shape
carat
color
clarity

Scientific Proof, DiamCalc and Gemadviser

Status
Not open for further replies. Please create a new topic or request for this thread to be opened.

Garry H (Cut Nut)

Super_Ideal_Rock
Trade
Joined
Aug 15, 2000
Messages
18,746
Scientific Proof for OctoNus Software
This is an interesting and long discussion between Dymonite, a medical scientist, and Sergey Sivovolenko, CEO of OctoNus and a Moscow State Uni CutStudy leading participant.
I have made slight editorial modifications.
To fully understand you should print this! it is long and complex.

Dear Serg,

Re the nature of the scientific method:

For the layperson, science is valid becasue it ''works''. (On the other hand science is not very good at giving ''answers'' in the social science and humanities such as history, psychology etc - these often require a different method of reasoning)

Nevertheless, science provides information which is practical and helps us to predict (some) events in this world. The scientist talks about testing the hypothesis and the repeatability of a test. In the end, they refer to this as ''proving'' a theory.


At the same time, we recognise that ''scientific theory'' needs to be continually revised. As mentioned, Einstein supersedes Newton. Nevertheless, for daily applications, Newton'' ''laws'' work very well. Engineers rely on these concepts all the time otherwise buildings and bridges would be falling down all the time! But how do we reconcile the fact today''s ''fact'' may become tomorrow''s ''fiction''?


Have you read Karl Popper regarding the philosophy of science? (We did this at medical school) His treatise was quite interesting. He states that power of science is not its ability to be ''proved'' but that the hypotheses generated allows it to be disproved. We can never prove anything because one day we may find an exception to the rule. However, without the ability to disprove a theory, there is no way forward to finding another theory to explain our observations.

Therefore:

From a layperson''s point of view - Does gemadviser ''work''? Does the contrast metric correlates or ''fit'' well with human perception?

From Karl Popper''s point of view - How will you know if gemadviser is wrong (and needs to be corrected? Is it possible for your theory to be tested (and possibly disproved)?

From a scientist''s point of view - Where is the literature/theory for your concepts of contrast? Can we see the literature so that we can offer our comments? Have you correlated gemaadviser''s reports with other people''s perceptions?

From a person point of view - I may never know what is it about a sunset that is so beautiful nor will I ever try to explain it. Neverthless, I am intrigued by how you have tried to define a comparatively simple (yet elusive) concept such as contrast and scintillation.

Regards,
Dymonite

Sergey’s response:
Dymonite,
Strangely enough, we had to study not only Engels’ and Hume’s works but also works by Popper. This was necessary to pass examinations. It seems to me that being first of all a philosopher, Popper addressed this paper to philosophers. (The question “Is philosophy a science or not?” is still under discussion). As for me, I’m not fond of Popper’s style: it’s rather dull and emasculated, many statements seem quite disputable, and the paper itself seems inconsistent. I also think that this paper is not quite scientific in the sense advocated by Popper himself, though maybe he was primarily interested in the problem how scientific hypotheses actually appear, and all the scientific approach criteria he systematized were only tools for solving this problem (of course, he failed to solve it).

I like Einstein’s and Vernadsky’s works on this subject much more than Popper’s ones. Mach is also worth reading, though this reading is much more difficult.

However, since you consider Popper as a primary authority, I’ll try to deal with Popper.

The phrase you cited is used by Popper to prove that the inductive proof is miserable with respect to the deductive proof. One of the ways of the latter is, as Popper wrote, “to check a theory by empirically applying corollaries (results, outcomes) that follow from it”.

You may use this approach even without the knowledge of our formula of contrast. Two of the three remaining ways of deductive proof are almost useless in this case, regardless of the knowledge of the formula. As for the criterion of novelty, you are not interested in it at the moment, at least as far as I understand.

The way of analyzing the corollaries (which sometimes represent a new knowledge, at least for researchers) is especially efficient when no formulas are known, because in this case the tests are more objective and less preconceived.

We always use this method to continuously improve our model. In particular, first studies made us conclude that psychophysiology of human perception should be taken into account for all the ''basic light responses'' and that the models of illumination sources and a photodetector (such as the human vision system) should be upgraded to compute Scintillation and probably Fire. The accuracy of the existing models is enough only for computing Brilliancy. (At the same time, for quite different reasons, the Q plot can be adequately used for a micro zone. As for the macro plot GIA referred to in their article, we informed GIA in communications before publishing about its incorrectness when they prepared their paper on Fire. We proposed they publish our new FIRE plot or take for their article our microanalyses plot for Fire, or publish the old plot with our comments, in which conditions of its usage should be outlined and we will explaining why our results for certain parameters should not be used. You may learn from this paper how GIA miss-used our information. If you wish, I can send you the full version of our note of protest, sent to GIA and their admissions and response.)

By the way, we introduced ‘contrast’ exactly because our experimental verification had revealed that high-LR stones correctly PREDICTED by our software had shown low Brilliancy. During the verification, we used no formulas but only the experimental results and their consequences, that is, a new knowledge we obtained by analyzing solely these results.

A second reason for performing the verification without using any formulas:

According to the technology we develop, Beauty and probably Brilliancy are planned to be computed with no use of static formulas, which can provide only “linear approximation” (i.e., a simplified description of various phenomena). This means that should you ask me to tell you the formula we use to compute Beauty, I will honestly answer that we haven’t such a formula, although the software runs and produces results.
The situation is similar to that you often face when driving your car and noticing that the ABS system activates to prevent the wheels from slipping and therefore to make your drive more safe. It can be said that ABS systems are based on some empirical rules, fuzzy logic, knowledge bases, and that to verify their reliability, it is required and sufficient for you, as a consumer, to know that a car equipped with this particular ABS system has been tested for a basic set of standard situations occurring on the road. Only competitors are interested in the internal scheme of the ABS system , while a consumer (nor an expert) does not need any formulas to test the system.

The third reason. The formulas used to compute the ''basic light responses'' are just the “tip of the iceberg” of the calculations and modeling our software performs. It is more important to verify the algorithms for computing the light intensity (to generate a photorealistic image) and the correctness of coding these by programmers. The efficiency of verification of a single computation stage without verifying the others is very low, even if it reveals some deviations from different “theoretical” studies.

(editors note: If you take diamonds and scan them to produce a 3D file (e.g. with Sarin) and export files to DaimCalc2 you can test the modelled appearance with the diamond in your hand. You can also test with Ideal-Scope views etc. Here in Melbourne we have done this many times and we are satisfied, even with sarin scan inaccuracy, that there is an excellent correlation)

So, if you are really going to verify our results, I suggest you to make some basic (characteristic) test samples, for which you can obtain more or less unambiguous grades by testing the human perception.

I answer “No” to all your questions about defining Brilliancy and Scintillation. I define these in a quite different way than you think I do .

During our studies, we invented a number of interesting ideas on Scintillation. The more I think about them, the more felicitous they seem to me. While Brilliancy is mostly due to the interaction between the cut and the secondary sources of light, both Fire and Scintillation are caused by the primary sources . If you supplement this banal idea with the concept of how the angular sizes of the pupil and diamond facets are important for determining the probability of observing the primary source and its visible intensity, you’ll get a good approximation for estimating these quantities. The following example may explain this.
(Editor: I will post this later after editing. It is a brain teaser and very interesting Einstein type thought experiment.)
MSUSerg
 
Status
Not open for further replies. Please create a new topic or request for this thread to be opened.
GET 3 FREE HCA RESULTS JOIN THE FORUM. ASK FOR HELP
Top