencing the third-party’s publication without the need to carry verification circuits. However, This assumption inevitably implies the degeneration of blockchain decentralization.</p><p>There is also another way that utilizes distributed key generation technology based on <strong>threshold cryptography</strong>, such as Powers-of-tau (Snarky ceremonies). Simply put, it is an event in which multiple participants contribute to publishing honest verification circuits over a certain period of time, for example, two weeks. This method eliminates the need for a trusted third party, but it requires a new assumption that <strong>at least one (or more than half, depending on the protocol) of the participants are honest</strong>. This leads to the degeneration of the blockchain security.</p><p>Here, blockchain security and decentralization are elements of the <a href="https://vitalik.eth.limo/general/2021/04/07/sharding.html?ref=bankless.ghost.io">Scalability trilemma</a> defined by Vitalik Buterin in 2021.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/464/0*gZsbiwWlNrVvMbBU.png" /><figcaption>The Scalability Trilemma (Image from Vitalik’s blog)</figcaption></figure><h3>Conclusion</h3><p>To summarize the story so far, the SNARK prover can act as a lossless compression algorithm for blockchain scalability, but this requires sacrificing either security or decentralization.</p><p>Then, is there any way to compress blockchain data without losing information as well as security? The approach of Optimistic Rollup is one such attempt. Optimistic Rollup compresses Layer 2 data using a traditional lossless compression algorithm (or a variation thereof) and then uploads it to Layer 1. Assuming the compressed data is not deleted (for example, there is no<a href="https://eips.ethereum.org/EIPS/eip-4844"> blobbing</a>), it would be the best way for compression without compromising security. Unfortunately, Their way does not provide a cost-effective feature that allows validator nodes in Layer 1 to validate the correctness of the original data without data restoration.</p><p>This conclusion is in no way to diminish the value of SNARK-based scalability solutions (e.g. Zk-rollup). I deeply agree with <a href="https://vitalik.eth.limo/general/2024/05/29/l2culture.html">Vitalik Buterin’s recent blog post</a> about the need for diversity in Layer 2. Since Layer 2 solutions also cannot escape the shackles of the scalability trilemma, in my opinion, there is great value in Layer 2 by giving users options for better performance at the expense of some security or decentralization. In other words, given the security guaranteed by Layer 1, Layer 2 can provide users with the opportunity to experience a wider variety of Dapps.</p><p><a href="https://medium.com/tokamak-network/project-tokamak-zk-evm-67483656fd21">Tokamak zk-EVM and Zk-rollup</a>, which we are developing, is a new approach that combines a SNARK and a consensus protocol. It is completely different from existing zk-rollup methods such as Scroll and Linea, which rely entirely on SNARKs with complex verification circuits, and it is also completely different from Optimistic rollup, which only relies on a consensus protocol called Challenging. We hope that our challenge will provide a new basis vector in the diversity space that can be defined over the scalability trilemma.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b763f5397332" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/can-snark-compress-blockchain-data-in-practice-b763f5397332">Can SNARK compress blockchain data in practice?</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SNARK는 정말 블록체인 데이터를 압축할 수 있을까?]]></title>
            <link>https://medium.com/@jake.j_23848/snark%EB%8A%94-%EC%A0%95%EB%A7%90-%EB%B8%94%EB%A1%9D%EC%B2%B4%EC%9D%B8-%EB%8D%B0%EC%9D%B4%ED%84%B0%EB%A5%BC-%EC%95%95%EC%B6%95%ED%95%A0-%EC%88%98-%EC%9E%88%EC%9D%84%EA%B9%8C-4be8fd05aa97?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/4be8fd05aa97</guid>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[information]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[layer-2-solution]]></category>
            <category><![CDATA[zkrollup]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Wed, 03 Jul 2024 05:59:44 GMT</pubDate>
            <atom:updated>2024-07-14T16:24:15.430Z</atom:updated>
            <content:encoded><![CDATA[<ul><li><em>주의: 저는 </em><a href="https://ieeexplore.ieee.org/document/8334251"><em>정보이론 (Information theory) 응용 연구</em></a><em>와 </em><a href="https://eprint.iacr.org/2024/507"><em>SNARK 설계 연구</em></a><em>를 수행한 경험이 있습니다. 이 글의 결론은 지난 경험을 조합하여 도출한 개인적 견해이며 학술적 근거는 없습니다.</em></li><li><em>배경 지식: SNARK와 검증회로에 관한 이해를 필요로 합니다. 관련된 간략한 설명은 저의 </em><a href="https://medium.com/tokamak-network/project-tokamak-zk-evm-67483656fd21"><em>이전 게시물</em></a><em>을 참고해 주세요.</em></li><li><em>An English version can be found </em><a href="https://medium.com/@jake.j_23848/can-snark-compress-blockchain-data-in-practice-b763f5397332"><em>here</em></a><em>.</em></li></ul><blockquote>SNARK는 정말 블록체인 데이터를 압축할 수 있을까?</blockquote><p>이는 지난 2022년 콜롬비아 Bogota에서 개최된 Devcon6에서 <em>Ian Miers</em>의 발표 ‘<em>Scalability is Boring, Privacy is Dead: ZK-Proofs, What are They Good for?</em>’ 를 현장에서 보고 떠오른 질문입니다.</p><p><a href="https://archive.devcon.org/archive/watch/6/scalability-is-boring-privacy-is-dead-zk-proofs-what-are-they-good-for/?tab=YouTube">Scalability is Boring, Privacy is Dead: ZK-Proofs, What are They Good for?</a></p><p>결론부터 말씀드리면, 이 질문에 대한 저의 대답은 다음과 같습니다:</p><blockquote>SNARK는 블록체인 데이터를 압축할 수 있지만, 블록체인 보안 (Security) 혹은 분산성 (Decentralization)의 희생이 반드시 따른다.</blockquote><p>이 글에서는 위의 대답에 대한 이유를 정보 이론 (Information theory) 관점에서 설명합니다.</p><h3>정보 이론이란?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/284/0*fjaZJLx5C1CVU0Tj.jpg" /><figcaption>Claude E. Shannon (사진 출처: Wikipedia)</figcaption></figure><p>정보 이론은 1920년대부터 연구되어 온 응용 수학으로써, 확률론적 변수 (Random variable)가 내포하는 정보의 양에 관한 연구입니다. 정보이론은 변수 <em>X</em>의 정보량 (Entropy)를 측정 (Measure)하는 것에서 시작합니다. 그리고 정보 이론의 아버지라 불리는 Claude E. Shannon은 정보량을 다음과 같이 정의하였습니다:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*pis6oBP493r_nIeuOUO5IA.png" /></figure><p>여기서 더하기의 범위는 <em>X</em>가 출력 가능한 모든 사건입니다. 정보량의 단위는 “bits”인데, 이는 우리가 컴퓨터 공학에서 word size를 측정할 때 사용하는 단위인 bits와 같습니다 (로그 함수의 밑을 다른 값으로 바꿈으로써 다른 단위를 사용할 수도 있습니다). 정보량은 변수가 더욱 불확실할수록, 즉 변수가 출력 가능한 사건들의 확률 분포가 균등할수록 더 커지게 정의되었습니다. 반대로, 변수가 결정적이면 (deterministic) 정보량은 0 bits 입니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/439/1*r05ElzVij1DN8SW-3kEXCA.png" /><figcaption>그래프: 0 혹은 1을 출력하는 랜덤 변수 X가 1을 출력할 확률에 따른 X의 엔트로피 (그림 출처: Wikipedia)</figcaption></figure><p>정보 이론은 비결정적 데이터의 압축에 대한 이론적 한계를 설명합니다: 비결정적 데이터가 내포하는 모든 불확실성을 잃어버리지 않고 표현하기 위해서는, 표현 수단의 차원이 적어도 데이터의 정보량보다 같거나 커야 한다는 이론적인 결과가 있습니다.</p><h4>정보 이론의 활용 예</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/267/1*nfd5a7DB6beQFp1jQD7dQw.png" /></figure><p>예를 들어, 면이 8개인 주사위를 던져 나온 가장 윗면의 값을 <em>X</em>라고 한다면, <em>X</em>의 정보량은 정의에 의해 3bits입니다. 정보 이론은 <em>X</em>의 측정 결과를 기록하기 위해서는 적어도 3-bit 길이의 이진 문자열이 필요하다고 말합니다.</p><p>또 다른 예를 들겠습니다. 어떤 변수 <em>Y</em>의 출력을 3-bit 길이의 이진 문자열에 기록하고 있는데, 이 출력물에서 다음과 같은 규칙들이 발견되었다고 가정하겠습니다: <em>1)</em> 세 비트 중 적어도 한 비트는 1이다; <em>2)</em> 세 비트 중 첫 번째 혹은 두 번째 비트가 1이라면, 그 다음 비트는 반드시 0이다. 이러한 규칙 하에서 발생 가능한 문자열은 001, 010, 100, 그리고 101뿐입니다. 이 출력물들이 균등 확률 (Equally-likely), 즉 각각 1/4의 확률로 발생한다고 가정하면, <em>Y</em>의 정보량은 2 bits입니다. 만약 <em>Y</em>라는 변수를 기록 및 저장하고 있는 기존의 3-bit 길이의 문자열에 어떤 처리를 가하여 용량을 줄이고자 한다면, 그 결과물의 용량은 2-bit 보다는 같거나 커야 합니다.</p><p>이러한 결과는 1차원 랜덤 변수가 아닌 다차원의 랜덤 벡터 (랜덤 변수의 벡터)에도 마찬가지로 적용됩니다. 어떤 벡터 (프로그래머의 표현으로는 배열)를 무손실 (Lossless) 압축 (Compression)할 수 있는 한계는 그 벡터의 정보량에 의해 결정됩니다. 다시 말해, 어떤 벡터를 정보량보다 더 작은 차원의 벡터로 압축하면 손실(Lossy) 압축이 될 수밖에 없고, 이는 원본 벡터로 다시 되돌릴 수 없음을 의미합니다. 우리가 잘 알고 있는 무손실 압축 알고리즘의 예는 zip, rar, 혹은 7z등이며, 손실 압축의 예는 고화질의 Tiff 포맷 이미지를 저화질의 JPEG 포맷으로 변환하는 것입니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/296/1*hU2uICE6gZeKxDQYSTtnSA.png" /></figure><h3>SNARK는 압축기?</h3><p>질문 “SNARK는 정말 블록체인 데이터를 압축할 수 있을까?”으로 다시 돌아오겠습니다. 이 질문에 대한 답을 도출하기 위해 먼저 SNARK의 증명기(Prover)가 하나의 압축 알고리즘이라고 가정하겠습니다. 그렇다면 SNARK 증명기는 손실과 무손실 압축 중 어느 범주에 속할까요? 이를 확인하기 위해 먼저 SNARK 증명기가 압축하려는 데이터인 “Witness”의 불확실성을 확인하겠습니다.</p><p>Witness는 SNARK의 증명기가 증거 생성 과정에서 생성하고 폐기하는 비밀스러운 데이터 벡터입니다. SNARK증명기가 하나의 압축기라는 가정 하에, Witness는 압축되기 전의 원본 데이터라 할 수 있습니다. 따라서 SNARK의 검증기(Verifier)의 관점에서는 Witness는 비결정적이며, 힌트가 없이는 전혀 예측할 수 없기 때문에 최대의 정보량을 가진다고 할 수 있습니다. 즉, 무손실 압축을 할 여지가 없습니다. 우리가 위에서 논의한 정보 이론에 의하면, SNARK 검증기가 증거로부터 Witness를 손실 없이 복원하는 것은 불가능해야 합니다.</p><h4>SNARK의 목표</h4><p>여기까지는 정보 이론의 입장이었고, SNARK의 목표는 다를 수 있으니 꼼꼼히 확인해 보겠습니다. SNARK 증명기가 목표하는 압축은 무손실과 손실 압축 중 어느 것일까요? 이는 SNARK의 3 가지 보안 정의인 Completeness, Knowledge soundness, 그리고 Zero-knowledge 중 두 번째인 Knowledge soundness의 정의를 보면 알 수 있습니다. Knowledge soundness의 쟁점은 “1에 가까운 확률로 아주 작은 차원의 압축된 증거에서 아주 큰 차원의 Witness를 복원할 수 있는 가상의 추출 알고리즘(Extractor)이 존재하느냐?” 입니다. 다시 말해 SNARK 증명기가 목표하는 압축은 무손실 압축입니다.</p><p>대부분의 SNARK 프로토콜들이 논문을 통해 knowledge soundness를 증명하였기 때문에, 언뜻 보기에는 이들이 정보 이론적으로는 무손실 압축이 불가능한 데이터를 다룸에도 불구하고 무손실 압축을 수행할 수 있는 증명기를 설계한 것 처럼 보이긴 합니다. 그리고 zk-rollup이라는 기술을 접해보신 분들 중 대부분이 그렇게 알고 계실 것입니다. 그런데 여기에는 치명적인 가정이 하나 숨어 있습니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/497/1*DlFGebrgs0TQWy6E8zpPjg.png" /></figure><h4>블록체인 압축기로써는 치명적인 가정</h4><p>SNARK의 가정은, 검증기가 Oracle polynomial들에 접근 가능하다는 것입니다. 이 Oracle polynomial들의 집합은 검증회로 (Circuit)라고 불리며 Witness에 규칙을 부여합니다. 검증회로 또한 정보량을 가진 변수입니다. 검증기가 검증회로에 접근 가능하다는 것은 검증기가 아주 큰 정보량을 이미 가지고 있다는 것을 의미합니다. 이러한 가정 하에 SNARK증명기로 압축된 증거에서 Witness를 복원하는 가상의 추출기가 존재 할 수 있는 것입니다 (존재한다는 것일 뿐, 검증기가 실제로 추출을 수행하지는 않습니다).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QU03UH8Ta0WCF_jO.png" /><figcaption>검증회로의 예</figcaption></figure><p>이는 정보 이론적으로도 설명이 되는데, 아무런 힌트가 주어지지 않은 Witness는 최대의 정보량을 가지는 반면, 검증회로라는 정보가 주어졌다는 조건 하에서는 Witness의 조건적 정보량이 0 bits가 됩니다. 이는 검증 회로가 Witness가 가질 수 있는 값들의 모든 규칙을 결정하기 때문입니다. 즉, SNARK증명기는 Witness와 검증회로의 상호정보량 (mutual information)을 활용하는 무손실 압축기입니다.</p><p>SNARK 증명기가 블록체인 Scalability를 위해 데이터를 압축하려는 상황에서 이러한 사전정보의 가정은 아주 비현실적입니다. 블록체인 Scalability에서는 검증 노드가 아주 작은 차원의 압축된 데이터만을 가지고도 큰 차원의 원본 데이터의 올바름을 검증할 수 있도록 하는 것이 목표 중 하나인데, 만약 검증노드가 SNARK 검증기라고 한다면, 검증노드는 이미 검증회로라는 큰 차원의 데이터를 지니고 있다는 모순이 발생합니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/589/1*Y8mjs6GNnDVqKFRdEAhN_A.png" /></figure><h4>SNARK를 블록체인 Scalability에 적용하는 실용적인 방법들</h4><p>이러한 문제을 극복하기 위한 실용적인 방안은 새로운 가정을 추가하는 것입니다: <strong>신뢰할 수 있는 제3자</strong>가 존재하여 이들이 올바른 검증회로를 게시한다는 가정입니다. 이 가정은 검증노드 각각이 검증회로를 지니고 있을 필요 없이, 제3자의 게시물을 참조하는 것으로도 제 역할을 수행할 수 있게 해줍니다. 그러나 이러한 가정이 실현된다면 <strong>블록체인 분산성 (Decentralization)의 저하</strong>는 불가피합니다.</p><p>또 다른 방안도 있습니다. Powers-of-tau (Snarky ceremonies)와 같은 Threshold cryptography 기반의 분산 키 생성 기술을 활용하는 것입니다. 쉽게 말해, 다수의 행사(Ceremony) 참여자가 일정 기간동안 올바른 검증회로 생성에 기여합니다. 이 방식은 이전 방식에서 필요했던 신뢰할 수 있는 제3 자를 제거합니다. 그러나 이 방식 또한 새로운 가정을 필요로 합니다: <strong>참여자들 중 적어도 한 명 (혹은 프로토콜에 따라 절반 이상)이 올바르다</strong>는 가정입니다. 이는 곧 <strong>블록체인의 보안 (Security)의 저하</strong>를 야기합니다.</p><p>여기서 블록체인 보안과 분산성이란 과거 Vitalik Buterin이 정의했던 <a href="https://vitalik.eth.limo/general/2021/04/07/sharding.html?ref=bankless.ghost.io">Scalability trilemma</a>의 요소들입니다.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/464/0*gZsbiwWlNrVvMbBU.png" /><figcaption>The Scalability Trilemma (그림 출처: Vitalik’s blog)</figcaption></figure><h3>결론</h3><p>지금까지의 이야기를 요약하면, 블록체인 Scalability를 위해 SNARK 증명기가 무손실 압축기 역할을 할 수는 있지만, 이를 위해서는 보안 혹은 분산성 중 하나를 희생해야만 합니다.</p><p>그렇다면 보안을 잃지 않으면서 블록체인 데이터를 압축하는 방법은 없을까요? 옵티미스틱 롤업의 접근이 이러한 시도 중 하나입니다. 옵티미스틱 롤업은 레이어 2의 데이터를 전통적인 무손실 압축 알고리즘 (혹은 그것들의 변형)을 이용해서 압축 후 레이어1에 기록합니다. 압축된 데이터를 삭제하지 않는다는 가정 하에 (예를 들어 <a href="https://eips.ethereum.org/EIPS/eip-4844">Blob</a>이 없다는 가정) 보안을 저하하지 않는 최선의 압축 방법을 선택한 것입니다. 그러나 안타깝게도 옵티미스틱 롤업에는 레이어1의 검증 노드가 복원 없이 원본 데이터의 올바름을 검증하도록 해주는 비용-효율적인 기능이 없습니다.</p><p>저의 결론은 결코 SNARK기반 Scalability 솔루션 (예를 들어, Zk-rollup)들의 가치를 평가절하하는 것이 아닙니다. 저는 <a href="https://vitalik.eth.limo/general/2024/05/29/l2culture.html">Vitalik Buterin의 최근 블로그 게시물</a>에 깊게 공감하는데, 그 내용은 레이어2의 <strong>다양성</strong>에 관한 것입니다. 레이어2 솔루션 또한 Scalability trilemma의 굴레를 벗어나지 못하므로, 보안은 레이어1이 보장하고, 레이어2에서는 사용자에게 보안 혹은 분산성을 일부 희생하는 대신 성능적 이점을 제공하는 선택지, 즉 사용자에게 더 다양한 Dapp을 활용할 기회를 제공해주는 것에 큰 가치가 있다고 생각합니다.</p><p><a href="https://medium.com/tokamak-network/project-tokamak-zk-evm-67483656fd21">우리가 개발중인 Tokamak zk-EVM 및 Zk-rollup</a>는 SNARK와 합의 프로토콜을 적절히 조합한 새로운 접근 방식입니다. 완전히 SNARK에만 의존하는 Scroll 및 Linea 등의 기존 Zk-rollup과는 전혀 다른 방식이며, Challenging이라는 합의 프로토콜에만 의존하는 Optimistic rollup과도 또 전혀 다른 방식입니다. 우리의 도전이 Scalability trilemma 하에서 발현 가능한 다양성 공간에 하나의 새로운 기저 벡터를 제공해 주기를 기대합니다.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4be8fd05aa97" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2024 Q1 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-2024-q1-report-5b31e4beb2f0?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/5b31e4beb2f0</guid>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[layer-2]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[rollup]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Wed, 24 Apr 2024 09:46:36 GMT</pubDate>
            <atom:updated>2024-04-25T04:59:11.867Z</atom:updated>
            <content:encoded><![CDATA[<h4>Tokamak zk-SNARK paper has been released</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZogqajtXWVJMMBn3EykGNg.png" /></figure><p>Hi, this is Jake, leading the Tokamak zk-EVM project. We share our progress for the first quarter of 2024.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>About the Tokamak zk-EVM project</em></a></li></ul><p><strong>The main achievement of the first quarter</strong> was the publication of a paper on Tomakak zk-SNARK.</p><ul><li>The paper is archived in Cryptology ePrint Archive, <a href="https://eprint.iacr.org/2024/507">https://eprint.iacr.org/2024/507</a>.</li><li>Compared to the previous SNARKs such as PlonK and Marlin, Tokamak zk-SNARK has reduced reliance on preprocessing by the verifier.</li><li>In other words, as a polynomial interactive oracle proof, we have reduced reliance on the polynomial oracle.</li></ul><h4><strong>What’s the benefit of Tokamak zk-SNARK?</strong></h4><ul><li>One challenging problem in zk-EVM development is how to instantiate or eliminate the verifier’s preprocessing required by a SNARK</li><li>Adopting universal circuits such as Scroll and Linea was a solution to eliminate the preprocessing at the cost of the very complex circuit size.</li><li>However, unlike them, our zk-EVM does not use a universal circuit to track the state transitions of EVM to keep the circuit much faster and lighter.</li><li>Tokamak zk-SNARK has greatly reduced the amount of information that the verifier needs to preprocess, and based on this, we will instantiate the preprocessing through fault-proof protocols or secure multi-party computation.</li></ul><h4><strong>Then, what kind of circuits does Tokamak zk-EVM use?</strong></h4><ul><li>Rather than a universal circuit, we deal with <em>field-programmable circuits</em>.</li><li>A circuit is field-programmable if it can be derived from a<em> subcircuit library</em> and uniquely characterized by a low-degree<em> wire map</em>.</li><li>A subcircuit library is a set of subcircuits, each of which is a circuit of a predefined arithmetic operation (Subcircuits are similar to custom gates in PlonK, but more general because subcircuits allow for more complex operations).</li><li>Here, each predefined arithmetic operation is a stand-alone execution of each arithmetic and logical instruction in the EVM (<a href="https://ethereum.github.io/yellowpaper/paper.pdf">instructions 0x01–0x20</a>).</li><li>A circuit is then represented by placing copies of the subcircuits and the wiring between them, where a wire map defines the wiring.</li></ul><h4>Progress of subcircuit library development</h4><ul><li>Apart from theory development, we have been implementing and optimizing our subcircuit library. You can find our work <a href="https://github.com/tokamak-network/circom-ethereum-opcodes">here</a>.</li></ul><h3>What’s next?</h3><ul><li>Development of secure multiparty computation for verifiable wire maps</li><li>Documentation of our zk-EVM specification</li></ul><h3>Contact us</h3><p>We are actively seeking creative researchers and developers, please contact us at <a href="/service/mailto: hr@tokamak.network">hr@tokamak.network</a>. Also, we welcome contacting <a href="/service/mailto:jake@tokamak.network">jake@tokamak.network</a> for technical discussions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5b31e4beb2f0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-2024-q1-report-5b31e4beb2f0">Tokamak zk-EVM 2024 Q1 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Tokamak zk-EVM 2023-Q4 Report]]></title>
            <link>https://medium.com/tokamak-network/tokamak-zk-evm-q4-report-d2985b4513b2?source=rss-ed01137b8bdc------2</link>
            <guid isPermaLink="false">https://medium.com/p/d2985b4513b2</guid>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[layer-2]]></category>
            <dc:creator><![CDATA[Jake Jang]]></dc:creator>
            <pubDate>Wed, 13 Mar 2024 01:02:44 GMT</pubDate>
            <atom:updated>2024-04-24T08:14:23.862Z</atom:updated>
            <content:encoded><![CDATA[<h4>Academic comparison of our SNARKs with other SNARKs and machine computations</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VJsoI1BZ4TWsz-jYUyouKw.png" /></figure><p>Hi, I’m Jake, the leader of the Tokamak zk-EVM project. We share the fourth quarter’s progress.</p><ul><li><a href="https://medium.com/onther-tech/project-tokamak-zk-evm-67483656fd21"><em>About the Tokamak zk-EVM project</em></a></li><li><em>This project is financially supported by Ecosystem Support Program of Ethereum Foundation through Academic Grants Round 2023.</em></li></ul><p><strong>The main achievement of the fourth quarter</strong> was academic studies. We conducted a formal proof of the security and efficiency of our SNARK, covering properties such as completeness, knowledge soundness, and zero-knowledge. Our findings indicated that the efficiency of our SNARK is comparable to the state-of-the-art SNARKs such as Sonic, Marlin, PlonK. Additionally, in terms of verifiable machine computation, we observed that our approach uses significantly shorter proof compared to the most recent related works, MUX-Marlin and SublonK.</p><p><em>For detailed notes on the technical differences between our SNARK and MUX-Marlin and SublonK, see our </em><a href="https://medium.com/onther-tech/tokamak-zk-evm-q3-report-69102605077b"><em>previous report</em></a><em>.</em></p><p>We expect to publish a paper pre-print containing all of our findings within 2024 Q1.</p><h3>The findings</h3><h4>Security Properties</h4><ul><li><strong>Statistical completeness:</strong> Our protocol is statistically complete. In other words, a proof generated from a valid statement is acceptable with overwhelming probability.</li></ul><p><em>The imperfection of completeness can be rectified to perfection with simple tricks that add negligible errors to knowledge soundness. For instance, PlonK employed such tricks.</em></p><ul><li><strong>Statistical knowledge soundness:</strong> Our protocol satisfies witness-extended emulation in generic group model. In other words, a proof generated from an invalid statement is not acceptable with overwhelming probability.</li><li><strong>Honest verifier Zero-knowledge: </strong>Our protocol can incorporate zero-knowledge without affecting completeness, knowledge soundness, and efficiency.</li></ul><h4>Efficiency</h4><p>We present two tables comparing efficiency as follows. The first table illustrates efficiency when our SNARK is utilized for ‘Setup universal reference string (RS) — Derive circuit-specific RS — Prove &amp; Verify.’ The second table displays efficiency when our SNARK is used for verifiable machine computation, specifically ‘Setup a RAM-specific RS — Prove &amp; Verify.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-Lr4PIFDZ7CRCr6v35NUEA.png" /><figcaption>Efficiency comparison of SNARKs</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aO5BD-j9Ok9Fm3oM17oxGQ.png" /><figcaption>Efficiency comparison of machine computation</figcaption></figure><p>Notations used in the tables are described as follows:</p><ul><li>“Univ.” indicates the universality of setup.</li><li>“Upd.” indicates the updatability of setup.</li><li>“dim(pre-input)” indicates the dimensionality of vectors that must be preprocessed by the verifier.</li><li>\mathbb{F} means field elements.</li><li>\mathbb{G} means group elements, defined by elliptic curves over \mathbb{F}.</li><li>\mathbb{F}_{Arith.} means arithmetic operations in the field.</li><li>\mathbb{G}_{Ex.} means exponentiations in the group.</li><li>“P” means bilinear pairings, defined over the group.</li></ul><p>The parameters used in the tables are described as follows:</p><ul><li>m denotes the number of constraints in a (optimized) circuit.</li><li>l denotes the length of public instances.</li><li>N denotes the maximum number of constraints that a circuit can have.</li><li>s_D denotes the number of predefined modules (= subcircuits).</li><li>s_max denotes the maximum number of modules that can be placed in a circuit (= depth of a circuit).</li><li>n denotes the maximum number of constraints that each module can have (= width of a circuit).</li></ul><h3>What’s next?</h3><p>After publishing an academic paper that includes all of our findings, our focus will shift to the following:</p><ul><li>Zk-EVM circuit development: We will modularize the operations of the EVM in our zk-EVM circuit, allowing for the derivation of circuits tailored to EVM applications.</li><li>Development of an interactive fault-proof protocol (IFP): To mitigate reliance on trust in the verifier’s preprocessed input, we will design an IFP that supports our SNARK. This protocol will enable IFP nodes to validate the preprocess of untrusted SNARK verifiers.</li></ul><h3>Contact us</h3><p>We are actively seeking creative researchers and developers, please contact us at <a href="/service/mailto: hr@tokamak.network">hr@tokamak.network</a>. Also, we welcome contacting <a href="/service/mailto:jake@tokamak.network">jake@tokamak.network</a> for technical discussions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d2985b4513b2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/tokamak-network/tokamak-zk-evm-q4-report-d2985b4513b2">Tokamak zk-EVM 2023-Q4 Report</a> was originally published in <a href="https://medium.com/tokamak-network">Tokamak Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>