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.…
le with you.
I am trying to achieve the minimal path algorithm of Steiners tree in Python using the minimal path algorithm.The syntax would be as followsFirst I need to create a cube of any dimension.
Then I need to specify one origin say point A and destination point say B.
Now for this point A,B I need to create a machine based network which will automatically enroute A to B.
Where the angle will be constant i.e 120, length can be a variable, triangular node(steiners tree)using these constraints it will create a network.
Now, I should iterate the program in such a way that I should specify the further points say like A1 and B1 so on.The program will contain a limit constraint where it will come out of iteration loop and start a new loop,forming the network.
By this I will get a dense network of 120 deg branches.
The branching gets denser the moment I add source and destination points.
There can be 100 iterations to reach from A to B but the algorithm chooses the one following the minimal path.
I would be highly thankful to you if you would please share the python syntax and grasshopper definitionCapture.JPG for the same
Thank you for your time in advance
I would be highly grateful if you help me through
warm regards
Arya
12.gifShortest%20path%20algorithm.gh
min-paths.jpgcc.henn.studyimagesminimalpaths.jpg …
y in English. ○Presenter
Robert (Bob) McNeel (McNeel & Associates founder) Robert (Bob) McNeel is the founder and president of Robert McNeel & Associates (RMA). Founded in 1978, RMA originally focused on developing accounting software for accounting, architecture, engineering, and other personal services firms. Within a few years, RMA expanded its services to include selling and supporting microprocessor-based engineering and design software including AutoCAD. By 1985, the main focus of the business had shifted to AutoCAD sales, service, training, and software development. Bob McNeel grew up in the mountains of southern Washington State on a subsistence dairy farm. To pay for college, he worked in construction as a carpenter, welder, and cement finisher. Bob has a BA in Accounting from Washington State University. Prior to founding McNeel & Associates, he was a practicing Certified Public Accountant and the comptroller for a large construction company in Spokane. Andrés González (Rhino Fablab director) Andrés is a software trainer and developer since the 1980s. He has developed applications for diverse design markets as well as training materials for different CAD and Design software including the community of training materialswww.Rhino3D.TV Andrés has been working with the Rhino Team since the very early stages. He is now the head of the McNeel Southeast US & Latin American Division. He is the worldwide director of the digital fabrication community called RhinoFabLabwww.RhinoFabLab.com as well as the Generative Jewelry & Fashion Design community GJD3D www.GJD3d.com and Generative Furniture Design community GFD3D www.GFD3d.com 1981 -1985 University of North Carolina at Charlotte N.C. - EE.UU. B.S., Bachelor of Science in Engineering
…
Added by Yusuke Oono at 9:28pm on October 16, 2013
on 2: I think the reason to draw a fitness landscape is to highlight graphically the presence of local minima, even in a simple optimisation problem. In architectural terms, this means getting an idea of how many sub-optimal solutions there are in a problem, which helps while exploring conceptual design proposals.
Have a look at this very basic example (which I published with two colleagues on "Shell Structures for Architecture", chapter 18): a shell footbridge (24m x 4m footprint), which is generated by two parabolic section curves (the two apex heights are the two design variables). The maximum displacement of the structure under gravity load and self-weight is the objective function. Simple example, but several local minima and interesting shell forms (image below).
@AB,
The expression used by David in the Number of Samples Input is a simple “x+1”. By grafting the Divide Curve Output, he got 81*81 lenghts (6,561 values). You have to make sure that number is divisible by the no. of samples. The second expression used for the Length output is only a scaling factor (my guess), to control the height of the fitness landscape drawing.
Cheers…
n account of the position of the sun and weather cannot be expressed in terms of a single set of luminous intensity values (which is what IES files do).
With regards to your example files, I agree with Chris. The primary reason for the low illuminance levels is that the light bounces are getting lost in the tube. Have you checked with the manufacturer/distributor if the location of the IES file should be inside the tube and not flush with the ceiling? Physically modelling such tubes in lighting software like Radiance (which is what HB uses) or AGI32 is a fairly expensive proposition. This is one of the reasons why manufacturers provide photometric data for such devices (however simplistic that data might be).
The candelamultiplier increases or decreases the luminous intensity values. So it will have a direct impact on the calculation. The primary reason for having that input was to enable users to do some testing with different lamp types and environmental factors such as dirt depreciation. You need not change them for your simulation. Assuming that the IES file is inside the tube, in order to make this calculation work inside HB you'd have to crank up the calculation settings to a very high level (start with -ab 10 -ad 4096).
Finally, due to shortcomings in the annual simulation software (Daysim), IES files will not work directly work with annual calculations. However, there is a fairly easy workaround for that issue. In case you are planning to run annual calculations with IES files, please let us know here.
Sarith…
ing illuminance and limiting exposure (lux hours). Hours with direct solar irradiance are likely to exceed the limiting illuminance thresholds, which range from (200 to 50 lux as per Table 3.4 in CIE 157:2004). It makes sense to consider direct illuminance (an ab=0 simulation in Honeybee) separately from a normal illuminance calculation.
Assuming that the museum exhibits have low to high responsivity to light, an ideal solution would minimize direct sunlight. For daylight from the sky and reflected light, it might be enough to keep the illuminance levels below the recommended thresholds and then sum up lux-hours.
Daysim, the annual daylighting engine used by Honeybee and DIVA, is not very accurate for direct-sun calculations. You will get more accurate results if you run your analysis with Radiance directly.
Instead of considering the horizontal illuminance grids, one can create grids that correspond to the dimensions of the exhibit and then average those values. I think single points, as shown in your gh file might not suffice. Calculating lux-hours is by far the simplest part of such a simulation. It will only require averaging these points, extracting them into an array and then summing up that array.…
ace Syntax." eCAADe 2013 18 (2013): 357.
http://www.sss9.or.kr/paperpdf/mmd/sss9_2013_ref048_p.pdf
The measure Entropy is newer. I hereby explain it (from my PhD dissertation):
Entropy values, as described in (Hillier & Hanson, The Social Logic of Space, 1984) and specified in (Turner A. , “Depthmap: A Program to Perform Visibility Graph Analysis, 2007), intuitively describe the difficulty of getting to other spaces from a certain space. In other words, the higher the entropy value, the more difficult it is to reach other spaces from that space and vice-versa. We compute the spatial entropy of the node as using the point depth set:
(11)
“The term is the maximum depth from vertex and is the frequency of point depth *d* from the vertex” (ibid). Technically, we compute it using the function below, which itself uses some outputs and by-products from previous calculations:
Algorithm 4: Entropy Computation
Given the graph (adjacency lists), Depths as List of List of integer, DepthMap as Dictionary of integer
Initialize Entropies as List(double)
For node as integer in range [0, |V|)
integer How_Many_of_D=0
double S_node=0
For depth as integer in range [1, Depths[node].Max()]
How_Many_of_D=DepthMap.Branch[(node,depth)].Count
double frequency= How_Many_of_D/|V|
S_node = S_node - frequency * Math.Log(frequency, 2)
Next
Entropies [node] = S_node
Next
…