THE POSSIBILON

QUANTUM COMPUTING PROGRESS BEING MADE FOR THE POSSIBILON NPC IN NEOS VR

Originally drafted: November 15-22, 2021 by Dr. Jeremy Owen Turner

Hello Quantum computing and social VR enthusiasts!

BBMC’s “Quantum Computing in NEOS VR” project – funded by the Canada Council for the Arts – is now gaining full steam and nearing the anticipated deadline of the end of 2021!

This particular post will discuss Dr. Jeremy Owen Turner’s technical progress towards preparing the mind of his NPC (The Possibilon) for querying usage by D-Wave’s 5,000 qubit Enterprise quantum annealing computer (QPU). 

WARNING: The following contains heavily technical information, so the entertainment value you get from reading the rest of this blog post will be entirely up to you 😉 

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 

With the technical assistance of ViARSys (a programming team working within NEOS VR), Alex Kahn, and Chicago Quantum, Jeremy has been preparing three different (yet conceptually inter-related) energy source matrices from which D-Wave’s QPU will ultimately perform an Ising-based energy minimization function. The purpose of this energy minimization from these 3 energy preference matrices is to allow the Possibilon’s virtual mind to discern and recommend a few possible NEOS VR worlds based on up to 114 inter-convoluted meta-tags logged inside the NEOS data environment and 6 tags representing different avatar user archetypes (stereotypes). Each virtual world-orb bubbling from  within the NEOS metaverse – whether already or possibly existing –  has its own kind of energetic properties. Jeremy has found ways to quantitatively harness the qualitative aspects of these world-recommendation properties by assigning numerical float-values to particular world meta-tags and their relationship(s) to other meta-tags within each energy source.   

D-Wave’s data-annealer will sort from a quantitative mixture of 3 energy sources:

1 – ENERGY SOURCE #1 – Energy preference values extracted from a raycast-scanning extraction procedure within NEOS VR. 

In this instance, the avatar appointee arrives to meet the Possibilon in NEOS VR.  The Possibilon NPC then uses a kind of laser-vision to scan that avatar. Using machine learning techniques (programmed by Guillermo Valle Perez aka. Guillefix) [SEE CAPTIONS 3-5], the Possibilon attempts to visually identify this avatar based on his/her/their stereotypical appearance. This visual identification is then categorized according to one of 6 possible (and sometime, paradoxical) avatar archetypes: furry, party-goer, staff member, newbie, old flame (i.e. NEOS veteran), and a special wildcard archetype known as a, “wizard”.   Once categorized archetypally, implicit presumptions about the avatar guest’s world-preferences are mapped to both the archetype’s preferences as well as to all the other possible world meta-tag variables. Jeremy manually typed in these archetypal world-energy preference variables in spreadsheet matrix format. Jeremy made his own subjective assignments for each of the variable float-values. You can see the entire spreadsheet below [CAPTION 1]. The 2nd image shows a zoom-in on part of this manually prepared spreadsheet [CAPTION 2].

CAPTION 1: ENERGY SOURCE #1 – Here is the full 120×120 variable energy (Hamiltonian) preference matrix  for D-Wave’s quantum annealing process over the cloud. 

This matrix is used as a reference for when the Possibilon NPC visually automatically scans an avatar questioner in NEOS VR. This pre-made matrix is used as an archetypal reference for determining a priori what kinds of world-preferences an avatar questioner SHOULD like. Each preference variable is quantitatively represented as some kind of float-value between -1 and +1 (with zeroes also allowable). The quantum annealer considers these polarities as part of an Ising formulation. For this representation, the negative values indicate a higher level of world-preference attraction, while the positive values indicate the tendencies towards world-preference repulsion. There is a holistic Boltzmann-like statistical temperature corresponding to each of these deeply inter-convoluted variable slots. The green slots indicate negative (attractive) preference values while the red slots indicate positive (repulsive) preference values.  The empty slots indicate zero values (i.e. no preference either way). The lighter gradations of either green or red indicate weaker attraction/repulsion strengths. Jeremy made subjective qualitative assessments of what each archetype SHOULD be attracted to and repulsed by within a particular possible NEOS virtual world.

CAPTION 2: Here is a zoomed-in image showing a portion of this world-preference energy matrix. Within this matrix, each variable considers levels of attraction/repulsion to another variable (and vice versa).

CAPTION 3:  Guillefix’s raycasting studio has successfully used machine learning to visually classify/identify Jeremy’s own avatar (aka. Ideatron) as a furry…Yaaay! 😀 The capabilities of this raycasting studio will eventually be repurposed into the virtual brain of Jeremy’s Possibilon NPC. The other creature in this picture is a mannequin for visual testing purposes.

CAPTION 4: Here is a picture of Guillefix having himself scanned in his raycasting studio…Yes, the machine classification still works! Guillefix can now relax know that others also identify him as a furry avatar. He does not need to self-identify anymore 😉

CAPTION 5: Guillefix and Jeremy are testing the raycasting studio. They managed to successfully identify the “party-goer” archetype.

CAPTION 6: One amazing byproduct of the NPC’s raycasting process is that it can also produce 8-bit portraits of avatar faces! Here is an example of an 8-bit portrait render of Jeremy’s face! This picture was taken in our group meeting space. Lucas Rossetti aka LucasRo7 (NEOS LogiX programmer) also makes an appearance in this photo.

CAPTION 7: Here is another view of Jeremy’s 8-bit avatar portrait! We have the ViARSys production team also featured in this photo. In the background, you can see LucasRo7 and Guillefix coding this raycasting mechanism coding everything entirely in-world! This is how powerful NEOS VR is! 

  1. ENERGY SOURCE #2 – Energy preference values extracted from avatar user-preferences within NEOS VR. 

After being scanned by the Possibilon, the avatar guest questioner will proceed to a user-interface console (designed by Jeremy and Sirkitree) [SEE CAPTION 8] where he/she/they can enter in their world-preferences.  The output of this user interface (UI) interaction will yield the next 116 variables for mixing into the final matrix. 

CAPTION 8: This picture shows Sirkitree doing the finishing touches on the user interface (UI) design for the Possibilon’s energy source #2. You can see the Possibilon NPC in front of the NPC. The Possibilon is eagerly waiting to have its brain strung up to D-Wave’s QPU 😉

  1. ENERGY SOURCE #3 – Energy preference values extracted from world-table activity within NEOS VR. 

In this instance, the Possibilon NPC takes a snapshot of the most current world-orb bubbling activity taking place at the local world-table within NEOS VR. Further machine learning classification is performed on this snapshot to determine and extract any of the 116 available meta-tag variables. The current popularity of these meta-tags on the world-table will determine their quantitative energies. These energy values will be mixed with the values extracted from energy sources #1 and #2 and then sent to D-Wave’s QPU for annealing.  The next image for this blog posting [CAPTION #9] shows what a world-table looks like and how world-bubbles appear on the table inside of NEOS VR. 

CAPTION #9:  Here is a photo of the quantum computer programmer, Alex Khan from Chicago Quantum ca. 2020.  It was during this session that we had discussed the importance of also considering the bubbling-up of world-table activity as part of the overall data-mixture (energy source) for quantum annealing. The world-bubble activity energy like the kind depicted in this pic is one of the three preference energy sources the Possibilon collects for its QPU querying session.  Guillefix is taking care of the actual recording and machine learning of this world-bubbling activity.  If you want to know more about this world-table activity, check out this recorded twitch stream from this time… 

LINK: https://youtu.be/ZUMmxz2Jag4 

That is the technical update for now! Stay tuned for more quantum computing milestones within NEOS VR and showing Jeremy’s Possibilon NPC and JE’s Virtual Poet NPC in action!

BONUS CAPTION:  Here is an early graph-visualization of energy source #1.