Engage in a consultation with CIPM

[dr-shorthair]

CIPM have requested a consultation with the UCUM governing organisation to explore options for use of UCUM in the context of Digital-SI (see briefing document). I suggest that we nominate members of the UCUM AB to meet with CIPM including at least one of

• a technical person (@chgessner, @clemmcdonald ?)
• an organizational person (Marjorie, @ehafeza , @aflackey ?)

Briefing document
https://drive.google.com/file/d/159QJncutSKxj3FT06uxTzoAcRC5tKvUi/view

[chgessner]

I am ready to join such a meeting. In the first place it will be adjusting mutual expectations, I believe.

[aflackey]

Suggesting 2 UCUM advisors + 2 RI members to ensure coverage of all technical, content, and operational perspectives.

[gschadow]

I would be in. This being on European time zone hopefully would not happen at 3 am my time.

[colin-e-hscic]

The internal architecture of UCUM is not consistent with SI, in that the two systems have different base units. Most of the time this doesn't have practical impacts, but sometimes it does, primarily around dealing with moles which are a base unit in SI and mereley a large constant (a "chemists dozen") in UCUM.

In theoretical terms I believe this also means UCUM is not a coherent system of units in the way that SI is.

I would be interested to know if the colloaboration might get around to considering these questions.

[gschadow]

The internal architecture of UCUM is not consistent with SI, in that the two systems have different base units. Most of the time this doesn't have practical impacts, but sometimes it does, primarily around dealing with moles which are a base unit in SI and mereley a large constant (a "chemists dozen") in UCUM.
[...]
I would be interested to know if the collaboration might get around to considering these questions.

This is certainly what would come up.

Indeed where I took these liberties departing from SI (and never trying to hide it) it was done for specific purposes and I have been fully vindicated in at least one case, i.e., that of the mol. More then a decade later an IUPAC article pointing out how the definition of the mol had changed to become simply Avogadro's number of particles. This 2018 article here is not the first, and the point was pushed by some Jacob Ingversen (and I know I have his name totally wrong because Google brings up nothing), but it's not a small name in the field.

This new definition of IUPAC ultimately invalidates the mol as a separate dimension. Or at least this should be discussed (if it was in some CIPM meeting I'd like to see the transcript.)

And it is so interesting to consider your own comment about the dimensionality of angle arguing my case for the angle as a dimension.

There are some interesting papers around on these questions. A good one (to the extent I can understand it) from Dr. Paul Quincey at the UK National Physical Laboratory seems to come down on the same side as @gschadow, concluding that Radians and cycles are not dimensionless, but rather natural units of angle.

You then also declare that:

In theoretical terms I believe this also means UCUM is not a coherent system of units in the way that SI is.

to which I disagree. UCUM is certainly a coherent system, such system being defined as

A coherent system of units is a system of units of measurement used to express physical quantities that are defined in such a way that the equations relating the numerical values expressed in the units of the system have exactly the same form, including numerical factors, as the corresponding equations directly relating the quantities. *

Need to think of how CGM and MKS are and are not coherent and how they are related and not related. The G vs. K in CGM and MKS being the 3rd point where UCUM diverges from SI in a completely benign way. For syntactical considerations we cannot keep the kilo- in kilogram as a base unit. And we should not worry that the Newton is defined with the kilo factor inside.

[chgessner]

The new SI redefinition moves from explicit-unit to explicit-constant definitions ( see https://en.wikipedia.org/wiki/2019_redefinition_of_the_SI_base_units ). This should be reflected in giving the seven constants chosen by SI some special rank (e. g. some tag like 'fixed', 'definitional', 'explicit'). And then define the SI base units in terms of the seven constants by standard UCUM formalism.

I described this earlier in my post #242 (comment) .

[chgessner]

Regarding the discussion on the nature of the "mole", here is a somewhat different take on the semantics of units in general. For this purpose I will look at the unit merely as some "label" that tells us something about the meaning of the associated numerical measurement result value (i. e. very much like a LOINC code). This view works on measurements expressed in SI units as well as any other reference systems, like e. g. the WHO international units:

• When measuring an instance f of quantity “frequency”, the (dimensionless) ratio between the measured f and the reference frequency (the ground state hyperfine structure transition frequency of the caesium-133 atom ΔνCs) divided by the number 9192631770 shall be labelled with the unit “1/s”.
• When measuring an instance v of quantity “speed”, the (dimensionless) ratio between the measured v and the reference speed (the speed of light c) divided by the number 299792458 shall be labeled with the unit “m/s”.
• When measuring an instance n of quantity “amount of substance”, the (dimensionless) ratio between the measured n and the reference amount of substance (the amount “1”) divided by the number 6.02214076×10^23 shall be labelled with the unit “mol”.

Here is the generic expression for this view:

• When measuring an instance “x” of quantity “X”, the (dimensionless) ratio between the measured “x” and the reference quantity “ref_X” divided by the number “c_X” shall be labelled with the unit “u_X”.
• This means that the unit “u_X” tells us which “ref_X” and which scale factor “c_X” were used for the (dimensionless) numerical value of measured “x”.

In this view all the acrobatics of unit conversions and potentially physical laws is moved to the relations between the attributes "ref_X" and "c_X" and the corresponding UCUM syntax in the "u_X", while the casual user may just use "u_X" in the sense of a label (i. e. a code), maybe from a value set fixed for a specific context.

[gschadow]

OK, I come here from this big discussion that Christof brings up on #242 might warrant a new ticket, but it is much related to the CIPM conversation.

Here we prepare a grand new release of UCUM, where we might fully take on board the approach with the constants. Even back in 1998 I had already talked about that approach. And in fact it is another difference with the SI base units:

We use charge rather than electric current as a base/kind of quantity for electromagnetic phenomena simply because electrons, and their elementary charge, are the first cause for electric phenomena including current. As explained above, with this change our system is still isomorphic with the SI. [*]

I think the 2019 change is an ideal moment to now shake this out and make a grand change on UCUM as well.

Part of me is inclined to go all the way making the constants the base. And grant no more special status to the "base units" so the question of charge and coulomb or current and ampere doesn't even occur any more and my original charge decision wins over SI with e as the underlying base unit (but then scaled back with a specified number to actually match the coulomb.)

This would shake out all dimensions other than candela and mol. But mol is no longer an issue as in this diagram:

you see how the mol is completely disconnected from everything else. It's still not even a unit any more, no dimension of its own. I wonder how CIPM deals with that IUPAC reorientation now?

Second one is the candela. It is a psychophysical unit, which is energy flux with a spectral weight. In the same way one should be able to deal with "phone" (and indeed dB(SPL) and dB(A), etc.) But I never argued against the cd and it stays in a base.

We can see that K and cd look very similar, essentially the are derived of energy via a constant (with or without psychophysical adjustment curve (what comes to mind about a psychophysical derivate of temperature is "wind chill temperature".)

So we put the constants into the base. and suddenly mass becomes a derived unit with 3 of the component of the dimension exponent vector being non-zero. Time will still have only one. Length will have two.

The result will be rather impractical. As certainly every one of my implementations have given some use of the dimension exponent vector to construct canonical unit terms (e.g. "g.m.s-2"). But perhaps this can still be accomplished by providing the constants-based exponent vector in the inner core, and then some mapping (matrices) to map that to the traditional SI base, the UCUM base, or even CGM base if desired. I think that would possibly resolve that old struggle with the SI base being different, by saying "hey, we use the same constants base and you can use one or the other matrix to map into whatever base you want, with SI and UCUM and CGM base being available out of the box.

Loose ends of this discussion are still

• mol (the talk about "elementary entities" is IMO obfuscating the fact that there is nothing "elementary" about glucose molecule or a shoe. Nothing that has any metrological impact.
• the candela was unfairly singling out the interests of photographers and interior designers over, for example, audiologists (but then again, it's just a sore thumb, not something I would yet address, although it suggests like some sort of plug-in extension of the base.
• the angle, should it be its own dimension? I think yes, as far as (and if it in fact) has a relationship with the steradian in the exponent.
• what about Christoph's re-phrasing

  When measuring an instance “x” of quantity “X”, the (dimensionless) ratio between the measured “x” and the reference quantity “ref_X” divided by the number “c_X” shall be labelled with the unit “u_X”.

[chgessner]

Note that the base units will remain being base units, not derived units. No change. The change is that even as being base units, they will receive a definition, i. e. in terms of the said "definitional constants". Those constants are NOT considered base units or derived units, they have a special meaning in being "definitional". This is the way to stay in sync with the existing view of base units etc.

[gschadow]

Yes, they don't change the base units, but I am thinking we could show them how they could do it. So we actually jump to base constants and we create base unit system matrices (?) that map that constant core to whatever base units you want to see. That way the old conflict with SI would be surpassed, because whether m or cm, or kg or g or A or C is in the base is not even an issue for UCUM semantics any more. It is implementations that could choose any one. Including possibly extend (to grant themselves their own candela-like psychophysical unit, maybe play games with moles and Einsteins, swap plain angle in or out of a base, and possibly extend your x X ref_X formalism for all those arbitrary units (which currently I implement anyway by an extension of the dimension vector, essentially it becomes a sparse vector of possibly hundreds of dimensions.

[gschadow]

One thing I am unsure with is how the constants would be updated, because they surely would at some point be updated, with more precision in the digits, no?

[chgessner]

And yes, following your matrices approach with the "inner" and "outer" base, this is probably consistent with what I want to say. In other words: Being a "base unit" would become merely an attribute (to be assigned carefully and consistently, of course). There could be a set of "SI base units", and another set of units labelled "CGM base unit", and other sets of base units as well. However, this approach has to be thoroughly verified - for now it's just an idea!

[chgessner]

My understanding is that those seven constants will NOT change in the future. The SI brochure states "The numerical values of the seven defining constants have no uncertainty." Any advances in precision from now on will be reflected only in the numerical values of OTHER constants, that are depending on the definitional constants through the physical laws. Note that there always were systems that fixed "arbitrarily" some of the constants (or the base units in the respective system), which introduces corresponding "uncertainty" in other constants (or "derived units" in the respective system).

[chgessner]

Choosing the base for the dimension vector was always possible, this will not change. And choosing a different base always introduces "wiggle" in some of the numerical values (while others were fixed in the respective system). The only new thing in SI is that there is no more reference to any material physical object (the Ur-kilogram in the case of SI, previously). This move made it possible to redefine the system in terms of arbitrarily fixing ONLY the constants, and NOT any of the units (which become indeed "derived" in that sense, yes). The way this "smooth base transformation" was performed made sure that there are no changes in the numerical values of OTHER "derived" constants within the CURRENT limits of experimental precision.

For more detail on this see SI brochure section 2.2.1 The nature of the seven defining constants

[gschadow]

"The numerical values of the seven defining constants have no uncertainty."

I find that amazing, mind boggling. How can we be so sure we have the speed of light measured exactly? Why are such universal constants known not to be irrational numbers like π or Euler's number?

[chgessner]

Hi Gunther,

it is not the measurement that is exact, it is the numerical value of "c" that scales the speed of light measurement to "m/s" that is fixed per definition now.

[gschadow]

But what if c was an irrational number? And what about the ratio between speed of light and gravitational constant, what if that has to be exact to the million's digit or else the universe would collapse?

[chgessner]

The SI transition is well explained here: https://en.wikipedia.org/wiki/2019_redefinition_of_the_SI_base_units. Neither the definition of the meter nor the second was changed by this recent transition. The fact that "c" has a fixed exact numerical value when expressed in m/s is also unchanged since 1983.

• definition of the meter is unchanged as before (as since 1983): "The metre is the length of the path travelled by light in vacuum during a time interval of 1/299792458 of a second."
• definition of the second is unchanged, as before (as since 1967): "The second [...] is defined by taking the fixed numerical value of the caesium frequency, ΔνCs, the unperturbed ground-state hyperfine transition frequency of the caesium 133 atom, to be 9192631770 when expressed in the unit Hz, which is equal to s−1."

[clemmcdonald]

Thank you Christof From: Christof Gessner ***@***.***> Sent: Wednesday, May 3, 2023 9:27 AM To: ucum-org/ucum ***@***.***> Cc: McDonald, Clem (NIH/NLM/LHC) [E] ***@***.***>; Mention ***@***.***> Subject: [EXTERNAL] Re: [ucum-org/ucum] Engage in a consultation with CIPM (Issue #255) The SI transition is well explained here: https://en.wikipedia.org/wiki/2019_redefinition_of_the_SI_base_units. Neither the definition of the meter nor the second was changed by this recent transition. The fact that "c" has a fixed exact numerical value when expressed in m/s is also unchanged since 1983. * definition of the meter is unchanged as before (as since 1983): "The metre is the length of the path travelled by light in vacuum during a time interval of 1/299792458 of a second." * definition of the second is unchanged, as before (as since 1967): "The second [...] is defined by taking the fixed numerical value of the caesium frequency, ΔνCs, the unperturbed ground-state<https://en.wikipedia.org/wiki/Ground_state> hyperfine<https://en.wikipedia.org/wiki/Hyperfine_structure> transition frequency of the caesium 133<https://en.wikipedia.org/wiki/Caesium-133> atom, to be 9192631770 when expressed in the unit Hz, which is equal to s−1." — Reply to this email directly, view it on GitHub<#255 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/A3VBCFLC4NRIUVBMQJDL74LXEJMJ7ANCNFSM6AAAAAASCUW2MQ>. You are receiving this because you were mentioned.Message ID: ***@***.******@***.***>> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and are confident the content is safe.

[gschadow]

Christoph, this might not really be about anything practical on UCUM, but I still wonder, if we fix the constants with a number of decimals, and we declare that they have no uncertainty, wouldn't that just hide the uncertainty in the definition of the base units? So, if we suddenly discover some vastly more exact method of measuring the speed of light, that would seem to make the real definition of the meter more exact, even if we decide to keep c exact to only those digits 299792458. I am not sure if I was able to make my intuition understandable that something shifts as we develop more exact measurement techniques, and fixing the constants by axiom seems to be a very harsh approach.

More importantly than looking at speed of light vs. meter (or any of the other constants), what about the ratio of those constants? Right now the Planck constant is defined as h = 6.62607015×10−34 Js, and declared to be exact. But what if we divide c/h and the exact value we are getting (regardless of scale) would be 662607015 / 299792458 = 2.21021909430423363085. First question is, are these extra significant digits significant? Are they correct? Is it really impossible in the future to find that the exact value would have a few more digits?

And finally, with universal constants being divinely inspired, why do we know that they are not irrational numbers? Would that not prohibit that we fix them to such a human limited small number of digits? What if they were irrational numbers? Is it not quite possible, perhaps even plausible, that like π or Euler's number, the ratio of speed of light and Planck constant (or any other) should have an irrational value?

[dr-shorthair]

The discussion is irrelevant. The SI and CODATA have spoken. The constants are exact.

The consequences are that the actual size of the meter, second etc might change. This is because the whole method of arriving at the values of the base units has been inverted - they are a consequence of the constants, and not the other way round. The decision has been made at a level more authoritative than UCUM. If you want to argue about this, then you must engage with CIPM/SI/CODATA.

[chgessner]

Not completely irrelevant: My original intention was to adapt the current SI conventions, and doing so within our existing UCUM framework. That's why I am pushing towards somehow highlighting the definitional character of those few constants with their fixed numerical values in some way better than it's done today.

[gschadow]

The discussion is not irrelevant at all. And I still feel blistering heat / animosity. Who "has spoken" doesn't change reality.

I am curiously interested in these transcendental questions. What if the ratio of these constants have irrational values?

Besides that I am ready to engage in any argument with any group, I don't think this must rise to a level of an argument at all, I just want to understand this. But if there results any point of argument from it, I am not intimidated, because I had foreseen the change of the mol as being nothing but Avogadro's constant back in 1998, when I created UCUM, and the SI ultimately followed this necessary logical change, correcting it's prior convoluted definition.