plan: summarize Bogdan's hand-edited review of v3 proposal

# Bogdan's hand-edited review of Genesis_11c_RadiaSoft_20260501_v3.pdf

**Source:** `/media/ssd2/Projects/Radiasoft/20260501122921.pdf` (8-page scan, 2026-05-01 12:29)

**Reviewing:** `/media/ssd2/Projects/Radiasoft/Genesis_11c_RadiaSoft_20260501_v3.pdf` (18 pages, generated 2026-05-01 08:19)

**Audience:** Proposal-writing team (RadiaSoft-led, PI Bruhwiler)

This file decodes Bogdan's handwritten annotations on the printed v3 narrative

into a list of concrete edits, organized by criticality so the team can triage

against the submission deadline.

---

## Bogdan's review agenda (cover page checklist)

He wrote a "TO REVIEW" list at the bottom of the cover page identifying his

four review priorities:

1. Obvious layout mistakes

2. Do tasks ↔ milestones?

3. Do milestones ↔ objectives?

4. Decision-gate metrics improvement

The items below report his findings against each.

---

## Critical (must fix before submit)

### 1. Typo: HIFR → HFIR (4 instances, global search/replace)

The High Flux Isotope Reactor abbreviation is consistently misspelled "HIFR"

throughout the narrative — it must be **HFIR** (F before I). Bogdan circled

each instance and wrote "HFIR" in the margin. Confirmed via text extraction

of the v3 PDF, the four occurrences are:

| Line | Context |

|------|---------|

| Page 1, Introduction | "the High Flux Isotope Reactor (HIFR) and" |

| Page 1, Introduction | "HIFR and the TOPAZ single-crystal diffractometer [3] at SNS" |

| Page 1, Project Objectives | "HB-2A powder diffractometer experiments at HIFR" |

| Page 4, Section 3.3 (Task 3) | "all of HIFR or the full suite of neutron beamlines at SNS" |

**Action:** Global replace HIFR → HFIR in the LaTeX source.

### 2. Section 6 Decision Gate Metrics — major restructuring

This is the largest content change Bogdan asked for. He rebuilds the section

in four ways:

#### 2a. Map every metric to a milestone

| Focus Area | Existing metric | Maps to |

|---|---|---|

| HB-2A | ≥20% reduction in estimated counting time via automation | **Milestone 7** |

| PIP-II | 3D viz latency <100 ms from data stream to UI rendering | **Milestone 6** |

| MLOps Integration | Verified EPICS-to-MLflow bidirectional model deployment | **Milestone 5** |

The current text doesn't explicitly tie each metric to a milestone; Bogdan

wants that mapping in the table itself.

#### 2b. Reframe metric language

The column header "Success Criteria / Metric" is annotated with "**ability

to**" and "**accuracy**" in his hand. He wants metrics phrased as a

capability ("ability to reduce counting time by ≥20%") rather than a static

threshold. Two specific deletions:

- **HB-2A row:** strike "via automation" — the metric should not assume the

  mechanism of how the reduction is achieved.

- **MLOps row:** strike "EPICS-to-MLflow" — keep the metric framework-agnostic.

#### 2c. Add a top-level Phase 1 → Phase 2 go-criterion

In addition to the existing "By end of month 6, all 3 metrics are on track to

be met at month 9," Bogdan adds (verbatim transcription of his handwriting,

lightly normalized):

> **Demonstrate functional-equivalent DT that behaves consistent with actual

> instrument given environmental physics, and produces equivalent data

> products.**

This is the proposal's ultimate Phase 1 bar — a DT that the instrument

scientists can't distinguish from the real instrument on the data products

that matter. It belongs as the headline criterion above the three

focus-area metrics.

#### 2d. Clarify the decision-gate scope

Bogdan adds a note: "Technically, the decision-gate (if it is to support a

6-month evaluation) should apply to **3 milestones**." Those three are

milestones 5, 6, 7 — restated in his hand:

- **Milestone 5:** DT with EPICS, model, and visualization

- **Milestone 6:** GPU-accelerated visualization of PIP-II data ✓ (already clear)

- **Milestone 7:** AI evaluation of HB-2A experiment that:

  - **(a)** identifies when an experiment can be **stopped** (sufficient-data criteria)

  - **(b)** identifies what instrument configurations would **reduce experiment time** (design-of-experiment support)

The (a)/(b) split for Milestone 7 is a meaningful tightening — it specifies

concretely what "agentic AI selection of physics-based models" produces for

the user, instead of leaving "compare with experimental data" open-ended.

**Action:** Rewrite Section 6 to incorporate (2a) through (2d). Update

Section 4 (Milestones) Milestone 7 to reflect the (a)/(b) split.

---

## Strong recommendations (content & framing)

### 3. Section 2 Project Objectives — "Achieving" → "Demonstrating" / "Illustrating"

Page 1, paragraph beginning:

> "Achieving the following project objectives during Phase I will reduce

> technical risk and clarify our scientific vision."

Bogdan's margin note: "demonstrate? / illustrate". Recommend:

> "**Demonstrating** the following project objectives during Phase I will

> reduce technical risk and clarify our scientific vision."

Rationale: "achieving" overpromises for a Phase 1 risk-reduction effort.

"Demonstrating" / "Illustrating" is the right register and matches the

"Go Faster, Show Faster" framing (see strategic note §10 below).

### 4. Section 3.6 Task 6 — "rather than" → "toward the goal of … in later phases"

Page 5, sentence currently reading:

> "These studies will focus on defining requirements and evaluating

> feasibility **rather than** implementing a fully autonomous control system

> in **Phase I**."

Bogdan strikes "rather than" and "in Phase I" and writes the replacement.

Recommend:

> "These studies will focus on defining requirements and evaluating

> feasibility **toward the goal of** implementing a fully autonomous control

> system in **later phases**."

Rationale: "rather than" reads defensive — like we're carving the work down.

"Toward the goal of … in later phases" reads as a planned trajectory with

momentum and a clear Phase 2 link.

### 5. Section 3.4 Task 4 — critique Option 2, add naming convention

Page 4, the four DT control-system approaches. Bogdan strongly objects to

**Option 2** ("Configure the EPICS PVs to include virtual and live

components") with the marginal note "X hate this..." — *with one

exception*: the staged "PrePIP" / pre-deployment scenario where the virtual

PV namespace is renamed distinctly. He proposes a **virtual-prefix naming

convention**:

- BL4B (real beamline) → **VL4B** (virtual)

- HB2A (real instrument) → **VB2A** (virtual)

The "V" prefix removes ambiguity between virtual and live PVs. This

effectively merges Bogdan's use case into Option 3 ("Create copies of EPICS

PVs that are named differently") with a clear convention.

**Action:** In Section 3.4, after the four-option list, add a clarifying

clause: we will avoid Option 2 in production but accept it for staged

PrePIP / pre-deployment rollouts under a clear V-prefix naming convention

(e.g., HB2A → VB2A, BL4B → VL4B).

### 6. Section 3.4 Task 4 — add enduring-capability statement

Page 5, end of Task 4 paragraph (before Section 3.5). Bogdan writes a

closing sentence in the margin:

> **The enduring capability of this effort is to allow easy deployment of

> the DT at arbitrary scale, for any purpose.**

This reframes Task 4's deliverable as a generalizable capability rather

than a one-off prototype, and gives the task a long-term value statement.

**Action:** Append this sentence to the end of Section 3.4.

### 7. Tighten Tasks 1 & 2 → Milestone 7 linkage

Bogdan's task→milestone cross-check (see §11 below) flagged item 4: the

text doesn't make crisp the path from Task 1 (model hardening) and Task 2

(model management) into Milestone 7 (apply AI-selected models to realistic

HB-2A data). One sentence at the end of each of Tasks 1 and 2 naming

Milestone 7 as the consumer would close the loop.

**Action:** Add a one-sentence Milestone 7 hand-off at the end of Sections 3.1 and 3.2.

### 8. Milestone 2 — add "statistics"

Page 5, Milestone 2:

> "Completed design of HB-2A scheduler with model-informed quantification

> of sufficient data **statistics**."

Bogdan adds "statistics" — converts soft language ("sufficient data") into

a measurable statistical bar.

---

## Polish

### 9. Figure 2 (SLACTwin UI) — dark background

Page 3, Figure 2. Bogdan's margin note: "consider white background — or

doesn't it look good?" The dark UI screenshot prints poorly on a white

page. Options:

- Re-capture the SLACTwin screenshot in a light theme if available.

- Crop or frame the figure to reduce the dark area.

- If the dark theme is the only available rendering, leave it but verify

  legibility on a printed page (his actual concern).

---

## Strategic framing — "Go Faster"

Bogdan notes a discussion with **Jon Taylor** (ORNL Neutron Sciences

Directorate) referencing **Dario Gil's** directive: **"Go Faster."**

The current draft hedges in places ("rather than implementing", "Achieving"

objectives in Phase I, vague success criteria). The "Go Faster" framing

argues for forward-leaning verbs and concrete, milestone-mapped deliverables —

which is exactly the throughline of the edits in §3, §4, and §2c (the

functional-equivalence go-criterion). A final read-through of the

introduction and objectives with this voice in mind is worthwhile.

---

## Cross-check: tasks → milestones → objectives

Page 5 bottom shows Bogdan working through the alignment of all 7 milestones

against the 6 tasks. Reproduced with his check marks:

| # | Milestone theme | Tasks that contribute | Status |

|---|---|---|---|

| 1 | Compare/contrast/eval (initial 3D viz approaches — M1) | Task 5 (design + documentation) | ✓ |

| 2 | Design model to predict data statistics — M2, feeds M7 | Task 6 (R&D) | ✓ |

| 3 | Requirements & specs — M3, feeds M5/M6/M7 | Tasks 1, 2, 4 | ✓ |

| 4 | Select physics model — M4, needed for M7 | Tasks 1, 2 | **flag** |

| 5 | Publish DT artifact — M5 | Tasks 3, 4, 5 | ✓ |

| 6 | Demo viz — M6 | Task 5 | ✓ |

| 7 | Demo system — M7 | All tasks | ✓ |

His conclusion: the cross-walk is mostly good. **Item 4 is the only one

flagged** — the Task 1/2 → Milestone 4 → Milestone 7 chain isn't explicit

in the current text. Edit §7 (above) addresses this.

He did not annotate a separate milestones-↔-objectives check, but his

re-statement of milestones 5/6/7 in concrete language under §2d implicitly

ties them back to the three Project Objectives bullets (scheduler, EPICS

↔ AI/ML interface, GPU-accelerated 3D viz).

---

## Suggested email triage for the proposal team

If splitting the email by urgency:

**Critical (must fix before submission):**

- §1 HIFR → HFIR global replace (4 instances)

- §2 Section 6 Decision Gate Metrics restructure (2a–2d) and Milestone 7 (a)/(b) split

**Strong recommended (clear improvements, easy to apply):**

- §3 "Achieving" → "Demonstrating"

- §4 "rather than … in Phase I" → "toward the goal of … in later phases"

- §5 Section 3.4 V-prefix naming clause

- §6 Section 3.4 enduring-capability sentence

- §7 Tasks 1 & 2 hand-off sentences to Milestone 7

- §8 Milestone 2 "+statistics"

**Polish (do if time allows):**

- §9 Figure 2 background

**Tone pass:**

- §10 "Go Faster" voice in intro and objectives

---

## Items not in this scope

Bogdan's review covered the 6-page narrative. He did not annotate the

appendices (Bibliography, Facilities, Equipment, Data Management, Synergistic

Activities, Foreign Connections, Other Attachment) — they are not flagged

for changes here. If those need a separate review pass, that should be

requested explicitly.