首页 > AI前沿 > TangentPDE: Learning Local Operator Responses with Sparse Finite-Difference Supervision

TangentPDE: Learning Local Operator Responses with Sparse Finite-Difference Supervision

arXiv机器学习 2026-11-19 02:23 12 阅读 查看原文
用户希望将一段纯文本新闻正文排版成语义化的HTML片段。 规则: 1. 只输出HTML片段本身,不要Markdown围栏,不要解释。 2. 每个自然段落用`

`包裹,长文用`

`/`

`划分小节。 3. 列表用`
  • `,引用用`
    `,代码用`
    `,重要句子可用``。
    4. 保留原文所有信息,不得增删改内容。
    5. 不要输出``/``/``等文档级标签。
    
    分析原文:
    原文是一段关于TangentPDE的学术新闻/摘要,共5个句子,实际上是一个长段落。可以将其拆分成几个自然段落,或者加一个小标题。
    原文:
    "Neural PDE operators are often optimized through state reconstruction, which largely constrains operator values at observed inputs but does not fully characterize their sensitivity to local perturbations. A more faithful approximation of operator behavior requires not only accurate state predictions, but also correct variations along local perturbation directions." (第一部分:背景与问题)
    
    "We present TangentPDE, which uses finite differences to approximate directional derivatives and constructs task-relevant finite-step responses from the state outputs of a reference solver at sparse paired inputs. Its objective jointly constrains state fidelity, response-direction alignment, and response-magnitude calibration, thereby learning both operator values and their local variations. Because the response labels are constructed solely from state outputs at paired inputs, the method requires neither explicit derivative labels, linearized or adjoint PDE solves, nor full Jacobians, and remains applicable to nonlinear solvers without derivative interfaces." (第二部分:方法介绍)
    
    "Building on this, we additionally propose a selectively executed adaptation mechanism to strengthen local operator representations under finite-difference supervision. Budget-matched and correspondence-perturbed controls isolate the effect of relational supervision from improvements attributable to additional state observations or increased data volume." (第三部分:机制与控制)
    
    "Designed as a plug-and-play training framework, TangentPDE integrates with diverse neural-operator backbones. Across four PDE systems and 20 task-backbone settings, TangentPDE lowers the three-seed mean state error in 18 settings and the corresponding Rollout@20 error in 18, while intrinsic finite-step response gains are strongest on DiffReact and remain task- and backbone-dependent elsewhere." (第四部分:框架与实验结果)
    
    排版设计:
    加一个 `

    ` 作为小节标题,比如 "TangentPDE:基于有限差分的神经PDE算子优化方法" (不行,规则4:不得增删改内容。所以不能加原文没有的标题文字。那可以不加h2,或者只用原文的词作为标题?或者不划分小节,因为原文并没有小节标题。如果必须用h2/h3,又绝对不能增加内容,那可能不需要h2/h3,直接几个`

    `即可,或者把第一句作为`

    `?不,"长文用

    /

    划分小节",如果原文没有小节,就不需要强行加,或者把每个部分用`

    `包裹。为了安全,直接分四个`

    `,或者给关键句子加``。 原文没有列表、引用、代码。 重要句子加``: - "We present TangentPDE" - "the method requires neither explicit derivative labels, linearized or adjoint PDE solves, nor full Jacobians" - "TangentPDE lowers the three-seed mean state error in 18 settings and the corresponding Rollout@20 error in 18" 段落划分: 段落1:Neural PDE operators... local perturbation directions. 段落2:We present TangentPDE... without derivative interfaces. 段落3:Building on this... increased data volume. 段落4:Designed as a plug-and-play... backbone-dependent elsewhere. 检查是否有需要加`

    `的地方。如果不加`

    `也是可以的,因为规则说“长文用

    /

    划分小节”,这篇不算特别长,且原文没有小节标题,强行加标题会违反“不得增删改内容”的硬性规则。所以只用`

    `和``。 段落1:

    Neural PDE operators are often optimized through state reconstruction, which largely constrains operator values at observed inputs but does not fully characterize their sensitivity to local perturbations. A more faithful approximation of operator behavior requires not only accurate state predictions, but also correct variations along local perturbation directions.

    段落2:

    We present TangentPDE, which uses finite differences to approximate directional derivatives and constructs task-relevant finite-step responses from the state outputs of a reference solver at sparse paired inputs. Its objective jointly constrains state fidelity, response-direction alignment, and response-magnitude calibration, thereby learning both operator values and their local variations. Because the response labels are constructed solely from state outputs at paired inputs, the method requires neither explicit derivative labels, linearized or adjoint PDE solves, nor full Jacobians, and remains applicable to nonlinear solvers without derivative interfaces.

    段落3:

    Building on this, we additionally propose a selectively executed adaptation mechanism to strengthen local operator representations under finite-difference supervision. Budget-matched and correspondence-perturbed controls isolate the effect of relational supervision from improvements attributable to additional state observations or increased data volume.

    段落4:

    Designed as a plug-and-play training framework, TangentPDE integrates with diverse neural-operator backbones. Across four PDE systems and 20 task-backbone settings, TangentPDE lowers the three-seed mean state error in 18 settings and the corresponding Rollout@20 error in 18, while intrinsic finite-step response gains are strongest on DiffReact and remain task- and backbone-dependent elsewhere.

    检查规则: 1. 只输出HTML片段本身,不要Markdown围栏、不要解释。 -> 满足 2. 每个自然段落用

    包裹,长文用

    /

    划分小节。 -> 满足(没有小节,所以不用h2/h3,或者可以把"TangentPDE"作为h2?不,增词了。直接p即可。) 3. 列表内容用
    • ,引用用
      ,代码用
      ,重要句子可用。 -> 使用了,没有列表引用代码。
      4. 保留原文所有信息,不得增删改内容。 -> 满足,完全一致。
      5. 不要输出//等文档级标签。 -> 满足。
      
      确认没有遗漏任何字符,标点符号也完全一致。
      原文:
      Neural PDE operators are often optimized through state reconstruction, which largely constrains operator values at observed inputs but does not fully characterize their sensitivity to local perturbations. A more faithful approximation of operator behavior requires not only accurate state predictions, but also correct variations along local perturbation directions. We present TangentPDE, which uses finite differences to approximate directional derivatives and constructs task-relevant finite-step responses from the state outputs of a reference solver at sparse paired inputs. Its objective jointly constrains state fidelity, response-direction alignment, and response-magnitude calibration, thereby learning both operator values and their local variations. Because the response labels are constructed solely from state outputs at paired inputs, the method requires neither explicit derivative labels, linearized or adjoint PDE solves, nor full Jacobians, and remains applicable to nonlinear solvers without derivative interfaces. Building on this, we additionally propose a selectively executed adaptation mechanism to strengthen local operator representations under finite-difference supervision. Budget-matched and correspondence-perturbed controls isolate the effect of relational supervision from improvements attributable to additional state observations or increased data volume. Designed as a plug-and-play training framework, TangentPDE integrates with diverse neural-operator backbones. Across four PDE systems and 20 task-backbone settings, TangentPDE lowers the three-seed mean state error in 18 settings and the corresponding Rollout@20 error in 18, while intrinsic finite-step response gains are strongest on DiffReact and remain task- and backbone-dependent elsewhere.
      
      拼接HTML:
      

      Neural PDE operators are often optimized through state reconstruction