try now to integrate Geco in an interdisciplinary architectural engineering studio: hoping we can show you some nice applications of your tool, I'll keep you update and sending now details by e-mail. Here the file (very welcome to be shared). It most probably contais trivial errors by me, thanks for helping and giving some tip! Gr. Michela
FILE:
Ok, right, I see the outputs update correctly. Origin of problems must be in some different mistake I do:
- Incident radiation: I am not sure I understand what is going on: why I get so many 'not a number' ? (The Galapagos report is full of NaNs).
Bio-Diversity: 0.887 Genome[0], Fitness=NaN, Genes [89% · 44%] { Record: Too many fitness values supplied } ...
Genome[7], Fitness=NaN, Genes [74%] { Record: No fitness value was supplied } ....
Genome[9], Fitness=NaN, Genes [37% · 11%] { Record: Genome was mutated to avoid collision Record: Too many fitness values supplied }
- Daylight calculations: the geometry accumulates withouth deleting the previous models. As a consequance, results almost do not change after few varations (so, outputs get updated but do not vary). In current daylight definition: the first object being imported is the one where the grid has to fit; its setting makes it cancelling all the other objects during import. All the others, do not delete anything when imported. When running loops (manual or GA) that vary parameters, the entire geometry do not get cancelled - so I guess the loop does not pass back by the cancelling step, but imports only the geometry which has been varied by the parameters using the setting of that import component only? I will then try again by changing the order of the operations, but if you have specfic tips, let me know.
THANKS!
…
edit 29/04/14 - Here is a new collection of more than 80 example files, organized by category:
KangarooExamples.zip
This zip is the most up to date collection of examples at the moment, and collects t
mplex the models are. If we are running multi-room E+ studies, that will take far longer to calculate.
Rhino/Grasshopper = <1%
Generating Radiance .ill files = 88%
Processing .ill files into DA, etc. = ~2%
E+ = 10%
Parallelizing Grasshopper:
My first instinct is to avoid this problem by running GH on one computer only. Creating the batch files is very fast. The trick will be sending the radiance and E+ batch files to multiple computers. Perhaps a “round-robin” approach could send each iteration to another node on the network until all iterations are assigned. I have no idea how to do that but hope that it is something that can be executed within grasshopper, perhaps a custom code module. I think GH can set a directory for Radiance and E+ to save all final files to. We can set this to a local server location so all runs output to the same location. It will likely run slower than it would on the C:drive, but those losses are acceptable if we can get parallelization to work.
I’m concerned about post-processing of the Radiance/E+ runs. For starters, Honeybee calculates DA after it runs the .ill files. This doesn’t take very long, but it is a separate process that is not included in the original Radiance batch file. Any other data manipulation we intend to automatically run in GH will be left out of the batch file as well. Consolidating the results into a format that Design Explorer or Pollination can read also takes a bit of post-processing. So, it seems to me that we may want to split up the GH automation as follows:
Initiate
Parametrically generate geometry
Assign input values, material, etc.
Generate radiance/ E+ batch files for all iterations
Calculate
Calc separate runs of Radiance/E+ in parallel via network clusters. Each run will be a unique iteration.
Save all temp files to single server location on server
Post Processing
Run a GH script from a single computer. Translate .ill files or .idf files into custom metrics or graphics (DA, ASE, %shade down, net solar gain, etc.)
Collect final data in single location (excel document) to be read by Design Explorer or Pollination.
The above workflow avoids having to parallelize GH. The consequence is that we can’t parallelize any post-processing routines. This may be easier to implement in the short term, but long term we should try to parallelize everything.
Parallelizing EnergyPlus/Radiance:
I agree that the best way to enable large numbers of iterations is to set up multiple unique runs of radiance and E+ on separate computers. I don’t see the incentive to split individual runs between multiple processors because the modular nature of the iterative parametric models does this for us. Multiple unique runs will simplify the post-processing as well.
It seems that the advantages of optimizing matrix based calculations (3-5 phase methods) are most beneficial when iterations are run in series. Is it possible for multiple iterations running on different CPUs to reference the same matrices stored in a common location? Will that enable parallel computation to also benefit from reusing pre-calculated information?
Clustering computers and GPU based calculations:
Clustering unused computers seems like a natural next step for us. Our IT guru told me that we need come kind of software to make this happen, but that he didn’t know what that would be. Do you know what Penn State uses? You mentioned it is a text-only Linux based system. Can you please elaborate so I can explain to our IT department?
Accelerad is a very exciting development, especially for rpict and annual glare analysis. I’m concerned that the high quality GPU’s required might limit our ability to implement it on a large scale within our office. Does it still work well on standard GPU’s? The computer cluster method can tap into resources we already have, which is a big advantage. Our current workflow uses image-based calcs sparingly, because grid-based simulations gather the critical information much faster. The major exception is glare. Accelerad would enable luminance-based glare metrics, especially annual glare metrics, to be more feasible within fast-paced projects. All of that is a good thing.
So, both clusters and GPU-based calcs are great steps forward. Combining both methods would be amazing, especially if it is further optimized by the computational methods you are working on.
Moving forward, I think I need to explore if/how GH can send iterations across a cluster network of some kind and see what it will take to implement Accelerad. I assume some custom scripting will be necessary.…
eventually found out about genetic algorithms on which I found extensive researches, projects,... ! I looked into it and ended up on a few papers which I believe are the jumpstart for my master thesis.
"Galapagos; on the logic and limitations of generic solvers" by David RuttenArticle in Architectural Design 83(2) March 2013
"Black-box optimisation methods for architectural design" by Thomas Wortmann and Giacomo NanniciniConference Paper: CAADRIA 2016, At Melbourne, AU, Volume: 177-186
So I started looking into alternatives to genetic algorithms in architectural design.So far, I've ended up on :
Thomas Wortmann's work with the surrogate(or model) based optimization approach!You can check out the tool he developped for GH (Opossum):http://www.food4rhino.com/app/opossum-optimization-solver-surrogate-models
Judyta Cichocka's work, specially with the Swarm approachYou can check out the tool she developped for GH (Silvereye):http://www.food4rhino.com/app/silvereye-pso-based-solver
And that's it !!! I've been researching through article references (mainly on "researchgate") but I'm now stuck in a loop of references I already visited!That probably means the litterature on the subject is not (yet) extended but I might probably be missing something.The keywords make it difficult to search : "optimisation", "algorithms", "architecture", send me most of the time to computational engineering and deep mathematics papers I unfortunately do not have the background knowledge to comprehend ! So there it is ! If you have any clue of where (or how ! ) I should be looking, please tell me :)I know Mr Rutten is pretty active on the forum so hopefully... (fingers crossed :p) !Also if you have any good tips for getting into algorithms in general (you think could help), I'd be glad to hear(read) it ! A book, tutorials maybe ?!So, autors, architects, projects books, articles, conferences I should go to,specialized architecture offices/studios (I'm also looking for an internship so ...).If you know about a more appropriate forum please let me know !If you want to get deeper into this, you can contact me at :
e1635331@student.tuwien.ac.at
tdissaux@student.ulg.ac.be
My master thesis is due for may 2018 but I have a paper to write for January 2018 in order to be elligible for a PHD program afterwards.What I mean by that is that if you read this message in 6 month, I'll still be open to discussion !
I am right now an erasmus student at TUWien (Vienna) but my main university is The university of Liège in Belgium.I can handle French, English, Italian litterature and eventually Dutch if really you think it's worth it ! I have access to most online libraries via my university's portals so access shouldn't be an issue !I'm very excited to hear from you I wish you all a great day,Cheers,Thomas
…
ld be the best UI.
I think difference is made by 'Slider = 10' vs 'Slider = 10.000' more than by simple input/component initialization so, why to stop when it could be even more powerful?
Slider = 0 To 5 --- Slider in [0, 5]
Slider = {3; 0 To 5}
Slider = {3;0;5}
Slider = 3;0;5
Slider = 3 0:5
Slider = 3,0,5
Slider = 3 0 5 --- Value and range (min max)
3 0.0 5 --- 3.0 0.0 5.0
3 0 5.0 --- 3.0 0.0 5.0
3.0 0 5 --- 3.0 0.0 5.0
-1 0 5 --- 0 0 5 (-1 -1 5)
6 0 5 --- 5 0 5 (6 0 6)
Slider = 0:2:6 --- Even numbers: 0, 2, 4, 6.
Slider = 1:2:7 --- Odd numbers: 1, 3, 5, 7.
0:2:5 --- 0:2:4 (or 0:2:6)
3:2:8 --- 3:2:7 (or 3:2:9)
3 1:2:7 --- 1 3 5 7 (value 3)
Bang! = 7 --- 7 outputs
Merge = 5 --- 5 inputs
What's your opinion about Bang! = 7? As it's setting number of inputs, should it use different format? Bang! 7? Bang! (7)? Bang! i7?
+ * - / \ % ^ & | ! = > --- Addition, Multiplication, Subtraction, Division, Integer Division, Modulus, Power, AND, OR, NOT, Larger than, &c.
= could be a problem.
\ Integer division or Set difference?
! could be NOT but also Factorial.
| could mean intersection.
& could mean concatenate.
1+ --- Addition: input A = 1
2* --- Multiplication: input A = 2
+{0,1,1} --- Addition: input B = {0,1,1}
0-, 1/, 2^, 10^, e^ have their own components
Flatten = {7} or Flatten = 7 --- Input P = {7} (off-topic: Why can’t P be a list?)
Pt = {1, 2, 3} --- Point XYZ, X = 1, Y = 2, Z = 3.
Swatch = 129,239,231 (102)
Swatch = 129 139 231 102
F2 = "x^2+y"
"List Length" and "List Insert" don't work properly: "Value List" is choosen. Why? What's the reason to this choice? Well, I'd like to know how the whole thing (search by keywords) works, David.
Name and nickname can be now used as keywords. "Larger" works for ">" but "greater" doesn't. Could it be improved? Could synonyms be used? Could a short description even be used (I know this could be a bit weird)?
more than --- >
more or less --- Similarity
more less --- Similarity
red green --- Sets.List components should be showed
lightning --- Split Tree
What about use Curve.Analysis or Math.Boolean to display those Tab.Panel components? Param, Math, Sets, Vector? Primitive, Special, Util? Tab, Panel, and Tab.Panel as keywords.
At the moment that I write this, I check that ignoring accents in keywords has almost been included (0.8.0009): p`anel, pañel, pánel --- panel (almost)
Shouldn’t 'Dom2' work for Dom²?
What about nested search? You type some keywords (say 'Params' or 'Params.Geometry', or 'red green', or 'lst') and then you make a fine-tunning search over previous results/keywords. Tab.Panel and/or nested geometry could be useful when search by plug-in is desired or when you want to search among .ghuser components (first 'ghuser' or 'Extra.MyPlugIn' or 'lst' keyword and then fine-tunning, specific, search).
Is 'list length' performing this nested search right now ('lst' > 'length')? Anyway, I am thinking about UI (graphical) changes; successive searches.
As I said, description (and even words from the help info) could be used to search. What about "some kind of tags"? I mean that if 'list l' to finally choose List Length has been used for a while, that could be learned. Eventually, an XML file could store these tags, so you could even edit them. That could implement description, name, nickname, help info, Tab.Panel, .ghuser, synonyms (lots of them), tags/shortcuts or wathever.
How could flatten/graft/reverse be used? Initialize graft+Simplify or graft+Bang! could be really useful.
What about expressions? I don't how could it be done properly: would Slider = x^2 (expression) work? I mean, aren't expressions parsed when initializing?
Is Panel somehow doing this? 'panel = wathever' always suppose that wathever is a string, so you can't use 'panel = <pi>'. Sets.Strings components also do this.
I've been about to write several paragraphs about height/width (resizable components: Panel, Graph Mapper, Slider, &c.), input/output names (Scripts, F components; or any component with editable input/output names), orientation (Scribble), type hint and access option, nickname, &c. but, to sum up: being able to set any property when initializing would be really useful. I'd like to know the best choice of syntax but I'm sure that, David, you're closer to the answer. What do you think about this?
Slider: 3 0 5 "MySlider" "Slider^2"
Panel: "This is the content" "This is the title"
VB: "N" List Integer 7 "r" Item Double
Addition: A 1 B 2
I guess that any unified syntax would be elegant and useful, but additional ad hoc syntax (per component) could be even better (cleaner).
What about use lists of values? I'm not sure about format: panel = ("Hello", "Bonjour", "Hola")? If any valid format/syntax is found, maybe more sophisticated fetaures could be achieved: panel = {0;0} ("A", "B", "C") {0;1} ("1", "2", "3") How would you like this to be implemented?
There is a much simpler and interesting feature that would be useful, in my opinion: being able to initialize more than one component. I mean say 7xSlider = 10.0 and get 7 sliders and I also mean multiline (multi-component) initialization: Ctrl+Intro when you want to start a new line and Intro (or even some Accept/Cancel buttons when you activate multiline mode) to initialize (every line/component), for example. I mean:
3 x Slider = 1
Panel
Mass addition
Panel
And the whole bunch of components that were in mind (pre-thinked definition) is initialized. It speeds up the workflow, making more dynamic to add components that are only available via the drop-down panels.
Should this multiplier be something like a text box adjacent to search field more than '7x'?
These are some of my thoughts about intitializing. Please let me know your opinion :]
…
Introduction to Grasshopper Videos by David Rutten.
Wondering how to get started with Grasshopper? Look no further. Spend an some time with the creator of Grasshopper, David Rutten, to learn the