diff --git a/.archive/3-99 Research/5 PhD Proposal Ideas/Robust Control Inspired Integrity Monitoring System.md b/.archive/3-99 Research/5 PhD Proposal Ideas/Robust Control Inspired Integrity Monitoring System.md index 72780f80..33d223cc 100755 --- a/.archive/3-99 Research/5 PhD Proposal Ideas/Robust Control Inspired Integrity Monitoring System.md +++ b/.archive/3-99 Research/5 PhD Proposal Ideas/Robust Control Inspired Integrity Monitoring System.md @@ -1,14 +1,25 @@ ---- -tags: - - Ideas ---- -## What are we doing: -Remember when Lance was talking about his monitor that would read a control system to see if a cyber-attack was taking place? This is in that vein. -The idea that I'm thinking of is that to tell if a control system is compromised, we should be able to look at the dynamics of the system to know if an attack is happening. A couple thoughts: -1. Sensor integrity: If we have a monitor that is an observer, this monitor should be able to know when the plant is diverging from what it would expect based on sensor values and it's model of the plant. When the error signal has a higher magnitude than expected, this is a sure fire sign that something might be wrong. - 1. This is sensitive to maintenance problems however. - 2. Monitor must be using data diodes and not connected to any outside sources. -2. A secondary, redundant control system: Assuming the perpetrator is tampering with signals, activate a secondary control system that a) latches the first system out of its control authority, and b) operates a safety shutdown mode. This controller doesn't need to be super fancy - it just needs to prevent damage. -## Why are we doing this: -This seems like a fun way to integrate control system math with cybersecurity. This is somewhat a more CIE topic, but actually does address some cybersecurity issues. This could also be implemented using KOs and the protected domains. -## Other details: \ No newline at end of file +--- tags: +- Ideas --- ## What are we doing: Remember when Lance was talking +about his monitor that would read a control system to see if a +cyber-attack was taking place? This is in that vein. The idea +that I'm thinking of is that to tell if a control system is +compromised, we should be able to look at the dynamics of the +system to know if an attack is happening. A couple thoughts: +1. Sensor integrity: If we have a monitor that is an observer, +this monitor should be able to know when the plant is diverging +from what it would expect based on sensor values and it's model +of the plant. When the error signal has a higher magnitude than +expected, this is a sure fire sign that something might be wrong. + 1. This is sensitive to maintenance problems however. + 2. Monitor must be using data diodes and not connected to any + outside sources. +2. A secondary, redundant control system: Assuming the +perpetrator is tampering with signals, activate a secondary +control system that a) latches the first system out of its +control authority, and b) operates a safety shutdown mode. This +controller doesn't need to be super fancy - it just needs to +prevent damage. ## Why are we doing this: This seems like a fun +way to integrate control system math with cybersecurity. This is +somewhat a more CIE topic, but actually does address some +cybersecurity issues. This could also be implemented using KOs +and the protected domains. ## Other details: diff --git a/.archive/300s School/ME 3100 - Engineering Research and Leadership Management/1. Goals and Outcomes.md b/.archive/300s School/ME 3100 - Engineering Research and Leadership Management/1. Goals and Outcomes.md index 279c17d7..956a27ef 100644 --- a/.archive/300s School/ME 3100 - Engineering Research and Leadership Management/1. Goals and Outcomes.md +++ b/.archive/300s School/ME 3100 - Engineering Research and Leadership Management/1. Goals and Outcomes.md @@ -1,22 +1,40 @@ #ERLM + # Goals and Outcomes -Review [[ERLM_Structure_of_Objectives_Page.pdf]] -_Begin with the end in mind_ -_--- Steven Covey_ -_You've got to be careful if you don't know where you are going, because you might not get there._ -_--- Yogi Berra_ +Review [[ERLM_Structure_of_Objectives_Page.pdf]] _Begin with the +end in mind_ _--- Steven Covey_ -_Alice: Would you tell me, please, which way I ought to go from here?_ -_Cheshire Cat: [smiling] That depends a good deal on where you want to get to._ +_You've got to be careful if you don't know where you are going, +because you might not get there._ _--- Yogi Berra_ -_What are you trying to achieve? Explain your outcomes using absolutely no jargon._ +_Alice: Would you tell me, please, which way I ought to go from +here?_ _Cheshire Cat: [smiling] That depends a good deal on +where you want to get to._ -The Goals and Outcomes of your proposal are your chance to set the vision for your research. It is important at the beginning of your research to make clear to the reader where you want to be at the end. The proposal is a plan for getting to that place, and everything after the Outcomes is about explaining where you start and what path you will take. +_What are you trying to achieve? Explain your outcomes using +absolutely no jargon._ + +The Goals and Outcomes of your proposal are your chance to set +the vision for your research. It is important at the beginning of +your research to make clear to the reader where you want to be at +the end. The proposal is a plan for getting to that place, and +everything after the Outcomes is about explaining where you start +and what path you will take. + +An outcome is a thing aimed at or sought; it's a goal. That idea, +however, creates confusion since a research goal is the broad +purpose, and outcomes are specific statements about what you want +to be able to achieve. For our purposes, we will distinguish +between the research goal and the research outcomes, but both +will be contained in the Goals and Outcomes section of the +proposal. -An outcome is a thing aimed at or sought; it's a goal. That idea, however, creates confusion since a research goal is the broad purpose, and outcomes are specific statements about what you want to be able to achieve. For our purposes, we will distinguish between the research goal and the research outcomes, but both will be contained in the Goals and Outcomes section of the proposal. ## Research Goal -You must write the proposal with the reviewer in mind. The reviewer will want to know from the beginning what the proposal is about, so start every proposal with + +You must write the proposal with the reviewer in mind. The +reviewer will want to know from the beginning what the proposal +is about, so start every proposal with "The goal of this research is to ..." @@ -24,57 +42,192 @@ You can also say "The purpose of this research is to ..." -This goal sets the vision for what you want to achieve; it is the circle on the map for where you want to go. Your goal should not be too broad or grandiose. Remember that your colleagues are trying to solve similar problems and know what can and cannot be achieved. Grand goals call for grand projects, and if the details of you project do not align with your goal, the reviewers will see the disconnect and decline funding. The goal should also not be too narrow or specific. Narrow projects are often overly constrained limiting inquiry, have solutions with limited impact, and lack vision. Researchers who lack vision often fail. They are unable to inspire teams, motivate performance, or create sustainable value.   The challenge, one that successful researchers often spend considerable time confronting, is to make the goal of the research "just right," thus creating a vision of a problem with sufficient intellectual merit and broader impact to be worthy of investigation. +This goal sets the vision for what you want to achieve; it is the +circle on the map for where you want to go. Your goal should not +be too broad or grandiose. Remember that your colleagues are +trying to solve similar problems and know what can and cannot be +achieved. Grand goals call for grand projects, and if the details +of you project do not align with your goal, the reviewers will +see the disconnect and decline funding. The goal should also not +be too narrow or specific. Narrow projects are often overly +constrained limiting inquiry, have solutions with limited impact, +and lack vision. Researchers who lack vision often fail. They are +unable to inspire teams, motivate performance, or create +sustainable value.   The challenge, one that successful +researchers often spend considerable time confronting, is to make +the goal of the research "just right," thus creating a vision of +a problem with sufficient intellectual merit and broader impact +to be worthy of investigation. -It can be challenging or even unwise to stuff a big vision into just one sentence, but you don't want to have it take up a page or even half of one. Keep the research goal one to three sentences in length. If you have to go longer than that, you probably don't have a goal that is clear enough to you, which means it will not be clear to the reader, or you are trying to over define your goal, in which case some of what you are saying might be part of the outcomes or impact. +It can be challenging or even unwise to stuff a big vision into +just one sentence, but you don't want to have it take up a page +or even half of one. Keep the research goal one to three +sentences in length. If you have to go longer than that, you +probably don't have a goal that is clear enough to you, which +means it will not be clear to the reader, or you are trying to +over define your goal, in which case some of what you are saying +might be part of the outcomes or impact. + +While your goal begins your vision of what you want to achieve in +your research, it may not be tangible or concrete. Reviewers will +want and need to see a more clear picture of where your research +will go. It would seem, since the point of the proposal is to +explain what you want to do in your research, that next you +should lay out what you would do in your research. The trouble is +that, while the reviewer does care what you will do, they care +more where you will end up.  Think of them as the one with the +money --- they sometimes are, but often are not --- and they want +to know what it is they get for their investment. What are they +buying? What is the product of your effort? What will you +achieve? _These questions are not about the tasks you will +undertake}_, but rather about where you will be if the research +is successful and what you will be able to do that is new. The +answer to these questions are the research outcomes. -While your goal begins your vision of what you want to achieve in your research, it may not be tangible or concrete. Reviewers will want and need to see a more clear picture of where your research will go. It would seem, since the point of the proposal is to explain what you want to do in your research, that next you should lay out what you would do in your research. The trouble is that, while the reviewer does care what you will do, they care more where you will end up.  Think of them as the one with the money --- they sometimes are, but often are not --- and they want to know what it is they get for their investment. What are they buying? What is the product of your effort? What will you achieve? _These questions are not about the tasks you will undertake}_, but rather about where you will be if the research is successful and what you will be able to do that is new. The answer to these questions are the research outcomes. ## Research Outcomes -The research outcomes are brief, clear, concise statements of what you should be able to do if the research is successful. These are like mini-goals, but they are more specific. They are not necessarily steps along the way to a larger goal; rather, they are the sign posts that, if achieved, together would meet the overarching goal. The outcomes provide clarity and definition of your goal, and make clear what the specific things you want to be able to do if successful. This vision is something that you will return to throughout the proposal as you make connections between your efforts and how they will help you achieve your outcomes. Since you will be making these connections, the reader should be able to keep your vision in mind as they read, and you need to facilitate this by making each outcome sticky. Do this by making them simple, concrete, and credible. You do not need all of the traits, but do as many as you can. + +The research outcomes are brief, clear, concise statements of +what you should be able to do if the research is successful. +These are like mini-goals, but they are more specific. They are +not necessarily steps along the way to a larger goal; rather, +they are the sign posts that, if achieved, together would meet +the overarching goal. The outcomes provide clarity and definition +of your goal, and make clear what the specific things you want to +be able to do if successful. This vision is something that you +will return to throughout the proposal as you make connections +between your efforts and how they will help you achieve your +outcomes. Since you will be making these connections, the reader +should be able to keep your vision in mind as they read, and you +need to facilitate this by making each outcome sticky. Do this by +making them simple, concrete, and credible. You do not need all +of the traits, but do as many as you can. Well formulated outcomes should do the following: - provide a clear purpose for the research; + - direct your choice for research activities; + - guide the assessment of the success of the research. -Each outcome should be a single sentence that starts with a verb --- remember you are stating what you should be able to do if the research is successful. Ideally, the verb should convey something verifiable. Verbs like 'understand'', 'know', 'comprehend', or 'make sense of' may describe something we want to achieve in research --- understanding is a general goal of all research. But, these are not observable or verifiable. Remember the eighth Heilmeir question asked "What are the mid-term and final 'exams' of the research?'' By making your outcomes observable and verifiable you setting yourself up to answer that question. +Each outcome should be a single sentence that starts with a verb +--- remember you are stating what you should be able to do if the +research is successful. Ideally, the verb should convey something +verifiable. Verbs like 'understand'', 'know', 'comprehend', or +'make sense of' may describe something we want to achieve in +research --- understanding is a general goal of all research. +But, these are not observable or verifiable. Remember the eighth +Heilmeir question asked "What are the mid-term and final 'exams' +of the research?'' By making your outcomes observable and +verifiable you setting yourself up to answer that question. Examples of an outcomes written to varying degrees. -**Goal:** The goal of this research is to generate artificial light. +**Goal:** The goal of this research is to generate artificial +light. -| **Quality** | **Outcome** | -| ---------------- | -------------------------------------------------------------------------------------------------------------------- | -| Good | Use electricity to generate incandescent light from a wire filament. | -| Vague | Use electricity to make light. | -| Unmeasurable | Understand how incandescent light can be made with electricity. | -| Verbose | Use AC or DC electricity to generate light as a result of heating a wire filament to sufficiently high temperatures. | -| Not goal related | Determine the relationship between filament temperature and the characteristic of the light emitted. | +| **Quality** | **Outcome** +| | ---------------- | +-------------------------------------------------------------------------------------------------------------------- +| | Good | Use electricity to generate incandescent +light from a wire filament. +| | Vague | Use electricity to make light. +| | Unmeasurable | Understand how incandescent light can be +made with electricity. +| | Verbose | Use AC or DC electricity to generate light +as a result of heating a wire filament to sufficiently high +temperatures. | | Not goal related | Determine the relationship +between filament temperature and the characteristic of the light +emitted. | -How many outcomes should you have? You want the reviewer to be able to remember your outcomes --- they should be able to recall your outcomes throughout the proposal--- so don't make too many, but don't make too few since that runs the risk of suggesting either a narrow scope for the research or overly broad outcomes that don't define the research sufficiently. Between three and five is about right. Less than three is too few --- the reviewer wonders what else you might be able accomplish. More than five is too much --- the reviewer thinks that you are over committing yourself. My preference is three outcomes, and there is room for more if you absolutely need them. +How many outcomes should you have? You want the reviewer to be +able to remember your outcomes --- they should be able to recall +your outcomes throughout the proposal--- so don't make too many, +but don't make too few since that runs the risk of suggesting +either a narrow scope for the research or overly broad outcomes +that don't define the research sufficiently. Between three and +five is about right. Less than three is too few --- the reviewer +wonders what else you might be able accomplish. More than five is +too much --- the reviewer thinks that you are over committing +yourself. My preference is three outcomes, and there is room for +more if you absolutely need them. -Presumably, your outcomes are not something that you were able to do before. That is, your research should extend the state of the art and redefine the limits of current practice. Who is the judge of the current art and limits? This depends upon the funding organization. If you are applying to the NSF, the state of the art may be much different from what a company sees as their state of the art. A large national funding agency sees their job as pushing the state of the art for the country, and you need to define an outcome that pushes the state of the art for the national science and engineering community. A small company may just want to innovate for their customers, and your research may be bringing new capabilities to the company and their customers but may not be as grand or far reaching it might be if you has applied, say, to the NSF. It is important to understand the organization to which you are applying and what their expectations are for what reasonable outcomes might be. +Presumably, your outcomes are not something that you were able to +do before. That is, your research should extend the state of the +art and redefine the limits of current practice. Who is the judge +of the current art and limits? This depends upon the funding +organization. If you are applying to the NSF, the state of the +art may be much different from what a company sees as their state +of the art. A large national funding agency sees their job as +pushing the state of the art for the country, and you need to +define an outcome that pushes the state of the art for the +national science and engineering community. A small company may +just want to innovate for their customers, and your research may +be bringing new capabilities to the company and their customers +but may not be as grand or far reaching it might be if you has +applied, say, to the NSF. It is important to understand the +organization to which you are applying and what their +expectations are for what reasonable outcomes might be. --- + # Peer Review: Goals and Outcomes 1 -Offer comment using Acrobat's comment feature.  Address the questions below, which  focus whether the content meets the intention of the section, but you can say more about whatever else you think will help.   Since it can be hard to know what the writer intended, it is best to highlight parts you find confusing or that you may have had to read several times before understanding.  Some questions will ask you to summarize your understanding of what was written.  This will help the writer see how well their message has been communicated. -A note about spelling, grammar, and typos:  Writers should make a point to eliminate these errors from what they write.  Typos, for example, distract the reader and detract from your writing.  As editors, it is not your job to ferret these mistakes out.  Focus on the message and substance of the writing, and only comment on these annoyances if you just can't stand it. +Offer comment using Acrobat's comment feature.  Address the +questions below, which  focus whether the content meets the +intention of the section, but you can say more about whatever +else you think will help.   Since it can be hard to know what the +writer intended, it is best to highlight parts you find confusing +or that you may have had to read several times before +understanding.  Some questions will ask you to summarize your +understanding of what was written.  This will help the writer see +how well their message has been communicated. + +A note about spelling, grammar, and typos:  Writers should make a +point to eliminate these errors from what they write.  Typos, for +example, distract the reader and detract from your writing.  As +editors, it is not your job to ferret these mistakes out.  Focus +on the message and substance of the writing, and only comment on +these annoyances if you just can't stand it. + +_When you offer a criticism, offer a remedy so the writer can +improve their work._ -_When you offer a criticism, offer a remedy so the writer can improve their work._ ## Research Goal -The goal sets the vision for what you want to achieve; it is the circle on the map for where you want to be if the research is successful. -- Is the goal a clear statement about what the research would achieve if successful?  Explain why or why not. +The goal sets the vision for what you want to achieve; it is the +circle on the map for where you want to be if the research is +successful. + +- Is the goal a clear statement about what the research would +achieve if successful?  Explain why or why not. + - Explain what you think the vision for the research is.  -- Is the scope of the goal of the research "just right'', not too grandiose or too narrow?  Explain how it could be made that way. -## Research Outcomes -The research outcomes should be brief, clear, concise statements of what could be done if the research is successful.  These are like mini-goals, but they are more specific. -- Are the research outcomes brief, clear, and concise statements of what should be able to be done if the research is successful?  Verify that the outcomes are not tasks or steps of the research. -- Explain how the outcomes, if achieved, meet the overarching goal. +- Is the scope of the goal of the research "just right'', not too +grandiose or too narrow?  Explain how it could be made that way. + +## Research Outcomes + +The research outcomes should be brief, clear, concise statements +of what could be done if the research is successful.  These are +like mini-goals, but they are more specific. + +- Are the research outcomes brief, clear, and concise statements +of what should be able to be done if the research is successful?  +Verify that the outcomes are not tasks or steps of the research. + +- Explain how the outcomes, if achieved, meet the overarching +goal. + - Are the research outcomes verifiable?  How so? + ## Reader's Perspective -- In your own words, describe the goals and objectives.  Do you have a clear picture of where the PI wants his research to go?  Has the writer transmitted his message to you? -- Has the PI provided sufficient explanation about the goal and objectives?  Are they over/under explained?  Explain what pieces would be helpful to clarify the picture. \ No newline at end of file + +- In your own words, describe the goals and objectives.  Do you +have a clear picture of where the PI wants his research to go?  +Has the writer transmitted his message to you? + +- Has the PI provided sufficient explanation about the goal and +objectives?  Are they over/under explained?  Explain what pieces +would be helpful to clarify the picture. diff --git a/.sessions/nvim_config.vim b/.sessions/nvim_config.vim index 2352e03d..984b26bc 100644 --- a/.sessions/nvim_config.vim +++ b/.sessions/nvim_config.vim @@ -16,7 +16,7 @@ endif badd +16 ~/.config/nvim/lua/custom/plugins.lua badd +8 ~/.config/nvim/lua/custom/configs/lspconfig.lua badd +6 custom/init.lua -badd +19 custom/language_specific_commands/markdown.lua +badd +22 custom/language_specific_commands/markdown.lua argglobal %argdel edit custom/language_specific_commands/markdown.lua @@ -32,12 +32,12 @@ setlocal foldnestmax=20 setlocal foldenable silent! normal! zE let &fdl = &fdl -let s:l = 9 - ((5 * winheight(0) + 31) / 62) +let s:l = 22 - ((21 * winheight(0) + 28) / 56) if s:l < 1 | let s:l = 1 | endif keepjumps exe s:l normal! zt -keepjumps 9 -normal! 066| +keepjumps 22 +normal! 06| tabnext 1 if exists('s:wipebuf') && len(win_findbuf(s:wipebuf)) == 0 && getbufvar(s:wipebuf, '&buftype') isnot# 'terminal' silent exe 'bwipe ' . s:wipebuf diff --git a/Zettelkasten/Permanent Notes/thesis-ideas.md b/Zettelkasten/Permanent Notes/thesis-ideas.md index 26b7a5bf..69485b98 100644 --- a/Zettelkasten/Permanent Notes/thesis-ideas.md +++ b/Zettelkasten/Permanent Notes/thesis-ideas.md @@ -15,13 +15,11 @@ ___________________________________________________________ ## **Integrating Shielding into Nuclear Power Control** ### Goal: - The goal of this research is to develop machine learning control algorithms for nuclear power applications with strict safety guarantees. ### Outcomes: - If this research is successful, I will have accomplished the following: @@ -34,53 +32,80 @@ following: 3. ??? ### Impact: - Machine learning based systems have been shown to be more -efficient than typical PID based controllers, and are able -to learn more complex objective functions than a typical -controller can. The problem with these controllers though is -that they are often unexplainable. This is not acceptable -for high assurance applications, where slight perturbations -on inputs can yield wildly different outputs. Shielding can -solve this problem, helping ensure safety of ML based -controllers while not limiting their development or -construction. +efficient than typical PID based controllers, and are able to +learn more complex objective functions than a typical controller +can. The problem with these controllers though is that they are +often unexplainable. This is not acceptable for high assurance +applications, where slight perturbations on inputs can yield +wildly different outputs. Shielding can solve this problem, +helping ensure safety of ML based controllers while not limiting +their development or construction. ### Relevant Papers - [[safe-reinforcement-learning-via-shielding]] [[evaluating-robustness-of-neural-networks-with-mixed-integer-programming]] ___________________________________________________________ -## **Formally Verified Control of Reactor Systems** +## **Formally Verified Neural Network Control of Control Rod System** ### Goals: -The goal of this research is to use formal methods to ensure -that a neural network based control rod controller will never violate -safety guarantees of a reactor trip system. +The goal of this research is to use formal methods to ensure that +a neural network based control rod controller will never violate +safety guarantees of a reactor trip system. To do this, a +satisfiability modulo theory method will be applied to +exhaustively search the network for potential failure modes. ### Outcomes: -If this research is successful, I will accomplish the -following. +If this research is successful, I will have accomplished the +following: + +- Build a neural network controller for real time control of a +control rod system. + +- Formalize safety guarantees of shutdown margin in a +satisfiability modulo theory embedding + +- Formally verify that the neural network based controller will +not violate any shutdown margin restrictions ### Impact: +SMT solvers and MILP formulations have been applied to neural +networks to ensure that the network is resilient to input +perturbations. I think we can expand this to more general +considerations of the state space, especially when there are a +relatively small number of states such as in power contexts. The +benefit of this system is that we would get closer to saying +neural network based systems can be high assurance for physical +systems. ### Related Papers: - +[[reluplex-an-efficient-smt-solver-for-verifying-deep-neural-networks]] +[[evaluating-robustness-of-neural-networks-with-mixed-integer-programming]] +[[formal-verification-of-neural-network-controlled-autonomous-systems]] ___________________________________________________________ -## **Temporal Logic Specifications for Autonomous Controller Synthesis** +## **Temporal Logic Specifications for Autonomous Controller Shield Synthesis** (3) ### Goals: +If this research is successful, we will be able to generate +autonomous controllers that provably adhere to specifications +written with temporal logic automatically. ### Outcomes: +- Create an intermediary shield that mediates signals between an + optimal control system and the physical plant + +- ### Impact: ### Related Papers: +[[enhancing-cyber-physical-system-dependability-via-synthesis-challenges-and-future-directions]] +[[safe-reinforcement-learning-via-shielding]] ___________________________________________________________ ## **Formally Verified Runtime Monitoring and Fallback**